The Analysis of Fatty Acids
Gas Chromatographic Analysis of Fatty Acid Derivatives
Sections E to G. Applications to natural samples and quantification
E. Separation of the Common Fatty Acids of Plant and Animal Origin on WCOT Columns
It could be argued that almost any polar stationary phase may be used for the analysis of the methyl ester derivatives of the fatty acids of the common seed oils of commerce, which tend to contain a limited range of C16 and C18 components. Yet even these will be found to have a number of trace constituents, when the attenuation of the gas chromatograph is turned up somewhat, and it may sometimes be necessary to utilise similar techniques for identification as are applied to the more complex samples of animal origin. If the analyst wishes to look hard enough, all tissues will be found to contain an enormous range of different fatty acids; for example, 437 different fatty acids, including positional and geometrical isomers, have been found in cow's milk fat . No single GC system could hope to resolve more than 15 to 20% of these, and a barrage of combinations of complementary techniques, of the kind described in Chapter 6, were applied to make the identifications. In this section, it is assumed that the goal of most analysts is to identify only those naturally occurring components present at a level of 0.1% or more in a sample, relying mainly on relative retention times. To illustrate this aspect, a relatively few "typical" examples of applications to real samples have been selected. It is assumed that appropriate methylation procedures will be used to eliminate artefact formation (see Chapter 4).
The "standard" WCOT column in general use nowadays is made of fused-silica, and is probably 25 m in length and 0.2 to 0.25 mm in internal diameter. Narrower-bore columns (0.1 mm i.d.) are now becoming available, and they hold promise of improved resolution. The author is not convinced of the value of wide-bore (0.5 mm i.d.) WCOT columns. As discussed in Chapter 3, a substantial increase in the length of a column is necessary before an appreciable improvement in resolution is achieved; there may be occasions when this is of value, but as a corollary there will be a substantial increase in the time required for an analysis. An alternative philosophy has been to make use of shorter columns, losing something of the resolution but gaining by reducing analysis times. For example, in a comparison of 100 m, 10 m and 2 m glass columns coated with SP-2340™ for the analysis of animal and hydrogenated vegetable fats, the 100 m column gave superb resolution as expected, but with analysis times of 2 hours or more, the 10 m column gave adequate resolution for the major components, with analysis times of under 30 minutes, while the resolution with the 2 m column was inadequate although analyses were completed in less than 4 minutes . A 15 m fused-silica column coated with Carbowax 20M™ gave acceptable resolution of the fatty acids of a rapeseed oil with a high erucic acid content in about 3 minutes .
Polar phases are used almost universally for fatty acid analysis, although the inherent resolution of WCOT columns is such that some remarkable separations can be achieved even with nonpolar silicone phases, which are more stable at elevated temperatures. Such columns are easier to manufacture than are polar ones. For example, the separation of a hydrogenated fish oil is a horrendous problem for any stationary phase, yet a published chromatogram obtained with a 44 m column coated with the nonpolar OV-73™ (175,000 theoretical plates) is probably as good as any in which polar phases have been used . A set of retention data has been published for cod liver oil fatty acids on a 50 m fused-silica column coated with SP-2100™; a large number of isomeric branched-chain monoenoic and polyenoic components were clearly resolved . A short column of this type was used for the analysis of plasma fatty acids .
The nature of the separation attained on nonpolar columns is rather different from that with polar columns as can be seen from Figure 5.7, in which a separation of pig testis fatty acids on a 12 m fused-silica column coated with BP5™ (a 5% phenylmethylsiloxane polymer) is illustrated. Unsaturated components emerge ahead of saturated fatty acids of the same chain length. Isomeric fatty acids differing in the positions of double bonds are usually clearly resolved, thus 18:1(n-9) and 18:1(n-7) are separated as are many of the polyenes. Indeed, the C22 fatty acids are possibly almost as well separated as on a polar column of the same length. Unfortunately, there are substantial overlaps among the C18 fatty acids, and 18:2(n-6) is not fully resolved from 18:1(n-9); 18:2(n-6) and 18:3(n-3) merge completely, and this is also true of the corresponding C20 and C22 compounds . As linoleate and linolenate are essential fatty acids with major nutritional importance, the deficiency in this aspect of the separation is likely to mitigate against a wider use of nonpolar columns. The order of elution of the C22 components is not that which might expected intuitively, i.e. it is 22:5(n-6), 22:6(n-3), 22:4(n-6) and 22:5(n-3). In this instance, it appears that the position of the double bonds has a greater effect on retention time than does the number of double bonds. Nonpolar phases do have advantages in specific applications, e.g. with fatty acids of high molecular weight or containing thermally labile functional groups, and in GC-mass spectrometry, where their stability at high temperatures, their considerable degree of inertness and their low rate of bleed are virtues.
|Figure 5.7. GC analysis of the fatty acids (methyl esters) of pig testis on a fused-silica column (12 m × 0.22 mm i.d.) coated with BP5 (a 5% phenylmethylsiloxane) (SGE Ltd., Milton Keynes, U.K.). Helium was the carrier gas at a flow-rate of 0.96 mL/min. The oven of the Packard Model 428 gas chromatograph, fitted with split injection (ratio 100:1), was maintained at 190°C for 3 min, then was temperature-programmed to 220°C at 1°/min. Peaks can be identified by reference to the numbers in Table 5.3.|
The separation of the methyl esters of the fatty acids of pig testis lipids on the Carbowax 20M™ column is illustrated in Figure 5.8. Each of the main chain length groups is reasonably well resolved. For example, three 16:1 isomers are seen and they are distinct from the C17 fatty acids. Similarly the important C18 components are clearly separated from a minor 19:1 fatty acid, and this is in a different region from each of the C20 unsaturated constituents. With the last, the only serious overlap problem is with 20:3(n-3), which co-chromatographs with 20:4(n-6); these are, however, just separable on a slightly more polar Silar 5CP™ column. Finally, all the biologically important C22 fatty acids are cleanly resolved.
Figure 5.8. Separation of the fatty acids (methyl esters) of pig testis on a fused-silica column (25 m × 0.22 mm) coated with Carbowax 20M™ (Chrompak UK). Hydrogen was the carrier gas at a flow-rate of 1 mL/min; the temperature was maintained at 165°C for 3 min, then was raised at 4°C/min to 195°C, where it was maintained for a further 23 min. A Carlo Erba Model 4130 gas chromatograph fitted with a split/splitless injection system was used with the split ratio set to 100:1. Peaks can be identified by reference to the numbers in Table 5.3.
An analogous separation of the fatty acids of cod liver oil is illustrated in Figure 5.9. Here the C16 region contains a wider range of fatty acids than does the previous sample, including branched-chain and polyunsaturated fatty acids, but all the main isomers are resolved. Similarly, the C18 polyenoic fatty acids are eluted before the C20 fatty acids begin to emerge, and 20:5(n-3) comes well before the first of the C22 fatty acids.
Figure 5.9. Separation of fatty acids (methyl esters) of cod-liver oil on a fused-silica column coated with Carbowax 20M™. Conditions and peak identification are as in the legend to Figure 5.8.
With the more polar CP-Sil 84™ column, there is again excellent resolution of the pig testis fatty acids (Fig. 5.10). Individual unsaturated esters are particularly well resolved, and for example there is near base-line separation of 18:1(n-9) and 18:1(n-7). On the other hand, 18:3(n-6) emerges after the minor C19 fatty acid. The C20 group are all well separated from each other but are beginning to run into an area occupied by C22 fatty acids. This last effect can be more troublesome with fish oils, which contain appreciable amounts of 22:1 isomers as in the cod liver oil sample (Fig. 5.11). The latter compounds emerge only just before 20:4(n-3). Similarly the C16 polyenes elute among the C18 fatty acids, and the C18 polyenes run into the C20 fatty acids. The C16 branched and monoenoic constituents tend to co-chromatograph. Nonetheless, with tissue lipids from plants and terrestrial animals especially, the polar column gives excellent results provided that care is taken in identifying the fatty acids.
These chromatographic traces lend support to Ackman's view that Carbowax 20M™ is the best general-purpose stationary phase for WCOT columns, and it is certainly recommended by this author to newcomers to the technique. (See also a more recent review of choice of column type at this website here...).
Figure 5.10. Pig testis fatty acids (methyl esters) separated on a fused-silica WCOT column (25 m × 0.22 mm) coated with CP-Sil 84™ (Chrompak UK). The oven was maintained at 150°C for 5 min, then was raised at 2°C/min to 180°C, where it remained for a further 10 min. Other conditions and peak identifications are as in the legend to Figure 5.8.
Figure 5.11. Cod liver oil fatty acids (methyl esters) separated as in the legend to Figure 5.10.
The separations illustrated here can by no means be considered the ultimate that can be achieved (Note that all are complete in about 30 minutes only). Rather, they are examples of what all well-organised lipid analysts should be able to attain in their own laboratories under routine conditions. By using longer columns, extending the analysis time and taking particular care to optimise the separation conditions, greatly improved resolution is possible. It is invidious to have to select examples from the literature to illustrate the "state of the art", but analyses of natural mixtures that fall into this category include plasma lipids [115,429], baby food formulae , and sea urchin lipids (Takagi and Itabashi, reproduced by Ackman ).
F. Some Applications to Less Common Fatty Acids
1. Short-chain fatty acids
When short and medium chain length fatty acids are present in a sample in addition to the common range of longer chain length components, it is possible to obtain an analysis by isothermal gas chromatography at two or more different temperatures, if a peak common to each chromatogram is used as a reference point for quantification purposes. It is generally much more convenient and accurate, however, to use temperature-programming in such circumstances, i.e. to start the analysis at a low column temperature and raise it at a fixed reproducible rate to an appropriate final temperature. The optimum temperature differentials and programming rates will depend on the nature of the sample and of the chromatographic column, and must be determined empirically in each instance. Ideally, members of homologous series should emerge at approximately equal time intervals, giving symmetrical peaks of approximately equal width on the chromatographic trace. Some valuable advice on the problems associated with the temperature-programmed analysis of the fatty acid components of some common fats of commercial interest, such as coconut oil or butter fat, is contained in a brief review article .
Because of its commercial importance, milk fat has provided many analysts with a challenge. As mentioned earlier, 437 different fatty acids have been reported as being present, but fortunately only a few of these have any nutritional significance and only 20 to 30 are likely to concern most analysts. The separation itself is a small part of the problem, and quantification presents more difficulties; the various steps in the process have been studied by several research groups [72,78,420]. First, it is essential that transesterification (to methyl or butyl esters) should be carried out by a method that ensures negligible losses, and that described by Christopherson and Glass , or appropriate modifications of it, should be used for the purpose (see Chapter 4 for details). Next, there should be no discrimination during the introduction of the sample onto the gas chromatograph; a non-vaporising on-column injection technique is strongly recommended, especially with WCOT columns. As cautioned above, the rate of temperature-programming during the chromatographic run should be such that the major components emerge at approximately equal time intervals. A typical chromatogram is illustrated in Figure 5.12. There is a larger than usual solvent peak, because of the low initial temperature and the presence of methanol, and the baseline is almost steady before butyrate emerges followed by each of the remaining fatty acids in turn. Perhaps the single most comprehensive analysis is that by Strocchi and Holman . Many different stationary phases have been used for milk fat analyses, and there is no clear favourite.
Figure 5.12. Milk fatty acids (methyl esters) separated on a fused-silica column coated with CP-Sil 84™. The oven was held at 30°C for 3 min, then was raised at 8°C per min to 160°C and was held at this point for a further 10 min. Other conditions are as in the legend to Figure 5.8.
Some workers who have studied the problem systematically indicate that small response factors only, and ideally the theoretical factors for flame ionisation detectors, calculated by Ackman and Sipos [27,28], are all that need be applied [78,197]. On the other hand, others claim that the theoretical response factors cannot be applied to short-chain esters [72,73]. It is certainly advisable to determine what response factors for 4:0 to 8:0 fatty acids are necessary under the analyst's own conditions. This is discussed in greater detail in Section G below.
2. Fatty acids of longer than usual chain length
When chromatographing the fatty acids of animal tissues, most analysts have tended to assume that they can terminate the analysis after the 22:6(n-3) fatty acid has emerged from the column, but it is increasingly being realised that components of even longer chain length will often be present. Their detection has been facilitated by the inertness of fused-silica WCOT columns and the high temperature stability of modern stationary phases, which also reduce the times necessary for elution. For example, Grogan  identified 26:4(n-6), 26:5(n-6), 28:5(n-6) and 30:5(n-6) fatty acids from rat testis on a fused-silica WCOT column coated with the highly polar SP-2340™ and temperature-programmed to 260°C. The ECL values obtained under isothermal conditions were consistent with the structures proposed. Fatty acids up to C26 were found in plasma by GC-MS , but more spectacular discoveries of C26 to C38 fatty acids with 4, 5 and 6 double bonds of both the (n-6) and (n-3) series have been found in the brain of patients with Zellweger's syndrome (a peroxisome deficiency) [733,838], in ram spermatozoa  and in vertebrate retina [64,66]. A nonpolar methylsilicone (BP-1™) in a fused-silica column, temperature-programmed to 320°C, resolved the brain fatty acids sufficiently well for GC-MS identification, as shown in Figure 5.13 . In contrast, polar phases in packed and WCOT columns were used for the unusual fatty acids from retina. Fatty acids of this type have long been known as constituents of marine lipids, and in one typical study 30:4(n-6) and 30:5(n-3) were identified on packed columns of Silar 10C™ and OV-101™ as components of the lipids of the sponge Cliona celata .
|Figure 5.13. Separation (part chromatogram) of very-long-chain fatty acids (methyl esters) from the cholesterol ester fraction of brain lipids from patients with Zellweger syndrome . A fused-silica WCOT column (12 m × 0.22 mm) coated with a methylsilicone phase (BP-1™) was temperature-programmed from 160 to 320°C, with helium at 1 mL/min as the carrier gas. (Reproduced by kind permission of the authors and of the Biochemical Journal, and redrawn from the original paper).|
Bacterial lipids may contain fatty acids with very long chain lengths, and the methyl ester derivatives of mycolic acids, silylated to reduce the polarity of the free hydroxyl group, from Corynebacterium diphtheriae were found to consist of more than 20 components with up to 36 carbon atoms and 4 double bonds ; these were resolved on a fused-silica column coated with a nonpolar phase (OV-1™) and temperature-programmed to 330°C. The record must, however, go to methyl ester/silylether derivatives of C66 to C84 mycolic acids from a Mycobacterium species, which were partially resolved on a short packed column (0.3 to 0.4 m × 3 mm) of 1% OV-101™, operated isothermally at temperatures of 320 to 340°C .
Mono-, di- and trimeric fatty acids, formed by thermal polymerisation, were resolved on a short packed column, containing 3% JXR™ (a methylsilicone) and temperature-programmed to 350°C .
The analysis of fatty acids of high molecular weight is one of the areas where HPLC with reversed-phase columns may have advantages in comparison to GC , and this technique was indeed also used to complement the GC analyses in some of the studies just described.
3. Acetylenic fatty acids
An isolated triple bond has a similar effect on the retention characteristics of a fatty acid as three methylene-interrupted double bonds, and methyl stearolate (methyl octadec-9-ynoate) is eluted with or slightly after methyl linolenate on columns packed with DEGS  or PEGA . Retention data on several different stationary phases have been published for the complete series of methyl octadecynoates (i.e. in positions 2 to 17) [85,323]. The ECL values tend to be parallel to those of the corresponding trans-monoene isomers, but differing by 0.6 to 0.7 units on DEGS and 0.1 to 0.2 units on Apiezon L™. In addition, ECL data are available for a number of synthetic C18 diynoates [326,533,545].
In certain species of moss, there are polyunsaturated fatty acids with acetylenic bonds separated from double bonds by single methylene groups. It proved possible to use separation factors and FCL values to predict the retention times of these compounds as with the more common range of unsaturated fatty acids as described above . Distinctive separations of acetylenic fatty acids are obtained on the cyanoalkylpolysiloxane phases, in that FCL values for triple bonds are appreciably lower than on more conventional polar phases .
4. Branched-chain fatty acids
While distinctive long-chain branched fatty acids occur in bacteria, fatty acids with simple methyl branches are encountered most often in microorganisms and in animal tissues. Generally, only a single methyl branch is present, but multi-branched fatty acids (including isoprenoids) are found in ruminant and certain other tissues, but especially in the preen glands of birds. Normally the acyl chain is saturated, but in some bacteria there may also be a single double bond. The separation of branched-chain fatty acids by GC has been reviewed elsewhere [6,114,854].
Retention data, including ECL values, were reported for the complete series of isomeric methyl-branched octadecanoates, and all are eluted before methyl nonadecanoate on both polar and nonpolar stationary phases [6,10,11]. In addition, analogous data for fatty acids of different chain lengths have been published . Elution patterns similar to those seen with the unsaturated series discussed above are found, in that components with methyl branches in the centre of the chain have the lowest retention times, while those remote from the carboxyl group have the highest. The iso- and anteiso-isomers, i.e. with the methyl branch on the penultimate and antepenultimate carbon atoms, respectively, are those most often found in nature as they are ubiquitous if minor constituents of animal lipids, and they are easily separated from each other. The iso-compound is eluted first, and the C17 fatty ester with this structure has an ECL value of approximately 16.5 on packed columns of 15 per cent EGSS-X™ and EGSS-Y™, while the related anteiso-compound has an ECL value of 16.7 on these columns (author, unpublished). Further data for WCOT columns are included in Table 5.3. Relatively small changes are seen for these compounds with stationary phases differing widely in polarity.
As with unsaturated fatty acids, FCL values obtained for monomethyl-branched fatty acids have been used to estimate ECL values for multi-branched acids with some success [9,11]. Similarly, Jacob [423,425] has tabulated FCL data for branching of various kinds in fatty acyl chains. However, GC-mass spectrometry (see Chapter 7) is probably the only reliable means of identification. Diastereomers of branched-chain acids have been separated on WCOT columns [9,20], and the problems of such analyses have been reviewed by Smith . Among fatty acids of this kind, most attention has been given to phytanic acid, 3,7,11,15-tetramethylhexadecanoic acid, a metabolite of phytol which can accumulate in the tissues of ruminants and of humans under certain conditions ; it tends to emerge very close to 17:0 on stationary phases of moderate polarity, and with 18:2(n-6) on Apiezon L™.
The difficulties involved in the analysis of complex mixtures of branched-chain fatty acids are perhaps best illustrated by some selected examples of actual analyses, and the reader may find illuminating those studies of such components in human milk , Vernix caseosa (mono-, di- and trimethyl-branched) , ruminant tissues [232,853], avian uropygial gland secretions (reviews) [423,425], the bacteria Streptomyces R61 and Actinomadura R39  and the bacterium Desulfovibrio desulfuricans (iso- and anteiso-methyl-branched and monoenoic) .
5. Cyclopropane, cyclopropene and other carbocyclic fatty acids
Cyclopropane fatty acids are common constituents of bacterial fatty acids and sometimes accompany cyclopropene fatty acids in certain seed oils. The ECL values of the complete series of methyl esters of C19 isomeric cyclopropane fatty acids have been recorded on polar (NPGS, DEGS and PEGA) and nonpolar (Apiezon L™) liquid phases; DEGS and PEGA were in packed columns and the remainder were in WCOT columns . In each instance, the ECL values are approximately one unit greater than those of the corresponding C18 monoenoic esters from which they are derived, synthetically or biosynthetically. Therefore, as with the monoenes, a cyclopropane ring in the centre of the chain has less effect on retention time than one at either extremity of the molecule. The 9,10- and 11,12-methyleneoctadecanoates, which are occasionally found together in bacterial lipids, were separable on the nonpolar WCOT column. Synthetic fatty acids containing a cyclopropane ring with a trans-configuration were eluted before the corresponding cis-isomers on packed columns of EGS and Apiezon L™ . The chromatographic properties of cyclopropane and other carbocyclic fatty acids have been reviewed .
Methyl esters of cyclopropene fatty acids are less easily analysed by gas chromatography as they tend to decompose or rearrange on GC columns to give spurious peaks. As cautioned in Chapter 4, it is necessary to avoid acidic conditions when preparing the methyl ester derivatives. If highly inert supports and silicone liquid phases are used, however, successful GC of the native esters is possible [258,759]. For example, methyl sterculate eluted after 18:0 and methyl malvalate just before 18:2 (and partially overlapping with it) on a packed column of 5% SP-2100™ (2 m × 2 mm); as little as 0.3% of each could be quantified . Some useful results have also been achieved on a glass WCOT column coated with Carbowax 20M™ [99,281]. The author has successfully chromatographed methyl sterculate on a well-conditioned EGSS-X™ column (packed) where it eluted just after methyl linoleate, although the same compound decomposed on a new EGSS-Y™ column. On the other hand, it might be easy to inadvertently miss a small amount of degradation on a GC column, since Conway et al.  found that some minor decomposition products co-chromatographed with other fatty acids on a Carbowax 20M™ WCOT column.
Most workers have preferred to prepare stable derivatives prior to analysis. For example, cyclopropene fatty acids can be subjected to hydrogenation, or reaction with silver nitrate  or methanethiol . Silver nitrate in anhydrous methanol reacts with cyclopropene rings in about 2 hours at 30°C to form predominantly methoxy ether but with some enonic derivatives, which appear as twin peaks (because of reaction on either side of the ring) on analysis by GC [99,241,281]. An application of this procedure to the analysis of kapok seed oil is illustrated in Figure 5.14. Alternatively, a brief reaction with hydrazine will selectively reduce the cyclopropene compounds to the more stable cyclopropanes; by examination by GC before and after the reaction, the small amounts of natural cyclopropane components can also be identified .
Figure 5.14. Analysis of kapok seed oil methyl esters after reaction with silver nitrate in anhydrous methanol, on a glass WCOT column coated with Carbowax 20M™, maintained at 190°C with hydrogen as carrier gas . Abbreviations: 18:CA, dihydromalvalic acid; 19:CA, dihydrosterculic acid; 18:CM, methoxy derivatives of malvalic acid; 19:CM, methoxy derivatives of sterculic acid; 18:CC, enone derivatives of malvalic acid; 19:CC, enone derivatives of sterculic acid. (Reproduced by kind permission of the authors and of Analytical Chemistry, and redrawn from the original paper).
Seed oils containing cyclopropene fatty acids have been successfully analysed by means of HPLC in the reversed-phase mode (reviewed elsewhere ) and this may now be the method of choice.
Other saturated and unsaturated cyclic esters are more stable to gas chromatography. Methyl 11-cyclohexylundecanoate, a minor constituent of rumen microrganisms and milk fat, is eluted with 18:2(n-6) on a packed column of PEGA and with 18:0 on one of Apiezon L™ . In addition, retention data have been published for some synthetic and natural ω-alicyclic fatty acids on polar and nonpolar phases [215,458].
Cyclopentenoic acids occur in seed oils of Hydnocarpus and related species of the genus Flacourtiaceae, and they have been subjected to GC analysis. The main components are chaulmoogric (13-cyclopent-2-enyltridecanoic), hydnocarpic (11-cyclopent-2-enylhendecanoic) and gorlic (13-cyclopent-2-enyltridec-6-enoic) acids, but higher and lower homologues exist. On packed columns of polar phases, such as EGS, the cyclopentene ring has a substantial effect on retention time but methyl hydnocarpate and methyl oleate elute together; on nonpolar phases, such as Apiezon M™, this pair are separated [75,876,1013]. By a careful use of polar and nonpolar phases, it is possible to estimate all the main components. Fortunately, the principal fatty acid constituents of such seed oils are now much more readily separated on fused-silica WCOT columns coated with phases such as Silar 5CP™ and Carbowax 20M™, from which methyl hydnocarpate elutes before methyl stearate as illustrated in Figure 5.15 ( and W.W. Christie, E.Y. Brechany and V.K.S. Shukla, Lipids, 24, 116-120 (1989)).
Figure 5.15. Separation of the methyl ester derivatives of the fatty acids of Hydnocarpus anthelminticus on a WCOT column of Carbowax 20M™ (see Fig. 5.8 for conditions). Abbreviations: 16cpe, 11-hendecylcyclopent-2-enoate; 18cpe, 13-tridecylcyclopent-2-enoate; 18cpde, 13-tridec-4-enylcyclopent-2-enoate; 20cpe, 15-pentadecylcyclopent-2-enoate; 20cpde, 15-pentadec-9-enylcyclopent-2-enoate.
6. Oxygenated fatty acids
Oxygenated fatty acids are not uncommon in plant and microbial lipids, and they are also found in animal tissues, where 2-hydroxy fatty acids especially are ubiquitous constituents of the sphingoglycolipids. Their GC characteristics have been reviewed by Vioque, who presents a substantial list of ECL data . Tulloch has reported ECL values for the complete series of methyl hydroxy- and acetoxystearates on polar and nonpolar liquid phases in packed columns . As is usual with substituted compounds, components with central functional groups are not easily separated, but where the substituents are close to either end of the molecule, positional isomers can be resolved. In this instance, the ECL values are highest for the 4- and 5-isomers, drop as the functional group approaches the carboxyl end of the molecule and rise appreciably close to the terminal methyl end. A hydroxyl group increases the ECL value of a fatty acid greatly, especially on polar columns, and on EGS and SE-30, methyl 12-hydroxystearate has ECL values of 26.25 and 20.00, respectively. Its acetate derivative has ECL values of 24.70 and 20.55, respectively, on these phases. In addition, the nature of the derivative has a profound effect on the elution profile for different isomers.
Compounds with free hydroxyl groups present several problems to the gas chromatographer. Adsorption on the support or on the walls of the column and hydrogen-bonding effects may come into play, so that unsymmetrical peaks are obtained and recoveries are incomplete. Losses can occur because of transesterification of the hydroxyl group with polyester liquid phases. Although such phenomena are less apparent with modern catalyst-free liquid phases and highly inert supports, and they are now much less troublesome with glass or fused-silica WCOT columns coated with nonpolar phases (c.f. ), it is still advisable to chromatograph the compounds in the form of a nonpolar volatile derivative such as the acetate , trifluoroacetate  or trimethylsilyl ether . With the last especially, sharper peaks, better recoveries and improved resolutions of positional isomers are obtained. n-Butylboronate derivatives are invaluable for characterising 2- and 3-hydroxy fatty acids (see Chapter 4 for details of preparation) [55,133].
The 2-hydroxy fatty acids of sphingolipids have been resolved on packed columns of SE-30™  and EGSS-X™  as the methyl ester-trimethylsilyl ether derivatives for identification by mass spectrometry. On the other hand, much better separations were obtained on a fused-silica WCOT column (25 m × 0.25 mm) coated with OV-101™, as shown in Figure 5.16 . In this instance, the 2-hydroxy acids were chromatographed simply as the methyl ester derivatives together with the nonhydroxy fatty acid constituents, although the two groups could also be resolved on their own following a TLC separation. Other natural hydroxy fatty acids (also containing isolated double bonds) occur in seed oils, and ECL data for several of these (and derivatives) have been published [4,482,907]. With fatty acids with conjugated unsaturation that also contain allylic hydroxyl groups, dehydration may occur leading to spurious peaks . The problem was apparently not alleviated if the acetate or methoxy derivatives were prepared, but it might be instructive to repeat this work with more modern catalyst-free liquid phases on inert supports, or better with a nonpolar phase on a WCOT column of fused silica. Certainly, TMS ether derivatives of conjugated hydroxy fatty acids have apparently been subjected to GC on WCOT columns successfully .
Figure 5.16. Separation of hydroxy and nonhydroxy fatty acid methyl esters from cerebrosides of bovine brain on a fused-silica WCOT column (25 m × 0.25 mm) coated with OV-101™ . The column temperature was 280°C and the flow-rate of the nitrogen carrier gas was 0.5 mL/min. (Reproduced by kind permission of the authors and of the Journal of Chromatography, and redrawn from the original paper).
Synthetic polyhydroxy esters have been subjected to GC in the form of the trimethylsilyl ethers , as trifluoroacetates  and as isopropylidene derivatives . Erythro- and threo-forms of vicinal diols can be separated on packed columns when they are converted to either of the last two derivatives and, as these compounds can be prepared quantitatively to a high degree of stereochemical purity from cis- or trans-olefins, respectively, this provides a basis for gas chromatographic separation and estimation of stereoisomers of unsaturated acids on packed columns . (Epoxide derivatives can be used in a similar manner as discussed in Sections D.1 and 2 of this chapter [116,243,244]). It appears to be a general rule that derivatives with the trans- or threo-configuration are eluted before those of the cis- or erythro-configuration, especially on nonpolar stationary phases. Polyhydroxy fatty acids (also with oxo and epoxyl groups) occur naturally in the form of polyesters in shellac and in plant cutins. For the latter, the composition, methods of analysis and GC retention data have been reviewed [391,929].
Optical enantiomers of the methyl esters of mono-hydroxy fatty acids can be resolved by gas chromatography if they are first converted to the (-)-menthyl, D-phenylpropionyl or related derivatives [348,468] (reviewed briefly elsewhere [456,858]). Particularly good results were obtained with racemic methyl 2-hydroxy palmitate and stearate, after conversion to the (-)-α-methoxy-α-trifluoromethylphenylacetate derivatives, on a WCOT column coated with OV-1™ as shown in Figure 5.17 ; it is probable that the D-form elutes first. Some limited success has also been achieved in separating diastereoisomers of polyhydroxy fatty acids .
|Figure 5.17. Separation of diastereomeric methoxytrifluoromethyl phenylacetate derivatives of methyl 2-hydroxypalmitate (A,B) and 2-hydroxystearate (C,D) on fused-silica WCOT columns (25 m × 0.32 mm) coated with OV-1™ . The column was temperature-programmed from 180 to 240°C at 2°C/min, and helium was the carrier gas. (Reproduced by kind permission of the authors and of the Journal of Chromatography, and redrawn from the original paper).|
Hydroperoxy fatty acids as such cannot be separated by GC as they decompose at high temperatures, and HPLC is probably the preferred method for their analysis . Nonetheless, there are times when it is advantageous to convert them to the hydroxy derivatives by means of sodium borohydride reduction and then to the TMS ethers for GC analysis, for example for identification by GC-mass spectrometry; products derived from linoleic [223,341,911,946,1005], arachidonic [124,407,568, 1005] and docosahexaenoic acids  have been examined in this way (the list is not intended to be comprehensive). Woollard and Mallet , in particular, have presented a comprehensive list of ECL data for compounds of this type. In addition, GC methods were used for the determination of the absolute configuration of hydroperoxides formed by lipoxygenase reaction [124,568,947].
Retention data for all the isomeric methyl oxostearates were recorded by Tulloch , and the elution pattern resembles that described above for the corresponding acetoxy derivatives. GC retention data for some conjugated polyenoic fatty acids with keto groups in position 4 have been recorded .
GC separations of synthetic epoxy fatty acids have been referred to briefly above. Epoxy fatty acids also occur naturally, in seed oils especially, and ECL data for a number of these have been published [191,735,871].
Furanoid fatty acids are present in some seed oils, but more interestingly perhaps they are found in the reproductive tissues of fish. The chromatographic analysis of these compounds has been reviewed by Lie Ken Jie . It is apparent that the methyl ester derivatives can readily be subjected to GC analysis, and retention data have been published for a number of synthetic [4,548,549] and naturally occurring compounds [300,301,335,790,825].
7. Other fatty acids
Deuterated fatty acids can be separated from non-deuterated in part at least on WCOT columns coated with polar stationary phases [693,695]. t-Butyldimethylsilyl ester derivatives (see Chapter 4 and Chapter 7) are useful for GC-mass spectrometric estimation of these compounds [692,709,1004]. Long-chain dicarboxylic acids have been separated by GC and identified by GC-MS in the serum of patients suffering from Reye's syndrome  and in the lipids of royal jelly [542,543]. Brominated vegetable oils are added to soft drinks to disperse flavouring agents; after acid-catalysed methanolysis, the methyl ester derivatives of fatty acids containing two, four and six bromine atoms have been separated on packed  or fused-silica  columns with nonpolar silicone stationary phases.
8. Free fatty acids
There may be times when it is necessary to analyse fatty acids by GC in the free (underivatized) form, and procedures for achieving this have been reviewed in some detail by Kuksis . While this is not easily accomplished on the more conventional polyester phases, it is not at all difficult if specially made packings are used, such as the FFAP™ and Carbowax 20M™-terephthalic acid phases discussed briefly in Section B.1 (Part 1 of this chapter), in glass or fused-silica WCOT columns (see for example [211,731]). Fused-silica columns containing a chemically bonded phase of the FFAP type are now available commercially available specifically for this type of analysis. It is also possible to achieve acceptable separations on fused-silica columns coated with methylsiloxane polymers, provided that the column is not overloaded .
G. Quantitative Estimation of Fatty Acid Composition
Although this is the last Section of this Chapter, it is certainly not the least important. Ackman  has suggested that many analysts might consider it indelicate if asked "When did you last verify the response quantification and linear range of your gas chromatograph?" Yet as data collection has become more automated, it has become easier to neglect this aspect because of the popular view that "computers don't make mistakes". In fact, errors or selective losses can be introduced during sample preparation, on injection into the column, while on the column, at the detector, and in measuring peak areas and relating them to sample quantity.
With modern flame ionisation detectors, there should be sufficient linear range to cope with most problems, although this can be abused if columns are grossly overloaded (a common fault with novices to the technique). It has recently been demonstrated that the effective linear response of detectors can be improved if a higher flow-rate is used than is required for optimum sensitivity . As discussed from a theoretical standpoint in Chapter 3, the areas under the peaks on the GC traces are within limits linearly proportional to the amount (by weight) of material eluting from the columns. Problems of measuring this area arise mainly when components are not completely separated, and there is no way of overcoming this difficulty entirely. When overlapping peaks have distinct maxima, the height multiplied by retention time method of area measurement is probably the most accurate manual technique in isothermal analyses, but computer analysis of peak shapes may improve the accuracy of the estimation. Where one component is visible only as a minor shoulder or broadening of a major peak, no manual or computer method is likely to give very precise results for the individual components, although electronic integration can at least give an accurate measure of the total amount of material present in a multiple peak. Wherever feasible, column conditions should be altered in an attempt to improve resolution, or another liquid phase can be tried. The magnitude of this problem can be much less with modern WCOT columns.
Errors arising through loss of components on the columns can be minimised with packed columns as discussed above, by using catalyst-free liquid phases and highly inert supports. With all column types, it is necessary to check at frequent regular intervals whether losses are occurring by running standard mixtures of accurately known composition through the columns. These standards should be similar to the samples to be analysed; for example, saturated standards should not be used to calibrate columns for the analysis of polyunsaturated fatty acids, and the calibration should be checked regularly. Methyl ester mixtures made originally to the specifications of the National Institutes of Health in the U.S.A. are available from several commercial sources for the purpose. Difficulties arise most often with polyunsaturated fatty acids, and it is possible to circumvent them in part by adding an appropriate internal standard. For example, it has been suggested that a 22:1 fatty acid be used in the determination of 22:6(n-3) in fish oils  and a 20:3 fatty acid for 20:4(n-6) in animal tissue lipids . Of course, with fatty acids of this type a major potential source of loss is autoxidation caused by faulty sample-handling technique. One simple and convenient means of checking this aspect is to analyse a sample before and after hydrogenation to compare the relative proportions of the components of various chain lengths (a suitable procedure is given in Chapter 4).
With WCOT columns, losses are most likely to occur through faulty injection technique (see Chapter 3 for a detailed discussion). The author is convinced of the value of "hot needle" and "cold-trapping" injection techniques, especially with split injection, and he used these in preparing the chromatograms in all those figures in this book from his own laboratory. Factors affecting accuracy and reliability with this type of injection system with specific reference to fatty acid analysis have recently been reassessed . A high speed of injection was found to be especially effective in avoiding discrimination when the sample was in the needle, while rapid vaporisation was achieved by using relatively dilute solutions of the smallest size of sample that could reasonably be analysed. In addition, an injection temperature much higher than is usually recommended, namely 375°C, improved the reproducibility appreciably and had no detectable effect on the recovery of highly unsaturated components. On-column injection techniques with WCOT columns of course present fewer problems to the analyst.
Even with packed columns, good sample injection technique can improve the recoveries of fatty acids and the resolutions attainable. The syringe needle should be inserted rapidly but steadily into the column until it reaches a predetermined depth, just below the surface of the packing material, when the plunger is pressed in firmly; the needle is left in place for about one second and is then removed rapidly.
If necessary, calibration factors may have to be calculated for each component to correct the areas of the relevant peaks in the mixtures analysed. When flame ionisation detectors are used, small correction factors can be applied, when high precision is required, to compensate for the fact that the carboxyl carbon atom in each ester is not ionised appreciably during combustion [8,10,27]. There are also small effects of this kind due to the absence of hydrogen atoms at double bonds. A list of these factors is contained in Table 5.4. The degree of correction necessary is obviously greatest for fatty acids of shorter chain length or with a high degree of unsaturation. In a recent reassessment of this aspect of quantification in the analysis of fatty acid methyl esters on WCOT columns, it was concluded that these theoretical detector response factors were indeed valid and applied equally to the analysis of short-chain and polyunsaturated components [38,78,79,197]. If further correction factors were found necessary by analysts, it was concluded that some aspect of the chromatographic technique has not been optimised. On the other hand, a deficiency in the carbon atom response for lower esters has been found by other analysts, and it appears that empirical response factors are necessary in the analysis of milk fats say [27,72,73]. Careful calibration with pure standards is necessary in this instance.
Theoretical response correction factors (RCF) for flame ionisation detectors to convert to weight percent methyl ester, and molecular weight correction factors (MWF) to convert to molar percent for some common fatty acids. To open this - click here..
In general, the proper approach to the generation of results of high accuracy is to optimise the equipment parameters and operational technique (sample preparation and injection) so that the true answer is obtained with a primary standard, rather than to introduce empirical correction factors to correct for faulty practice.
It is not easy to give an objective assessment of the standard of accuracy that should be possible in routine analyses of fatty acids, but in a collaborative study of IUPAC methodology for fatty acid analysis, typical coefficients of variation (%) at various concentrations were 15 (2% level), 8.5 (5% level), 7 (10% level) and 3 (50% level) . As might be expected, the greatest errors were experienced in estimating butyric acid in milk fat.
When an analysis has been completed, the results can be expressed directly as weight percentages of the fatty acids for presentation in tabular form. On the other hand, it is often necessary to calculate the molar amounts of each acid as, for example, in most lipid structural studies (positional distributions and molecular species proportions). This is performed simply by multiplying the area of each peak by an arithmetic factor, obtained by dividing the weight of a selected standard ester (say 16:0) by the molecular weight of the component, followed by re-normalising to 100%. For convenience, these factors are also listed in Table 5.4. It should be noted that if fatty acid compositions are calculated on a weight percent basis, it is not always necessary to positively identify each compound; this cannot be avoided if molar proportions are required.
The following abbreviations are employed at various points in the text of these chapters:
amu, atomic mass units; BDMS, tert-butyldimethylsilyl; BHT, 2,6-di-tert-butyl-p-cresol; CI, chemical ionisation; DNP, dinitrophenyl; ECL, equivalent chain length; ECN, equivalent carbon number; EI, electron-impact ionisation; FCL, fractional chain length; GC, gas chromatography; GLC, gas-liquid chromatography; HPLC, high-performance liquid chromatography; IR, infrared; MS, mass spectrometry; NMR, nuclear magnetic resonance; PAF, platelet-activating factor; ODS, octadecylsilyl; TLC, thin-layer chromatography; TMS, trimethylsilyl; UV, ultraviolet.
This document is part of the book Gas Chromatography and Lipids by William W. Christie and published in 1989 by P.J. Barnes & Associates (The Oily Press Ltd), who retain the copyright.
Return to Part 1 of this Chapter
Updated June 27, 2011