The Lipid Library

GAS CHROMATOGRAPHY AND LIPIDS

Chapter 5 - Sections A to D

GAS CHROMATOGRAPHIC ANALYSIS OF FATTY ACID DERIVATIVES

A.  INTRODUCTION

The advent of gas-liquid chromatography revolutionised the analysis of the fatty acid components of lipids, and it is undoubtedly the technique that would be chosen in most circumstances for the purpose. It is now possible to obtain a complete quantitative analysis of the fatty acid composition of a sample in a very short time. Individual fatty acids can usually be identified by GC with reasonable certainty from their relative retention times, especially if the analysis is carried out with a variety of stationary phases, and taking into account the large body of knowledge that now exists on the compositions of specific tissues or organisms. On the other hand, there are many circumstances when it must be recognised that GC analysis permits a tentative identification only. In the first analysis of any new sample, for example, confirmation of fatty acid structures may have to be obtained by unequivocal chemical degradative and spectroscopic procedures, although some of these may also benefit from an involvement of GC.

The only technique to compare with GC for the analysis of fatty acid derivatives is HPLC in the reversed-phase mode with UV-absorbing or fluorescent derivatives (reviewed elsewhere [168] and briefly in the Chapter 6). Both the capital and running costs of this technique are appreciably higher than for GC, and identification of components emerging from HPLC columns is rarely easy, because of the complex nature of the separation process. On the other hand, HPLC certainly has advantages for the isolation of specific components on a small scale for structural analysis or for radioactivity measurements, and for the analysis of fatty acids with labile functional moieties such as hydroperoxy or cyclopropene groups.

GC procedures only for the analysis of fatty acid derivatives are considered in this Chapter, and the topic is discussed in terms of both packed and WCOT columns. The superb resolution attainable with WCOT columns can present the analyst with the problem of identifying large numbers of minor components, many of which have little metabolic or nutritional relevance. Analysis with packed columns gives simplified chromatograms, but often with much of the essential information. Several review articles have appeared recently on the subject [14,163, 428,430,508,546,922]. Alternative or complementary methods for identification of fatty acids are described in Chapter 6 and Chapter 7 (mass spectrometry).

Precautions should be taken at all times to prevent or minimise the effects of autoxidation (see Chapter 2).

B.  COLUMN AND INSTRUMENTAL CONSIDERATIONS

1.  Liquid phases

The liquid phase used in a GC column is the principal factor determining the nature of the separations that can be achieved. Non-polar silicone liquid phases, such as SE-3O™, OV-1™, JXR™ or QF-1™, permit the separation of fatty acid esters mainly on the basis of their molecular weights, when in packed columns. However, there can be separation of unsaturated fatty acids of the same chain-length with WCOT columns (c.f. Figure 5.7 below) or when the amount of the stationary phase on the support (in packed columns) is low (1-3 %). Non-polar phases are of value in the analysis of fatty acid derivatives of higher than normal molecular weight, especially with WCOT columns.

Liquid phases consisting of high molecular weight hydrocarbons, such as the Apiezon™ greases (of which the most popular is Apiezon L™) also separate saturated and unsaturated components of the same chain-length, unsaturated esters eluting before the related saturated compounds, but in packed columns there is very little separation of esters of a given chain-length differing in the number of double bonds in the molecule; these phases are now less used than formerly having been superseded by methylsilicone polymers.

Polar polyester liquid phases are much more suited to fatty acid analysis as they allow clear separations of esters of the same chain-length, but with zero up to six double bonds, unsaturated components eluting after the related saturated ones. These phases can be subdivided into four main classes: group a, the highest polarity phases, e.g. alkylpolysiloxanes containing various polar substituents including nitrile groups, marketed under trade designations such as Silar 10C™, Silar 9CP™, SP 2340™ and OV-275™; group b, highly polar phases, e.g. polyethyleneglycol succinate (EGS), polydiethyleneglycol succinate (DEGS), EGSS-X™ (a copolymer of EGS with a methyl silicone), CP-Sil 84™ and CP-Sil 88™; group c, medium polarity phases, e.g. polyethyleneglycol adipate (PEGA), polybutanediol succinate (BDS) and EGSS-Y™ (a copolymer of EGS with a higher proportion of the methysilicone than in EGSS-X™); group d, low polarity phases, e.g. polyneopentylglycol succinate (NPGS), EGSP-Z™ (a copolymer of EGS and a phenyl silicone), Carbowax 20M™ and Silar 5CP™. In work with packed columns, EGSS-X™, EGSS-Y™ and newer phases of equivalent polarity are widely accepted as the most useful representatives of the second and third groups respectively because of their relatively high thermal stability, particularly in packed columns. Low polarity phases are utilised principally in WCOT and SCOT columns, because saturated and monoenoic components of the same chain-length are poorly separated when these phases are used in packed columns. In packed column work, the phases from the group of highest polarity are stable at temperatures above those possible with EGSS-X™ (providing that oxygen is rigidly excluded), for example, and they afford excellent separations of polyunsaturated fatty acids, complementing those obtained with the more common stationary phases; they are of particular value in the separation of cis- and trans-isomers. Examples of the properties of many of these phases in specific applications are given below.

It is occasionally advantageous to be able to subject fatty acids in the unesterified form to GC analysis. In this instance, acidic liquid phases such as DEGS containing 3 % phosphoric acid, FFAP™ ("free fatty acid phase"), Carbowax 20M™-terephthalic acid or the structurally related phase, SP-1000™, are used (see Section F.8).

The quality and to some extent the nature of the separations achieved is influenced by the amount of liquid phase applied to the inert support in packed column GC. Low levels of polyesters (1-3 % by weight relative to that of the support) are occasionally suggested for use with highly inert supports, as methyl esters of long-chain polyunsaturated fatty acids will elute from them at comparatively low temperatures. On the other hand, greater amounts of polyester (10-15 % by weight) offer more protection to polyunsaturated esters at high temperatures and this is generally preferred. The retention times of fatty acid esters relative to a chosen standard ester (usually 16:0 or 18:0) tend to decrease as the amount of liquid phase on the support is decreased and therefore, for reproducible work, it is advisable to determine the optimum amount of liquid phase necessary for a given separation and to standardise the chromatographic conditions accordingly. Unfortunately, columns age with use as the stationary phase polymerises further or bleeds from the column, and some changes in the retention characteristics of esters inevitably occur.

When the technique was first introduced, there were a number of reports of losses of esters of polyunsaturated fatty acids on GC columns. These were attributed partly to the use of too active support materials and partly to trans-esterification of methyl esters with the polyester liquid phase. The latter effect was caused by residues of the catalyst required for the preparation of the polyester, so that those components remaining longest on the columns suffered the greatest losses. Now, polyester liquid phases are made without the aid of catalysts and such losses should not be significant.

As long as packed columns continue to be employed for fatty acid analysis, it would aid inter-laboratory comparisons, if more general use were made of a select number of phases; the author has found that three columns containing 15 % EGSS-Y™, EGSS-X™ and Silar 10C™ on 100-120 mesh acid-washed and silanized supports cover a sufficient range of polarity for the analysis of most polyunsaturated fatty acids. Fatty acids which co-chromatograph on a given column will often be separable on one of the others.

With the advent of WCOT columns of fused silica, the number of different stationary phases in use for the analysis of fatty acid derivatives has appeared to diminish, although phases covering the extremes of the polarity spectrum do have specific uses. Polyglycol phases based on Carbowax 20M™ seem to have been used in a high proportion of recent published papers. Many of these are sold under trade designations so that their chemical derivation is not immediately obvious, but Carbowax-20M™, FFAP™, Supelcowax-10™, and SP-1000™, for example, are very similar. Ackman [14] has shown that these phases, variously in glass, fused silica or stainless steel WCOT columns, have very similar effects on the relative retention times of a wide range of fatty acid derivatives. Similarly, cross-linking or bonding to the support did not appear to affect relative retention times greatly, although others found that small but significant improvements in the resolution of certain critical pairs could be obtained with a cross-linked and bonded phase [493]. In polarity, these stationary phases fall into the last of the groups described above, but the inherent resolution of WCOT columns is such that satisfactory resolution is achieved for most practical purposes. Indeed, they can have distinct advantages in many applications (see below).

Ackman [14] has proposed that phases of the Carbowax 20M™ type should be utilised in the "'standard' reference WCOT column for inter-laboratory studies as well as for application in its own right". For the moment and until any new phase with demonstrable advantages is introduced, this suggestion seems eminently sensible. The author has made considerable use of fused silica columns coated with Silar 5CP™, a phase which is slightly more polar than Carbowax 20M™.

2.  Carrier gases

It was demonstrated in Chapter 3 that hydrogen had undoubted advantages as a carrier gas in WCOT columns in terms of efficiency. On the other hand, the potential consequences of a leak of hydrogen within the oven of a gas chromatograph should be obvious to all. In modern instruments with solid state electronics, the risk is perhaps less than it was formerly, but it is still advisable to install leak detection equipment as a precaution. Helium is much safer, if rather costly outwith the U.S.A., and still gives excellent results. Both gases should be passed through oxygen and moisture traps in advance of the column to give greater base-line stability at high sensitivity and to prolong column life.

With packed columns, nitrogen (containing less than 5 ppm of oxygen) affords adequate resolution at the standard flow-rates. Argon is more expensive than nitrogen, but it is generally obtainable with a very low oxygen content so has advantages in the analysis of sensitive compounds.

3.  The oven of the gas chromatograph

It should go without saying that the oven in a gas chromatograph should have a highly sensitive means of controlling the temperature, and that the temperature should be uniform in all regions. Unfortunately, when WCOT columns of fused silica were introduced, it became apparent that higher standards were needed in some commercial instruments. The reasons for this are still not entirely clear, but it may be related to the thinner walls or the rate of transmission of heat through fused silica rendering such columns more sensitive to minor temperature fluctuations or uneven heat distributions within the oven. In practice, the effects were manifested by poor peak shapes, often exhibiting spiking. Ackman [14] solved the problem in his laboratory by enclosing the column in a simple chamber made from disposable aluminium pie plates. The author merely wrapped aluminium foil round his columns to resolve the difficulty (at the suggestion of the instrument manufacturer). New equipment ought not to suffer from this problem.

C.  PROVISIONAL IDENTIFICATION USING STANDARDS OR RETENTION TIME RELATIONSHIPS

Lipid analysts soon acquire an intuitive understanding of the relationship between the retention times of peaks on a GC trace and their identity. For example, a typical fingerprint of the fatty acids from animal tissue phospholipids would have the 16:0 component standing in relative isolation, followed by the three peaks for the C₁₈ components (18:0, 18:1 and 18:2), then a gap to the next substantial peak for 20:4(n-6), followed by a further gap to the C₂₂ components, the last of which is 22:6(n-3). Many of the minor peaks can be identified tentatively according to their proximity to these major components.

Please note that in this section and in the remainder of the Chapter, it will be assumed that fatty acids are being subjected to GC in the form of the methyl ester derivatives, unless it is stated otherwise.

When the fatty acids of a simple material such as maize (corn) oil is analysed in a laboratory for the first time, there should be no problems of identification, because its composition has been so well documented in the literature. This may also be true of some more complex lipid samples, such as extracts of rat liver. However, problems of identification can arise whenever any new or unknown sample is analysed, or when trace components are seen in otherwise familiar samples, now a relatively common situation with WCOT columns. If the interpretation of metabolic events in tissues hinges on the recognition of a particular fatty acid, intuitive labelling will not suffice. Components can be identified with much more certainty from a systematic study of retention time relationships on particular liquid phases, as described in the remainder of this section. Ultimately, it may be necessary to use the methods described in the Chapter 6 and Chapter 7 (mass spectrometry) for complete certainty.

Standard mixtures containing accurately known amounts of methyl esters of saturated, monoenoic and polyenoic fatty acids are available commercially from a number of reputable biochemical suppliers. These are invaluable for checking the quantification procedures used (see Section G), and also for the provisional identification of fatty acids by direct comparison of the retention times of their methyl esters with those of the unknown esters on the same columns under identical conditions. Comparisons should be made on at least two columns with phases of different polarity.

The lipids in animal tissues usually contain a much wider spectrum of fatty acids than is available commercially. It is then helpful to obtain a secondary external reference standard consisting of a natural fatty acid mixture of known composition. This can be a common natural product that has been well characterised or a mixture of natural esters, the composition of which has been accurately established by the procedures described in Chapters 6 and 7. Ideally, it should be similar to the samples under investigation; for example, Ackman and Burgher [15] used cod liver oil in this way to identify the fatty acids of other marine animals, and Holman et al. [392,393] have used the fatty acids of bovine and porcine testes in the same manner in analyses of animal tissues. Rat liver fatty acids are frequently used for the latter purpose. In work with packed columns, the author used a simple mixture made up of the fatty acids of pig liver lipids and cod liver oil together with a little linseed oil, as this contains significant amounts of all the major fatty acid classes (saturated and mono-, di-, tri-, tetra-, penta- and hexaenoic components of both the (n-3) and (n-6) families), including most of the fatty acids likely to be encountered in animal tissues; the ingredients are readily obtainable from local shops.

Figure 5.1 illustrates separations of such a standard mixture, on 15 per cent EGSS-X™ and EGSS-Y™ in packed columns. Both phases give excellent separations of fatty acid esters of a given chain-length that differ in degree of unsaturation. Separation of esters which differ only in the positions or configurations of the double bonds, where these are approximately central, is not easily achieved with monoenoic acids, but is possible with esters of polyunsaturated fatty acids. For example, on both of these columns, 18:3(n-3) and 18:3(n-6) are separated, as are three isomers of 20:3, two isomers of 20:4 and two isomers of 22:5 as is apparent in Figure 5.1. With the methyl esters of the more common families of polyunsaturated fatty acids, the shorter the distance between the last double bond and the end of the molecule, the longer the retention time of the isomer.

Figure 5.1. GC analysis of a complex mixture of natural fatty acids (as the methyl ester derivatives) on packed columns with EGSS-X™ and EGSS-Y™ as stationary phases (see footnote to Table 5.1 for further chromatographic details).

The principle disadvantage of these columns is that there is some overlap of fatty acids of different chain lengths. For example on EGSS-X™, 18:3(n-3) and 20:1(n-9) coincide, as do 20:4(n-6) and 22:1(n-9). On EGSS-Y™, these pairs can be separated, but 20:0 and 18:3(n-3), 18:4(n-3) and 20:1(n-9) or 22:1(n-9) and 20:4(n-3) are not separable. If Silar 10C™ had been selected as a stationary phase rather than EGSS-Y™, a different range of separations again would be seen, complementing those on EGSS-X™. Again, by a judicious use of two or more columns differing in polarity in this way, most of the fatty acids of metabolic importance can be separated and estimated.

With WCOT columns, the inherent resolution is such that there tend to be fewer problems of overlap of major components. On the other hand, a multiplicity of peaks may be revealed, so compounding the identification problems. It is possible to eliminate the difficulties with overlapping components of different chain-lengths by using low polarity polyester liquid phases such as those of group d above (but not with packed columns in which the resolution is markedly inferior to that obtained with more polar liquid phases).

As cautioned earlier, it should be noted that the retention times of esters and the separations achieved are all dependent on the precise column conditions used and may vary with such factors as the temperature or the age of a column and the amount of stationary phase on the support (with packed columns).

The absolute retention time of an ester on any GC column has very little meaning as a measure of its elution characteristics, because slight changes in the operating conditions or in the character of the packing material (in its origin or on ageing) can affect this parameter drastically. In contrast, the retention time of a fatty acid ester relative to that of a chosen standard commonly occurring component (usually 16:0 or 18:0) has a greater absolute significance and is a quantity more suited to inter-laboratory comparisons, i.e. the relative retention time (r_18:0) of an ester is its retention time divided by that of 18:0.

Theoretically, retention times should be measured from the time of emergence of an unretained sample on the gas chromatographic column to the time when the peak is at its maximum, but as very large peaks may be skewed, it has been suggested that the distances should be measured from the point at which the solvent first emerges to that where the tangent drawn to the leading edge of the peak intercepts the base line [7]. Also, as the retention times of esters are influenced to some extent by components eluting immediately adjacent to them, relative retention times should be measured on pure compounds or on simpler fractions isolated by silver ion chromatography (see Chapter 6) say, whenever this is possible. The retention times relative to that of 18:0 (r_l8:0) of the component esters of the external reference standard, separated on packed columns of EGSS-X™ and EGSS-Y™ (as illustrated in Figure 5.1), are listed in Table 5.1, as are similar values for a Silar 10C™ column. An interesting feature of the Silar 10C™ column is that the retention time of 22:6(n-3), the last component to emerge in most analyses of animal lipids, relative to that of 18:0 is much lower than that obtained with the other polyester columns. In practice, this can mean that shorter analysis times are possible while maintaining a good spread of peaks in the chromatogram in isothermal analyses especially. Relative retention times vary somewhat with the conditions of the column packing materials and with other operating parameters, such as temperature or flow-rate, but these variations are comparatively small and are in the same direction for all components.

Kovats' retention indices are more generally accepted as a standard means of recording GC retention data, but have been little used for the esters of fatty acids. Analogous parameters known as equivalent chain-lengths (abbreviated to ECLs) [615] or carbon numbers [1003] (the latter term is now little used because of its ambiguity) have considerable utility, however. ECL values can be calculated from an equation similar to that for Kovats' indices [436], but are usually found by reference to the straight line obtained by plotting the logarithms of the retention times of a homologous series of straight-chain saturated fatty acid methyl esters against the number of carbon atoms in the aliphatic chain of each acid (Figure 5.2). (Semilogarithm paper is particularly convenient for the purpose). The retention times of the unknown acids are measured under identical isothermal operating conditions and the ECL values are read directly from the graph. The ECL values of the esters of the component acids separated as illustrated in Figure 5.1 are listed in Table 5.1, and were obtained in this way. Such values have more obvious physical meaning than relative retention times and are more easily remembered. It should be noted, however, that outwith the normal range of chain-lengths (C₁₄ to C₂₂), a straight line relationship between log retention time and number of carbon atoms may no longer hold [436].

Figure 5.2. Plot of the logarithms of the retention times of the methyl ester derivatives of a homologous series of saturated fatty acids against the number of carbons in the aliphatic chains (Equivalent Chain-Lengths (ECL)), on a packed column of EGSS-X™ (see footnote to Table 5.1 for further chromatographic details). The elution times of some unsaturated fatty acids are indicated.

The increment in ECL value of a given ester over that of the saturated ester of the same chain-length, sometimes known as the fractional chain-length (or FCL) value, is dependent on the structure of the compound, and it is influenced by the number of double bonds in the aliphatic chain and the distance of the double bonds from the carboxyl and terminal ends of the molecule. From studies with synthetic fatty acids, such as the complete series of C₁₈ monoenoic (2-18:1 to 17-18:1) [85,323], methylene-interrupted dienoic (2,5-18:2 to 14,17-18:2) [158] isomers and many others [326,387], a picture began to emerge and was well documented by Jamieson (but with special emphasis on packed columns) [435] and more recently by Ackman [13]. A single double bond in the centre of a long aliphatic chain gives an FCL value, when compared with that of the corresponding saturated compound, of about 0.4 to 0.5 on a DEGS column, and as the double bond nears the carboxyl end of the molecule, the FCL value tends to increase slightly; as the double bond nears the terminal end of the molecule, the FCL value increases somewhat more rapidly. Proximity of a functional group of any kind to the terminal end of the molecule appears to have a greater effect on FCL values than when it is a similar distance from the carboxyl end.

The data on which this scheme was based may now be superseded by new work with fused silica capillaries, but the principle still holds true; the absolute values may change, but the relative order of elution does not. These matters are discussed in much greater detail in the following section. Although the principle introduced here is discussed in terms of double bonds, it applies equally to all substituent groups, including methyl-branches, ring systems and oxygenated moieties.

With any homologous series of fatty acids that contain a substituent in the alkyl chain, the distance of the substituent from either the proximal or the terminal end of the chain must vary. The logarithms of the retention times of the esters of such series plotted against the numbers of carbon atoms in the chain do not therefore lie in straight lines unless the series are short. Deviations are greatest for short-chain esters, but with longer-chain compounds, the FCL values obtained for esters with similar groups of double bonds and terminal structures are approximately constant. Using the data in Table 5.1 for the EGSS-X™ column, 20:4(n-6) has an ECL value of 22.43 (i.e. 20.00 + 2.43) and that of 22:4(n-6) is 24.45 (i.e. 22.00 + 2.45); 20:4(n-3) has an ECL value of 23.00 (i.e. 20.00 + 3.00) so we can predict that 22:4(n-3), which does not occur in the standard mixture, will have an ECL value of close to 25.00 (i.e. 22.00 + 3.00). With a suitable secondary reference standard and packed columns, the ECL values of the esters of most of the fatty acids likely to be encountered in animal tissues can be measured or predicted. With WCOT columns, a simple arithmetic procedure of this kind will certainly indicate the general area in which a specific component can be expected, but may not be sufficiently precise in all regions of a chromatogram for positive identification unless particular care is taken in making the measurements.

Marine oils, which may contain more families of polyunsaturated fatty acids, present more complicated identification problems and there are difficulties in applying FCL factors to esters of shorter chain-length. Ackman [7] has proposed the use of an alternative series of systematic separation factors, which can be calculated from the relative retention times of esters, to assist in overcoming these problems, and these are discussed below.

Again it must be emphasised the ECL values may vary a little with column conditions (e.g. temperature and carrier gas flow-rates), with the nature of the support, with the amount of liquid phase and with the age of columns. ECL values obtained in allegedly similar circumstances in other laboratories may be taken as a guide, but should be applied with caution. A method of overcoming some of the difficulties of inter-laboratory comparisons has been developed in Jamieson's laboratory [436,440]. Polyester liquid phases are considered as a single class of substances varying continuously in polarity, and the ECL value of methyl linolenate (18:3(n-3)) on a given column is a function of this polarity. If the ECL value of 18:3(n-3) is known, those of other fatty esters can be obtained from tabulated values, or from a simple equation and computer-derived constants. It is usually stressed that ECL values must be obtained under isothermal conditions. However, it has been shown that the relationship between ECL values and temperature is a rectilinear one, and that in linear temperature-programmed analyses with gradients of 0.5 to 3.5ºC/min, this can be expressed by -

ECL = A + BT

where A and B are constants depending on the nature of the solute and the chromatographic system [497]. This finding may be useful when isothermal analysis is impracticable.

Of the many ancillary techniques that can help with provisional identification of fatty acids, silver ion chromatography is probably the most useful or cost-effective (see Chapter 6). Mass spectrometry of course has enormous advantages, when it is available, for unequivocal identifications (Chapter 7).

D.  POSITIONAL AND CONFIGURATIONAL ISOMERS OF UNSATURATED FATTY ACIDS

1.  cis- and trans-Monoenoic fatty acids.

Most lipids of animal origin contain a wide range of isomeric fatty acids in which the positions of the double bonds differ. In addition, isomeric fatty acids are generated during commercial processing of fats and oils, especially during the partial hydrogenation step which is employed to raise the melting point of fish and vegetable oils in margarine manufacture. Many different isomeric fatty acids thus enter the food chain and appear in human tissues. The magnitude of the analytical task thus varies with the nature of the sample, and may not always be soluble by GC methods alone. Before discussing practical examples, it may be instructive to examine some data obtained with pure compounds. It should perhaps be reiterated at the outset of this discussion that ECL values from a particular laboratory under specified conditions can rarely be reproduced in another, a problem compounded by the changes in the nature of column materials in recent years. Nevertheless, while the specific numerical values have no absolute significance, the order of elution of particular components does have considerable relevance. Ackman [13] has compiled an exhaustive list of retention data for unsaturated fatty acids. A relatively few examples only are listed here.

A complete series of C₁₈ cis-monoenoic fatty acids (2-18:1 to 17-18:1) was synthesised first by Gunstone and colleagues in St Andrews, and they obtained retention (ECL) data for the methyl ester derivatives on several different columns [85,323]. To illustrate the pattern, some of this data is depicted graphically in Figure 5.3.

Figure 5.3. Variation of ECL values for the methyl ester derivatives of the isomeric octadecenoic acids (cis and trans) with double bond position on WCOT columns coated with Apiezon L™ and DEGS [85].

While the absolute values differ for each stationary phase, increasing with the polarity of the phase, the elution patterns are broadly similar. Thus the ECL values are lowest when the double bonds are approximately central, i.e. in positions 8 or 9. They increase relatively rapidly as the double bonds near the terminal (methyl) end of the molecule, reaching a maximum with 16-18:1, before falling slightly for the 17-isomer (the geometrical configuration of which cannot be designated). Similarly, the ECL values increase, although rather more slowly, as the double bonds near the carboxyl group and reach a second lower maximum at 3-18:1, before dropping once more for the 2-isomer. In addition, the ECL value of the 6-isomer is sometimes slightly out of line, being higher than those of adjacent isomers. The phenomenon has been explained theoretically in terms of the shapes of molecules and the opportunities for interaction between the double bonds and the walls of a WCOT column [92]. Analogous data for these and many other monoenoic fatty acids have been published [13,22,565,815,828,845]. The author recently had the opportunity to re-determine ECL data for the C₁₈ cis-monoenes on WCOT columns of fused silica and coated with one non-polar (a 5% phenylmethyl silicone) and three polar stationary phases (Carbowax 20M™, Silar 5CP™ and CP-Sil 84™), and this is listed in Table 5.2 [170].

As isomers differing in ECL value by about 0.04 should be separable on most WCOT column, it would be expected that those fatty acids with central double bonds (about 4-18:1 to 9-18:1) will not be easily resolved; petroselinic (6-18:1) and oleic acids occur together in some seed oils and are not readily resolved by GC (although this is possible with reversed-phase HPLC [168]). In the monoenoic fatty acids from animal tissues, there tend to be isomers in which the double bond positions are two carbon atoms apart, because they are formed biosynthetically from homologous fatty acids by chain-elongation or by beta-oxidation, in each instance the difference being two carbon atoms (see Chapter 2). Thus 16:1(n-9) and 16:1(n-7), 18:1(n-9) and 18:1(n-7), and 20:1(n-11), 20:1(n-9) and 20:1(n-7) are frequently found together and they are usually separable. A good example of this is seen in Chapter 6 (Figure 6.4(B)). As the separation improves with increasing distance of the double bonds from the carboxyl group, resolution of isomers from particular biosynthetic families tends to improve with increasing chain-length. Monoenoic acids with double bonds in even numbered positions are not usual constituents of animal tissues, and the presence of 12- and 14-16:1 (separated on a WCOT column) in rat hepatoma was seen as evidence of an error of metabolism, for example [993]. In hydrogenated fats, double bonds are found in the even positions as often as in the odd, and this is discussed further below.

ECL data for the complete series of C₁₈ monoenes with double bonds of the trans-configuration have also been published [85,323], and they are likewise depicted schematically for phases of medium and low polarity in Figure 5.3. The pattern tends to resemble that for the monoenes in that ECL values are lowest when the double bond is in the centre of the chain, and increase as it nears either end of the molecule. In this instance, the 2-isomer often has the highest ECL value. For a given double bond position, the trans-isomer nearly always elutes a little before the corresponding cis compound on polar phases, the difference tending to increase with increasing polarity [845,882]. On non-polar phases, cis-isomers elute before trans [323,870].

In practice then, cis- and trans-isomers of mono-unsaturated fatty acids are not easily separated on the more widely used polyester stationary phases, but some excellent resolutions have been recorded with the newer highly polar cyanoalkylpolysiloxane phases such as Silar 10C™, SP-2340™ or OV-275™. For example, near base-line separations of methyl oleate and methyl elaidate have been achieved on longer than usual packed columns (6 to 7 m x 0.2 mm i.d.) containing these phases [193,222,297,682,706,949]. Obviously, much better analyses of the pure compounds can now be obtained with the same phases in WCOT columns. In addition, excellent resolution of the cis- and trans-isomers of 11- and 13-22:1 fatty acids was achieved on a glass WCOT column (44 m x 0.3 mm) coated with the non-polar phase OV-73™ [870]. Acceptable separations of isomers have been achieved on a liquid-crystal phase [679].

It should be noted, however, that many of the published separations have been accomplished with simple model mixtures. With many lipids of potential commercial interest, such as partially hydrogenated vegetable or fish oils, there is a wide range of positional as well as configurational isomers. Ruminant fats also contain many different positional isomers, as by-products of biohydrogenation of dietary fatty acids by microorganisms in the rumen [162]. The problems involved in this type of analysis have been reviewed [190,813,882]. It appears that acceptable results can be attained by GC with the more polar phases, especially with hydrogenated vegetable oils, as opposed to fish oils; the former are used almost exclusively in the U.S.A., while the latter are employed extensively in much of the rest of the World for margarine manufacture. For example, in a large collaborative trial [297], a method described originally by Perkins et al. [706] gave consistently better results than the more traditional infrared spectrophotometric procedure. A glass packed column (6.1 m x 2 mm i.d.) with 15% OV-275™ stationary phase on 100-120 mesh Chromosorb P™ (acid-washed and silanized), helium or nitrogen as the carrier gas and an isothermal column temperature of about 220ºC were recommended. The nature of the separation is illustrated in Figure 5.4A. Although individual isomers of the cis- and trans-monoenes are not separated, this is probably not necessary in many circumstances.

Figure 5.4. A. Separation of cis- and trans-isomers of 18:1 on a packed column (6 m x 2 mm i.d.) containing 15% OV-275™ on Chromosorb P AW-DCMS™ (100-120 mesh). Helium was the carrier gas and the operating temperature was 220ºC [297]. (Reproduced by kind permission of the authors and of the Journal of the Association of Official Analytical Chemists, and redrawn from the original paper). B. As A, but with a WCOT column (100 m x 0.25 mm) coated with SP-2560™, with hydrogen as the carrier gas and at a temperature of 175ºC (redrawn from the original figure) [891].

Many published separations testify to the resolving power of WCOT columns in analyses of hydrogenated vegetable fats and of ruminant fats [115,535,551,683,785,851,891], and one such is illustrated in Figure 5.4B; cis- and trans-isomers elute as series of partially resolved peaks [891]. Regrettably, it is not easy to state exactly where the division between the two groups of peaks lies as trans-components with double bonds close to the terminal part of the molecule may overlap with isomers with cis-double bonds nearer to the centre. Van Vleet and Quinn [949] have argued that many workers have exaggerated the quality of the separation that can be achieved, since trans-11- and cis-9-18:1 are not readily resolved on a WCOT column coated with SP-2340™; they contend that the best results are obtained with lengthy packed columns of OV-275™. Similar conclusions were reached by Strocchi et al. [885]. Hydrogenated fish oils contain a wider spread of isomers and of chain-lengths, and determination of the cis- and trans-content of these cannot yet be achieved by GC methods alone, in spite of occasional claims to the contrary [16,680,829,845,870,894].

The author believes that even those claims for packed columns may prove to be illusory on rigorous examination, and suspects that other procedures, especially silver ion HPLC, will ultimately prove to be more reliable. On the other hand, it is probably true to say that there is no method currently available that can be recommended with complete confidence. The best GC method for the determination of the relative proportions of the total cis- and trans-isomers of monoenoic esters involves epoxidation of the double bond followed by analysis of the products, trans- eluting ahead of the corresponding cis-isomers [243,244,514]. (A suitable procedure for the reaction is described in Chapter 4). Individual positional isomers are not resolved, but near base-line resolution of the configurational isomers is possible, especially with WCOT columns.

2.  Dienoic fatty acids

By far the most abundant dienoic fatty acid in nature is linoleic acid (18:2(n-6)), but other methylene-interrupted dienes are found in tissues, and dienes with more than one methylene group between the double bonds occur in some organisms. ECL data for the complete series of synthetic methylene-interrupted C₁₈ dienes (2,5- to 14,17-18:2) have been published for packed and capillary columns [158]. In this instance too, the author has been able to re-determine the data for modern columns of fused silica, at the same time as for the C₁₈ monoenes, and this is also listed in Table 5.2 [170]. With each of the phases examined, the ECL values tended to increase with the distance of the double bonds from the carboxyl group, though there are discontinuities for the 3,6- and 13,16-isomers, where the ECL values are higher than those of adjacent compounds (c.f. the data for the 3- and 16-monoenoic isomers). If the FCL values for the complete series of monoenes [323] are used to predict the ECL values for these dienes, the calculated values are somewhat lower than those actually found. Thus in the earlier work, there was found to be a mean difference between the actual and predicted values of 0.13 on the Apiezon L™ column and 0.16 to 0.18 on more polar phases. This probably means that there is some interaction (possibly homo-conjugation) between the double bonds or with the diallyl methylene group, that increases the dipole moment of the unsaturated system. Comparable results were obtained in other studies, and the same principle held whether the double bonds were of the cis- or the trans-configuration [22,815,828]. Using the newer data of Table 5.2, the discrepancy between the actual and found values tends to vary with the position of the double bonds as well as with the stationary phase, and is highest for double bonds in positions 5 to 11 (0.10 to 0.15) and diminishes towards either end of the molecule. The FCL values for monoenes together with factors for the interaction with the appropriate methylene groups (the difference between the actual and predicted results for the dienes) have been used by Ackman and co-workers especially for the prediction of ECL values; some examples are given below.

Isomeric forms of linoleate with trans-double bonds are frequently found in hydrogenated fats. The four possible isomers, i.e. 9-trans,12-trans-, 9-trans,12-cis-, 9-cis-,12-trans- and 9-cis,12-cis-18:2, have been separated in the order stated on WCOT columns coated with polar cyano-polysiloxane phases, as illustrated in Figure 5.5 [489,535,683].

ECL data have also been obtained for a number of synthetic octadecadienoates with more than one methylene group between the double bonds [22,326,532,533,545,815,816,828], and some newly acquired information is listed in Table 5.2. Fatty acids of this type occur at trace levels in a variety of natural sources, and can be major components of the lipids of marine invertebrates. Although data for cis,cis-, trans,trans- and some cis,trans-dienes is available, only the first will concern most analysts. Once more, if FCL values from the monoene data are used to calculate ECL values for these components, the difference between the actual and predicted results was found to be small, i.e. 0.07, when there are two methylene groups between the double bonds; it becomes negligible, i.e. 0.00 to 0.02, when there are more than two methylene groups [326]. Again, this is confirmed by the newer data in Table 5.2, although there appear to be some differences according to the polarity of the stationary phase. Similar results were reported by others [21,22,828].

The presence of conjugated double bond systems in the alkyl chain increases the retention time of an ester considerably over that of a similar compound with methylene-interrupted double bonds. Methyl 9-cis,11-trans-octadecadienoate, for example, is a common minor constituent of ruminant and other tissues and has ECL values of 20.48 on EGSS-X™ and 20.24 on EGSS-Y™, i.e. appreciably greater than the corresponding values for methyl linoleate [161]. In addition, the configuration of the double bonds has a much more pronounced effect and some separation of geometrical isomers of conjugated esters may be possible on columns containing the conventional polyester phases as well as on the cyanoalkylpolysiloxanes. ECL data for further conjugated dienoic isomers have been published [19,226,326].

Again, the epoxidation procedure described for monoenoic acids in the previous section has been employed in the GC analysis of configurational isomers of dienes and trienes. However, the author has observed (unpublished) that the method does not appear to work with fatty acids having conjugated double bond systems, and such compounds are formed during commercial hydrogenation reactions.

3.  Polyenoic fatty acids

C₁₈ and C₂₀ fatty acids of the (n-9), (n-6) and (n-3) families are the most abundant trienoic fatty acids in animal tissues; they elute in this order on polar stationary phases and some ECL data are contained in Table 5.3. All of the eight possible geometrical isomers of 9,12,15-octadecatrienoic acid (α-linolenic acid) have been prepared by nitrous oxide-catalysed elaidinisation [21-23,815] and by total synthesis [747]. Ackman and Hooper [21-23] were able to predict their ECL values, from data for the appropriate monoenes obtained on a WCOT column coated with Silar 5CP™, by applying diethylenic coupling constants for each pair of double bonds. Some of these isomers were detected in deodorised vegetable oils [24]. On the other hand, any dubiety about the order of elution was removed by having compounds of defined structure, and Rakoff and Emken [747] were able to demonstrate resolution into six components, eluting in the order ttt (ECL = 19.93), ctt and tct (20.11 and 20.12), cct and ttc (20.20 and 20.23), tcc (20.38), ctc (20.39), and ccc (20.52), on a WCOT column of Silar 10C™ as shown in Figure 5.6. ECL data for some C₂₀ trienes, prepared by partial reduction of 20:5(n-3) with hydrazine, have been published [828].

Figure 5.6. Separation of the geometrical isomers of 9,12,15-octadecatrienoic acid on a glass Quadrex™ WCOT column (50 m x 0.25 mm i.d.) coated with Silar 10C™ [747]. The carrier gas was helium and the operating temperature was 170ºC. (Reproduced by kind permission of the authors and of Chemistry and Physics of Lipids, and redrawn from the original figure).

In some early work, conjugated trienes were reported to undergo cis-trans isomerization and double bond migration on packed columns containing polyester stationary phases [609,619]. On the other hand, when an "all-glass" WCOT system was used with the non-polar stationary phase, OV-1™, it proved possible to separate a number of geometrical isomers, eluting in the order ctc-9,11,13- (ECL = 18.95), ctt-9,11,13- (18.99), ttc-8,10,12- (19.09), ttc-9,11,13- (19.10), ttt-9,11,13- (19.36) and ttt-8,10,12-18:3 (19.39), without any losses [900]. Comparable results have been obtained by others [226,282,394]. Indeed, somewhat better resolution was obtained with a WCOT column coated with Carbowax 20M™ (ECLs = 21.50 to 22.19), from which components eluted without loss in the same order as on an OV-1™ column [282].

Methylene-interrupted tetra-, penta- and hexaenoic fatty acids of the (n-6) and (n-3) families, eluting on GC in this order, are ubiquitous components of animal tissues and present no problems in analysis provided that glass or fused silica WCOT columns are employed. Some ECL data for the more frequently encountered components of this type, and obtained on modern WCOT columns of fused silica, are given in Table 5.3, and additional information can be found in most of the review articles cited above (see also Table 5.1). Even with compounds of this kind, it is possible to predict ECL values from the FCL values of the appropriate monoenes by applying correction factors for interactions with the methylene groups, and 18:5(n-3) was identified as a component of a marine alga in this way, for example [25].

The data in Table 5.2 can be used to illustrate the principles involved [170]. Thus, an ECL value for an 18:4(n-3) fatty acid on Carbowax 20M™ is equal to 18 + the FCL values (0.18 + 0.16 + 0.30 + 0.56 = 1.20) + the methylene group factors (0.13 + 0.12 + 0.05 = 0.30), i.e. ECL = 19.50; the actual value found by direct measurement is 19.45 (Table 5.3). With the silicone phase for this acid, the calculated and actual values were only 0.02 units apart, while with the more polar phases, the difference was 0.07 in each case. It therefore appears that this approach to the identification of unknowns is of some value, provided that the primary data are of sufficient accuracy.

ABBREVIATIONS

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.

Go to Part 2 of this Chapter

This document is part of the book "Gas Chromatography and Lipids" by William W. Christie and published by the Oily Press, Bridgwater in 1989.

Updated: 7/8/2006 Credits/disclaimer

Scottish Crop Research Institute (and MRS Lipid Analysis Unit), Invergowrie, Dundee (DD2 5DA), Scotland