The Chromatographic Column

3.

The Chromatographic Column

The success of a gas chromatographic analysis depends above all on the chromatographic column in which the separation process takes place. This part, often described as the "heart" of the chromatograph, can be designed as the pivot of the gas chromatographic system. The choice of its type, dimensions and stationary phase determines the feasibility, quality and duration of the analysis. In their original work, James and Murtin used columns prepared from tubes of 4 mm I.D. with lengths of 1.2-3.3 m, packed with a stationary phase consisting of a liquid distributed on the surface of relatively inactive porous solid particles. Since then, packed columns have been applied which, as early as 1958, could be optimized to an efficiency of 30000 theoretical plates for a 16 m length and an I.D. of 2.2 mm [70]. By the introduction of more inert supports and very low loadings, the range of application of packed columns could be extended to the analysis of steroids and other biologically important compounds [71]. An extraordinarily significant breakthrough in increasing the column performance is due to Go&, who was the first to report on open-tubular columns [72], a column type revolutionizing high-resolution gas chromatography and offering the facility for achieving column efficiencies of many hundreds of thousands of theoretical plates. Further important steps in column technology were the development of support-coated open-tubular columns [73] with increased sample capacity, of glass [74] and even flexible fused-silica open-tubular columns [75] and of cross-linking the liquid phase on the inside walls of open-tubular columns [76]. A detailed report on the historical development of columns was given by E m [77]. Before dealing in detail with stationary phases themselves, the column, although not being a major topic in this book, should be discussed in more detail than any other part of the chromatograph, as it is the vessel containing the stationary phase. Its function is to expose the stationary phase to the mobile phase such as to guarantee an optimum separation by maximum repeated partition steps between these two phases. There are several possibilities for achieving this goal, each having pros and cons. They will be considered in this section, including the description, where necessary, of pre-treatments and other preparation steps. We shall consider the details of different column types successively.

3.1.

Packed Columns

Originally, these classical columns had lengths of 1-3 m and inner diameters of 4-6 mm and were packed with coated supports having particle diameters of 0.2-0.4 mm. The column materials were copper, aluminium, stainless steel, and glass. As a result of numerous investigations and of the development of detectors of higher sensitivity, column inner diameters and particle diameters could be decreased and column lengths were increased, resulting in highperformance columns. At present the commonly used classical columns have the following characteristics. 3.1.1.

Column Materials

The most commonly used column materials are stainless steel, glass, quartz, nickel and polytetrafluorethylene (PTFE). It is important to note that the gaseous or vaporized sample com-

44

3. The Chromatographic Column

ponents come into contact with the column walls (and the walls of the connecting tubing). Hence, on metallic inner surfaces undesirable effects might occur, especially at higher temperatures, such as decomposition, conversion or adsorption of labile or polar compounds. Such substances should be analysed using glass, especially borosilicate glass, or quartz columns. On the other hand, metallic columns are more advantageous because they are more rigid, have good thermal conductivity and are more easily handled and processed. Nevertheless, the choice of the column material should primarily depend on the type and properties of the sample compounds. Columns made of stainless steel are predominantly applied. They are rigid and commercially available with various lengths and diameters. Before being used they should be washed with ethyl acetate, methanol and distilled water. Subsequently they should be filled with 50% nitric acid and allowed to stand for 10 min, then rinsed with distilled water to neutrality and with methanol and acetone. Finally they are dried in a stream of nitrogen. Lable compounds may be decomposed, e.g., steroids and pesticides, presumably owing to the presence of transition metals in the steel alloys. Glass columns are more inert provided that they do not contain alkali metals or calcium on the inner surface. In the analysis of labile compounds the inner surface must be treated by leaching the metal ions with hydrochloric acid, washing with distilled water to neutrality, then with methanol, drying, filling with a 5% solution of dimethyldichlorosilane in dried toluene, allowing to stand for several hours, draining, rinsing several times with dry toluene and methanol and drying. Another procedure for deactivating glass surfaces [78] involves heating for 4 h at 90°C with concentrated nitric acid, cooling, rinsing with distilled water and allowing to stand for 1h at ambient temperature with concentrated ammonia solution. The column is then rinsed with water and allowed to stand for 1h at ambient temperature with glacial acetic acid, then rinsed again thoroughly with distilled water. Ammonia treatment results in an activated porous surface, fit for the subsequent silylation, and acetic acid removes metal oxides from the surface. The active sites are subsequently blocked with a 20% solution of octadecyltrialkoxysilane in tert.-butanol and diacetone alcohol. Instead of the trifunctional silane, difunctional silanes (dimethyldichlorosilane) and siloxanes and monofunctional silanes (trimethylchlorosilanes), silazanes (hexamethyldisilazane, tetraphenyldimethyldisilazane) can also be used, the success of the procedure depending on the structure of the sample components and the conditions applied, especially temperature [79-871. Soda-lime glass columns, containing more than 20% of sodium oxide + calcium oxide, are less suitable for analysing polar and labile substances than columns made from borosilicate glasses which, however, also contain alkali, at lower levels, but also requiring the above treatment. The treatment with dimethyldichlorosilae is necessary in order to deactivate silanol groups, which are always present on glass surfaces and exhibit harmful adsorption effects. @or a detailed discussion of such surface phenomena and treatments see Section 3.5.1.) With the exception of aggressive fluorine compounds, most organic and inorganic samples, especially those containing strongly polar or sensitive constituents, can be better analysed in glass columns than in columns made of other materials. They also have, moreover, the advantage of being transparent and thus to allow the packing of the column to be inspected and cavities to be seen. A disadvantage is their fragility, especially in the column oven when using hydrogen as the carrier gas, because of the risk of explosion. Glass-lined metal tubing (GLT) columns (Scientific Glass Engineering, Melbourne, Australia) consist of an inner borosilicate glass tube encased in a stainless-steel sleeve, resulting in a column that retains the chemical inertness of conventional glass columns but which can withstand far more careless handling and offer better uniformity of diameter than can be expected from conventional glass tubes. By applying a temperature of about 800°C it can be coiled. The maximum column temperature is otherwise 500°C.

3.1. Packed Columns

45

Nickel columns, if pre-treated in the same way as stainless-steel columns, are almost as inert as glass columns, and steroids, alkaloids, analgesics, barbiturates and catecholamines are readily analysed [88]. PTFE columns are most suitable for investigating aggressive substances, e.g., HCl, Cl,, HF and ClF3, since this material is chemically inert. Nevertheless, caution is required, because owing to their processing, such tubes might be microporous, and traces of ambient air might enter the column and give rise to undesirable effects (oxidation, additional N2and 0,peaks [891). Even organic compounds may be adsorbed and retarded [90]. Owing to its small surface energy and to its relative neutrality [90], quartz offers certain advantages. The most important application of this material in gas chromatography, however, is in open-tubular columns. A detailed discussion is given in Section 3.4.1. Columns made of tantalum, tantalum-tantalum oxide or zirconium-zirconium oxide are appropriate for the analysis of corrosive samples [91]. Finally, it should be mentioned that deactivation even of metallic surfaces has often been carried out with alkylchlorosilanes or dazanes. A temporary improvement may occur, as chemisorption and adsorption sites seem to have been neutralized. However, one should be aware that if a reaction of metallic OH groups with the S i - C l groups of the chlorsilane has taken place, the resulting metal-0-Si bonds are hydrolyzable, and traces of water, which cannot easily be avoided, will soon re-establish the previous state.

3.1.2.

Column Dimensions

For solving not too complicated separation problems, a column length of 1-3 m is sufficient, corresponding, depending on the quality of the packing, to a number of theoretical plates between 500 and 6000, or with high-performance columns even 10000. Simple analyses, e.g., of sample components the boiling points of which are far apart, may be carried out with even shorter columns, having lengths of 0.3-1 m. For the solution of complicated problems gas chromatography offers two possibilities: increasing the efficiency (n) or the selectivity (r), of which the latter, if realizable, is always favoured. A pure or mixed stationary phase giving the best r values for the compounds under investigation is packed into a column of the required length. Originating from eqns. (38) and (93) describing the almost complete separation of the most crucial solute pair, an expression for the required minimum column length Lhcan be derived:

The general column length is 1-6 m. Longer columns are needed for the separation of multicomponent mixtures. Lengths exceeding 12m for packed columns are hardly suitable. It should be emphasized again, and illustrated by an example, importance of the selectivity: let k be m, then when rz,l= 1.05 and h = 0.05 cm a column length of 7.56 m would be necessary, whereas with r2,1= 1.20 only 0.54 m would be required. The resolution can be optimized by varying the column length and temperature by a method called length-temperature time normalization chromatography [92, 931. The basic equation of time normalization is

where L is the column length, r7 is the average linear carrier gas velocity, k is the capacity ra-

46

3. The Chromatographic Column

tio and subscripts A and B identify two sets of chromatographic conditions. As we have seen

L

in eqn. ( l l l ) , the term Y (1 + k ) corresponds to the retention time, and eqn. (112) only exU

presses the fact that tIu = fRB, i.e., the retention time of a certain compound is the same on both columns. If the carrier gas velocity remains constant and if the columns are operated in the vicinity of the Van Deemter equation, we can draw the following conclusions and apply the following relationships. As column B is lengthened with respect to A, the temperature at which B is operated must be higher than that for A in order to decrease the value of kB, and thus keep eqn.'(ll2a) true equality. The required capacity ratio, at any length L,which will keep tR constant, i.e., normalize the system, is calculated from eqn. (112) (subscripts B will henceforth be dropped):

The temperature required to give that k can be obtained from eqn. (114) [94]:

T= where

T

R

AHV

TA 1 - (NAH")T A In (Wkd = absolute temperature, = gas constant, = molar vaporization enthalpy

ized one.

(1 14)

of the last eluted compound, i.e., the normal-

For each length change, be it an increase or a decrease, the normalizing temperature can be easily calculated via eqns.(ll3) and (114). AH" can either be approximated from Pictet-Trouton's rule or according to eqns. (28) and (29) from runs on column A at two different temperatures [8]. Gnishka [92] has shown that the optimum capacity ratio is given, using the fundamental relationship of eqn. (lOO), by

where r2.1 is a pair of substances difficult to separate, kptis the optimum capacity ratio giving the maximum resolution, a is a constant that can be determined [95] by running two analyses:

at different teperatures and b is another constant. Once the optimizing k value is known, eqns. (113) and (114) allow the calculation of the column length and temperature that will yield kptand thus maximize the resolution. A convenient procedure is as follows [93]. Install a column of any convenient length in the oven, operate at any sensible temperature, find roughly the velocity for the maximum efficiency and observe the resolution. If it is not sufficient, change the temperature for a second run. This will allow the calculation of the required parameters to solve the left-hand side of eqn. (115). This in return will yield the optimized capacity ratio, kpt, thus allowing the calculation, via eqns.(ll3) and (114). of the maximized lengths and the temperature at which it should be run. Then install in the chromatograph a column of the above length and operate it at the normalized temperature just calculated. The resolution thus obtained should be maximum at the analysis time. The two parameters temperature and length of the column are changed

3.1. Packed Columns

41

concurrently. Although one might be tempted just to increase the column length or decrease the temperature independently, Grushku [931 has shown that frequently a reduction in the column length and temperature might substantially increase the resolution at a constant analysis time. A limitation of the above equations is that they can only be used to give an indication of the parameters to be optimized and not to calculate their actual magnitudes, as complex and not binary mixtures usually have to be separated and as, ideally, the optimization should take into consideration both the column parameters (type, length, diameter, stationary phase, phase ratio) and the operating parameters (temperature, carrier gas nature and velocity). The column inner diameter of classical packed analytical columns ranges from 1.5 mm to 6 mm. Although the limits between the dimensions of classical packed and microbore columns are controversially defined, in this book only columns with diameters d, 5 1 mm are called microbore columns. An important parameter concerning d, is the particle diameter, d , , of the packing material, and the ratio d,ld, should not be lower than 0.03 or higher than 0.3. Each value offers different advantages, depending on the aim of the analytical application. Columns with a ratio of d,/d, near 0.3 are highly permeable, owing to the column permeability, which is proportional to the square of the particle diameter, according to eqn. (59). On the other hand, the efficiency is almost inversely proportional to the mean particle size (which is present explicitly and implicitly in eqn. (62)), and hence smaller particles and columns with narrower bores would be more advantageous. This has been emphasized by Huber et al. [96]. An important restriction in attempting to prefer fine particles is the higher column inlet pressure. Therefore, short columns have been used in such instances. An important factor that should not be overlooked, is the load. At the same impregnation rate, the load is directly proportional to the column cross-section and hence to the square of the inner diameter, i.e., load = constant d : . This is why higher sample weights can be analysed on columns with larger diameters. The sample weight ranges from approximately 3 mg for 5-mmcolumns to 1 mg for 3-mm columns, if other dependences (on temperature, weight of stationary liquid, and type of sample) are disregarded. Hence numerous factors have to be considered when choosing a suitable column diameter. Suitable column inner diameters might be 2-3 mm if the columns are packed with support materials of particle size dp = 0.12-0.15 111111. The shape of the column is most often a helix. Metal columns can easily be coiled by an appropriate device [97]. For the various forms, tubing connectors, etc., see ref. [97].

3.1.3.

Preparing of the Packing and Packing Procedures

In packed columns, the packing may itself be the stationary phase, if it is an adsorbent (GSC),or an appropriate solid support, being as inert as possible, which may be coated with a liquid stationary phase (GLC).Both the adsorbents and the inert solid carriers, and also the

liquid stationary phases, will be discussed in later chapters, whereas the coating of the sup-

port and the packing of the column with the stationary phase will be dealt with here.

The carrier has to be coated with a liquid film that is as uniform as possible. The wettability of a surface can be expressed in terms of the “critical surface tension” yc [98], which is a value above which liquids with surface tensions yi show a finite contact angle (8) on a smooth surface. The value of y, can be obtained from a plot of cos 8 vs. yi, which is usually a straight line; the intercept with the Cos 8 = 1 line giving yc [99]. On a rough surface, as exists on the usual porous support particles, the (macroscopic) contact angle is decreased (and cos 8 increased) because the liquid phase penetrates into the scratches, holes and pores. The degree of roughness is characterized by rf, the roughening factor [loo]:

3. The Chromatographic Column

48 cos 8’

rr=-=-cos8

s’ s

where

8 = contact angle on the smooth and B on the rough surface, s = macroscopic surface area, s‘ = microscopic surface area. This equation indicates that a porous surface, because s’> s and hence cos 8’ > cos S, can better be coated than a smooth surface. We shall return to this fact when discussing open-tubular columns and column support materials. The weight of the stationary liquid (mJ that is coated on the support material to give a certain impregnation rate depends on several factors. Basically, the weight has to be chosen so as to avoid agglomeration; the material must not be sticky. The area and structure of the support determine the impregnation rate. High impregnation rates have two advantages: the residual activity of the support is less troublesome, and the sample weight can be increased. On the other hand, some disadvantages have to be considered: the efficiency decreases, the h value depends strongly on the gas velocity and the analysis time increases. As a guide for diatomaceous-type supports, 5-20% by weight is advantageous; for other materials, see Chapter 7 and ref.[101]. The optimum value for a certain separation problem can be found [lo21 as follows. The most appropriate stationary liquid is coated on to the support using an impregnation rate of about 5% by weight, and the stationary phase is packed into the column, which serves as a test column. The column temperature that gives the optimum resolution of the investigated sample constituents is determined and applied to the actual column, which will give, independent of the impregnation rate, the most advantageous resolution. The impregnation rate, on the other hand, is selected so as to give the required analysis time. Decreasing the weight of liquid is limited, however, because of possible adsorption effects on the uncovered surface, on the solid-liquid interface and in the pores, which contribute to the retention of polar compounds. However, provided that the support has been well deactivated previously, low impregnation rates are successful even in the analysis of high-boiling compounds. This applies to diatomaceous-type supports, previously deactivated, impregnated in the range ca. 0.1-3% [102]. Nevertheless, one must be aware of the following phenomena. The real stationary phase is not a single-phase but a multiphase system, and the net retention volume is not only determined by solution processes obeying eqn. (15), but additionally by adsorption on the gas-liquid [43] and gas-solid [lo31 interfaces and by adsorption exchanges between liquid and sample molecules interacting with the active sites of the support [104]. Such effects have a relatively greater influence with low impregnation rates because of the lower V, values and hence the relatively higher contributions of the additional effects expressed by the second term of eqn. (83 a) [45]: m

(for definitions, see Section 2.5). The adsorption on the liquid surface is favoured by the adsorption power of the support, which becomes apparent with very thin films. Such a modified layer may develop if a support with a specific surface of a.3 m2/g is less coated than 1% [105]. The minimum possible liquid loading depends not only on the activity of the support, but also on the polarity of the liquid and sample molecules, i.e., when polar compounds are investigated by using non-polar stationary phases, the danger of adsorption processes has to be taken into account. This applies especially to the calculation of retention indices, and differences in experimentally determined retention indices can often be attributed to a non-uni-

3.1. Packed Columns

49

form coating. Retention index differences of 5-10 have been observed with badly deactivated supports [106], whereas relatively inactive supports when coated with 2 2 % of non-polar poly(dimethylsi1oxane)did not indicate a retention index dependence on the liquid loading; merely a decrease from 2 to 1 resulted in a retention index increase of 10 units [107]. Finally, it should be pointed out that if the dimensions of the impregnation rate are as usual weight/ weight, then supports with higher bulk densities (glass beads, Chromosorb G) would contain a larger amount of liquid phase per unit support surface area, and also per unit column packing volume, than supports with lower bulk densities. Hence the degrees of impregnation should rather be related to the support’s surface area in order to be comparable. Supports with a small specific surface area and porosity (glass beads, quartz particles) must be loaded with less than 2% (w/w). There are many coating techniques. Generally the liquid stationary phase is dissolved in an appropriate solvent, from which it is deposited as a film on the support by means of one of the following procedures. The degree of impregnation (the loading) is given either in weight of liquid per 100 g of support (percent by weight) or in weight of liquid per unit packing (i.e. percent liquid stationary phase referred to the weight of impregnated support). Provided that the surface area of the carrier is known, it can also be a basis for the impregnation rate [108], the dimensions of which in this instance would be weight of liquid per square metre of support surface area. Before outlining the various coating procedures, it should be mentioned that detailed discussions have not been included, but only summaries of the principles, with short evaluations and descriptions of techniques to be recommended. For details the reader is referred to a textbook [lo91 and original papers. Filtration technique. The stationary liquid is dissolved in a suitable solvent and the support is added and briefly mixed. After filtration the residue is dried, e.g., by the fluid-bed technique [110-1131. This method is recommended with low loadings (<3%, w/w), the value of which ensues from the difference in added liquid and residual liquid contained in the filtrate. Frontal technique. The support is packed into the column and a solution of the stationary liquid is passed through the packed column until the concentrations of added and draining solution are equal [114]. The solvent is expelled from the column by heating in an inert gas flow. Evaporation technique. A weighed amount of stationary liquid to be coated is dissolved and, after addition of the support, the solvent is evaporated. The simplest way to carry out the evaporation is to use a water-bath and subsequently a steam-bath and a drying cupboard. However, although this method is often applied, it cannot be recommended, for several reasons. The main reason is the danger of oxidation of the liquid stationary phase. If we consider that a small amount of liquid is spread over a large surface area and present as a very thin film, in presence of oxygen partial oxidation of the liquid is likely to occur. For example, a packing with a loading of 5% (w/w) is prepared for a 6 m X 3 mm I.D. column, corresponding to a volume of ca. 42 cm3. Assuming that the support would have a packing density (in the column) of 0.5 g/cm3 and a specific surface area of 4 m2/g,the 1.05 g of stationary liquid would cover a surface of 84 mz. It can be calculated that the average film thickness would be ca. 12 MI, provided that the surface area is covered completely, and we can imagine this thin film to be exposed to atmospheric oxygen at oven temperatures, as often used, above 200°C and hence easily subject to oxidation processes with serious consequences in view of the quality of the packing. Therefore, the evaporation should be performed out of contact with air. A short-cut method con-

50

3. The ChromatographicColumn

sists in evaporating the solvent in a rotary evaporator. At ambient temperature a water-jet vacuum is applied for a few minutes until the support no longer evolves air bubbles. The vacuum is filled up with nitrogen, the bath temperature is elevated to ca. 40°C, then the pressure is reduced cautiously while rotating the flask and the solvent is withdrawn. Once the material is flowing freely, the rotation is stopped and the residual solvent is issued, maintaining the vacuum, at a bath temperature of 80°C [115]. Vacuum-vibration technique The stationary liquid is heated to 20-50K above its maximum operating temperature and by vacuum and vibration impregnated on the support [116]. A maximum loading of 14%is attainable and the occurrence of degradation products is unavoidable. Recommended procedures (a) Prior to coating, the support should be dried in a round-bottomed flask for 8 h at 300°C under vacuum. The cooled flask is vented with dry nitrogen and the solution of the liquid stationary phase is added. Residual oxygen is removed by a partial vacuum, until air bubbles no longer occur. The solvent is mainly evaporated by rotary evaporation and frnally under vacuum at a temperature of 100°C below the maximum operating temperature of the liquid phase overnight [117]. (b) In a special flask (hedgehog flask) the support mixed with the solution of the liquid stationary phase is cautiously evacuated and degassed and after ca. 10min disconnected from the vacuum pump and connected with a rotary evaporator. The temperature of the heating bath should be 20°C above the solvent's boiling temperature. Pure nitrogen is introduced into the flask at ca. 1Ymin. The solvent is vaporized while slowly rotating the flask. A visible reflux should be maintained, thus rinsing the suspension downwards from the walls of the flask [118]. (c) The support is placed in a Pyrex tube (480 X 23 mm) fitted with a sintered glass disc, closed at one end and evacuated at room temperature by connecting the other end to a vacuum pump. After several minutes, the tube is placed in a furnace at 450°C for 4-24 h while maintaining the vacuum, and removed from time to time and shaken to mix the powder. After heating, the tube is cooled to room temperature while still under vacuum. The tap connected to the pump is closed and the dissolved liquid is run into the Pyrex tube via a funnel connected to the tap. This tap is opened and the solution flows over and coats the powder. The tube is allowed to stand overnight. Then the cap, which closed the other end of the tube, is removed and, after the upper surface of the solution has reached the carrier surface, the remaining liquid is forced out under a gentle stream of nitrogen until the solid becomes freeflowing. The purpose of this procedure is to avoid the formation of water, which when using coated support material not been subjected to vacuudheat treatment could be eliminated by condensation of silanol groups to form siloxane bridges by heating in the column. This water could be trapped underneath the liquid phase and thus might account for the gradual deterioration of some types of packed columns [119]. Determination of the degree of impregnation The simplest way to determine the loading is Soxhlet extraction using the same solvent as applied in the coating step [120]. It should be noted that a polar solvent might leach soluble constituents from the support, thus falsifying the impregnation values. The loading with organic stationary liquids can be established by a simple ash determination provided that the liquid is quantitatively oxidizable at 800°C to form only volatile products of combustion such as C02 and H20[121]. Silicone liquid loadings cannot be investigated by this procedure; instead C and H can be determined by elemental analysis.

3.1. Packed Columns

51

It should not be neglected to run a blank with the uncoated carrier, especially ifit has been previously treated with organochlorosilanes or -silazanes. Carboranesiloxanes or organomet a l k phases can be determined by a boron or metal determination, respectively. A method that can be applied to most organic and to poly(dimethylsi1oxane) phases is based on the vaporization and degradation of the stationary liquid phase with exclusion of oxygen. With a temperature programme from ambient to 800°C within 2 h the sample is heated in a flow of suprapure nitrogen. After a further hour the “vaporization” is finished, without the formation of carbon. As water would affect the results, the non-impregnated support has to be analysed in the same way [122]. Finally, it should be pointed out that the determination of the degree of impregnation should be carried out only when the stationary phase was previously conditioned, because during the conditioning residual solvent and vaporizable constituents of the liquid phase could be released, which therefore would not contribute to the retention and must not be taken into account when, e.g., calculating V,. Details concerning the conditioning are given in Section 3.1.4. Packing procedure In a simple way U-tubes are packed by connecting both ends with small funnels through which small portions of the impregnated support are poured alternately while vibrating and slightly tamping until the upper surfaces of the packing can be seen 1 mm underneath the column ends. The ends are loosely closed with deactivated glass-wool in such way that the particles of the packing will be held back but the resistance to flow will not be increased. Packing of coiled columns has to be carried out differently, by tap-filling the precoiled column with a nitrogen flow at the column inlet and a vacuum applied at the column outlet [117, 1231. As the permeability of the packing decreases during its growth, the gas flow-rate produced by the vacuum also decreases, and the packing procedure is prolonged. Therefore, commercially available devices for pressurized packing should be used. To overcome the disadvantage of such appliances, that columns packed in this way are inefficient, a more favourable device has been designed for pouring the stationary phase slowly and continuously under a gas pressure into the column, so ensuring a homogeneous and tight packing [124]. It is especially suitable for laboratories in which many columns have to be packed. Coated PTFE supports should be packed into the column at 0°C to avoid electrostatic charges [125], using vibration [126]. A further recommended procedure applies pressure and ultrasonics. It has been developed especially for the optimum packing of columns with a coated support of the Volaspher type [127]. Finally, it should be emphasized that after finishing the filling of the columns, the pressure and vacuum must be reduced only slowly, as otherwise the tight packing would be loosened and the efficiency diminished. The efficiency of packed columns may also be improved if the packing is completed after having been heated to the maximum column temperature (which depends on the liquid stationary phase) [117].

3.1.4.

Column Conditioning

If the column is used for performing analyses immediately after the packing, then the retention times, selectivity and efficiency will change gradually and the detector will be contaminated. If the packing is not packed densely enough it will become tighter under the influence of the carrier gas flow. Solvent residues, volatile constituents of the liquid stationary phase and its degradation products will evaporate to various extent. At higher temperatures, the stationary phase might be degraded or cross-linked. In order to avoid these and other difficulties, any gas chromatographic column has to be conditioned before operation. Basically, the

52

3. The Chromatographic Column

stationary phase is heated, either in a special tube [128] or in the column itself, whilst (purified, especially oxygen-free) carrier gas is passed through it at a higher temperature than the subsequent maximum operating temperature. When the column itself is used, its effluent end has to be disconnected from the detector. In detail, the following procedures can be recommended: (a) The column containing the packed stationary phase is installed in the oven of the gas chromatograph at ambient temperature; it must not be connected to the detector. A flow of purified carrier gas, which especially has to be oxygen-free, is begun at a flow-rate of 10-20 ml/min. Purging the column for 30-45 min allows air to be removed from the column, which might otherwise cause oxidation of the stationary phase. The column is then heated at 60" for 30 min and subsequently programmed at 2 Umin to the desired upper limit. This temperature, the value of which should be 25OC above the desired operating temperature (but not exceeding the upper temperature limit of the stationary phase), is maintained for at least 12 h. The length of the conditioning time depends on the type of stationary phase, the impregnation rate, the maximum temperature of the analysis relative to the upper temperature limit of the stationary phase and on type and necessary sensitivity of the detector being used, and can require as long as 1 week if extreme parameters (operation at the maximum temperature limit of the stationary phase and at maximum sensitivity of the detector) are inevitable [129]. (b) The packed column is COMeCted with the injector, but not with the detector, [both as in (a)], and a flow of purified carrier gas is begun at the same flow-rate as in (a). After 30 min at ambient temperature the column temperature is raised to 60°C and maintained there for 1 h. After cooling to room temperature the carrier gas flow is stopped, the effluent column end is closed gas-tight and the column is heated to the upper temperature limit of the stationary phase and maintained for 14 h. The column is then cooled, the tube seal is disconnected and the carrier gas is allowed to flow through the column, which is heated to a temperature 20°C below the upper temperature limit of the stationary phase for further 14 h. This method eleminates troublesome surface activity from the support and column walls, especially when using glass columns packed with silicone stationary phases [130]. A shortcoming of both methods (a) and (b) is the lack of control of weight loss of stationary phase, as the relatively high weight of the column generally does not allow exact weighing. 3.1.5.

Column Testing

The column walls and connections must not permit any trace of carrier gas to pass as otherwise losses of carrier gas and sample would occur and, if the carrier gas is hydrogen, an explosion hazard may arise. The column is checked by sealing the effluent end and applying first an Nzpressure which, after interrupting the N2supply, must remain constant for a longer period, and then repeating the procedure with a Hzpressure. By the determination of the permeability, the mobile time, the efficiency, the selectivity, the time of analysis and the degree of impregnation according to the previous chapters and sections, a decision can be made as to whether the column comes up to expectations or preferably should be re-prepared. The peak shape, especially of polar compounds on non-polar stationary phases, indicates residual adsorption sites in the system, and deactivation procedures or a change of support would be required unless the amount of sample is too high for the available loading (in this instance the peak shape, however, would differ from the unsymmetric shape caused by adsorption phenomena). Further details concerning this subject will be dealt with when discussing open tubular columns (Section 3.3) and support materials (Chapter 7).

3.1. Micro-Packed Columns

53

The capacity of the column for the amount of sample can be established by determining the number of theoretical plates as a function of increasing sample volume. If the efficiency is only 90% of the value found for the smallest sample volumes, one has arrived at the maximum capacity. The upper temperature limit and its determination will be described in a subsequent chapter. The packing density (g/ml) can be calculated from the amount of packing and the volume of the column, which was previously determined with any liquid the density of which is known. If water is applied, the remarks in Section 3.1.1. should be considered.

3.1.6.

Pre-columns

A special type of packed columns are pre-columns. These are short columns, having lengths between 2 and 20 cm, and with inside diameters between 2 and 8 mm. They can, generally, be backflushed and are easily interchangeable. Their function consists in saving the column (and detector) from slowly or not eluting sample constituents and from changes in selectivity caused by strongly polar sample constituents having low vapour pressures, in that these are eliminated before arriving at the column by backflushing. Further, the analytical column can be saved from losses of stationary phase in the initial part of the column by using a pre-column with a high degree of impregnation. Moreover, they enable samples that contain nonvaporizable residues, e.g., polymers or salts, to be analysed without changing the quality of the analytical column. It should be borne in mind that the pre-column diameter affects the efficiency of the main column. The higher the pre-column inside diameter, the greater is the loss of efficiency of the column. This loss can partially be diminished by using of smaller diameter solid support particles, but a better method is to increase the pre-column temperature [13Oa]. Pre-columns are packed with an inactive support impregnated with 20-30% of an appropriate silicone phase, and they are connected gastight with the analytical column and operated isothermally at the highest possible temperature, the magnitude of which depends, of course, on the problem to be solved. The principle and installation have been described by Kaiser [131]. Conventional packed columns, if coupled with open-tubular columns, can be considered as pre-columns if they are used to remove the main sample components, thus allowing the trace components to be optimally separated in the capillary column. For trace analysis, this is an important possibility. Both column types are complementary to one another, the packed column with its high sample capacity (2-3 orders of magnitude higher than that of capillary columns) and the open-tubular column with its outstanding performance.

3.2.

Micro-Packed Columns

Owing to their very promising parameters of mass capacity, phase ratio and resolution with respect to the required analysis time, micro-packed columns have increased in importance. Their main features are microbore tubes with inside diameters (4)between 0.3 an 1mm and lengths of ca. 1-15 m, packed with particles of 0.007-0.3 mm diameter (dp).If the ratio of particle diameter to column inside diameter is d,/d, = 0.25, the columns can be packed regularly and densely. For dp/dc> 0.25, the packing becomes increasingly irregular and less dense. Transitions between both classical packed and micro-packed columns and regular and irregular micro-packed columns obviously occur, and an exact classification must therefore be arbitrary. Nevertheless, packed columns ought to be subdivided according to the following

54

3. The ChromatographicColumn

Table 3. Characteristics of Packed Columns

Tube i.d. (d,) [mm] Particle size (d,)

1-1 dP dc

Sample capacity Cg] h [-I L [ml n Preferred applications

Classical packed columns

Regular micro-packed columns

Imgular micro-packed columns

22 0.12-0.30

0.3-1.5 0.04-0.3

0.3-0.5 0.05-0.15

0.04-0.10

0.07-0.25

0.25-0.5

> 1000

1-10 0.15-0.40 5 15 5 50 000

5-20 0.4-1.2

>o.s

<15 < 10 OOO*)

Simple separation problems; trace analysis; coupled techniques with spectrometry

Multi-component analysis; high-resolution coupled technique with MS; trace analysis

s6

5 15000

Short columns permit high-speedanalyses; coupled techniques with MS

9 n = 30000 for 16 m length obtained by Scott (70) is the exception, not a standard

parameters, as their merits and hence their application fields differ from one another, as shown in Table 3. It should be realized, however, that the values are only guidelines. The term “packed capillary columns”, used repeatedly in the chromatographic literature, has been deliberately avoided in this book in order to prevent misunderstandings. In this we agree with Stnrppe, who also suggested a subdivision into these three types of packed columns [132]. Owing to their high efficiency combined with a sufficient sample capacity, micro-packed columns offer excellent possibilities for the analysis of multi-component mixtures and trace constituents, and, especially, for coupling with mass spectrometry. They further combine high efficiency with short analysis times, thus permitting high-speed analyses. The stationary phases, as in classical packed columns, consist of both adsorbents (GSC)and liquid phases on supports, the particle sizes of which are given in Table 3. With smaller particle diameters (< 160 pm) the above packing procedures have to be modified by using ultrasonic vibration, maintaining an inert gas pressure [133-1351 or at least using electromechanical vibration of 20-100 Hz [136-1381. Apart from the detailed advantages a disadvantage should not be ignored, namely that the pressure drop, depending on particle size and packing density, ranges from 0.03 to 0.3 MPa, thus permitting, in the case of the higher values that occur with smaller particles, only relatively short columns. However, this disadvantage is not serious: owing to the higher efficiency and to the essentially lower C, term of the Van Deemter equation (eqn. (63)) compared with classical packed columns, even short micropacked columns yield high numbers of theoretical plates (e.g., up to 5000-10000 per metre [117, 1351 if operated at high pressures), and higher carrier gas velocities can be applied (lower C, term because of the smaller particle diameters) without seriously increasing the height equivalent to a theoretical plate, both advantageous conditions for high-speed analyses. It should be noted merely that longer columns packed with small particles require modified injection devices applicable to gas inlet pressures above 0.5 MPa [117, 1351.

3.3.

Open-Tubular Columns

Since the first detailed reports by Coluy [20, 211 on open-tubular columns, i.e., tubes that do not contain a packing, and in which the liquid phase is coated on the inside wall and the

3.3. Open-Tubular Columns

55

shape of the free gas cross-section is maintained over the whole length of the column, this gas chromatographic technique has developed considerably and its applicability has widened tremendously. This development is based on the fact that gas chromatography with open-tubular columns represents the most efficient method for separating mixtures of vaporizable constituents regarding information content, resolution and analysis time, and there is no doubt that the future of gas chromatography, disregarding preparative GC, special cases of trace analysis and coupling methods if larger amounts of the sample constituents are required, will lie in the application of open-tubular columns with either thin films or thin layers.

3.3.1.

Tube Material

Essentially, the tube used for the preparation of open-tubular columns consists of stainless steel, glass or fused silica glass, each having advantages and shortcomings. Stainless steel: This is commercially available as tubing with the required internal and external diameters, can easily be bent and connected to the other parts of commercial gas chromatographs, has good thermal conductivity and does not break. On the other hand, the quality of its inner wall is often rough, dirty and active towards polar and labile compounds. This disadvantage is especially important with wall-coated columns, whereas in support-coated columns the inner surface, being covered with porous particles, cannot be reached by the sample molecules. Recently, a rather inert metal column has been developed by Chrompack, HT-SIMDIST, especially for high-temperature application (to 450 "C) [138a]. Glacis: Glass is cheap, easily deformable and has a homogeneous surface. Nevertheless, because of some disadvantages, its application has been restricted to highly skilled chromatographers: glass capillaries are fragile and their dead volume-free connection to the other parts of the gas chromatograph is complicated. The main restriction is the poor wettability, i.e., the difficulty of forming a coherent and stable film of stationary phase on the inner wall of the glass tube, and only after about ten years of intensive investigations by numerous chromatographers did glass open-tubular columns become generally accepted. The production of glass tubes can be implemented by means of a glass capillary drawing machine, invented by Desly et al. [139]. Most often soda-lime glasses are drawn, as capillaries made from this glass type are more elastic than those from borosilicate glass, although the latter are more inert. Fused-silica glacis: Fused-silica glass is a column tube material with outstanding properties. It exhibits flexibility and great inertness, the latter being due to the low metal oxide content compared with soda-lime and borosilicate glasses. Unlike normal glasses with a wall thickness of about 0.3 mm, fused-silica glass capillaries are drawn to give a very thin wall of about 0.05 mm thickness and externally coated, by analogy with the fibre optics process, with an appropriate polymer, e.g., polyimide, to impart flexible tensile strength. The outside coating is thermally stable up to 340°C and has a film thickness of about 0.05 mm. Two types of vitreous silica glasses have been used as starting materials: (a) Conventional fused quartz derived from Brazilian quartz rock crystals by electric or flameCusion [76]. It consists of ca. 99.99%SiO, with a few hundred ppm of N 2 0 3as the main contaminant [76], the content of which can be lowered by further chemical purfication. (b) Fused silica has been produced by gas-phase hydrolysis of a very high purity silicon tetrachloride and fusion of the S O z formed. This synthetic product contains only lppm of impurities. The drawing process is carried out at the melting temperature, which ranges, depending on the composition, between 1700 and 2100°C. An appropriate drawing machine has been described by Lipsky et al. [76].

56

3.3.2.

3. The Chromatographic Column

Column Dimensions

Stainless-steel capillaries for gas chromatography commonly range in length from 5 to 100 m and have outside diameters from 1.0 to 1.6 mm, most often 1.59 mm (1/16 in.), and the inside diameters may vary from 0.05 to 1.55 mm, but commonly only from 0.2 to 0.3 mm (wall coated) or from 0.2 to 0.5 mm (support-coated open-tubular columns). It should be emphasized that the column diameter must be constant all over the column length, thus requiring good column drawing technology. From eqn. (63), it can be derived that the column inside diameter or radius, r,, strongly influences the efficiency (expressed by h), i.e., increasing r, increases h, and hence decreases the efficiency. On the other hand, increasing r, means increasing the amount of stationary phase in the column, because VL= 2 nr, L d,

(118)

(provided r, a dJ, where VL = volume of the stationary phase, L = column length and 4 = average film thickness of the stationary phase, and a larger amount of stationary phase permits the amount of sample to be increased without overloading the column. This is important when coupling techniques are to be applied or trace analyses are to be carried out. A further fact should not be neglected, namely the influence of variations in the column diameter on the phase ratio. This term has been defmed by eqn. (23):

As VG can be expressed by VG = nra L

(118a)

[provided r, % dl because otherwise VQ would be VQ= TI (r, - dd2* L] we obtain, on combining eqns.(23), (118) and (118a),

With jl= K L / k (eqn. (24)) we obtain from eqn. (119)

This means that the smaller the column radius, the higher will be the capacity ratio, k = tlR/tM, or, in other words, increasing the column radius would increase the phase ratio and decrease k (as KLdepends only on the stationary phase and the temperature and does not depend on the column dimensions). Decreasing k would, however, according to eqn. (99), require more theoretical plates to achieve the same resolution of two adjacent peaks. As a consequence, the column diameter has to be suited to the analytical problem. As an average, standard column diameters of 0.25-0.30mm (wall-coated) or 0.5 mm (supportcoated columns) can be recommended, especially for screening purposes. When volatile compounds have to be analysed, the phase ratio must be decreased, which can be achieved by increasing the film thickness or decreasing the column diameter, and vice versa when analysing high boiling samples, where the phase ratio has to be increased. The diameters of glass and fused-silica glass capillary columns are similar to those made of stainless steel. The inside diameter commonly ranges from 0.10 mm to 0.20 mm for high resolution analyses (only split mode) and for mass selective detectors. For split/splitless injec-

3.3. Open-Tubular Columns

57

tion, column diameters of 0.25-0.32 mm can be recommended. In spite of their lower separation efficiency (about one third compared with columns of 0.18mm I.D.), wide bore columns (0.53-0.75 mm I. D.) are favoured for trace analysis (owing to their expanded sample capacity of ca. 2000 ng, i. e., a factor of >20 at comparable film thickness!), for high speed analyses in the case of less complicated separation problems and for packed column conversion. Depending on the inside diameter, the outer diameter is 0.20-0.29 mm (0.10-0.18 mm I. D.), O . ~ ~ I I I I I I (0.25 mm I. D.), 0.40-0.50111111 (0.32I.D.) and 0.66(0.53I. D.). Returning to the column length, which, as it is directly proportional to the number of theoretical plates [eqn.(38)] and to the analysis time [eqn.(lll)], is an essential column parameter, we must carefully consider necessity and possibility. In order to save time when analysing a large number of samples, the column length should be chosen only so as to achieve the necessary resolution of the worst separable pair of sample constituents. This possibility, which saves analysis time contrary to the concept of "open-tubular columns = exceeded columns length = superior efficiency", has often been overlooked. Thus, unless very complex mixtures have to be analysed, relatively short columns can also solve many less complicated problems, in a shorter time. For calculating the required column length and for length-temperature time normalization, see Section 3.1.2.

3.3.3.

Wall-Coated Open-Tubular (WCOT) Columns

Wall-coated open-tubular (WCOT) columns, introduced into gas chromatography by Golay in 1958 [20, 721, contain the stationary phase in form of a thin film deposited on the more or less smooth internal surface of the tube wall. This doubtless most promising column type in addition to the support-coated open-tubular columns, has been developed particularly by Diksfra and De Goey [140], Desty [141], Kaiser and Sfruppe [142], Scoff [143], Condon [144], Zlatkb and Louelock [145], Schomburg [146], Grob [147], Tesaiik and Novofny [148], Eftre [149] and Guiochon [150]. 3.3.3.1.

Properties and Pre-treatment of the Inner Tube Surfaces and the Formation of Films

An essential requirement for high separation efficiency is the formation of a thin and uniform film along the total length of the column, as otherwise the formation of droplets might occur, which would decrease the efficiency of the column to unacceptably low levels, and problems connected with active sites would be even more in evidence. The applicability of a WCOT column stands or falls with the homogeneity and stability of this film on the wall of the column. The wettability of the surface depends, in one respect, on the surface energy of the tube wall, and in the other on both the surface tension of the liquid stationary phase and the (mostly unknown) interfacial tension between liquid and solid. A measure of wettability is the cosine of the contact angle, which is correlated with the mentioned quantities by eqn. (121) [90, 981: Ys = Y L S + YL cos @

where

ys = surface energy of the solid, 12. = surface energy of the liquid stationary phase,

(121)

interfacial tension between liquid and solid, 0 = contact angle. In order to wet a surface sufficiently, 0 must be as low as possible and cos 0 as near as possiy~~ =

58

3. The ChromatographicColumn

ble to unity, and ys must be higher than fi. We have seen in Section 3.1.3 that the contact angle can be decreased by roughening the surface [eqn. (117)],and hence the film formation can be improved. Let us now consider the different surfaces. Stainless steel has relatively rough walls, which is advantageous with regard to film formation, as we have just seen. However, this material is active towards polar and labile compounds, and the activity must, prior to or simultaneously with the coating procedure, be reduced, e.g., by the addition of surfaceactive agents [151].Because of their lower activities, siliceous tube materials have been preferred for numerous gas chromatographic applications with the exception, perhaps, of the analysis of hydrocarbons, which is most often carried out with stainless-steel capillaries. Nevertheless, one should be aware of the fact that even glass and to some extent also fusedsilica glass have some activity. Its structure, composition and history of temperature treatment during the manufacture affect its behaviour, and especially its surface, the composition and structure of which influence column performance much more than the bulk glass characteristics [152],and this will be discussed in detail now. If we consider glass surfaces, we can expect metal and boron ions, which can serve as Lewis acid sites, causing adsorptive tailing. Owing to the substantially minor concentrations of these ions on fused-silica glass surfaces, this material is originally more inert [75].To improve the surface of soda-lime and borosilicate glasses, the oxides of metals (Ca, Mg, Al, Fe) and boric oxide have to be removed by leaching or etching procedures. Leaching involves, according to Grob and Grob [154],extensive hydration of the SiOz lattice, which permits soluble ions to be extracted with an appropriate solvent. The procedure consists in, e.g., filling of the tube with 20% HC1(90%of the tube length), sealing of the inlet and outlet and heating for 14 h at 160°C.After cooling the tube it is rinsed with water to neutrality [155].The surface is now a silica gel, which has to be dehydrated, but without removing most of the silanol groups because these are needed for the subsequent deactivation steps (preferably silylation). Therefore, the tube is purged with nitrogen and dried, applying a vacuum at temperatures between 150°C [156]and 300°C [155, 1571. In contrast to leaching, in the etching process the entire surface, including the lattice ions, is attacked and more or less dissolved [154].Etching reagents are gaseous HF or fluorine compounds that liberate HF at elevated temperatures, e.g., 2-chloro-1,1,2-trifluoroethyl methyl ether or NH4HF2[158,1591. The reagent is introduced, the column ends are sealed and the column is placed in an oven, the temperature gradient of which is small, heated (in the case of the ether) to 400°C and maintained at this temperature for 24 h. By the reaction of HF with the glass surface, silicon fluorides are formed, which are decomposed in the closed system to form silicon dioxide, which in turn is deposited in the form of whiskers. This uniform layer of silica whiskers provides, owing to the lack of the metal ions, and to the increased surface area and roughness, a suitable substrate for the liquid stationary phase. Wettability and the achievable (increased) sample capacity favour this procedure and its results. On the other hand, its disadvantage may not be overlooked. In particular, the subsequent, necessary deactivation seems to be difficult [160]. So far we have only considered the conversion of glass surfaces by leaching, rinsing and etching processes to produce a pure silica surface. The physical and chemical state-porosity, ruggedness, SiOH content and entrapped residual reagents-cannot be expected to be the same in each treated column. This is the reason why on the one hand excellent columns can be produced from conventional glasses, but on the other hand some columns are obtained that have to be rejected, as differences in surface density and porosity may affect the chromatographic behaviour [1611.Fused-silica glass columns, in contrast, possess only negligible, if any, Lewis acid sites and exhibit by nature a pure silica surface without any previous treatment and have a high degree of uniformity from column to column. The development of the fused-silica column has thus offered a better understanding of the role of silica surfaces with respect to deactivation and coating procedures, and treatments prior to the coating pro-

59

3.3. Open-Tubular Columns

cess are much more predictable and reproducible and less expansive than those required for conventional glass columns. Nevertheless, owing to the presence of surface SiOH groups, there are active sites on fusedsilica column surfaces and also leachedhinsed or etched conventional glass column surfaces. In an excellent survey, Jennings dealt with such surface phenomena, their origins and methods of deactivation [161]. Hydroxyl groups occur on those lattice silicon atoms, whose interatomic distances preclude the formation of additional siloxane linkages or which are formed after re-reaction of highly strained siloxane bridges with water. Different types of such silanol groups can be present: Lone silanols (a), vicinal silanols (b), geminal silanols or silanediols (c). OH

HO

I

- Si

HO

I -Si-0-5-

I

\ /

I

I (b)

(0)

OH

OH OH

(d

Free silanols are acidic and can react with ion-pair electron donors and hence they exhibit strong adsorption sites. When the interatomic distances of adjacent oxygen atoms are less than 2.8 A (2.8 10-lo m or 0.28 nm), the silanol groups are hydrogen-bonded to one another

(d)

(e)

[161] (d) and exhibit only weak or no adsorption, unless, through interaction with water, they become strong adsorption sites [162, 1631 (e). Heating to ca 165°C results in the desorption of water, which in turn can be readsorbed on cooling. Condensation reactions of spatially appropriate silanol groups to form siloxane bridges,

take place between 150 and >800"C. At first, on increasing the temperature from ambient to ca. 200"C, water from the condensation of geminal hydroxyl groups (c) is degassed, followed by water stemming from types (b), (d) and (e). This dehydration is reversible on heating to 400°C and partially even to 800"C, and increases with increasing temperature. Even above 800°C silanol groups still disappear, and strained, reactive siloxane bridges form. This reaction becomes reversible once again at much lower temperatures and exposure to room air and moisture [164]. We can now conclude that there exist in each glass column (after having removed Lewis acid sites) and fused-quartz and fused-silica glass columns (where Lewis acid sites, if any, would exist in negligible concentrations) a series of gradations from completely free silanols to fully bound silanols and from non-reactive to highly strained, very reactive siloxane linkages and bridges. These surface properties and compositions determine the possibility and degree of reactive and adsorption tendencies and also wettability. A column, just fused, exhibits a very high surface energy and hence can, according to eqn. (121), easily be wetted. On the other hand, a high surface energy means a high concentration of active sites, which has to be avoided in order to produce an inert column. Between these oppositing demands a compromise has to be reached. This is the goal of the following column surface pretreatments, and of the recently developed coating procedures described below and in Section

60

3. The Chromatographic Column

3.5.2., which avoid droplet formation by using gum phases, by bonding phases and by immobilizing the stationary liquid by means of cross-linking reactions. Because of their lesser importance in present, several pre-treatments, even though not long ago frequently applied and investigated, will be mentioned only briefly. In addition to the carbonization of methylene chloride and deposition of the formed carbon black on the inner column walls [165], the controlled deposition of sodium chloride [166, 1671 and barium carbonate [168] have been thoroughly investigated. In lieu of salt deposits, which have been found to decrease the column activity to a certain extent, thin layers of silica can be deposited by the hydrolysis of silicon tetrachloride at high temperatures [169]. These layers increase the surface area and wettability, but simultaneously the column activity is distinctly increased, for obvious reasons, and subsequent deactivation treatments are in any event necessary. Procedures meant to block active sites by surface active groups, by polar polymers or by thermal degradation products of polyethylene glycols sometimes seem to be successful (e.g. ref. [170]). However, the thermal stability is limited, and the column activity is frequently regenerated at elevated temperatures; further, abstraction of reactive sample constituents may occur, and the retention behaviour may be affected. One of the most promising ways of deactivating the inner walls of glass and fused-silica glass columns (which must be free from Lewis acid sites!) is chemical reaction of the surface silanols and highly strained siloxane bridges with thermally stable silylating reagents or other organosilicon compounds when reacted at relatively high temperatures. Let us first consider the true silylation reaction. A reactive silanol reacts with a silylating reagent to form silicon-oxygen-silicon bonds or, in other words, the active hydrogen atom of the SiOH is exchanged by a silyl group:

I I

-SOH

+ A-SiR3

+

I I

-Si-0-SiR3

+ HA

where A-Si is a reactive group (Cl-Si, CH,O-Si, HN-Si, etc.) and R is alkyl (most frequently CH3)or phenyl or, occasionally, if A is C1, a further C1 atom. For details of such reactions, see refs. [171-174, 85 and 1551. Most often hexamethyldisilazane is applied: 2 SiOH + (CH3),SiNHSi(CH3), --* 2 SiOSi(CHJ3 + NH, which, at high temperatures [85], in addition to silylation, might due to attack on the lattice by the liberated ammonia form fresh SiOH groups. However, this seems innocuous, as at such lattice points where the reaction SiOSi + NH, + SiOH + H2NSi might have taken place, in successive reactions, including water from condensation processes, excessive silylation might occur: SiNH2+ H20 + SiOH + NH3 and 2 SiOH would react as above to form 2 SiOSi(CH3),. Hexamethyldisilazane, of course, can also react with water to form hexamethyldisiloxane: (CH3)3SiNHSi(CH3)3 + H20 + (CHJ3SiOSi(CHJ3 + NH3 but this phenomenon is not of great consequence because of the excess of the reagent, unless the amount of ammonia formed is to be measured for elucidating the silylation rate. In lieu of hexamethyldisilazane, its phenyl and fluoroalkyl substitution products, R3SiNHSiR3 (R = CH,, CF3CH2CH2,CsHS),have been applied, which, when using modified conditions, also deactivate well [155, 156, 1751and provide a wettable surface for weakly and medium polar phases, as they increase the surface energy of the column walls [compare eqn.(l21)]. If organic compounds are to be silylated, often a mixture of hexamethyldisilazane and trime-

61

3.3. Open-Tubular Columns

thylchlorosilane is used [173] because of its higher silylation power. This is also true for silylating surfaces under mild conditions. However, such conditions are not sufficient for good deactivation, and high-temperature silylation should rather be used as the presence of additional trimethylchlorosilane is not necessary. Chlorosilanes themselves react with silanols, the reaction rate depending on the number of chlorine atoms in the molecule: CH3SiCI3 > (CH3)2SiC12 % (CH3)3SiC1. Monofunctional chlorosilanes R3SiCI (R = alkyl, phenyl, most often methyl) react according to

I I

I + C1SiR3 + -Si-O-SiR3

-SOH

+ HCl

I

the liberated hydrogen chloride being suspected of affecting the silica lattice. With dimethyldichlorosilane, one reaction course corresponds to the above-mentioned equation:

c1 I I I --SiOH + C12Si(CH3)*4 -Si-OSi(CH3)2 I I

+ HCI

one reactive Sic1 group surviving and being capable of consecutive undesirable reactions. However, as dimethyldichlorosilane is a difunctional chlorosilane, it may react with two properly spaced silanol groups: OH

I

-Si-Si

I

OH

I

1

CI

-+

\

Si

VH3

0-Si-0

/

__c

/ \

CH3

- II

CI

I

CH3

I

I

CH3 -

I1

1

+ 2 HCI

I

This type of reaction is also applicable to a trifunctional chlorosilane, but here unreacted C1 atoms will survive to a greater extent: I

CI

I

-Si-0-Si-CH? I I CI

and

CI I 0-Si-0

I

I

CH3

I

and have to be removed, as with dimethyldichlorosilane, by an additional reaction with moisture or methanol (producing new SiOH groups or hydrolyzable SiOCH3groups in the case of methanol). Regarding these restrictions, the following procedure seems more promising [155, 1611. After the leaching step (for opening of a maximum number of silanol groups) the column is thoroughly dried in order to remove traces of water (but without condensing silanol groups!) and rinsed with hexamethyldisilazane-diphenyltetramethyldisilazane(1:1) in two parts by volume of pentane, subjected to a vacuum, flame sealed and heated at 400°C for ca. 12 h. With fused-silica glass columns, the oven has to be purged with nitrogen in order to protect the polyimide that covers the outer walls of the column. One sealed column end is opened under toluene, which is allowed to fill some of the coils, then toluene is replaced with methanol and finally with diethyl ether. The amount of reacted reagent, recognizable by the ammonia liberated, is determined by measuring that part of the column which remains un-

3. The Chromatographic Column

62

filled when the pressure has equilibrated (as mentioned earlier, this only applies if water, which also could have reacted, can be excluded). According to Grob et al. [155], the proper degree of silylation has been reached when a third to a half of the column is filled with NH3, which prevents it filling with solvent. Further phenyl-substituted silylating reagents were used by Grob and Grob [176]. Deactivation can also be achieved using organosiloxanes. Schomburg et al. [177, 1781 coated dynamically the leached glass column with a poly(dimethylsiloxane),filled it with inert gas, sealed it and heated it at 450°C for 2-20 h. The degradation products formed under these severe conditions were considered to react with surface silanol groups or be adsorbed. The column was then opened and subjected to solvent extraction to remove that part of the decomposed polysiloxane which was not bound. It is well known in silicone chemistry that by thermal degradation of poly(dimethylsiloxanes), in addition to linear fragments of varying chain length, oligomeric cyclosiloxanes (hexamethylcyclotrisiloxane,octamethylcyclotetrasiloxane, and in decreasing amounts the higher homologues) are the main products that are formed. It can be assumed, and this would be analogous to ref. [179], where cyclosiloxanes were applied for deactivation of silanols on capillary walls from the start, that at elevated temperatures the cyclosiloxanes react with the silanols via ring opening of the cyclosiloxane and substitution of the reactive SiOH proton:

,

IjI.." P-EjiR2

- si -0

I

.... SiRZ I

0

I

0 -SiR2

--

I

-Si-O-SiRZ

I

-0SiRZOSiR20H

on the surface. These groups exhibit less active sites and, on the other hand, create a better wettable surface for non- or weakly polar stationary phases, and are thermally stable with regard to chromatographic requirements. Similar ways of deactivating the inside wall and simultaneously improving the wettability have been the subject of recently published and current investigations [180- 1851. Murkides et al. [180] treated fused silica columns (0.20 mm i.d.) as follows. A plug of halfconcentrated hydrochloric acid ( ~ 1 8 % HC1 w/w) (10% of the column length) was forced through the tube by nitrogen until it had left the end of the tube. Both ends were carefully sealed and the capillary was heated at 140°C. The water may substantially increase the number of silanol groups on the surface and hence increase the reaction between these groups and functional groups of the consecutively applied organosiloxanes, being silanol-terminated or having become silanol-terminated by ring-opening reactions. The acid was displaced by one capillary length of dilute HCl (PH3), followed by one capillary length of methanol. Dehydration was carried out at 260°C for 5 h under a slow stream of nitrogen. The column was then dynamically coated (for deactivation!) (for true coating procedures see below) with 20% (w/v) solution of bis(cyanopropy1)cyclotetrasiloxane in CH2C12at a constant rate of 20 mm/s in order to cover the silica walls with an even film. After evaporation of the solvent (6 h), the tube ends were carefully sealed under vacuum, preventing any trace of air from entering the column. Chemical modification was achieved at 395°C (the outer surface of the fused-silica tube has to be taken into account, however, when choosing the temperature, as with a polyimide outside coating immediately after the drawing process its thermal stability is limited to ca. 340°C unless a protective gas (N,) is purged through the oven) for 2 h, the oven being heated and cooled slowly (5 Wmin). The column end was opened under methylene chloride, which was used to rinse the column free from excess of cyclosiloxane. The column was dried and was then ready for the true coating procedure with the stationary phase. In lieu of cyclosiloxanes, which have to undergo ring opening during the treatment prior to the reaction with surface silanols, silanol-terminated organosiloxanes can be directly applied for covalent bonding to the inner walls of the column.

3.3. Open-Tubular Columns

63

Verzele et al. [1811 synthesized an a,w-oligo(methylpheny1siloxane)diol CH3

I

HO-(Si-O)nH

I

C6H5

and deactivated the tube by dynamically coating it with a 0.3% solution of the diol in CHzC12 at a speed of 2 cm/s. The capillary was evacuated, sealed and heated to 300°C at 3 K/min and held at this temperature for 15 h. Excess of reagent was removed as above [180]. Even with untreated fused-silica glass tubing exhibiting only relatively few silanol moieties on the surface, specially prepared non-polar, high-molecular-weight silanol-terminated polysiloxane stationary phases can be bonded covalently to the walls [182, 1831. Untreated fusedsilica glass columns were purged with nitrogen at room temperature before use to remove residual HC1 (unreacted Sic& present in the various procedures used in the drawing of narrow-bore tubing above 2000°C is converted by moisture into HCl, which clings to the inner surfaces of the tube unless preferentially removed [183]. For a 25-m column, all traces of hydrogen chloride disappeared, as indicated by moist pH paper, in ca. 5 min. Non-polar silanol-terminated poly(dimethylsi1oxanes) or weakly polar poly(methylphenylsi1oxane) mixtures with poly(dimethylsi1oxane) (5% phenyl), partially cross-linked by heat before use, were dissolved (30% in CHzC12for dynamic coating or 0.3-1.2% in pentane or CHzC12for statical coating). Traces of solvent were subsequently removed by gently purging the column with nitrogen for 2-3 h at 60°C and, with N2still flowing, the column was allowed to cool to ambient temperature. The ends were placed under vacuum for 30 min and then carefully sealed. The temperature of the column was slowly increased at 2-4"C/min to 370°C and maintained at this temperature for 5-15 h. After cooling, the seals were broken and the column was again purged with nitrogen for 1 h and rinsed with methylene chloride and pentane. The polymers, contacted directly to the untreated surface of the fused-silica capillary tubing at these high temperatures, readily deactivate and wet the column inner surface by distributing themselves as a homogeneous film throughout the whole length. Following the extraction with methylene chloride and or pentane, 85-95 % of the film appeared to be cross-linked in the case of partially gummified polymers prior to coating; if not previously cross-linked, only 50-60% of the condensed film was immobilized. From numerous results (e.g., refs.[177-186]), it can be concluded that no single pre-treatment can be used for the preparation of the surface even of as a unique glass as fused-silica which would enable one to coat it effectively with stationary phases of any polarity. Thus, surface pre-treatment has to be selected carefully, depending on the stationary phase to be employed. A step forward in this direction is the surface deactivation of untreated fused-silica glass with non-polar and polar prepolymers having various chemical compositions and subsequent cross-linking of the prepolymer to obtain the proper stationary phase film. 3.3.3.2.

Coating Procedures

As already emphasized in Section 3.1.3, a detailed discussion of coating procedures is not given in this book; only summaries of the principles and brief mentions of their pros and cons are included. For more details the reader is referred once again to handbooks (e.g., ref. [187]) and to the original papers cited. A practicable procedure is the dynamic technique introduced by Dijkstra and Goey [40]. The stationary phase is dissolved in a suitable, carefully purified solvent to give a concentration of 5-30% w/w. The solution is poured into a coating reservoir according to Schomburg and Husmann [188] together with mercury in order to produce reproducibly homogeneous films.

3. The Chromatographic Column

64

The coating device is connected with the tube, and by means of nitrogen, the flow-rate of which is controlled by a precision pressure regulator, a plug (10-30 cm long) of the impregnation solution, followed immediately by a mercury plug (3-15 cm long which corresponds to 0.02-0.11 g of mercury at a tube i.d. of 0.25 mm), is forced through the tubing at a velocity of 1-2 cm * s-*. The nitrogen flow is continued, even if both plugs have left the column, in order to remove the solvent; if the boiling point of the solvent is not too high, purging with nitrogen for 3 h at 60°C will be sufficient. For details of this method, see refs. [187-1911. The advantage of this method is that it is easily and quickly practicable. The disadvantages are that: The amount of the stationary phase in the column cannot be determined exactly; deviations of the inside diameter of the tube result in variations of the film thickness; and the efficiencies are lower than with static procedures. The static vacuum method first applied by Bouchd and Venele [191] consists in slowly vaporizing the solvent from the stationary phase solution in the column. The dilute solution (5l%),prepared with highly purified solvent, is degassed and sucked into the column. After complete filling, one column end has to be sealed, avoiding any air bubbles, e.g., by means of a mechanical device [192], and the column is placed in a suitable thermostat the temperature of which depends on the solvent. This is volatilized slowly, cautiously maintaining a vacuum. When the volatilization has finished, the sealed end is opened and traces of solvent and residues are removed by an inert gas stream. For details, see refs. [192-1981. The advantages of this method are that it is especially applicable to high-viscosity stationary phases, high-efficiency columns are produced; and the amount of stationary phase and the film thickness can be calculated. The disadvantage is that it is time consuming. The static evaporation procedure was introduced by Golay [20]. As above, the impregnation solution is sucked into the column, carefully avoiding evaporation of the solvent. After having sealed one end, the column, starting with the open end, is moved through a relatively hot zone into a thermostat, thus avoiding re-condensation of the already vaporized solvent. In the hot region the stationary phase film develops. Several parameters (velocity of the column movement through the heated zone, the applied temperature etc.) have to be optimized. For details, see refs. [20, 187 and 199-2031. The advantages of this method are that the amount of stationary phase in the column can be measured and the film thickness easily be calculated; viscous liquids can be coated; and it is not time consuming. The disadvantage is that occasionally inhomogeneous films are formed with low viscosity stationary phases (owing to non-uniform evaporation of the solvent).

3.3.3.3.

Bonding and Cross-Linking of Stationary Phases

Recent advances in bonding phases on to the inside walls of fused-silica glass columns and in immobilizing stationary phases in fused-silica glass columns have led to enhanced stationary phase stability. Owing to its outstanding advantages compared with the usual coated open-tubular columns, in all likelihood this method will, in conjunction with fused-silica glass column material, substantially extend the application of WCOT columns to increasing numbers of laboratories and fields. This is why this procedure, although still in mid-development, will be discussed here in detail. Let us first define the suitable chemical reactions taking place (a) on the wall surface between the surface silanol groups or strained siloxane bridges and appropriate stationary phase reactants, and (b) within the stationary phase itself prior to, during or after coating. Type (a), which we have already dealt with in the previous section, is the actual bonding reaction, which can be assumed to proceed according to one of the following simplified schemes:

65

3.3. Open-Tubular Columns

-SiOH

I

+

Si - R

HO-

I

-SiOH

I

+

CI

I

- Si-R I

--

I

SiOH + H - 3

I

I

1

I

+(R2SiO)nt)

R

18

+ HCI

I I

-R

-

R

I

I

+ H2

-SiOSiR

R

-SiOH

H20

R

R

I

I

-SiOSiR

R

I

+

-5iOSiR

R R

-

-

I I

R

-

followed by ++I

R = C 3 z n + 1 , C ~ H> SClzHg, CnH7 CF3CHzCH2 CN(CHz)n, OH 9

9

*) cyclic Siloxanes **) thus linking two surface silanol groups, which otherwise would for steric reasons not react with them-

selves, via a short diorganosiloxane chain [small number n of D-units (R2SiO)].

\SI "'Si

/

\o/

/'

\

+

HOSiR,

-

OH

OSlR3

\ ISI - 0 - 5 - '

/ \

/

(highly strained)

\s/o\si/ H20 \o/ \

/

+

OH

OH

\ I si -0 - si I/ /

\

/I

(being reactive os above shown) Si R2

66

3. The Chromatographic Column

PH

(being reactive as above shown) *) Cyclic siloxane

Type (b) is a reaction taking place in the liquid stationary phase. Several terms have been used, e.g., curing, vulcanization, immobilization, cross-linking, in situ cross-linking, polymerization and condensation of prepolymers, free radical cross-linking, auto cross-linking, gummification, in situ vulcanization and production of non-extractable (or insoluble) films. Some terms stem from polymer chemistry, others from elastomer and rubber technology and two can be traced back to gas chromatography. In this book, we prefer in general to use those terms that describe the chemical reaction (cross-linking, polymerization, condensation) if the reaction itself is dicussed. The main reasons for the extensive investigations in this field are that on glass and fused-silica glass walls in open-tubular columns, column deterioration was often observed as a consequence of a physical rearrangement of the stationary phase (formation of droplets) at elevated temperatures and that the wettability of the surface with polar stationary phases, if any, was considerably obstructed. The first attempts to bond the phase to the surface in combination with cross-linking were carried out by Grob [165], Bossurf [203] and Mudani [204] and soon afterwards by Blornberg and co-workers [205-2071. So far, bonding and cross-linking has chiefly been restricted to silicone-type phases. This is based on the fact that these phases have been most widely used in gas chromatography because of their numerous merits, which will be discussed later in this book, and not least by the possibilities that they offer for tailormaking stationary phases. Bonding reactions have already been discussed and will only be described here when experimental details prior to the cross-linking reaction have to be given. Increasing the viscosity (i.e., increasing the chain length in the case of linear polymers) has been shown to improve film formation and film stability of liquid stationary phases [208], and polysiloxane gum phases with long chains have been synthesized, thus increasing the column efficiencies of coated fused-silica open-tubular columns [209-2 111. In spite of these improvements, the bulk phases are not yet immobilized, even when partially bonded to the surface. By introducing a slight degree of cross-linking, however, in the stationary phase itself, better immobilization can be achieved [205-207, 1211. Occasionally, a,o-(polysiloxane)diols have been partially bonded to the surface or/and partially coated and subsequently heat-cured after addition of a catalyst, e.g., tetramethylammonium hydroxide. Such coatings are less extractable, but the reaction having taken place is a polycondensation, leading to high-molecular-weight materials, and is not a desired cross-linking reaction. Cross-linking is a decisive step for immobilization. With the mentioned siloxanediols, tri- or tetrafunctional silanes, e.g., RSi(OR)3 or Si(OR)4, have to achieve a true cross-linking reaction (184, 185, 204-207) by condensation of hydroxy and, e.g., alkoxy groups with elimination of water, alcohol or ether and formation of Si-0-Si cross-links.

67

3.3. Open-Tubular Columns

Even if the thermal stability, due to the stability of the Si-0-Si bond, is outstanding, this type of cross-linked stationary phase has crucial disadvantages. High cross-linking levels, which give rise to diminished solubilities of sample molecules, and an undesirable activity of the column, are caused by residual SiOH and SiOCzHScontent [213].Further, residual alkoxy groups are susceptible to hydrolysis. Hence it is much more promising at present to use free radical generators to form carbon- carbon cross-links, as very little cross-linking (0.1-1%) is necessary to change long polymeric chains into insoluble rubbers, hence requiring only low levels of cross-linking agents [213].Free radicals can be generated by decomposition of organic peroxides, of azo compounds or by the use of high-energy radiation. The linking occurs according to the following scheme in the case of pure poly(dimethylsi1oxanes) (R' = radical):

)

R' R'

I

I

s

CH3- Si - CH3

T

)

2R'

CH3 -5- CH3 -CH3 I

P

)

-Si -CH2

I

>

- CH2-Si

I

+2 R H

-CH3

Groups other than methyl present in the poly(diorganosi1oxane)molecule have a decisive effect on the cross-linking rate and the nature of the cross-linked product. If vinyl groups are present (in addition to the bulk of methyl groups), the methyl-to-methyl cross-linking is no longer dominant, but rather methyl-to-vinyl cross-linking [213,2141:

1

CH3-Si-CH I

R'

= CH2

2

R'

2

}

I

I

CH3-Si -CH~-+CH3-Si-CH2-CH~-C~-Si-CH3 T

I

P

P

+ R'

P

Because of the greater tendency of vinyl groups to cross-link, as is well known in silicone chemistry [215],lower levels of free radical generators are required, in order to achieve the same or even a higher degree of cross-linking. Numerous free radical generators have been used for the initiation of cross-linking. First, peroxides are most effective in forming insoluble stationary phases. Benzoyl peroxide, bis (2,4-dichlorobenzoyl) peroxide, dicumyl peroxide and di-tert.-butyl peroxide have, for example, been applied, of which dicumyl peroxide can be considered to give decomposition products that affect the column stability less than the other peroxides, and hence is the most often used. It would be beyond the scope of this book to review the numerous investigations in this area and therefore only a few bibliographic details are cited [210,212, 213, 216-2311. Second, in situ generation of free radicals can alternatively be accomplished by using azo compounds, e.g., azo-tert.-butane, azoisobutyronitrile, or azo-tert.-dodecane. Owing to its main decomposition products, nitrogen, isobutane and isobutene, which are non-polar and do not react with the polysiloxane chain, the azo-tert.-butane, in contrast to peroxides, cannot form acidic products which would attack the column. Unfortunately, it is less reactive than the peroxides and requires higher temperatures to obtain acceptable free radical generation rates, and hence comparable cross-linking [213].Nevertheless, apart from dicumyl peroxide, azo compounds can be used; for details see refs. [182,211, 213, 227 and 232-2371. A third procedure for cross-linking silicone, especially polar silicone, stationary phases consists in applying radiation, either gamma radiation from a 6oCosource [227,238, 2391 or accelerated electrons from a Van de Graaff generator [240].It offers the advantages that no chemicals with harmful effects have to be added, that the reaction takes place at ambient tempera-

68

3. The Chromatographic Column

ture and that the columns may be tested before cross-linking. Owing to the expansive equipment required, its application will be limited, however. More readily available is the equipment for a fourth method, in situ cross-linking with ozone [241, 2421. Immobilization can be achieved at low temperatures (ambient for dimethylsiloxanes) or medium temperatures (150°C for phenyl- or cyanopropyl-substituted siloxanes). Nevertheless, we feel that this reaction type should be treated with caution. Even if silicones have proved to be relatively resistant towards oxidation, the presence of ozone must have some effect. The absence of an infrared carbonyl band, or at least only a very weak absorption, observed after curing, [242], may not be admissible evidence, nor are the marginal differences in Kovhts retention indices [242]. Naturally, this also applies to peroxide-initiated cross-linking. More thorough investigations need to be carried out if small alterations, which could affect the analysis of labile or trace compounds, are to be recognized. A few cross-linking procedures will now be given in detail.

Procedure (a), according to Wright, Peaden and Lee [213]. Deactivation consists in rinsing the fused-silica capillary tubing with 5-10 ml of methanol at room temperature, then purging with Nz for several hours to remove any traces of methanol. Next, octamethylcyclotetrasiloxane (D4 in organosilicon chemistry nomenclature) is dynamically coated by filling ca. 20% of the column and then rapidly pushing the D4 plug through the column with Nz until it has been expelled. After sealing both ends, the column is heated for 2 h at 420°C (whilst protecting the polyimide outer coating with N2!),and subsequently the column is purged with Nz for about 30 min at 350°C to remove any residual D4. The column is statically coated. The long-chain poly(dimethylsi1oxane) containing vinyl groups is dissolved in purified pentane and the concentration is selected to give a film thickness of 0.1-0.5 pm. Solid dicumyl peroxide, dissolved in methylene chloride (l%),is added to the coating solution 30 min prior to coating to give a concentration of 0.28%.After the usual coating the column is purged with nitrogen, sealed and temperature-programed from 40 to 175°C at 4 W rnin and held at 175°C for 15 min. After cross-linking, the column is washed for 30-60 rnin with 5-10 ml of methylene chloride. Subsequently it is conditioned with a slow carrier gas flow for 1h at 40°C to desorb any residual solvent from the stationary phase left from the washing procedure and then temperature programmed to 260°C at O.S"C/min and held there for 8 h. When using azo-tert.-butane, which is a liquid at room temperature, the column is coated in the usual way, without the free radical generator prior to cross-linking. The coating is followed by saturation of tke stationary phase with the vapour of azo-tert.-butane by bubbling nitrogen through the azo compound and purging the coated column at 40°C for 2 h. Subsequently, for dynamic curing the column is attached to an argon manifold in an oven and heated at 5-1O"Umin to 220°C and maintained there for 15 min; the argon linear velocity is 10 c d s . After curing, the column is rinsed with 10-25 column volumes of methylene chloride-acetone (5050,v/v) and then conditioned with a rapid carrier flow for 30-60min at room temperature. After being reconnected to the argon manifold, the column is heated at 5"C/min to 350°C and held there for 4 h with an argon velocity of 25-30 cm/s. Procedure (b), according to Lee and co-workers [210, 2361. First, a mixture of dimethyldichlorosilane, methylphenyldichlorosilane, diphenyldichlorosilane, methylvinyldichlorosilane and 1,4-dimethyl-1,1,4,4-tetrachlorodisilethyleneis prepared for synthesizing a polymer with the desired phenyl and vinyl contents. Hydrolysis of the mixture is accomplished by dissolution in an equal volume of acetonitrile and adding an equal volume of water, followed by an equal volume of methylene chloride. The methylene chloride layer is then extracted with distilled water to neutrality. The methylene chloride is finally removed by gentle warming under a nitrogen purge. Polymerization

3.3. Open-Tubular Columns

69

is accomplished by adding 0.05 wt-% of the catalyst tetramethylammonium hydroxide and heating at 110-130°C until the viscosity of the polymer ceased to change. The catalyst is then rendered inactive and the polymer is end-capped by the addition of trimethylchlorosilane. As the polymerized product contains both low- and high-molecular-weight moieties, fractionations have to be performed in order to remove the lower-molecular-weight materials. This is achieved by dissolving the polymer mixture in four times of its volume of CHzClzand adding an equal volume of methanol, which causes the higher-molecular-weight moiety to precipitate. After repeating this precipitation four times, the final precipitate is collected. In case of a 50% phenyl-, 49% methyl-, 1%vinyl-polysiloxane, the solvent used to prepare the coating solution is ratio n-pentane:methylene chloride 2:1, v/v. An appropriate amount of the polymer is dissolved to give the desired film thickness. The column is coated statically. To give a film thickness of 0.25 pm in a 15-m column of 0.1 mm I.D. in a reasonable time, the coating temperature can be chosen to be as low as 28°C. After coating, the column is purged with nitrogen for 30 min at room temperature. For cross-linking, azo-tert.-butane is purged through the column for 2.2 h using a special purging device [236]. Cross-linking twice, each time after purging azo-tert.-butane from opposite ends of the column, increases the successful cross-linking rate. After azo-tert.-butane purging, the column is sealed and temperature programmed from 40 to 220°C at 4"C/min and held at 220°C for 1 h for cross-linking. After cross-linking twice, the column is normally temperature programmed from 40 to 250°C at 0.5 Wmin under a gas stream (N2or He) and held at 250°C overnight. The finished column is then evaluated as usual and subsequently washed with 50-100ml of methylene chloride by means of an HPLC pump. Washout data can be obtained from the difference in k values before and after washing. Using this procedure, nearly 100%non-extractability can be achieved. Procedure (c), according to Venele et a1 [181]. Although not a cross-linking reaction, the following procedure is described here for two reasons. First, because free radical generators are avoided, decomposition products cannot occur and hence cannot give rise to column activity. Second, the polarity and selectivity are well defined and equal to those of, e.g., OV-17, because the prepolymer does not contain any vinyl or tolyl groups, which otherwise are necessary to achieve cross-linking of phenylsiloxanes initiated by free radical generators. Further, it is easily practicable. A prepolymer 1, synthesized by alkaline hydrolysis of methylphenyldichlorosilane, is used to deactivate the previously leached fused silica column. Deactivation is achieved by dynamically coating the column with a 0.3% solution of prepolymer 1 in CH2C12at a velocity of 2 cm/s. The column is then evacuated, sealed and heated to 300°C at 3 K/min and held isothermally at 300°C for 15 h. Excess of reagent is removed by rinsing the column with 10 ml of CHzClz.Prepolymer 1 is further polymerized to a semi-gum, called prepolymer 2, by heat-curing under nitrogen whilst continuously controlling the viscosity to avoid complete gummification and loss of solubility. The deactivated column is then statically coated with a 0.1-0.3% (w/v) solution of prepolymer 2 in CHzClz.The silanol-terminated prepolymer 2, coated on the column wall, is immobilized by in situ gummification by heat-curing. This is achieved by heating the column repeatedly from 150 to 250°C at 3 K/min for 15 h under a low flow of carrier gas (0.1 ml/ min). The column is then rinsed with 5 ml of CHzClzand conditioned. The immobilization yield was 100%. Residual terminal silanol groups can be capped by hexamethyldisilazane at elevated temperatures. Procedure (d), according to Markides, Elomberg, Buijten and Wannman [223, 223al. Bis(cyanopropy1)dichlorosilane is hydrolyzed with acid catalysis to form octakis(cyanopr0py1)cyclotetrasiloxane [212], which is coated dynamically in a 20% (w/v) or 2-5% (w/v) [223a] solution in methylene chloride at a constant rate of 20 mm/s on the inner walls of a fused-silica column. Prior to the coating, the column has to be leached with 20% hydrochloric acid at

70

3. The Chromatographic Column

100°C for 12 h, rinsed and dried [224, 1551 as described in Section 3.3.3.1, as a fused-silica surface seems to contain too few silanol groups to be sufficiently modified by this cyclosiloxane. To avoid any contamination, leached glass vessels, high-purity chemicals and freshly prepared solutions should be used for the leaching procedure. The dynamic coating has to be performed carefully in order to form an even film along the length of the capillary, and no air or solvent residues should be present. During the dynamic coating, a buffer capillary is connected to the column end. Immediately after the coating plug has left the column, the flow of dry nitrogen passing through the column is drastically increased, and the evaporation of the solvent is allowed to proceed for 6 h. The capillary is then evacuated and both ends are sealed. The sealed capillary is heated in an oven to 395°C at 5 Wmin and held at the fmal temperature for 1.5 h, after which the oven is allowed to cool slowly to room temperature. One end of the column is opened under the surface of CHzClz, thus forming a plug with which the column is rinsed using nitrogen. Finally, the column is dried by flushing with dry nitrogen. Now, the column is modified for good wettability with polar cyanopropylsiloxane phases. Prepolymers with a high degree of cyan0 substitution are prepared from bis(cyanopr0pyl)dichlorosilane, methyl(toly1)dichlorosilaneand dimethyldichlorosilane by basic reversed hydrolysis (in methylene chloride solution, according to Patnode and Wilcock [223b]). A linear methyl(viny1)pentasiloxane is synthesized from methyl(viny1)cyclopentasiloxane by ring-opening with boiling butanol [243]; 1.7-3 mol-% of this vinyl siloxane is included in the reaction mixture to incorporate vinyl groups into the gums as methyl(viny1)siloxanesegments rather than single methyl(viny1)-siloxane units in order to facilitate cross-linking of gums containing bulky groups. Further, 0.5 mob% of 1,4-dimethyl-1,1,4,4,-tetrachlorodisilethylene is added [analogous to procedure (b)] in order to introduce a slight degree of crosslinking in the prepolymer. The synthesis is carried out in acetonitrile. Polymerization is carried out at 110°C for 10 min using ammonia solution as catalyst [215, 2441. The reaction is performed with stirring under nitrogen. The gum is dissolved and purified by washing with dilute hydrochloric acid, followed by water, to remove catalyst residues. Finally, after drying the gum with calcium sulphate, residual silanol groups are capped by reaction with 1,3-divinyltetramethyldisiloxanein refluxing acetonitrile for 6 h under an atmosphere of nitrogen. This polymer is dissolved in acetonitrile-diethyl ether (3:2), and 5% of dicumyl peroxide, calculated from the amount of stationary phase, is added to the coating solution. Before filling the column, the coating solution is filtered and centrifuged; only freshly prepared solutions are used. Coating is performed by the static method and with the column immersed in a horizontal position in a water-bath. The coated column is opened under an atmosphere of dry nitrogen, 1min before disconnecting the vacuum, and is then directly purged with dry nitrogen for 30 min. The gum phase is then cross-linked in situ by dynamic curing in a GC oven, programmed from 40 to 170°C at 5 Wmin, the final temperature being maintained for 40 min. During curing the column is rinsed with a slow stream of dry hydrogen (0.1 ml/min). After cross-linking, the column is rinsed with 5 ml of methylene chloride and conditioned in a gas chromatograph programmed to 250°C at 1Wmin. Cyanopropylsilicone rubber, cross-linked via Si-0-Si, can be obtained according to ref. [212], and trifluoropropylsilicone rubber with cross-links of the type Si-C-C-Si according to ref. [220]. Conclusions concerning bonding and cross-linking of stationary phases With the development of immobilized phases by bonding and in situ cross-linking reactions during the past 5 years, the overall quality of WCOT columns has been vastly increased. The following advantages have become apparent: enhanced film stability involving long column lifetimes and high efficiencies of both non-polar and polar stationary phases on the col-

3.3. Open-Tubular Columns

71

umn wall; minimal phase stripping from sample injection solvents, thus allowing the injection of large amounts of sample, including samples dissolved in both non-polar and polar solvents, even in water; the possibility of washing the columns to remove non-volatile compound deposits, provided these are soluble; higher film thicknesses while maintaining stable films can be chosen; in spite of cross-linking, no significant differences in diffusion coefficients compared with conventional stationary phases [245] have been observed; cross-linking suppresses the glass transition mechanism of poly(dimethylsiloxanes), thus allowing efficient separations of, e.g., CI-C5 hydrocarbons at sub-ambient temperatures, even at - 70°C; and reduction of column bleeding can generally be observed as a consequence of cross-linking (including conditioning) and end-capping [which converts terminal silanols (which would give rise to degradation reactions) into innocuous SiOSiR,-groups, in the case of silicone stationary phases]. There are also a few disadvantages: temporarily, the immobilization is, with the exception of poly(ethy1ene oxides), restricted to silicone phases at present, but as different functional groups can easily be attached to the siloxane skeleton (C6H5,CN(CH2),, CF3(CH&, C6HS, C6H4,etc.), a wide range of selectivity can be achieved; cross-linking of polar silicone phases is more difficult than cross-linking of poly(dimethylsi1oxanes) and generally requires additional vinyl or tolyl groups and higher amounts of the free radical generator, e.g., peroxides; and decomposition of the cross-linking initiators, especially of peroxides, may cause undesirable column activity. In spite of these disadvantages, it can once again be concluded that fused-silica glass capillaries of different diameters, containing bonded and/or cross-linked tailor-made stationary phases, of different film thicknesses will increase the applicability of WCOT columns tremendously, at the expense of packed columns. 3.3.3.4.

Film Thickness of the Stationary Phase

We have seen in eqns. (62) and (63) that it is preferable to choose a stationary phase film that is as thin as possible, as both the C, and C, terms, i.e., terms describing the resistance to mass transfer from and in the stationary phase, respectively, depend on the stationary film thickness, d , . The first term is proportional to d : , and the second term is proportional to d , . Hence a reduction in d, will result in a decrease in C, and, owing to the square dependence, a greater decrease in C,, hence reducing hmin,i.e., increasing the column efficiency. However, this is only one side of the picture. On the other side we have to take into account, that a reduction in d, leads to an increase in the phase ratio, B, as

p , 2Ldl

(119)

and a higher B value means, according to eqn. (24) ( B = K L / k ) ,a smaller k value (because KL is a constant). A smaller k, in turn, would require, corresponding to eqn. (99), more theoretical plates to achieve the same resolution. This would be especially serious when analysing low-boiling compounds, which exhibit very small capacity ratio values. Therefore, we have to reach a compromise depending on the tube material (which might be active!), on the column temperature (which can be reduced when the film thickness is reduced) and on the physical and chemical properties of the sample components. If we return to the phase ratio, B = VG/VL [eqn. (23)], it is clear, that it can easily be adjusted by changing either the film thickness (which is contained in VJ or the column diameter (expressed as V,). Volatile compounds, for example, can well be separated in columns with an increased film thickness of up to 5 pm, compensating for the lower separation efficiency by an increase in

72

3. The ChromatographicColumn

column length, whereas the separation of high-boiling compenents can be accomplished on shorter columns with relatively thin stationary phase films of dl = 0.1 pm [2471. Although 5 and with film hitherto the exception, open-tubular columns of inside diameter ~ 0 . mm thicknesses of d, = 2 to 7 pm, rendered possible by cross-linking, can be prepared, thus permitting the injection of larger sample amounts but nevertheless maintaining better performances than with packed columns, especially shorter analysis times, lower adsorption of polar compounds and higher thermal stability [248, 2491. The lower limit of the film thickness may be 0.1 pm; the usual range on glass or fused-silica glass walls is between 0.2 and 0.3 pm and on stainless steel walls owing to their activity, between 0.5 and 0.6 pm. Using the static coating method, the film thickness can easily be calculated from dl = 5r,c

(122)

where

dl = film thickness [pm], r, = column radius [mm] and c = concentration of the stationary liquid phase in the coating solution [%, w/w], as all of the stationary phase being contained in the coating solution will remain in the colUmn. If the column is to be coated dynamically, the film thickness may be predicted by one of the following equations (Fairbrother and Stubbs [250]):

dl = rcc

where

4 4 + tl/n

d, = film thickness [pm], r, = column radius [mm], u, = coating velocity [mm/s], 9 = viscosity of the coating solution

Pa. s],

[

yc = surface tension of the coating solution 10-

c

= concentration

4

of the stationary liquid phase in the coating solution [%, w/w] (Nouotny et al. [251])

or (Guiochon [252])

4=

134r c

u,q 3 ( 7 T .)

The best accord with experimental results is given by eqn. (124), except with very thin films or very low viscosities of the coating solution, where eqn. (125) is to be preferred (2521. The film thickness can, however, also be determined from gas chromatographic data (Crumers et al. [245]): d-I-

where

kr, 273 2YBeL T

k = capacity ratio, V, = specific retention volume of a solute at the column temperature T [K], eL = density of the liquid stationary phase at the column temperature.

13

3.3. Open-Tubular Columns

The values of r, and eL are known, T and k can be measured and V, can be obtained advantageously from, e.g., corresponding data from packed columns.

3.3.3.5.

Quality Tests of WCOT Columns

In addition to the usual determination of the values of k [eqn. (21)], B [eqns.(23) and (24)], n or N [eqns.(36), (37), (40) and (41)], the following characterizations of the column ought to be carried out: (a) Determination of hdn (smallest experimental h value by plotting h =f(@ at Copt.(see Fig. 4). (b) Calculation of the coating efficiency, i.e., the ratio of the theoretical and experimental minimum height equivalent to a theoretical plate [192]: CE =

hmh (theor.)

(exp.) using the k value of a standard compound, the selection of which depends on the operating temperature. hmin(theor.) can be calculated from hmin

hmin(theor.) = r,

JW,

r, = internal column radius, k = capacity ratio.

Hence it follows that

or, if CE is to be given as a percentage, (128a) Instead of CE, the term UTE (utilization of the theoretical best efficiency) has also been used, see, e.g., ref. [223]. At this point it should be noted, that eqn. (127) is a simplified expression, neglecting both the pressure drop in the column and C, [eqns. (62) and (63)]. Provided that the dif€usion coefficients are known, an expression developed by Cramers et al. [253] can be utilized, which has been proved to correspond better with the experimental data [252]. (c) Determination of the loading, i.e., of the amount of sample that can be injected without a serious loss of efficiency. An approximation is [254] SL [g] = 0.05 M .$(1

where

+ k)

*

(129)

SL = sample loading = molecular weight of the standard compound, d, = column inside diameter, k = capacity ratio of the standard compound, which should be chosen such as to give k > 10.

M

A simple experimental means of obtaining the maximum solute load consists in injecting increasing amounts of sample and measuring the peak width at half-height until the value of

74

3. The ChromatographicColumn

wH [eqn.(33)] has increased to 5% above the average wH value. This value is the maximum solute load (MSL). (d) Determination of undesirable column activity. A serious problem, especially in the trace analysis of polar compounds, has been the possible residual column activity. The column therefore has to be checked by an appropriate test method. One must be aware of the fact, however, that deleterious effects might be caused by the injector, injector sleeves, the tube connectors, etc., unless these parts of the instrument have been properly deactivated. The inertness of a column can be tested by two approaches. Reversible adsorption can be observed by peak tailing of selected compounds having special functional groups. This tailing can be quantitied by an asymmetry factor [255]: A, =

where

a+b

(a + b ) - ( b - a )

factor, half of the peak, measured from the perpendicular drawn through the peak maximum, b = back half of the peak. A totally symmetrical peak would have, according to this equation, an A, value of 1.00, and tailing on the rear edge, being an indication of reversible adsorption would give asymmetry factors greater than 1.00. Irreversible adsorption and partial decomposition are even more detrimental when carrying out trace analyses. This effect, which would falsify the analytical results, can be detected by comparing the ratios of the peak areas of selected compounds with that of a hydrocarbon contained in the test mixture, as compounds being subject to such effects would exhibit reduced peak areas. This method holds true, however, only for the investigated concentration range. Hence it is recommended that as small amounts of the test mixture as possible are injected as at higher concentrations the effects might become less evident. The composition of the test mixture can be chosen according to Table 4. A, (I

= asymmetry = front

Table 4. Standard Test Mixtures Schombug et al. 1178.2581

Rohrschneider I2561

MeReynolds

benzene ethanol methyl ethyl ketone nitromethane pyridine

benzene n-Clo, n-Cll, 1-butan01 n-C12alkanes 2-pentanone G o - , c11-, nitropropane C12-acid methyl pyridine esters 2-methyl-2-pen- 1-octanol tan01 n-octylamine 1-iodobutane 2,6-dimethyl2-octyne aniline 1,4-dioxane dicyclohexylcis-hydrindane amine 2,6-dimethylphenol

12571

Gmb

[W ~~~~

n-Clo, n-CI1-al- n-Clo, nkanes C12-alkanes 1-0ctanol c10-9c11-, Clz-acids octanediol methyl esters octanoic acid 1-octanol trichlorophenol nonanal nitrophenol 2.3-butane-diol n-Cs-, n-Clo-, 2,6-dimethylani- n-Clz-amines line naphthalene 2,6-dimethylphe- biphenyl no1 dicyclohexylamine 2-ethylhexaneacid

n-Cs-C12-alkanes dibutyl ketone nonanal nonylamine nitrohexane 4-tert-butylpyridine nonanol naphthalene 4-propylphenol

3.3. Open-Tubular Columns

75

Owing to different causes of adsorption, it is necessary, as can be seen from Table 4, to test compounds with different structures. In any event, an m i n e ought to be present in the test mixture, because amines are very sensitive indicators of adsorption effects. Pyridine, used in the test mixtures of Rohrschneider and McReynolds (Table 4), seems to be less suitable. This is not surprising, as both of these test mixtures were developed for other reasons (see Chapter 4) and not to evaluate adsorption. Nevertheless, both of these mixtures have also been applied for this purpose occasionally, because they are necessary for determining the column selectivity, and dominant adsorption phenomena may occur. (e) Long-Term Stability Test Depending on numerous parameters the column can be subjected to changes during the operation. Therefore, several checks ought to be carried out at suitable intervals, e.g., once per week. An easily measured value is the capacity ratio, k, of a standard, which would give information on eventual losses of the stationary phase. Occasionally the Kovhts retention indices (see Chapter 4) of the Rohrschneider or McReynolds test mixture ought to be measured. Changes in the retention index of any of the components would indicate changes in column selectivity. Finally, any increase in adsorption effects can be realized by injecting a standard mixture according to (d) (Table 4).

3.3.4.

Porous-Layer Open Tubular (PLOT) and Support-Coated Open-Tubular (SCOT) Columns

In order to increase the sample capacity of open-tubular columns while decreasing the stationary phase film thickness, porous layers were first prepared on the inside walls of the column tubing by Halasr and Horuath [260, 2611 and thoroughly investigated and made commercially available by Ettre et al. [262, 2631. These porous layers increase the surface area of the original geometric area of the inner walls of the tubing, thus increasing the amount of stationary phase and with it the load of the column, in spite of the simultaneous reduction of the film thickness, in order to increase the efficiency according to eqn. (63) (decreasing both the C, and C, terms). Compared with WCOT columns, the phase ratio is much smaller, because the amount of stationary phase will be 10- to 50-fold higher (maintaining the high permeability and the efficiency of WCOT columns!). An additional advantage of this type of column is that the layer surface can be better coated than the normal wall surface, as the contact angle on the rough surface to be wetted by the liquid is smaller than that on smooth surfaces [compare eqn. (117)l. Porous layers can be applied to the inner tube wall by static coating with a suspension of particles of, e.g., graphitized carbon black [264, 264a], activated charcoal [265], modified silica [266], silanized silicic acid [268, 268a], molecular sieves or Chromosorb-based support materials [263, 2671, porous polyaromatic polymers [267a].Another possibility consists in the formation of porous layers direct from the column wall material by etching procedures (especially in glass capillaries), which will lead to silica gel-type layers [269, 2701, or to silica whiskers [158] (discussed in Section 3.3.3.1). The layer may be very thin, or thick, may be more or less active, and may be coated with a liquid stationary phase or may be used unmodified as an adsorbent, e.g., for the separation of non-or weakly polar low-boiling compounds. All these feasibilites exist when using the general term porous-layer open-tubular (PLOT) columns. In order to take into account the type and the geometric dimensions of the layer, we shall subdivide this type of column into subgroups: PLOT columns (strictly), SCOT columns and thick-layer columns; although this cannot be claimed to be scientifically exact, it is expedient for practical purposes.

76 3.3.4.1.

3. The Chromatographic Column

PLOT Columns (strictly)

Principle of preparation. Open tubes, most often having an inside diameter of 0.5 mm, are coated with adsorbent particles finely dispersed in a vaporizable liquid, which is subsequently evaporated so as to leave the particles as a thin porous layer on the column walls. Occasionally, the porous adsorbent layer can be modified by previously dissolving, e.g., a polar stationary liquid, in a dispersion, leaving, after evaporation of the solvent/dispersion liquid, a wet porous layer. The tube has first to be rinsed with purified solvents, e.g., in the order methylene chloride, methanol, acetone and R113 (CFZC1-CFClz),and subsequently filled with the solvent used for the dispersion (e.g., R113, carbon tetrachloride, chloroform or another vaporizable liquid of relatively high density). The particles generally have sizes of about 1 pm or less but larger particles have also yielded good efficiencies provided that the particle size distribution is narrow. Usually, adsorbents are applied, e.g., graphitized carbon black, activated charcoal, silica or porous polymers. They are added to the vaporizable dispersing liquid and dispersed by rapid stirring or sonication. The dispersion is forced through the tube by means of pressurized nitrogen, similar to the static coating procedure described in Section 3.3.3.2,avoiding the formation of gas bubbles. After being filled, one end of the tube is sealed, and the tube is slowly, beginning with the unsealed end, drawn or pushed through an oven, the temperature of which depends on both the tube length and the boiling point of the liquid to be vaporized. When the evaporation is completed, the column is purged with nitrogen to remove residual liquid. The particles adhere to the column walls by Van der Waals forces. The thickness of the layer depends on the particle size and structure, and generally ranges from 1 to 30 pm. If a wet, modified porous layer is to be prepared, the stationary liquid is added to the dispersion prior to stirring. This procedure for preparing porous layers allows a fairly unrestricted selection of particle size, layer thickness and composition, whereas etching of the inner walls to form the layer, produces porous layers, the structure of which depends on the tube material [269,270].These chemically produced layers (e.g., by the action of ammonia water or sodium hydroxide solution) may also act as an active porous layer per se or, if coated with a stationary liquid, as a modified adsorbent. 3.3.4.2.

SCOT Columns

This type of column differs from the previous type in the porous material used and by the fact that the layer, formed on the inside column walls, will always be wetted with a stationary liquid. The particles consist of relatively inert column support material (e.g., Chromosorb) having d, 5 1 pm. Silanized silicic acid has also been used [267](e.g., Silanox 101). They are coated on to the inner walls of the tube, together with the stationary liquid, both dispersedl dissolved in an appropriate solvent having a density L 1.5 g/cm3 and a boiling point between 50 and lOO"C,e.g., R113 (1,1,2-trifluoro-1,2,2,-trichloroethane), analogously to the procedure described in Section 3.3.4.1. Two further coating procedures are described in detail below. (a) Two-step procedure (according to Van Hour et al. [267].A plug consisting of Silanox 101 (silanized silicic acid) dispersed in a dilute solution of a polar phase in chloroform is forced through the column at a rate of 5-8mWs. The dispersion is prepared, immediately prior to use, by dissolving 0.1g of polar phase in 100 ml of chloroform, then adding 1.0g of Silanox and sonicating the mixture at 35°C for 15 min. A plug of about 1 ml is propelled through the capillary, and the solvent is removed by a flow of nitrogen through the column for 2 h. A thin layer of the polar phase containing Silanox remains on the walls. In the second step, a solution of the polar phase in acetone is prepared free from dissolved gases by dissolving 72 mg of the polar phase in 40 ml of acetone and removing 10 ml of the

3.4. Properties and Comparison of the Main Column Types

77

solvent by vacuum. The capillary is immediately filled with the solution using reduced pressure. One end of the column is warmed to expel a drop of the coating solution and immediately sealed, avoiding trapping of air. The opposite end of the tube is connected to a vacuum pump, and the solvent is slowly evaporated at room temperature. A 20-m column requires 36-48 h for solvent evacuation. (b) One-step procedure (according to Chauhan and Darbre [271]) A glass capillary previously treated with hydrogen fluoride is deactivated with benzyl triphosphonium chloride and pre-coated with poly(dimethy1oiloxane). A solution/dispersion is then prepared by dissolving 5% stationary liquid phase and intensively dispersing 5% of Silanox 101 or Chromosorb R-6470-1. A plug of this dispersion having a length of one quarter of the column length is propelled through the column at a rate of 4 c d s . After this dynamic coating the solvent is removed by a flow of nitrogen at 15 ml/min for 4 h. These SCOT columns, which belong, besides WCOT columns, to the most efficient gas chromatographic columns, exhibit lengths generally in the range 10-40 m, most often 15-20 m. The inside diameter is usually 0.5 mm, occasionally lower (down to 0.25 mm), and the layer consisting of support stationary liquid has a thickness from less than 1to a few pm. Their advantages and disadvantages in comparison with WCOT columns are discussed in Section 3.4. Due to the development of cross-linked surface-bonded liquid stationary phases, the application of SCOT columns has considerably decreased. 3.3.4.3.

Thick-Layer Columns

A special case of porous layer open-tubular columns, when considering only the presence of a layer, is the thick-layer column. One could argue, however, over the group of columns to which this type ought to be assigned. Regarding the layer type, either the PLOT or the SCOT type would be possible; regarding large layer thicknesses, with respect to the column inside diameter, the term open-tubular would no longer hold good, and these columns ought to be considered as a kind of micro-packed column. Nevertheless, as they are often thought to fill a gap between open-tubular and packed columns and as their properties, disregarding the thickness of the layer, are essentially the same as those of PLOT and SCOT columns, they are listed here. Their preparation [272-2741 differs from that of the other open-tubular columns. A glass tube with an inside diameter of, e.g., 3 mm is loosely filled with relatively coarse particles (d, = 60-90 pm) and drawn over a steel core of 0.3 mm to give a capillary tube with an inside diameter of, e.g., 0.5 m. The particles adhere tightly to the walls, and form a layer the thickness of which is about 100 pm. The particle material may be an adsorbent, and thus the column can be used in the unimpregnated form, or it may be an appropriate support material, which has to be coated with a stationary phase. Owing to the essentially smaller phase ratio, the amount of sample injected can be considerably larger than that with either of the other open-tubular column types, hence facilitating the method of sample introduction. Conventional gas chromatographs are applicable, on-column injection can easily be arranged, and routine analyses are simple. Nevertheless, the decrease in efficiency caused by the thick layer must not be overloocked.

3.4.

Properties and Comparison of the Main Column Types

At the beginning it must be emphasized that each column type has advantages and disadvantages. An absolute answer to the question of which column type is to be preferred cannot be given, as the decision is dependent on the equipment available, on the expenditure that can be devoted to research and development and, especially, on the analytical problem to be

78

3. The ChromatographicColumn

solved. In order to facilitate such a decision, several characteristic values of the different column types are listed in Table 5 and the pros and cons and the field of application of the most important column types are discussed below. The magnitudes in Table 5 are only average values, and should not be thought to represent all experimental results that have been obtained so far: nor was it possible to take all dependences into account. Table 5. Characteristic Properties of Different Column Types

Phase ratio, B Characteristic /J values k i n

1-1

[cm/sI Optimum practical gas velocity [ c d s ] L [ml Column inside diameter d , &p,

1-1

Film thickness, dl, or layer thickness, dl, , Cm] Sample capacity Cg] Column permeability, x [lo-’ cm2] n (maximum) AZmh, Minimum retention index difference necessary for two compounds to be separated at optimum column length (R,= 1.5)

3.4.1.

Conventional packed Columns

Micropacked Columns

Columns

SCOT

PLOT

WCOT

10-30 15 0.5-2 2-10 8-12

10-100 50 0.2-1 10-20 20-50

50-150 80 0.5-1 10-20 20-80

20-150 80 0.5-2 10-20 20-80

80-500 250 0.2-1 8-15 15-30

2-12 2-4

1-15 0.3-1

0.5

0.5

10-30

10-100 0.05-0.7

d10.01-0.3

d10.01-0.1

d1,O.S-S

dl,0.5-30

d10.1-1

1000 1-10

1-50 5-40

0.3-10 750- 1000

0.3-10 500-800

10 000 10

50 000 3

60 000 2

60 000 2

0.02-2 10- 1500, mostly 200 500 000 0.5

10-30

Columns

Columns

Packed Columns

Advantages: Inexpensive, can be laboratory-made, easy to use, require only simple apparatus, high sample capacity, no injection problems. Shortcomings: High pneumatic resistance (low permeability), restricted column length, low efficiency, low resolution, long analysis time. Application fields: Routine analysis provided that the mixtures to be separated are not too complex; simple separation problems; trace analysis; coupled techniques; process gas chromatography.

3.4.2.

Micro-packed Columns

Advantages: High efficiency enables short columns to be operated at high gas velocities and hence high-speed analysis to be implemented. Inexpensive, can be laboratory-made, sufficient sample capacity. Shortcomings: Pressure drop generally between 0.03 and 0.3 MPa may cause instrumental problems; e.g., at 0.5 MPa a special injector is necessary. Application fields: Multi-component analysis, high-resolution coupled techniques, trace analysis, high-speed analysis using short but efficient columns, high performance gas solid chromatography.

3.4. Prouerties and ComDarison of the Main Column Types

3.4.3.

79

Support-coated Open-tubular Columns

Advantages: High efficiency at high gas velocity permits high-speed analysis; higher sample capacity, lower column bleed, simpler wettability, smaller phase ratios than WCOT columns; splitless injection possible; low detection limits. Shortcomings: Less inactive and shorter column lengths than WCOT columns. Application fields: High performance gas chromatography, high-speed analysis, trace analysis of complicated mixtures, multi-component analysis, high-resolution coupled techniques, analysis of low-boiling compounds.

3.4.4.

Wall-coated Open-tubular Columns

Advantages: Highest efficiencies; possibility of optimization of analysis by changing 4 , 4 and L;owing to the low flow resistance long columns are applicable; analysis time 1-2 orders of magnitude shorter than with conventional packed columns; high-resolution at short analysis times, separability of compound pairs having small retention index differences; lowest adsorption of polar compounds; wall-coated fused-silica open-tubular columns, when combined with the most advanced GC hardware and when the stationary phase is immobilized, offer an almost ideal gas chromatographic system permitting highly reproducible retention values. Shortcomings: Complicated injection, lower sample capacity and more unfavourable phase ratios than other GC columns; only limited trace analysis practicable ( 210 ppm for 4 5 0.5 pm). Application fields: General purpose GC separation system for screening complex mixtures of different structure compounds; high-speed, high-resolution analysis; analysis of very complex mixtures the separation of which would otherwise be impossible; highly reproducible retention index measurements that will offer the possibility of identifying unknown components only by means of a comprehensive retention index library.