The Lipid Library

GAS CHROMATOGRAPHIC ANALYSIS OF FATTY ACID METHYL ESTERS WITH HIGH PRECISION

Summary: High precision in the gas chromatographic analysis of fatty acid methyl esters is possible with careful attention to detail during sample preparation, injection, chromatography and data collection. Small theoretical correction factors only should be necessary in most circumstances.

One of the commonest tasks undertaken by lipid analysts is the determination of the fatty acid compositions of samples by means of gas chromatography (GC). Obviously, samples with a high proportion of polyunsaturated fatty acids, such as fish oils, or those containing short-chain components, such as milk fat, are not as easy to analyse as are the common range of fats and oils of commerce. On the other hand, what standard of accuracy should we be aiming for? An industrial analyst with kilograms of material available to him has an easier time of it than someone in a clinical laboratory with micrograms of sample, yet there is no reason why the latter cannot routinely get excellent results.

Sample Preparation

There are two strands to the problem - sample preparation and gas chromatography per se. Both aspects were investigated at length in a series of important papers from John Craske's laboratory in Australia [c.f.1-3]. I cannot go into great detail on extraction procedures and storage of samples here, but it should be apparent that the methods chosen should not lead to any degradation of the lipids by autoxidation or other means. Lipids should therefore be stored in a non-polar solvent such as hexane with a small amount of an antioxidant in an atmosphere of nitrogen at −20ºC. Next they should be transesterified by some method that ensures quantitative conversion to methyl esters and recovery of these derivatives from the reaction medium. Appropriate methods were described in earlier articles in this series and in a major review article [4], also available here. Please consult these if necessary. I would, however, like to add one practical point. If a stream of nitrogen is to be used to evaporate solvents containing methyl esters, it is very easy to lose selectively some of the components of lower molecular weight - especially myristate (14:0) but even palmitate (16:0). With small samples, it is advisable to use a rotary evaporator to remove excess solvents in order to optimize results.

Gas Chromatography

In the early days of GC, losses of esters of higher molecular weight were obtained because of chemical interactions with polyester liquid phases in packed columns. These problems were eliminated as catalyst-free polyesters became available together with improved solid support materials. Perhaps the best objective assessment of the standard of fatty acid analysis possible with this older technology came in a collaborative study of IUPAC methodology in 1979 [5]. For a common range of fats and oils of commercial interest, typical coefficients of variation (%) were quoted as 15 (2% level), 8.5 (5% level), 7 (10% level) and 3 (50% level). Poorer data were obtained with difficult samples like milk fat. In a comparable study with modern equipment, the results were only a little better [6]. It should be possible to obtain much better data, but why is this not being achieved?

Most analysts will now be working with capillary columns of fused silica and instruments designed to take these. It is not so long since this type of column was seen by instrument manufacturers as an optional extra and they did not produce equipment that was entirely satisfactory. Those days should now be behind us and most modern gas chromatographs should give excellent results, provided that the columns are installed correctly. While this proviso may seem self-evident, poor positioning of columns in the injector and bad seals are often major sources of error.

Sample Injection

It is sometimes argued that the results may depend of the type of injection system. While it is often said that fewer problems are encountered with "on-column" injection, my experience suggests that this is not true. Excellent results are obtainable with injectors of the "split/splitless" type, and this is also more suited to automatic injection. There is little doubt that this is the best method of injecting samples when instruments are equipped with this facility.

With injectors of the latter type, Craske and co-workers recommended a much higher injector temperature than is normally considered, i.e. 375ºC, for optimum results [3], but I would be concerned with its application to fish oils. In addition, a high speed of injection was recommended to eliminate discrimination in the injector, and they facilitated rapid evaporation by employing as dilute a solution of the sample as was feasible.

Good manual injection technique may seem to consist of obvious or trivial points, but it can help greatly. Ackman, for example, recommends drawing a small plug of fresh solvent into the needle ahead of the sample to ensure that the latter is flushed completely out of the needle. I prefer a "hot needle" technique, i.e. the sample is drawn completely into the syringe barrel leaving the needle empty; the needle is then inserted firmly into the injection port and left for five seconds to heat up before the plunger is pressed. Also, I like a "cold trapping" method in which the sample is injected with the column (not the injector) at the boiling point of the solvent containing the sample (70ºC for hexane, for example) so that the latter is concentrated in a small volume at the top of the column; only when the solvent peak emerges is the oven temperature raised rapidly to the starting point for the analysis. It is just as easy to train staff to use these simple techniques routinely as any other.

Data Collection

The data collection system is obviously crucial. There is no doubt that some form of electronic integration is essential if optimum results are to be obtained. However, there is room for error if this is not set up correctly, for example with inappropriate settings for peak widths, threshold values and so forth. I have noticed that novice analysts can assume that data on a computer print out carries a cachet of infallibility.

A further contentious point in fatty acid analysis is the potential requirement for correction factors. With flame ionization, it is necessary to recognize that the carboxyl carbon in each ester is not ionized appreciably during combustion [7,8]. For maximum precision, small correction factors must be applied to compensate for this, and these vary from 1.02 for methyl palmitate (16:0) to 0.97 for methyl docosanoate (22:0). At least for fatty acids with the common range of chain lengths (say C₁₄ to C₂₂) and up to three double bonds,  Craske and colleagues [1-3] affirm  that NO  other correction factor should be necessary! I agree, at least for fatty acids with the common range of chain lengths (say C₁₄ to C₂₂) and up to three double bonds, but I believe that we have to allow for some instrumental or human frailty.

If in practise other correction factors are found to be required to obtain results for standard mixtures that agree with those specified, it means that there is something wrong either in the preparation of the sample, the setting up of the instrument, the injection technique or the integrator. The analyst should respond by correcting these problems rather than by introducing "fiddle factors" into calculations.

Milk Fat and Fish Oils

Milk fat is an important exception to the suggestion that only theoretical response factors for the flame ionization detector should be necessary as this does not appear to true for short-chain fatty acids, such as butyrate or hexanoate. In this instance, empirical response factors, which are determined experimentally, may have to be used [8].

The other important exception is for lipids rich in the highly unsaturated eicosapentaenoic (20:5) and docosahexaenoic (22:6) acids, such as fish oils [9]. There is ample evidence that the cis-double bonds in these fatty acids tend to isomerize to trans at temperatures above 180ºC, as often occurs in the injector or column during chromatography. The result is that a number of small minor peaks separate from the main one and may be ignored during quantification. Again, it is advisable to determine experimental response factors. If in doubt, I suggest hydrogenating a sample, and summing the C₂₀ and C₂₂ components before and after this treatment as a check.

References

1. Bannon, C.D., Craske, J.D. and Hilliker, A.E. Analysis of fatty acid methyl esters with high accuracy and reliability V. Validation of theoretical response factors of unsaturated esters in the flame ionization detector. J. Am. Oil Chem. Soc., 63, 105-110 (1986).
2. Craske, J.D. and Bannon, C.D. Gas-liquid chromatography analysis of the fatty acid composition of fats and oils: a total system for high accuracy. J. Am. Oil Chem. Soc., 64, 1413-1417 (1987).
3. Bannon, C.D., Craske, J.D., Felder, D.L., Garland, I.J. and Norman, L.M., Analysis of fatty acids of methyl esters with high accuracy and reliability. VI. Rapid analysis by split injection capillary gas chromatography. J. Chromatogr. A, 407, 231-241 (1987).
4. Christie, W.W. Preparation of ester derivatives of fatty acids for chromatographic analysis. In: Advances in Lipid Methodology – Two. pp. 69-111 (Ed. W.W. Christie, Oily Press, Dundee) (1993).
5. Firestone, D. and Horowitz, W. IUPAC gas chromatography method for determination of fatty acid composition: collaborative study. J. Assoc. Off. Anal. Chem., 62 709-721 (1979).
6. Beare-Rogers, J.L. and Dieffenbacher, A. Determination of n-3 and n-6 unsaturated fatty acids in vegetable oils and fats by capillary GLC. Results of a collaborative study and the standardized method. Pure Appl. Chem., 62 795-802 (1990).
7. Ackman, R.G. and Sipos, J.C. Application of specific response factors in the gas chromatography analysis of methyl esters of fatty acids with flame ionisation detection. J. Am. Oil Chem. Soc., 41, 377-378 (1964).
8. Ackman, R.G. and Sipos, J.C. Flame ionisation detector response for the carbonyl carbon atom in the carboxyl group of fatty acids and esters. J. Chromatogr., 16, 298-305 (1964).
9. Badings, H.T. and De Jong, C. Analysis of fatty acid methyl esters (FAME) with high accuracy and reliability. J. Am. Oil Chem. Soc., 65, 659 (1988).
10. Ackman, R.G. Losses of DHA from high temperatures of columns during GLC of methyl esters of long-chain omega-3 fatty acids. J. Am. Oil Chem. Soc., 83, 1069-1070 (2006).

This article has been updated appreciably from one by the author that first appeared in Lipid Technology, 3, 97-98 (1991).

Updated: 27/3/2007 Credits/disclaimer	Scottish Crop Research Institute (and MRS Lipid Analysis Unit), Invergowrie, Dundee (DD2 5DA), Scotland	Right click on the icon to open in a new window