An Introduction to Lipids and Gas Chromatography

Chapter 3. Theoretical Aspects

 

A.  Introduction

The technique of gas chromatography (GC) or gas-liquid chromatography (GLC) is a form of partition chromatography in which the mobile phase is a gas and the stationary phase is a liquid. A sample is injected into the gas phase where it is volatilised and passed onto the liquid phase, which is held in some form in a column; components spend different times in the mobile phase and the stationary phase, depending on their relative affinities for the latter, and emerge from the end of the column exhibiting peaks of concentration, ideally with a Gaussian distribution. These peaks are detected by some means which converts the concentration of the component in the gas phase into an electrical signal, which is amplified and passed to a continuous recorder, and perhaps to an integrator, so that the progress of the separation can be monitored and quantified.

Since the technique was first described in the 1950s (see Chapter 1), it has been subjected to continued development in every single aspect and indeed progress is still being made. In the second edition of the author's book Lipid Analysis [163], which was published as recently as 1982, only two pages were devoted to GC with capillary or open-tubular columns. This was not because the potential of these was not readily apparent or that little work had been done with them, but because they had disadvantages of high cost coupled to a limited working life; in addition, there tended to be substantial losses of polyunsaturated components on the stainless steel walls of the only columns available commercially at that time. The development of first glass and then fused-silica columns has entirely removed all these disadvantages, and there would now be few lipid analysts who would consider purchasing a new gas chromatograph without facilities for capillary columns. As this new methodology has been introduced so rapidly, there must be many lipid analysts who are not fully conversant with the pleasures (and also the pitfalls) of the technique. The major emphasis in this book must therefore be on GC with capillary columns, although this does not mean that GC with packed columns can now be neglected. Rather, it is hoped that this book will stimulate interest in the newer advances and lead to a further increase in our knowledge of lipids.

A large number of books have been published on the theory and practice of chromatography, and the author has found those cited to be of particular value [267,314,427,444,728,791]. In addition, there have been a number of review articles or books on the application of this theory to specific lipid problems, and these are listed in the appropriate Chapters below.

 

B.  Theory

1.  Some basic considerations

Most analysts appear to manage perfectly well with only a smattering of knowledge of the theory of gas chromatography. On the other hand, some understanding of the concepts involved can be beneficial in many circumstances, if only in allowing him/her to cast a more sceptical eye over the claims of competing column manufacturers.

A basic gas chromatograph has three essential components, i.e. some form of inlet through which the sample is introduced onto the column, the column itself which contains the stationary (liquid) phase and through which the mobile (gas) phase is passed continuously, and a detector. In this Section in which the theory of the separation process is described, the column is the main component of interest, and there are three main types. The workhorse of gas chromatography until relatively recent times was a packed column, consisting of a glass or metal tube, generally 2 or 4 mm in internal diameter and 1.5 to 2.5 m in length, coiled to fit the oven unit and filled with an inert solid support coated with the liquid phase. Support-coated open-tubular (SCOT) columns enjoyed a brief vogue, and are constructed of narrow bore tubing (0.5 to 1.25 mm i.d. in lengths of 10 to 15 m) and contain a finely powdered solid support coated with a liquid phase. Finally, wall-coated open-tubular (WCOT) columns consist of narrow bore tubing (0.1 to 0.3 mm i.d. and 25 or 50 m in length commonly), of glass or increasingly of fused silica, the inner wall of which is coated with the liquid phase. A conventional packed column is simple to construct and pack with the liquid phase on its particulate support, it is rugged in use and can take large sample sizes; however, resolution is limited. WCOT columns afford superb resolution, but the size of the sample and the method of its introduction are critical.

The function of the column is to allow partitioning of the constituents of the sample between the stationary and mobile phases, and this is aided by having the liquid phase as a thin film with a large surface area accessible to the flow of the gas phase. As the sample (solute) passes down the column, the molecules of each component partition between the liquid and gas phases according to a distribution coefficient or constant, KD, i.e.

Formula for distribution coefficient

This is a true equilibrium constant, which is specific for a given solute and liquid phase at the temperature selected. As the gas phase is moving continuously, solute molecules continue to dissolve in fresh liquid phase in relation to the equilibrium constant, while those molecules which have already dissolved overcome the various forces involved, re-emerge into the gas phase and pass further down the column.

As long as a molecule is in the gas phase, it travels down the column at the same speed as the carrier gas. When a mixture of components is present in the solute, these diffuse into the liquid phase to varying degrees according to their individual equilibrium constants, and so travel down the column at different rates. In other words, the retention times on the column are different and the components tend to separate.

Of course, this is a rather simplified account, and the resolution can be marred by various factors, which cause band broadening. In packed columns, for example, there is a multiplicity of different flow paths around the particulate packing material, causing changes in the rate of flow of the gas in discrete areas. The particles are uneven in shape and size and the liquid phase will also consist of regions of different depth; molecules diffusing farther down into the liquid phase will emerge again much later than those which diffuse into a thin area. In a WCOT column, there is only one flow path in essence and the liquid phase is usually more uniform in thickness, so that the contribution from such factors to band broadening is minimal. Components emerge from packed columns in wide bands relative to the time spent on the column, while those emerging from a WCOT column are relatively narrow. Therefore, the efficiency of a column is dependent on the degree of band broadening which it imposes upon a solute as it passes along, in a given time. When a component emerges from the column, the shape of the concentration peak is controlled by band broadening (width) and the absolute concentration in the gas phase (height), i.e. by the equilibrium constant. Under conditions of constant temperature, the KD does not change and the efficiency of a given column can be related to peak widths and retention times. With many practical problems, however, it is necessary to increase the column temperature during an analysis in order to bring off less volatile solutes in a reasonable time, and then the calculation of column efficiency is much more complex.

Under isothermal conditions therefore, and assuming that the sample is introduced to the column in an appropriate manner and at a suitable concentration (i.e. not overloaded) so that solutes separate in accordance with their equilibrium constants, peaks should emerge with a Gaussian distribution. Column efficiencies are then defined in terms of numbers of theoretical plates (n), the retention time (tr) measured from the point of injection until the peak reaches its maximum, and the width of the peak measured at half its maximum height (wh) by using the equation:

n = 5.54 x (tr/wh)2

The same units must be used to measure tr and wh. These parameters are illustrated in Figure 3.1. Column efficiencies may be described in terms of theoretical plates per metre, or as the column length (plate height) equivalent to a single theoretical plate. Ideally, retention volumes rather than times should be used, but the two are obviously related in a given column.

Calculation of column efficiency

Figure 3.1. A chromatographic peak with an ideal Gaussian shape. Calculation of column efficiency.

 

A certain volume of gas is required to carry a component that does not dissolve in the liquid phase through the column, and this is known as the hold-up volume, although in practice, it is more convenient to measure hold-up time (tm). This is usually measured by injecting methane onto the column and determining the time for the leading edge of the methane peak to emerge. Every solute molecule must spend the same time (tm) in the gas phase, so the true time spent in the liquid phase is therefore an adjusted retention time (t1):

t1 = tr - tm

The time spent in the gas phase should not affect the quality of the separation. In WCOT columns, tm is relatively great compared to that in packed columns, and it is necessary to take this into account in defining the true efficiency in terms of effective theoretical plates (N) [217,741], which are calculated as -

N = 5.54 x (t1/wh)2

This figure is still mutable to some extent because it is related to the nature of the solute used as a test compound for the measurement. The factor for the solute is termed the partition or capacity ratio (k), and is the ratio of the weight (not concentration) of the solute in the liquid phase to the weight in the gas phase, and is proportional to the time spent by the solute in the liquid phase and that spent in the gas phase, i.e.

k = t1/tm

One of the principal differences between packed and WCOT columns lies in the relative availability of the liquid and gas phases to a solute molecule, and this is defined as the phase ratio, β , i.e.

Formula for phase ratio

and in essence in a WCOT column -

β = ro/2df

where ro is the radius of the capillary and df is the mean depth of the liquid film. In WCOT columns, the values range from 100 to 300 typically (and from 5 to 20 in packed columns). As described above, the distribution constant, KD, is based on solute concentration, but this can also be stated in terms of the weight of solute and the volume of each phase:

Formula for partion constant

- where k is the partition or capacity ratio defined above.

This basic equation is of great importance in optimising all the parameters that can affect resolution in capillary GC.

 

2.  Practical implications

The efficiency of a given column is dependent on a number of factors, including the nature and flow-rate of the carrier gas, column dimensions, liquid-phase thickness and column temperature. By optimising these, it may be possible to increase the resolution attainable quite considerably. On the other hand, this improved resolution may be bought at the expense of increased analysis time. In practice, it may be desirable to compromise and select conditions for an analysis that give adequate resolution in a reasonable time.

The nature and velocity of the carrier gas are primary considerations. At high velocities, the opportunity for band broadening through longitudinal diffusion of solute molecules (along the length of the column) is diminished, but there may then be insufficient time for them to pass into the liquid phase. When the flow-rate of the mobile phase is low, there is an increased opportunity for band broadening through longitudinal diffusion. Efficiency therefore tends to fall off both when the flow rate is too high and when it is too low. The nature of the carrier gas is also important, and hydrogen and helium but especially the former, because of their high diffusivities or low resistance to mass transfer, are greatly to be preferred to nitrogen, say. It is also noteworthy that column efficiency varies much less with gas velocity over the useful working range when hydrogen is utilised, so that precise flow calibration is less critical in practice. This can be illustrated by a so-called Van Deemter plot of the variation in the height of an effective theoretical plate with carrier gas velocity for hydrogen, helium and nitrogen (Fig. 3.2).

Van Deempter curve for carrier gases

Figure 3.2. Plot of height of an effective theoretical plate (HETP) against carrier gas velocity (Van Deemter curve) for hydrogen, nitrogen and helium, obtained on a 25 m WCOT column (0.25 mm i.d.).

 

Of course if hydrogen is used as the carrier gas, it is necessary to guard against the risk of explosions, for example, through an electrical spark igniting a buildup of gas caused by a leak within the instrument.

As the temperature is increased during temperature-programming, the efficiency of the column decreases, although components emerge as sharper peaks, because the vapour pressure of the solute increases as does the ratio of the concentration in the gas phase to that in the liquid phase, i.e. the capacity ratio is inversely proportional to column temperature. However, the effect can be small in relation to the advantages in terms of the sharpness of peaks and the speed of analysis, provided that sensible temperature-programming rates are employed. As the flow of gas through capillary columns is usually controlled by pressure regulation rather than by flow per se, the flow-rate through the column will actually decrease with increased temperature because of an increase in the viscosity of the gas. Therefore, a temperature-programmed analysis should ideally be started at a higher gas velocity than might be used for an isothermal run. Again, this effect tends to be less with hydrogen than with other carrier gases.

There are three factors which are dependent on the physical characteristics of the column, i.e. column length, column internal diameter and film thickness. Of these, column length is least important as it can be established that resolution is proportional only to the square root of column length. Thus to improve by a factor of two on the resolution attainable with a 25 m column, it would be necessary to move to one 100 m in length, and this would inevitably mean that the analysis time would be increased by a factor of four.

In contrast by reference to the last equation above, it is apparent that changes in the column internal diameter or in the film thickness have a marked effect on the partition ratio. If the column diameter is reduced, this brings about a decrease in the phase ratio and a proportional increase in the capacity ratio. Retention times of solutes and the overall analysis time are increased, however. A decrease in film thickness corresponds to an increase in the phase ratio and results in a decrease in the partition ratio. In this instance, the retention time of the solute is decreased. In practice, the nature of the solute must also be considered, thick films being preferred for solutes of high volatility.

 

C.  Columns

1.  Packed columns

Packed columns have been in use since the technique of gas chromatography was first developed. For many years, some analysts continued to use columns made of stainless steel, although there was some danger of rearrangement of fatty acids on the metal wall. Most prefer columns made of glass, because these are almost inert, it is easy to see how well the column has been packed and whether any gaps in the packing appear during use; deterioration of the top of the column packing at the inlet end is immediately seen. Also, glass columns are more easily emptied for reuse. It is certainly true that they are relatively fragile, but the main opportunity for breakage tends to be when columns are either mounted into or removed from the oven of the gas chromatograph; if particular care is taken when these tasks are performed, columns should last for many years.

The solid supporting materials for the liquid phases are generally diatomaceous earths, graded so that the particles are of uniform size (usually 80-100 or 100-120 mesh), and deactivated by washing with acid and by silylation in order to prolong the life of the liquid phase and to minimise any adsorptive effects on the solutes. Supports pre-coated with liquid phases are available commercially at prices close to those of the starting materials. If the analyst wishes to make his own, it is claimed [403] that the most uniform coating of the liquid phase on the support is obtained if a solution is filtered through a bed of the support and the whole spread out as a layer to dry; the amount of liquid phase that remains on the support must be determined by experiment and depends somewhat on the nature of the support. It is also possible to achieve satisfactory coatings by evaporating a solution of the liquid phase in the presence of the support in an indented flask (to stir the solid material) on a rotary evaporator, but great care is necessary to ensure that no damage occurs to the surface of the support.

A column is packed by adding the coated support in small amounts via a funnel, while tapping gently and applying a vacuum to the exit end. When it is filled, a glass wool plug (acid-washed and silanized) is placed on top of the packing at the inlet end to consolidate it. If columns are too tightly packed, they may block or the injection syringes may plug easily; if they are too loosely packed, the separations are poor. The column must be conditioned for up to two days at a temperature just above the maximum at which it is to be operated during routine analyses, before being used. A well-packed column (1.5 to 2 m) containing a support with 10-15% (w/w) liquid phase should have an efficiency of 3000 to 5000 theoretical plates. Such columns can have an operating life of well over a year, and this can be prolonged by periodically repacking the top 3 to 5 cm with fresh packing material.

 

2.  Glass WCOT columns

It took some time for the pioneers in the use of glass WCOT columns to convince others that these were sufficiently robust for routine employment in the laboratory. There was never any doubt as to the resolution attainable or about the inertness of the surface, the latter being of particular importance to lipid analysts concerned with any potential isomerization of polyunsaturated fatty acids. Interest was heightened when glass drawing machines for capillary production became available commercially at reasonable cost.

Before glass capillary columns can be satisfactorily coated with a thin uniform layer of a liquid phase, the surface must first be treated chemically to ensure adhesion. Otherwise, surface tension in the liquid will cause it to contract into droplets and leave the surface of the glass bare. Initially a variety of different methods of treating the surface were tried including the use of wetting agents, carbonisation and silylation, but it appears to be established that gaseous etching or aqueous leaching treatments are to be preferred. All of these methods work in essence by turning the smooth surface of the glass, at the microscopic level, to a much rougher one thereby increasing its wettability. This simply increases the contact angle between the glass and liquid phase, as pits and crevices are filled and the spreading of the film is assisted. The etching process also thoroughly cleans the glass, while chemical treatment can change the surface so that the liquid film is not deposited on silica but on salt crystals.

The mildest gaseous etching process consists in treatment with HCl or HF gas, the former only with soda glass and the latter with borosilicate glasses such as Pyrex™. With hydrogen fluoride, the silicate structure is attacked to form silicon tetrafluoride, which forms crystals covering the surface. With hydrogen chloride, the reaction is extremely complex and involves a preliminary leaching of alkali metals from the surface. The surface generated in this way is then deactivated by one of several procedures before the liquid coating is applied by either dynamic or static techniques [267]. While this is indeed done by many analysts, it appears that there are some "tricks of the trade", and patented procedures are applied by commercial companies to ensure uniform reproducible products, especially with liquid phases of high polarity. As with packed columns, a new WCOT column must be conditioned for a time before use.

Glass capillary columns are much less fragile in use than might be anticipated. Again the greatest risk of breakage pertains when columns are mounted in or are removed from the oven of the gas chromatograph. One major drawback is that the column ends must be straightened before the column can be fixed in place, an operation which requires some practice and not a little manual dexterity.

 

3.  Fused silica WCOT columns

Fused silica has proved to be an excellent medium for the manufacture of WCOT columns. They consist of an amorphous silicate material, which is free of metal oxides and is therefore very inert. Such columns are very flexible and there is no need to straighten the ends. As supplied, they are covered externally with a polymeric material which prevents brittle fractures. Coating of these columns with liquid phase does, however, appear to be a specialised task and most analysts are content to leave it to commercial manufacturers. Nowadays, the liquid phase does not merely coat the surface and adhere by physical forces, but rather it is bonded chemically to the surface by various means. In addition, the individual molecules of polymeric liquid phases are cross-linked by chemical methods, when the film is in place, to improve their stability at high temperatures. It is then possible even to remove organic impurities from the stationary phase by passing a little solvent through it. Such are the advantages of these columns indeed, that they have now completely supplanted those of glass in commerce.

 

D.  The Liquid Phase

1.  Selectivity

The principal requirement of a stationary phase is that it provide the correct degree of selectivity for the separation required. At the same time, it should have reasonable chemical and thermal stability in order to prolong the working life of a column. The selectivity of a liquid phase is a product of several factors, and as it is not easy to define from first principles, it must be determined experimentally. In practice, this is accomplished by comparing the retention times of a series of standard test substances on a column containing the phase of interest against those on a nonpolar phase. The Rohrschneider/McReynolds indices (named after the originators of the system) assist in identifying groups of phases with similar properties, and the choice may then depend on a factor such as temperature stability. To lipid analysts, the system has its limitations since it aids in selecting phases that separate different groups of compounds, rather than of homologous series or closely related isomers, for example.

A major factor influencing separation according to degree of unsaturation is the polarity of the liquid phase. When packed columns only were available, this appeared to be a rather variable property, because of differences in the loading of the liquid phase, on the nature of the solid support, and on operating factors such as temperature and column age. Improvements in the manufacture of liquid phases helped a little but did not eliminate the problems. On the other hand, the thickness of the liquid film has very little effect on polarity in WCOT columns, so that the selectivity of phases can be defined with greater confidence.

In practice, lipid analysts have investigated the properties of particular stationary phases in some detail and certain have emerged as favourites for specific purposes. These are discussed in the chapters that follow in relation to each lipid class. No doubt new phases will be developed to challenge those in regular use, but changes are more likely to be made by direct comparison than by applying scientific concepts of selectivity.

 

2.  Prolonging column life

In regular use, the stationary phase on a column can deteriorate for a number of reasons, but usually as a result of chemical attack. WCOT columns are more susceptible than packed columns, if only because they contain less stationary phase. Most polar liquid phases are very sensitive to oxygen and water, and it is strongly advised that all traces of these be removed from the carrier gas by positioning traps containing suitable molecular sieves and oxygen scrubbers (available commercially) between the gas cylinder and the column. The author once saw several columns being destroyed by a day or so of accidental exposure to an inferior grade of helium, which contained a relatively high proportion of oxygen; the traps in place could not cope. Polar solvents (e.g. chloroform, alcohols, carbon disulfide, etc.) and traces of polar impurities introduced to the columns with samples may slowly react with the liquid phase or perhaps displace it from the column, and any adverse reaction will of course be exacerbated by exposure of the column to excessive temperatures. The flow of carrier gas should never be allowed to stop as long as the column is being heated. In addition, involatile materials injected onto a column along with the components of interest may gradually build up and alter the characteristics of the liquid phase.

Most damage tends to occur in the first few coils of the column, where the liquid phase can be displaced entirely or at least suffer appreciable degeneration. When the damage is minimal, reversing the column may be all that is required. Otherwise, depending on how extensively the deterioration has progressed, anything from a few centimetres to one or two coils of the column can be broken off. Such column shortening has very little effect on resolution.

As mentioned above, packed columns can be rejuvenated by replenishing the top few centimetres of the packing material.

 

E.  Detectors

1.  Flame ionisation detectors

A large number of detectors operating on different principles have been developed for use in gas chromatography, but only a few of these continue to be used to a significant extent. The flame ionisation detector is now almost universally adopted as it can be used with virtually all organic compounds, and has high sensitivity and stability, a low dead volume, a fast response time and the response is linear over an extremely wide range. This detector is simple to construct and operate, and it is highly reliable in prolonged use. Only inert gases, compounds with a single carbon bound to oxygen or sulfur (e.g. carbon dioxide, carbon disulfide, etc.) and a few other volatile substances do not give a substantial signal.

The principle of the detector is that ions are generated by combustion of the organic compounds as they emerge from the column in a diffusion flame of hydrogen and air. The carrier gas from the column may be premixed with hydrogen, although with WCOT columns, it is usual to position the outlet of the column at the orifice of the combustion jet in a chamber through which an excess of air is passed. The collector electrode is cylindrical and is placed just above the flame, and the ion current is measured by establishing a potential between the collector and the jet tip. In order to prevent the ions from recombining, a potential is selected in the saturation region, where increasing the potential has relatively little effect on the ion current. The signal current is passed to an amplifier, which must have a linear range to match that of the detector itself, and thence is transmitted to a recorder.

The response is subject to some experimental variables, chief among which are the flow-rates of hydrogen, air and the carrier gas. The instrument manufacturer's instructions should be followed closely to optimise these, although there is usually some leeway, as it is not easy to give general guidelines for instruments of different makes.

The nature of the response in relation to specific lipids is discussed in subsequent chapters.

 

2.  Electron-capture detectors

In the electron-capture detector, a radioactive source is used to bombard the carrier gas with β particles as it passes through an ionisation chamber. Each β particle can generate up to a thousand thermal electrons, which are collected by applying a voltage potential. When solutes containing electron-capturing moieties enter the cell, they interact with thermal electrons and a diminution in the background current is seen and can be measured with high sensitivity. In addition to this sensitivity, the chief virtue of this detector is its specificity, as molecules containing halogen atoms, for example, give a very marked response. Although few natural lipid molecules contain halogens, it is possible to convert lipids to halogen-containing derivatives to make use of the high sensitivity and specificity of the detector.

 

3.  Mass spectrometry

While mass spectrometry (MS) is rarely used for detection per se, it has become an invaluable tool for the identification (and adventitiously for the detection) of lipids separated by gas chromatography. The principle of the technique in its simplest form is that organic molecules in the vapour phase are bombarded with electrons and form positively charged ions, which can fragment in a number of different ways to give smaller ionised entities. These ions are propelled through a magnetic or electrostatic field and are separated according to their mass to charge (m/z) ratio; they are collected in sequence as the ratio increases, the ion current is amplified and it is then displayed by some means. The largest (or base) peak is given an arbitrary intensity value of 100, and the intensities of all the other ions are normalised to this. The ion from the parent molecule is termed the molecular ion (M+). With instruments of low resolution, peaks appear at unit mass numbers, but at high resolution the masses of individual ions can be measured with sufficient accuracy for the molecular formula of each to be determined.

Molecules do not fragment in an arbitrary manner but tend to split at weaker bonds, such as those adjacent to specific functional groups, or according to certain complex rules, which have been formulated empirically from studies with model compounds. Frequently, it is possible to deduce the structure of the original compound from first principles from the nature of the fragments produced. With other compounds when the results are equivocal, the spectrum can be compared with those of compounds with similar properties (nowadays with the aid of computer search facilities) until a good fit is obtained. The combination of mass spectral and GC retention data may also serve to eliminate alternative structures. In GC-MS applications, the total ion current produced from the column effluent is recorded continuously, and a trace is obtained resembling that from other detectors; spectra are also recorded continuously and can be related to specific peaks.

Of course, this description greatly oversimplifies the technique, especially as many different ionisation systems, in addition to electron-impact, are available which alter the extent and nature of the ionisation and fragmentation processes. The reader should consult one of the many authoritative texts on the subject for a more comprehensive account. There is further discussion in Chapter 7.

In interfacing a gas chromatograph with a mass spectrometer, a primary requirement is that the pressure must be reduced from appreciably above atmospheric in the column to about 10-4 torr in the ion source. The interface must be as inert as possible, there should be no cold spots where sample condensation can occur, and the extra-column volume should be as small as possible to minimise band broadening. The development of fused silica capillaries simplified the problem considerably, because a new generation of diffusion pumps had the capacity to extract helium at the same rate as it emerged from a column. It was thus possible to design the ion source so that the outlet of the capillary column passed directly in to it. The alternative is to use devices such as Teflon membrane or molecular jet separators to concentrate the solute relative to the carrier gas. With this and other systems, there is inevitably some loss of the resolution achieved on the WCOT column.

 

F.  Injection Systems

1.  Injection technique

The choice of an injection system for packed columns is entirely straightforward. Most analysts prefer simply to inject the sample in a small volume of solvent (1 to 5 μL), via a syringe and through a septum, just into the top of the column packing. The sample is thus volatilised in the presence of the liquid phase so that the risk of isomerization of double bonds or of other unwanted side reactions is minimal.

With WCOT columns, the choice of one of the many injection systems available from commercial sources may confuse the newcomer to the subject. The properties of the major types are discussed below, although it should be noted that models from different manufacturers that operate on similar principles may differ in some details. Of course, the analyst may have little choice in the matter, having to use whatever equipment is provided. It is then incumbent upon him to be aware of the potential pitfalls so that the results can be optimised. The nature of the sample is also relevant, and the topic is raised again in later chapters in relation to specific analytical problems.

During the injection process, it is extremely important that the sample should not change in composition nor should there be discrimination for or against any particular component. Ideally, thermal degradation or rearrangement should be negligible, no loss of column efficiency should be introduced, the solvent peak should not interfere with the detection of the solutes, and retention times and relative peak areas should be highly reproducible.

In addition to instrumentation, it is necessary to consider all aspects of injection technique. Problems can arise from volatilisation when the sample is in contact with the metal surface of the needle or from premature evaporation of the sample before the needle is fully inserted into the injector. In order to minimise these effects, the following method (the "hot needle" technique) is recommended:

"The sample (0.1 to 1 μL) is drawn up completely into the syringe barrel, leaving the needle empty, the needle is then inserted firmly and smoothly into the injection port, and it is allowed to remain in place for about 5 seconds to allow it to warm up, before the plunger is pressed rapidly".

Some analysts prefer to use a "solvent flush" method, which is similar to the above except that a small plug of fresh solvent is drawn up ahead of the sample, and is used to push the latter into the evaporation chamber. Superficially, attention to such detail may appear trivial, but the benefits can be substantial.

 

2.  Split injection

WCOT columns have a limited sample capacity and it is relatively easy to overload them by introducing a sample in too large a volume of solvent or at too high a concentration. Many injection systems have been developed to circumvent the problem, and split injectors are simple to use and can give excellent results in many types of analysis. A schematic diagram of an injector of this kind is shown in Figure 3.3. In this, the sample is vaporised in the carrier gas, which is divided into two streams, one of which is directed onto the column and the second of which is vented to the atmosphere. The flow through the latter is regulated prior to injection by a control valve to give the desired split ratio, usually from about 1:20 to 1:200. As the flow-rate through the column is commonly about 1 to 2 mL/min, the gas flow through the injector is very high (100 to 200 mL/min) and the vaporised sample is present in the injector only momentarily. This short time means that the sample is introduced onto the column as a narrow plug and is followed by the pure carrier gas.

Schematic diagram of a typical split injection system Figure 3.3. Schematic diagram of a typical split injection system for WCOT columns.

Unfortunately, the preset split ratio differs from the true sample ratio in a complex manner, which is dependent on parameters such as sample volume, solvent, and volatility, syringe handling, and injector and column temperatures. When the sample is volatilised, it produces a pressure wave, which introduces some sample onto the column before the pressure falls back. During the second phase, little sample enters the column and most is vented. The splitter ratio in fact controls the magnitude of the pressure wave, and through this the actual sample size. Thus if the same amount of sample is injected in different volumes of solvent, sample peak areas may not be the same.

The vaporisation chamber contains a glass or quartz liner, which can easily be removed for cleaning, and which provides a relatively inert surface for vaporisation. Some workers attempted to pack these with inert materials to promote mixing, but the increased surface area tended to cause some degradation.

The principal drawback of this injector is that it may discriminate against the higher-boiling components of a sample, so that quantification can be problematical. It is the author's experience that this type of injector can give excellent results in analyses of the normal range of fatty acids encountered in animal and plant tissues, for example, although it is necessary to check this with suitable calibration mixtures at regular intervals. Less satisfactory quantification is obtained with samples with a wider spread of volatilities, such as in the analysis of the fatty acids of milk fats or in the chromatography of intact triacylglycerols. As a high proportion of the sample is wasted, this injection mode is not suitable when sample size is a limitation.

For optimum results, the nature of the solvent and its volume should be kept constant, the injection technique described above should be used, the syringe needle should always penetrate to the same spot just above the column inlet and the initial column temperature should be reproduced accurately. It also helps if a "cold trapping" technique is employed during injection, i.e. the column temperature is maintained at about the boiling point of the solvent; the sample re-condenses as a narrow band, and when the solvent peak is seen to emerge, the oven is heated up rapidly to the normal analysis temperature. Some commercial gas chromatographs are now designed so that cold trapping can be carried out automatically.

 

3.  Splitless injection

The splitless sample injection method is designed to make use of the "solvent effect". In this mode, the sample is injected in a solvent with a high boiling point relative to the column temperature (but lower than the sample, of course) into a chamber through which only the carrier gas for the column is flowing. If the conditions are correct, i.e. such that the column temperature is at least 30°C below the boiling point of the solvent, the latter will re-condense in the column inlet and act as a temporary thick-film stationary phase, assuming a shape in which the column phase ratio, β, decreases continuously in the direction of migration. In accord with the final equation in Section B above, the front edge of each band moving into that section must slow down faster than the rear of the band, and the sample is concentrated into a narrow volume. After a set time, roughly 1.5 times that required for the carrier gas to sweep out the injection chamber or 1 to 2 minutes, a purge valve is opened. Gas flow is directed to the bottom of the inlet, where it is divided so that one stream continues as the carrier gas while the other sweeps any residual sample from the injection chamber. A further purge stream ensures that the septum is never contaminated.

A schematic representation of a typical split/splitless injection system, which can be used in either mode, is shown in Figure 3.4. The technique has most value in the analysis of trace components.

Figure 3.4. Schematic diagram of a typical split/splitless injection system for WCOT columns Schematic diagram of a typical split/splitless injection system

4.  On-column injection

In on-column injection, the sample is injected in a solvent directly onto the column. The sample is concentrated by "cold trapping" or a "solvent effect" at the head of the column, so must be injected at a column temperature near the boiling point of the solvent. The column inlet does not have a septum but rather a "duck-bill" valve, made of a soft elastomer. This consists simply of two plastic surfaces which are pressed together by the pressure of the carrier gas in the injection port. During injection, a needle of fused silica is merely slipped between the two surfaces and is guided into the top of the WCOT column. (Some commercial systems use a more complex solenoid valve).

As no sample splitting occurs, there should be no discrimination of sample components. The sample is not vaporised instantly, so it is not stressed as much as in other procedures. In consequence, a high degree of precision can be attained in quantitative analyses

As with other injection procedures, some care must be taken to ensure that the analyst gets the most out of it. The sample dilution must be gauged correctly so that the column is not overloaded. The mechanical problems of feeding the needle into the WCOT column should be minimal with a properly designed injector, provided that the needle dimensions are correct for the column, especially when both are made of fused silica. Correct injection technique is vital, and in particular the syringe plunger should be depressed rapidly so that the sample is "sprayed" into the column. With this method, it is especially important that care is taken during the preparation of samples for chromatography, in order to ensure that they are free from involatile materials which might accumulate on the column and bring about some degeneration of the stationary phase.

 

5.  Programmed-temperature injection

As direct on-column injection can lead to a relatively rapid deterioration of the column through interactions of the liquid phase and the injection solvent or involatile impurities in the sample, an alternative has been devised in which a specially designed injection port takes over the function of the top part of the column. The sample is introduced into this inlet, which is cooled initially, and then its temperature is raised at a controlled rate so that the sample components are selectively vaporised. Injection systems of this type are relatively new and appear to be well suited to the analysis of lipids of high molecular weight, such as intact triacylglycerols. This is discussed further in Chapter 8.

 

G.  Quantitative Analysis

In all aspects of chromatographic analysis, from sample preparation to the separation process itself, there are opportunities for errors to occur. In assuming everything that is possible has been done to minimise this, there can be little doubt that electronic digital integration is the most accurate and reproducible, not to mention the most rapid and convenient, means of quantifying chromatographic peaks. Yet even here, there are potential sources of error, and it is essential to ensure that the various instrumental parameters on the integrator, especially those defining the sampling rate, are appropriate to the elution volumes (peak widths) at various times during analysis. If this is not done carefully, according to the manufacturer's instruction manual, it is possible to negate the advantages of the technique. The author has noted before that it is not always easy to convince a novice analyst that a computer printout of his results is fallible. In analyses with WCOT columns, electronic integration is the only method suitable for the quantification of complex samples.

In analyses with packed columns, electronic integration is also desirable, especially in temperature-programmed analyses. On the other hand because peak width is greater than with WCOT columns, there are some manual methods available that can give acceptable results, provided that the peaks are symmetrically shaped. One such method consists in drawing tangents to the sides of each peak to produce a triangle with the base line as the third side; the area is then proportional to the product of the height and width of the triangle. Alternatively, the height of the peak itself and its width at half the maximum height can be measured, and again the area is proportional to the product of these. Although peaks are never truly symmetrical, there have been such marked improvements in the quality of the column packing materials in recent years that these methods are more accurate than reports in the earlier literature might indicate.

A third method, which is suited only to isothermal analyses, was once much used by the author. It consists in multiplying the retention time of a component by the peak height, this quantity also being proportional to peak area. The advantages are that two relatively large distances are measured, so errors at this stage are minimised, reasonably accurate results are obtained with incompletely resolved peaks, and it is simple and rapid. A correction factor [123] can be applied to increase the accuracy of the method and compensate for the fact that the sample is applied to the column as a finite band rather than as a point source as theory requires, i.e. the widths of the peaks at half-height are plotted against the retention times of the components and the resultant straight line intercepts the base line at a point in front of the actual point of injection. Retention times or distances are thereafter measured from this point. (In practice, the errors of extrapolation are such that the point can perhaps more easily be found by testing in the appropriate region until the results with a simple standard mixture best match the known composition.)

The next step in the quantitative analysis of a sample consists in determining whether any response factors need to be applied to correct the experimental results. These must be determined by using standard mixtures, which are as similar as possible to the samples to be analysed, and comparing the analytical results under the standard chromatographic conditions with the known compositions. With most lipid samples and flame ionisation detection, these should only vary slightly from unity; if large correction factors are necessary, some stage in the chromatography has not been properly optimised. Some specific examples are described in later Chapters.

In many analyses of lipids, the results will be presented simply in terms of the relative proportions (expressed in percentage terms) of each component. There may also be instances, where the absolute amount of each component must be measured. This is best accomplished by adding a precise quantity of an internal standard, ideally a substance which resembles those to be analysed in its chromatographic properties, but does not occur naturally in the sample. If possible, this should be added when the tissue is first extracted, so that it is carried through the extraction, group separation and derivatization steps, as well as through chromatography. This ideal situation is rarely encountered in practice, and it is usually sufficient to add the internal standard immediately prior to a derivatization step. The areas of all the sample peaks can then be related to that of the internal standard, the absolute amount of which is known. As an example, the absolute amounts of lipid classes as well as the fatty acid compositions are frequently determined by adding the methyl ester of an odd-chain fatty acid to the sample, prior to transesterification and GC analysis.

When GC is used in routine analytical applications, it is important to set up a proper system of quality control in order to ensure that the equipment functions correctly, and that it is not subject to gradual deterioration or excessive random variation. To do this, it is necessary to establish regular checks on procedures by testing them with defined primary standards. The results should then be evaluated objectively by statistical methods. Such systematic checks may also indicate whether reagents are deteriorating with age, or whether a faulty batch of solvent, say, has been received. It might help in picking up unplanned changes in methodology introduced by unskilled technical support staff. Of course, these comments could equally be applied to most other aspects of lipid analysis. Quality control in the lipid laboratory has been reviewed by Naito and David [659].

 

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.

 

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.

Updated June 27, 2011