The Analysis of Lipids other than Fatty Acids
Sections A and B. Introduction and High-Temperature Gas Chromatography of Triacylglycerols
In nature, lipid classes do not exist as single pure compounds, but rather as complex mixtures of related components in which the composition of the aliphatic residues varies from molecule to molecule. In some lipids, such as cholesterol esters, only the single fatty acid moiety will vary; in others, for example triacylglycerols, each position of each molecule may be esterified by a different fatty acid. Sphingolipids contain a number of different long-chain bases which may be linked selectively via an amide bond to specific fatty acids. A complete structural analysis of a lipid therefore requires that it be separated into molecular species that have single specific alkyl moieties (fatty acids, alcohols, ether-linked aliphatic chains, long-chain bases, and so on) in all the relevant portions of the molecule. With those lipids that contain only one or two alkyl groups, this is now often technically feasible. With lipids which have more than two alkyl groups, means have yet to be developed for physically separating all the possible species that may exist. However, if stereospecific enzymatic hydrolyses are performed on fractions separated by the available methods, it may be possible to at least calculate the amounts of a very high proportion if not all of the molecular species that are present. The analyst must at the moment be content to isolate simpler molecular fractions rather than single molecular species in such instances.
Alkyl-, alkenyl- and acyl-forms of a given lipid are strictly speaking not molecular species of it and can themselves be fractionated into molecular species, and they should therefore be isolated separately before an analysis of this kind is begun.
Ideally, it would be preferable if lipids could be separated into individual molecular species without being modified in any way so that, for example, the biosynthesis or metabolism of each part of the molecule could be studied with isotopically labelled components. The technical problems of the analysis can often be greatly reduced, however, if the polar parts of complex phospho- and glycolipids are rendered nonpolar by the formation of suitable derivatives, or if they are removed entirely by chemical or enzymatic means. Whatever approach is adopted, it is frequently necessary to apply combinations of different chromatographic procedures to achieve effective separations.
The chromatographic methods used for the analysis of molecular species of lipids differ little in principle from those used for simpler aliphatic molecules such as the fatty acids. When they are applied to the isolation of molecular species of more complicated lipids, the separations achieved depend on the combined physical properties of all the aliphatic residues. If triacylglycerols are considered to illustrate the magnitude of the analytical problem, a triacylglycerol with only five different fatty acid constituents may consist of 75 different molecular species. High-temperature GC has until recently been used largely to separate molecular species simply according to the combined chain lengths of the fatty acid moieties. However, improvements in technology have led to separations according to degree of unsaturation also. In essence, high-temperature GC is simply an analytical technique, but one which is capable of a high degree of precision. It can be married well with mass spectrometry.
Of the alternatives to GC, adsorption chromatography will permit the separation of molecules containing three normal fatty acids from those containing two normal fatty acids and one fatty acid with a polar functional group in the chain, from those containing one normal fatty acid and two polar fatty acids, and so forth. Silver ion chromatography will separate those molecules containing three saturated fatty acids from those with two saturated fatty acids and one monoenoic acid, and these are in turn separable from other distinct fractions containing molecules with fatty acids of a progressively higher degree of unsaturation, thus complementing separations by high-temperature GC particularly well. HPLC in the reversed-phase mode is used to separate triacylglycerols by their partition number, a double bond reducing the effective chain length of a fatty acid by the equivalent of about two carbon atoms. The principles of these alternative methods as they apply to molecular species separations are described further in Chapter 9. Such procedures can be applied on either an analytical or a semi-micro preparative scale, but quantification is not always easy or convenient.
It should be remembered that it is almost always advisable to calculate molar rather than weight proportions (or percentages) of any molecular species isolated. Precautions should be taken to minimise the effects of autoxidation (see Chapter 2). Procedures for the analysis of molecular species of lipids have been dealt with in several monographs [163,168,506,510] in addition to many shorter reviews on specific topics (see below).
In the discussion that follows, ether analogues of specific lipids are discussed in the same sections as the acyl forms. Separations of individual pure lipid classes are treated first as distinct topics, before analyses of complex natural mixtures are described in the final section. Methods for the preliminary isolation of lipid classes are described in Chapter 2 and elsewhere [163,168].
B. High-Temperature Gas Chromatography of Triacylglyerols
1. Separations on packed columns
As triacylglycerols are the main component of virtually all the fats and oils of commercial importance, a great deal of effort has been applied to their analysis. Because of the low volatility of intact lipids, GC is beset with a number of difficulties and, indeed for some time, it was thought that molecules such as triacylglycerols, with molecular weights up to 900, would pyrolyse at the temperatures required to elute them from GC columns. Work from the laboratories of Kuksis and Litchfield principally showed that this need not occur. Useful separations of such compounds are now achieved routinely, although the conditions necessary to elute them from the columns approach the limits of thermal stability both of the stationary phases and of the compounds themselves. The technique has been the subject of several review articles [342,500,503,504,507,509,553,590,642]. Here the separation of the pure lipid class is described; separations of total lipid extracts containing triacylglycerols are discussed in Section I.
For packed-column use, any modern gas chromatograph should be suitable, but it is essential that it have a flame ionisation detector for maximum sensitivity and facilities for accurate temperature-programming up to at least 350°C. In addition, it should be of a construction such that on-column injection is possible, although a pre-heater is necessary to warm up the carrier gas before it reaches the column packing. The dead volume between the end of the packing and the detector flame should be as small as possible, and the flame jet should preferably be wider than normal so that comparatively high flow-rates of the carrier gas can be used. Automatic flow controllers for the carrier gas are a useful accessory as the flow-rate in short columns can change markedly during temperature-programming. As bleeding of even those phases which are most thermally stable occurs at high temperatures, better results are obtained with dual-compensating columns than with single-column instruments, unless an integrator with comprehensive base-line correction facilities is available. The analyst should not be discouraged from using equipment that does not meet all these criteria, especially for separating compounds of intermediate molecular weight such as diacylglycerols or their derivatives, as patience and skill can compensate for many instrumental deficiencies.
Helium or nitrogen may be used as the carrier gas with packed columns and excellent recoveries of triacylglycerols have been obtained with both, although the former is to be preferred when feasible, as better resolutions are attainable at high flow-rates. Nitrogen is used most often for reasons of cost, but it is essential that it be of very high purity as traces of oxygen or water will destroy liquid phases at elevated temperatures.
Better results are obtained with glass columns than with those of other materials, but there can be technical difficulties in obtaining effective seals at each end of the column; O-rings of Viton™ or other materials generally become brittle and crack if used at temperatures above 250°C for any length of time, but graphite seals will stand up to such conditions for long periods. It is also possible to avoid this difficulty by using columns with direct glass-to-metal seals, and these are available for specific instruments. All-metal columns are certainly not recommended. Narrow-bore columns give the best resolutions, and those of 2 mm (i.d.) diameter are preferable to those of 4 mm; the length of the column selected will vary with the nature of the sample, but it is usually necessary to compromise resolution by using short columns (50-100 cm) in order that compounds are eluted in a reasonable time.
The most useful liquid phases are silicone elastomers of high thermal stability, such as SE-30™, JXR™, OV-l™ and Dexsil 3000™, with which separations are achieved solely on the basis of molecular weight. Silanized solid supports (80-100 or 100-120 mesh) are essential and they are generally coated with low levels (1-3%) of stationary phase. Care is necessary in preparing the columns, which must be packed firmly to ensure adequate resolution, but not too tightly, otherwise higher temperatures than are advisable may be necessary to elute samples in an acceptable time, and losses of components of higher molecular weight may occur. Finally, the column is sealed with a plug of silanized glass wool and is conditioned for four hours at a temperature at least 25°C higher than that at which it is to be used. It may be necessary to attempt the preparation of suitable columns several times before success is achieved, and the analyst should not be discouraged by initial failures. A low bleed-rate is especially important if the technique is to be used in conjunction with mass spectrometry.
The precise operating conditions and the resolutions attainable will vary with the nature of the samples to be analysed. As cautioned above, short columns are essential and the column temperature must be programmed in the range 180 to 350°C at 2 to 5°C/min, depending on the nature of the sample. Slow temperature-programming rates give improved resolution generally. The optimum flow-rate of the carrier gas will vary with the dimensions of the column and the amount of stationary phase on the packing material, but will normally be of the order of 100 mL/min. By means of a syringe, a solution of the sample should be injected directly on to the column packing at a point within the oven (rather than in the flash heater), and at a temperature about 40°C below that at which the first component emerges from the column. In this way, all the sample is vaporised, but it remains as a narrow band at the top of the column until temperature-programming is under way. Samples containing about 20 μg of the most abundant component provide the optimum load. The entire analysis should be completed in 25 to 45 min. With samples of intermediate molecular weight such as diacylglycerol acetates, wax esters or cholesterol esters, longer columns can be used to improve the resolutions attainable and the upper temperature limit in the analysis will be lower than with triacylglycerols.
The silicone stationary phases, which must be employed for GC analysis of high molecular weight compounds, do not in general permit the separation of saturated from unsaturated components of the same chain length. Separations are then based solely on the approximate molecular weights of the compounds and for example, tripalmitin and myristopalmitoolein elute together. Components that differ by two carbon atoms in the combined chain lengths of the alkyl moieties must be separable before the columns are considered satisfactory. This can usually be achieved with column efficiencies of 500 to 1000 theoretical plates per foot of packing material, and indeed, with well-packed columns, components differing in molecular weight by one carbon atom can often be resolved completely. Two examples of analyses of natural triacylglycerol samples from the author's laboratory are shown in Figure 8.1.
Figure 8.1. The separation of intact triacylglycerols of (A) coconut oil and (B) pig adipose tissue on a glass column (50 cm × 4 mm i.d) packed with 1% SE-30™ on Chromosorb W™ (acid-washed and silanized; 100-120 mesh). Nitrogen at 50 mL/min was the carrier gas and for separation (A), the oven was temperature-programmed from 230 to 330°C at 2°C/min, while for separation (B) it was programmed from 280 to 330°C at 2°C/min.
A shorthand nomenclature is in common use to designate simple glycerides separated in this way; the total number of carbon atoms in the aliphatic chains of the compounds (but not in the glycerol moiety) are calculated and this figure is used to denote the compound. As an example, tristearin, triolein and trilinolein are referred to as C54 triacylglycerols or as having a carbon number of 54.
It is perhaps invidious to select a single example from the wide literature on the subject, but in one of the better published separations a glass column only 30 cm in length by 2 mm i.d. was packed with 3% SE-30™ on Chromosorb G™ (100-120 mesh; acid-washed and silanized); the temperature was programmed from 275 to 350°C at 4°/min, while nitrogen at a flow-rate of 25 mL/min was the carrier gas .
During isothermal operation, as discussed in Chapter 3, there is a logarithmic relationship between the retention times and carbon numbers of components of a homologous series. When linear temperature-programming is used, there is a rectilinear correlation between the logarithm of the retention time and the reciprocal of the absolute temperature for short series of homologues . As a result, the elution temperature rather than the elution time is sometimes quoted to described the retention characteristics of a given compound. With longer homologous series, the relationships begins to break down, and improved resolutions are obtained with nonlinear (concave) temperature-programming profiles . As few commercial gas chromatographs are equipped with this facility, the refinement has not been widely adopted.
When flame ionisation detectors are used, the detector response is, within limits, proportional to the weight of material eluting from the columns (see Chapter 3 and Chapter 5), and the amount of each component can be calculated from the areas of the peaks on the GC recorder trace. There is no simple relationship between area, retention time and peak height for temperature-programmed analyses, so it is necessary to measure the area of each peak by means of an electronic digital integrator. Because of the high temperatures necessary for GC of intact lipids, there is always a danger that losses will occur on the columns as a result of pyrolysis, of reaction with the column materials and of condensation. It is, therefore, essential to check that acceptable reproducible recoveries are obtained and to calibrate the columns to compensate for any losses. The absolute recoveries from columns are not easily checked, as this requires a preparative collection facility or means of counting radioactive samples as they elute; it is, however, possible to check that recoveries are linearly related to the amount of material injected by inserting known quantities of standards (say 1-20 μg) into the columns and measuring the detector response. Alternatively, it can be assumed that in all but the very worst of columns, recoveries of tricaprin (C30) or trilaurin (C36) will be essentially complete, so that standard mixtures of these triacylglycerols and the higher molecular weight compounds can be analysed and the losses relative to the standard determined.
The losses that can be accepted will vary with the degree of difficulty of the analysis but, as a rough guide, recoveries of the highest molecular weight component of a mixture of triacylglycerols (usually C54), for example, should be at least 90% relative to tricaprin. In practice, the most efficient way of optimising the chromatographic conditions is to vary each of them in turn, especially the flow-rate and the rate of temperature-programming, selecting those which give the maximum responses . When pyrolysis occurs on the column, peaks for pure standards are often preceded by broad humps of decomposed material. Such effects can be minimised by adding to the sample a triacylglycerol of higher molecular weight than is normally present and this presumably decomposes preferentially; triarachidin (C60), for example, can be used in many circumstances . By this means, recoveries and quantification of minor components are greatly improved.
As a wide range of unsaturated compounds is unlikely to be available for standardisation purposes, there are a number of advantages to be gained by hydrogenating all samples prior to analysis by high-temperature GC in packed columns. If this is done, there are no selective losses of unsaturated relative to saturated components by degradation on the column, peaks on the recorder trace are sharper, resolutions are improved and quantification of components is made easier. In the analysis of molecular species of lipids, it is necessary to know the molar proportions of all components separated, and the weight responses of the detector must be corrected by multiplying by appropriate arithmetic factors obtained from the molecular weights of the compounds, as described for methyl esters of fatty acids (Chapter 5). A further advantage of hydrogenation prior to analysis then is that it greatly simplifies the range of factors required and removes any dubiety about their numerical values. This is of course particularly important with samples containing polyunsaturated fatty acids. A suitable hydrogenation procedure is described in Chapter 4.
Samples may be injected on to columns in carbon disulfide, diethyl ether or hexane solution; chloroform has also been used on occasion but tends to strip the stationary phases from the packing and damages the flame ionisation detector. While hydrogenated lipids tend to be less soluble than the unsaturated compounds in most solvents, they will usually dissolve on warming.
High-temperature GC with packed columns is then used to obtain separations of triacylglycerols simply on a molecular weight basis. Although species with carbon numbers up to 68 have been successfully resolved , there is little margin for error in the preparation of the column and, in most laboratories, it is considered a sufficient achievement to separate components with carbon numbers up to 56 or 58. Individual peaks are recognised by their carbon numbers relative to those of authentic standards. As single acid triglycerides (C42, C48 and C54, for example) only are available commercially, intermediate points are found by mathematical interpolation. Separation into homologues differing in carbon number by two units is achieved with relative ease, but more care is necessary to resolve components differing by one unit, as when odd-chain fatty acids are present in the sample .
Returning to Figure 8.1, the kind of separation obtainable with two natural triacylglycerol samples is illustrated, i.e. A - coconut oil, which contains a high proportion of fatty acids of medium chain length, and B - pig adipose tissue (lard), which contains mainly C16 and C18 fatty acids. For these analyses, a glass column (50 cm × 4 mm) was packed with 1% SE-30™ on Chromosorb W™ (acid-washed and silanized), and nitrogen was the carrier gas at a flow-rate of 50 mL/min. Efficient separations of the lower molecular weight components of the former sample (carbon numbers 28 to 48) are readily attained, but it is less easy to separate effectively the higher molecular weight components in the latter (C48 to C54). The author was not always able to reproduce the separation illustrated here with new columns. It is usual to see some convergence of peaks as the temperature is increased. Again, the compounds were hydrogenated prior to the analysis, to improve the resolution and recoveries. Poor resolutions are inevitable when packed columns are used with natural triacylglycerols that contain odd- and branched-chain fatty acids as well as those with wide ranges of chain lengths and numbers of double bonds, as in ruminant milk fats or fish oils, for example.
As cautioned earlier, quantification must be checked carefully with standard mixtures to ensure that losses of high molecular weight components relative to those of lower molecular weight are as low as possible. Losses of higher molecular weight triacylglycerols such as trierucin (C66) are inevitable, however, and must at present be accepted, but all such losses can be compensated for by determining calibration factors with standard mixtures, provided that the factors are checked regularly. Data should again be converted to molar proportions. The results of an international collaborative study to establish a standard method for triacylglycerol analysis have recently been published . High temperature GC has been employed for the analysis of triacylglycerols from a wide variety of natural sources, e.g. marine oils [17,18,555,557], seed oils [97,98,409,558] and animal lipids [408,511,634,849,989], and those cited are selected as representatives only of innumerable papers. In addition, the technique has been recommended for the determination of cocoa butter equivalents in the confectionery industry [687,1010]. Note that silver ion chromatography complements separations of this kind particularly well, since it provides the resolution by degree of unsaturation not achievable otherwise (see Chapter 9). Used sequentially, the two techniques permit separation of many more fractions than would be possible by either on its own, and this approach was favoured in many of the papers cited above.
Some separation of triacylglycerols according to degree of unsaturation in addition to chain length has been achieved on packed GC columns containing polar stationary phases [54,503,898]. Whether the limited resolution obtained is sufficient to be of value to lipid analysts is doubtful, however, especially as such columns will probably have a short lifetime. With phases of higher polarity, WCOT columns are greatly to be preferred (see below).
Although some preparative applications have been described, GC of triacylglycerols is now used exclusively as an analytical technique. HPLC procedures are certainly much better for small-scale preparative purposes (see Chapter 9).
High-temperature GC with packed columns has also been employed to separate ether analogues of triacylglycerols, i.e. the alkyldiacylglycerols, into molecular species. This was first accomplished for such compounds isolated from tumour tissue on a short column packed with a 3% JXR™ stationary phase [1000,1002]. Alkyldiacylglycerols tend to be minor compounds in tissues of terrestrial animals, but they can be major components of the harderian gland  and of some marine species [18,556], and GC proved to be of value in the studies cited here. Because they have one fewer oxygen atom than the corresponding triacylglycerols, alkyldiacylglycerols tend to elute the equivalent of one methylene unit earlier.
2. Separations on WCOT columns
The potential benefits of glass WCOT columns for the analysis of intact lipids must have been obvious from an early stage, and indeed the first such separations with Dexsil 300™ as the stationary phase were published in 1972 . Most of the more common stationary phases tended to bleed rather easily from WCOT columns at elevated temperatures, however, and further applications did not appear until chemically bonded and cross-linked phases became available together with fused-silica capillaries from about 1979 onwards. The use of WCOT columns for triglyceride separations has also been reviewed elsewhere [284,591].
Many different gas chromatographs have been used for the purpose, and most modern instruments appear suitable, but it is evident that the nature of the injection system can be of crucial importance. The minimum requirement is for some form of on-column injection. Grob , for example, demonstrated that techniques based on sample vaporisation in the injector are not suitable for intact lipids as discrimination in favour of the less volatile constituents occurs. With splitless injection, most losses were found to be a consequence of insufficient elution from the syringe needle; split injection gave even worse results, although the reasons for this were not clear, and only cold on-column injection gave acceptable recoveries. Any involatile material ("dirt") on the column from previous analyses affected the injection because of adsorption effects [309,310]. In addition, the flow-rate of the carrier gas can have a marked effect on sample loss and discrimination, and thermal decomposition takes place to change the sample composition [306,592]. Cold on-column injection eliminates many of these problems, although other factors then come into play [307,308,312].
The principal disadvantage of on-column injection is the contamination of the stationary phase that inevitably occurs, leading to peak broadening . This effect can be minimised by using effective clean-up procedures for the triacylglycerols during sample preparation, such as by preparative TLC or HPLC , or more conveniently by using a short column of Florisil™ or silica gel (0.3 to 0.5 g), from which the lipid is eluted with hexane-diethyl ether (4:1, v/v; 10 mL). In addition, it is possible to insert a length of deactivated fused-silica tubing (1 to 3 m), sometimes termed a "retention gap", in front of the column to collect any impurities . When this pre-column is beginning to show signs of contamination, it is replaced. Unfortunately, the low starting temperature required with an injection technique of this kind means that the analysis time is lengthened and there is additional opportunity for thermal decomposition to happen.
Various solutions have been suggested for the problem, including an independently thermostatted inlet section of the column  and a moveable on-column injector [283,286]. With the latter, the injector and inlet part of the column are moved up out of the oven, where they cool to room temperature. The sample is injected, most of the solvent is allowed to evaporate in the stream of carrier gas, then the injector is moved down into the oven so that the column inlet region heats up very rapidly to the initial oven temperature and the sample vaporises. As an alternative, a cold on-column injection system equipped with secondary cooling has been used; the solvent evaporates in a portion of the column cooled to 70°C, then the cooler is switched off so that the sample is heated to the oven temperature in about 30 seconds . This method makes use of commercially available equipment and can be adapted for automatic injection.
Another approach has been to use a programmed-temperature vaporiser, which in essence is a split/splitless injector that is maintained at a temperature close to the boiling point of the injection solvent, so that the sample is transferred to the column in a liquid; when the solvent has evaporated, the injector is heated at a rate of 14°C/sec to the minimum column temperature required [383,384]. The manufacturers (Perkin Elmer) claim that this injector is the closest to a universal system yet to be developed.
While WCOT columns constructed from both glass and fused silica have been used for the separation of intact triacylglycerols, there is now no doubt that the latter are to be preferred. The length of column used will be a compromise between the optimum in terms of resolution with a need to limit the exposure time of the solute to high temperatures to the minimum; commonly the length is 5 to 25 m with an internal diameter of 0.2 to 0.32 mm. Columns with a strengthened outer coating are now manufactured to better withstand temperatures above 300°C.
Initially, nonpolar stationary phases only (of the methyl silicone type) were used in high-temperature GC, and cross-linking and chemical bonding improved the properties of the columns appreciably. More polar bonded phases, consisting of phenylmethyl silicones, later came into use and are available commercially. At present, these have a temperature limit of about 360°C, and while this will no doubt be improved, the ultimate limit may depend on the pyrolysis temperature of triacylglycerols. The optimum thickness of the liquid film for high-temperature GC is about 0.1 to 0.12 μm.
As discussed earlier for packed columns, the rate and shape of the temperature-programming profile can have a marked effect on column efficiency. A nonlinear (concave) rate of temperature-programming is preferable whenever this is feasible. The lower the rate of temperature-programming, the lower is the elution temperature of a given compound, but the longer is its elution time. In practice, the optimum temperature limits and the rate of programming must be determined empirically for a given sample and column.
Hydrogen has important advantages as a carrier gas with WCOT columns, in that efficiency is less dependent on linear gas velocity, as discussed in Chapter 3. It permits the elution of components at lower temperatures or elution times than with other gases, so that there is less opportunity for thermal degradation to occur, especially of more sensitive components containing polyunsaturated fatty acids. It also promotes a longer working life for the column. In addition to the possible danger of explosion with hydrogen as a carrier gas (as discussed earlier), there is one report of hydrogenation of unsaturated lipids at elevated temperatures on a polar stationary phase . However, this effect has not been found by others apparently, and it has been suggested that it may have been a consequence of some impurity in the sample, the carrier gas or the stationary phase . As with packed columns, all air and moisture must be rigidly excluded from the carrier gas to extend column life and efficiency.
High-temperature GC of triacylglycerols with modern WCOT columns was accomplished by several research groups in different parts of the world virtually simultaneously. It is therefore not always possible to treat the subject from a historical standpoint or to establish scientific priority. When nonpolar phases are used in WCOT (as with packed) columns, triacylglycerols are separated according to molecular weight essentially and there is ordinarily no useful resolution by degree of unsaturation, although some partial separations may be seen. As an example, a separation of an interesterified palm oil on a 6 m glass WCOT column coated with a methylsilicone phase is illustrated in Figure 8.2; temperature-programming was from 250 to 350°C . Components varying in carbon number from 44 to 56 are clearly resolved, and there is some evidence for the presence of intermediate species containing odd-chain fatty acids. An improvement in resolution over Figure 8.1 is obvious. Similar results were obtained with other vegetable oils. In addition, comparable analyses have been reported by others with seed oils [204,305,313,951], algal triacylglycerols , beeswax , and butter [42,313,455,641,806,961], plasma [542,592,647] and other animal lipids [313,654]. The technique has been used in comparisons of triacylglycerols and chloropropanediol diesters in milk fat from goats [514,652], and of triacylglycerols and alkyldiacylglycerols in human milk .
|Figure 8.2. Separation of interesterified palm oil on a glass WCOT column (6 m x 0.4 mm) coated with a methylsilicone phase . Helium at 6 mL/min was the carrier gas, and the oven was temperature-programmed from 250 to 350°C at 4°C/min. The numbers above each peak refer to the carbon number of the component. (Reproduced by kind permission of the authors and of Revue Française des Corps Gras, and redrawn from the original paper).|
In the hands of a skilled analyst and with a good column, some partial resolution is possible according to degree of unsaturation or because of variation of the chain lengths of fatty acids within a molecular species of a given carbon number [291,768,923,924]. A separation of coffee oil on a 15 m glass column coated with OV-101™ is illustrated in Figure 8.3; temperature-programming was from 310 to 330°C . The component of carbon number 54 here, for example, is separated into three fractions according to the number of unsaturated fatty acids in the molecule, but not by the number of double bonds within each acid. With care, four fractions can sometimes be seen, eluting in the order UUU, UUS, USS and SSS, where S is a saturated and U an unsaturated C18 fatty acyl residue, i.e. unsaturated species elute before saturated. Unfortunately, it is doubtful whether the resolution is quite good enough to be of real analytical value. In an attempt to obtain more meaningful data on the relative proportions of saturated and unsaturated molecular species, triacylglycerols were subjected to high-temperature GC after ozonolysis of double bonds followed by reductive cleavage .
Figure 8.3. Separation of coffee oil on a glass WCOT column (15 m × 0.3 mm) coated with OV-1™ . Hydrogen was the carrier gas, and the oven was temperature-programmed from 200 to 310?C at 4°C/min. Abbreviations: P, 16:0; S, 18:0; U, a C18 unsaturated fatty acid. (Reproduced by kind permission of the authors and of the Journal of Chromatography, and redrawn from the original paper).
More recently, some remarkably effective separations of triacylglycerols have been achieved on WCOT columns coated with more polar (or "polarizable") silicone phases containing a high proportion of phenyl groups, mainly in the laboratories of Geeraert and Sandra, who developed the moveable on-column injection system discussed above. It is then possible to separate triacylglycerol species according to the number of double bonds in each fatty acyl residue within a given carbon number.
Excellent resolutions of seed oil triacylglycerols especially have been obtained on a WCOT column (25 m × 0.25 mm i.d) of fused silica coated with a methylphenyl silicone polymer (RSL-300™), containing 50% phenyl groups [287-290]. As an example, a separation of palm oil is illustrated in Figure 8.4; temperature-programming was from 340 to 355°C over only 16 minutes . It can be seen that the C52 species is separable into seven fractions, while the C54 species splits into six. In a more unsaturated seed oil, fractions emerge in the order - SSS, SSO, SSL, OOO, SLO, OOL, SLL, OLL, LLL and LLLn, where S = 18:0, O = 18:1, L = 18:2 and Ln = 18:3. In this instance, unsaturated fractions elute after saturated ones. The quality of the separation and the speed of the analysis will ensure that this technique is widely used, especially in the oils and fats industry, assuming that the precision of quantification is comparable to that with apolar phases.
Figure 8.4. Separation of palm oil on a WCOT column (25 m × 0.25 mm) of fused silica coated with a 50% phenylmethylsilicone phase . Hydrogen was the carrier gas, and the oven was temperature-programmed from 340 to 355°C at 1°C/min. Abbreviations: M, 14:0; P, 16:0; S, 18:0; O, 18:1; L, 18:2. (Reproduced by kind permission of the authors and of the Journal of High Resolution Chromatography and Chromatography Communications, and redrawn from the original paper).
Within a given carbon number group, some resolution is achieved for combinations of fatty acids of different chain lengths. In Figure 8.5, a separation of a hydrogenated butter fat is illustrated . The C46 fraction, for example, may contain MPP, MMS, LaPS, CSS and many more species, where M = 14:0, P = 16:0, La = 12:0 and C = 10:0. Intermediate fractions containing odd-chain and branched-chain fatty acids are also well resolved. It is not easy to identify the components within particular peaks without access to mass spectrometry (see below). Similar separations of butter fat and vegetable oils [384,912] have been reported with stationary phases containing up to 65% phenyl moieties in the polymer, and analyses of other fats and oils are described in a review article . In addition, triacylglycerols have been hydrolysed to diacylglycerol derivatives for analysis by high-temperature GC on polar phases in order to obtain more information on molecular structure (see Section C.2).
Figure 8.5. Separation of hydrogenated butter fat on a WCOT column (25 m × 0.25 mm) of fused silica coated with a 50% phenylmethylsilicone phase . Hydrogen was the carrier gas, and the oven was temperature-programmed from 280 to 355°C at 3°C/min. The numbers above each peak refer to the carbon number of the component. (Reproduced by kind permission of the authors and of the Journal of High Resolution Chromatography and Chromatography Communications, and redrawn from the original paper).
The key to a wider acceptance of high-temperature GC of triacylglycerols on WCOT columns is the precision that can be attained in quantification. It is virtually essential that electronic integration be applied for peak area measurements, ideally with some form of automatic base-line correction. The response of the detector should in theory be quantitative, in that it is linearly related to the amount of material eluting from the end of the column. However, if some of the sample is selectively lost during injection (discussed above) or if losses occur through degradation on the column, the overall efficiency of the process can fall off. Good injection technique and clean samples can eliminate some of the losses. There is little that can be done to prevent thermal degradation entirely, but it can be minimised by careful optimisation of the operating conditions, and reproducible if not quantitative recoveries can be attained.
In a detailed study of the factors affecting the quantification of intact triacylglycerols with WCOT columns coated with a nonpolar phase, Mares and Husek  demonstrated that the recovery of the higher saturated homologues was dependent on such factors as the injection technique, column quality, the flow-rate of the carrier gas, the weight of the solute and its molecular weight. Column quality is not easy to define, and the analyst is to a considerable extent in the hands of the suppliers. During use, the stationary phase begins to thin out and bare patches can appear, and there can be contamination by residues of previous samples. Such factors will inevitably lead to a loss of resolution and worsening recoveries. It is well documented that the flow-rate of the carrier gas changes during temperature-programming (see Chapter 3), so the detector response must change also. Optimisation to minimise the weight correction factor that is required for higher molecular weight species, such as triarachidin, must be carried out empirically for each new column. As long as good peak shape is maintained (no overloading), the larger the sample the better the response tends to be for triarachidin, but deleterious effects are much less apparent with C54 triacylglycerols. If sufficient care is taken in optimising the system and in the measurement of weight correction factors, excellent reproducibility can be achieved with nonpolar phases . Others appear to have a less sanguine view, while still recommending the technique .
Less information is available on quantification with WCOT columns coated with the more polar phases. In addition to the factors mentioned above, Mares  found that the response diminished with the length of the column and perhaps more importantly with the degree of unsaturation of the solute. However, the losses were reproducible so that good quantification was reportedly possible, except for highly unsaturated seed oils, with careful calibration. Others consider that the relatively greater bleed from the polarizable phases causes some quenching of the response to triacylglycerols of higher molecular weight, although reproducible results are again obtained after calibration . In contrast, Geeraert  reported that recoveries were complete and that the detector response was directly proportional to the carbon content of each molecule, except for the most highly unsaturated species found in such samples as fish oils. With the common range of vegetable oils, such as soybean or palm oil, and for confectionery fats, such as butter, cocoa butter (and substitutes) and coconut oil, there was a uniform response factor of unity. It is possible that the quality of the last results are due in some measure to the special injection system and columns used by Geeraert, but further objective studies are obviously required.
3. Gas chromatography-mass spectrometry
Such is the complexity of the GC traces of intact triacylglycerols that some additional means of identifying components is necessary, and foremost among these is mass spectrometry (MS). The topic has been briefly reviewed [386,483,642]. In most of the early work, samples were introduced into the mass spectrometer with direct probe insertion, but improvements in technology have made it possible to introduce triacylglycerols via a GC column. On the other hand, it is probably true to say that more work is now being done with HPLC interfaced to MS (reviewed elsewhere [168,515]).
Barber et al.  were the first to describe the EI mass spectrum of a triacylglycerol, and this was soon followed by other systematic studies [2,537]. The mass spectrum of 1,2-dipalmitoylolein is illustrated in Figure 8.6 . Usually, there is a very small molecular ion only, in this instance at m/z = 832, followed by a unique peak for an ester at m/z = 814 (M−18 or loss of water). There are, however, intense ions that are characteristic of the various fatty acyl residues and these fall into two classes, i.e. those containing two acyl residues and those with only one. In the first class, there is an ion equivalent to the loss of an acyloxy group, i.e. [M−RCOO]+, together with a related ion but minus a further hydrogen atom. In the example shown here, the loss of a palmitoyl acyloxy group (equivalent to 255 amu) gives ions at m/z = 577 and 576 (having one palmitoyl and one oleoyl residue), while the loss of the oleoyl moiety (equivalent to 281 amu) gives ions at 551 and 550 (having two palmitoyl residues). The relative intensities within each pair are dependent on whether the ion fragment contains an unsaturated residue (when the smaller ion is more intense). The other important class of diagnostic ions containing the individual fatty acid moieties are of the form RCO+, though if the fatty acid group is unsaturated an additional hydrogen atom is lost. Thus in Figure 8.6, the oleoyl moiety produces an ion at m/z = 264, while that from palmitate is at m/z = 239. Related ions with an additional 74 amu corresponding to the glycerol backbone, are found at m/z = 339 and 313, respectively; in this instance, the presence of a double bond has no effect. When the triacylglycerol contains three different fatty acids, there are three ions in each class. Metastable-ion mass spectrometry can be of great value for the recognition of specific ion fragmentations .
Figure 8.6. The EI mass spectrum of 1,2-dipalmitoylolein.
EI-MS with direct probe insertion was used, for example, to determine the structure of unusual tetraacylglycerols containing an allenic acid  and triacylglycerols containing sorbic acid . Molecular species of mixed triacylglycerols have also been analysed in this way ; the data were fed into a computer programmed to recognise key fragments, apply response factors and calculate the relative proportions of each fraction. One advantage of such a procedure is that large numbers of samples can be handled routinely, provided that suitable equipment is available.
When GC-MS was attempted with packed columns interfaced to mass spectrometers via molecular separators, difficulties were obtained with the recovery of triacylglycerols, but useful data were obtained from several natural samples in this way . The task of analysing triacylglycerols of marine origin containing isovaleric acid was perhaps a relatively easy one . Capillary columns of fused silica can now be introduced directly into the ion source of some instruments and this eliminates many of the technical problems (c.f. [196,806]). In addition, it is recognised that better spectra with substantial molecular (or quasi-molecular) ions are obtainable if soft ionisation methods, such as chemical ionisation [274,630,635] or field desorption [247,274,539,607,818] MS, are used. Only the former of these can be used in conjunction with gas chromatography, and the technique has been employed successfully with packed  and with WCOT columns; with the latter, triacylglycerols from algae  and butter fat  were analysed. Ammonia was employed as the reagent gas initially [630,635], but better results appear to be attainable with methane [768,961]. By this means, it is possible to recognise and quantify individual species in a single chromatographic peak and not resolved by GC alone.
1-O-Alkyl-2,3-diacylglycerols have similar mass spectra to triacylglycerols . The mass peak is small, and that for [M−18]+ is smaller than in triacylglycerols. In the high mass range, a peak corresponding to [M−(O−alkyl)]+ is always prominent, and there are also ions corresponding to the loss of the acyloxy groups. Further fragmentations lead to ions equivalent to [M−(HO−acyl+acyl)]+.
4. Supercritical fluid chromatography
Supercritical fluid chromatography is a rapidly developing technique, which is in effect a hybrid between GC and HPLC, using much of the instrumentation of the former while the mobile phase is a liquefied gas, commonly carbon dioxide. Much effort is being expended in developing the instrumentation and applications, and some interesting separations of molecular species of triacylglycerols have been described [740,969]. While there appear to be serious doubts about reproducibility at the moment, it seems probable that improvements will be made. The general technique is the subject of a recent monograph .
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.
Go on to Part 2 of this Chapter
Go on to Part 3 of this Chapter
Updated July 8, 2011