Ernst Kenndler: Gas Chromatography GAS CHROMATOGRAPHY ERNST KENNDLER Institute for Analytical Chemistry, University of Vienna 1 THEORETICAL ASPECTS OF GAS CHROMATOGRAPHY ............................................................ 2 1.1 DISTRIBUTION CONSTANT, SEPARATION SELECTIVITY ............................................................................... 2 1.1.1 Temperature Dependence of Distribution Constant........................................................................ 4 1.1.2 Retention index IR ............................................................................................................................ 5 1.2 DISPERSION IN CAPILLARY GC .................................................................................................................. 7 1.2.1 Golay equation .............................................................................................................................. 11 1.2.2 Plate height vs. retention factor. ................................................................................................... 13 2 PRACTISE OF GAS CHROMATOGRAPHY....................................................................................... 15 2.1 CARRIER GAS ........................................................................................................................................... 15 2.2 SAMPLE INLET ......................................................................................................................................... 17 2.2.1 Split and splitless injector ............................................................................................................. 18 2.2.2 On-column injection...................................................................................................................... 20 2.3 COLUMNS ................................................................................................................................................ 21 2.3.1 Stationary phases .......................................................................................................................... 22 2.3.2 Rohrschneider / McReynolds index............................................................................................... 24 2.4 COLUMN OVEN ........................................................................................................................................ 26 2.5 SPECIAL DETECTORS ............................................................................................................................... 26 2.5.1 Flame ionisation detector.............................................................................................................. 26 2.5.2 Electron capture detector.............................................................................................................. 28 2.5.3 Alkali flame ionisation detector .................................................................................................... 31 3 FURTHER READINGS............................................................................................................................ 34 © Ernst Kenndler Version 19/01/2004 1 Ernst Kenndler: Gas Chromatography THEORETICAL ASPECTS OF GAS CHROMATOGRAPHY 1 This text should be read in context with “Introduction in Chromatography”, where a fundamental discussion of the migration and dispersion phenomena occurring in the chromatographic separation system is given in a general manner. In the present contribution the theoretical discussion is mainly directed towards gas liquid chromatography in capillary columns. This chapter about theory of gas chromatography is followed by a considerably detailed presentation of practical aspects of this method. 1.1 Distribution constant, separation selectivity In Gas Liquid Chromatography the analytes are distributed between a liquid stationary phase and an ideal gas as the mobile phase as schematically shown in Figure 1. <v> cim (t) GAS LIQUID cis (t) Figure 1 Schematic presentation of a gas-liquid chromatographic system. <v> is the average linear velocity of the mobile phase. cim (t) and cis (t) are the concentrations of the analyte, i, in the mobile and the stationary phase, respectively. Both are functions of time. This distribution is determined by the partition constant, given as usual by Ki = c il c im (1) However, for practical reasons the concentration in the liquid is expressed by the mole fraction, xi, and that in the gas phase by its partial pressure, pi. For the concentration in the liquid we express the vapour pressure of the solute, i, over the binary mixture consisting of liquid phase and analyte by Henry`s law 2 Ernst Kenndler: Gas Chromatography p i = α H p i0 = γ i0 x il p i0 (2) Where αH is the Henry constant, xil the mole fraction of analyte, i, in the liquid phase, γi0 is the activity coefficient (at infinite dilution), pi0 is the vapor pressure of the analyte (as pure compound) at the given temperature. The partial pressure, pi, of the analyte in the ideal gas phase is given by Dalton`s law pi = n ig RT Vg (3) T is the absolute temperature and R is the gas constant. Substitution of the volume concentrations in eq. 1 (for infinite dilution) gives the expression for the partition constant in GLC: K io = RT p γ ioVs ,mol (4) o i where Vs,mol is the mean molar volume of the stationary phase. Two most important parameters occur in this expression: ⇒ p io , the vapour pressure of the analyte as pure compound at temperature T γ i0 ⇒ the activity coefficient of the analyte in the stationary liquid (at infinite dilution). Separation selectivity of two consecutively eluting components, i and j, is defined in chromatography by the selectivity factor, αji, the ratio of the distribution constants, K. However, measurement of the values of K is complicated compared to k` values. As in a certain chromatographic system the phase ratio is the same for all components, the selectivity coefficients can be expressed by the ratio of the k` values of the pair of separands α ji = k j, (5) k i, The selectivity coefficient for GLC using eq. 4 is thus α ji = pio γ io p oj γ (6) o j It can be seen that selectivity in GLC is determined by two ratios: the ratio of the vapour pressures of the analytes as pure compounds (at working temperature) 3 Ernst Kenndler: Gas Chromatography the ratio of the activity coefficients the in stationary phase (at infinite dilution). Whereas the temperature only influences the first, the second reflects the difference in the chemical interactions of the two separands in the stationary liquid. Insofar it is in fact a kind of molecular recognition, which determines selectivity. This is exactly what makes the use of different stationary phases meaningful. 1.1.1 Temperature Dependence of Distribution Constant Eq. 4 might lead to the erroneous conclusion that the distribution constant in GLC increases with increasing temperature. In fact the contrary is the case, because this linear dependence of K on T is more than overcompensated by the exponential increase of the vapour pressure, pio , with temperature, according to the Clausius-Clapeyron equation. For this reason the distribution constant, and the capacity factor as well, decreases strongly with increasing temperature. In fact the following linear approximation can be found log K io resp. log k i' ∝ 1 T (7) The experimental dependence of the capacity factor on the temperature is shown in Figure 2 for two analytes. lok k´ 10 etbenzoat 1 C7ol 0,0023 0,0024 0,0025 0,0026 0,0027 0,0028 0,0029 1/T Figure 2 Relation of the logarithm of ethylbenzoate and 1-heptanol on the reciprocal of the absolute temperature. 4 Ernst Kenndler: Gas Chromatography 1.1.2 Retention index IR In contrast to HPLC, gas chromatography possesses a useful and generally accepted parameter for the characterisation and identification of analytes. This parameter is based on the finding that log k` values of the members of a homologous series of organic compounds are linearly depending on the number of carbon atoms, n, in their molecules. Applied to the homologous series of n-alkanes this means that log k n, = A + B. n (8) log k` This relation can be graphically represented by 18 16 14 12 10 8 6 "n" = 11,63 4 4 6 8 10 12 14 16 18 number of C atoms Figure 3 Logarithm of the retention (capacity) factor, k`, of the homologues series of the nalkanes as function of the number of carbon atoms For each analyte with a certain capacity factor a pair of n-alkanes exists, between that the analyte is eluted in the chromatogram. This analyte is considered as a fictive n-alkane with a hypothetical number of C-atoms, in the example in Figure 3 with 11.63 C-atoms. This number (and the number of C-atoms of the n-alkanes as well), multiplied by 100, is the retention index. The analyte in the example given consequently has the retention index 1163. It is obvious that n-alkanes have always retention indices with full hundreds, n-hexane e.g. 600, nundecane 1100, n-eicosane 2000, etc. Obviously the retention index of analyte i is normally determined with higher accuracy than it would be possible by graphical interpolation. For this reason the following equation is used that can be derived simply from Figure 3 by comparison of similar triangles: I Ri = 100 z log k i, − log k n, + 100n log k n, + z − log k n, (9) Substitution of the capacity factors by the net retention times 5 Ernst Kenndler: Gas Chromatography t RiN = t Ri − t R 0 (10) leads to an expression convenient for measurement and determination of the retention index log I Ri = 100 z log t RiN N t Rn t RN( n + z ) + 100n (11) N t Rn Here (n+z) and n are the numbers of C-atoms of the n-alkanes eluting after and before the analyte. Usually, but not necessarily, z is 1. If e.g. due to economic reasons only the (cheaper) even-numbered higher n-alkanes are used (e.g. C26 and C28), z is 2 in that case. It is the advantage of the retention index that it is independent of certain, often varying experimental parameters: velocity of the mobile phase phase ratio length of the column. It depends on: kind of the stationary phase (column temperature) It is therefore a very well suited number for the characterization of an analyte on a certain stationary phase. Retention indices are used for identification of an analyte by comparison of the value of the index with either that found in tables, or determined previously with the help of reference compounds. Retention indices of an unknown compound on different stationary phases allow further conclusions about the polarity of the analyte. Table 1 Retention indices and ∆IR –values of differently polar C8 compounds on apolar polydimethyl siloxane (OV1) and polar polyethylenglycol (PEG) as stationary phase Compound IR (OV1) IR (PEG) ∆IR =IRpolar – IRunpolar n-octane 800 800 0 n-dibutylether 864 966 102 n-hexylacetate 963 1101 138 n-octanon-2 957 1295 338 n-octanol-1 1038 1545 507 6 Ernst Kenndler: Gas Chromatography For this purpose the indices are determined at two columns with different polarity, e.g. with methylsiloxane and polyethylenglycol as stationary liquids. Obviously n-alkanes have always a ∆IR -value of zero. The more polar the functional group is, the larger is the ∆IR –value. ∆IR –value values of selected reference compounds are used to describe the polarity of stationary phases as well (see concept of the Rohrschneider / McReynolds index, chapter 2.3.2.). Conclusion for the use of retention indices: Retention indices at one phase n-alkanes always full hundreds, n-hexane 600, etc. homologues differ by 100 at one stationary phase Retention indices at two phases ∆IR =IRpolar – IRunpolar zero for n-alkane ∆IR – the larger, the more polar functional group ∆IR –values of selected reference compounds are used to describe the polarity of stationary phase 1.2 Dispersion in capillary GC Peak dispersion in chromatography is discussed in general in “Introduction to chromatography”. We will here concentrate on dispersion in capillary GC. Peak broadening is described by the model of the theoretical plate height, H. Note that 4 processes were found to contribute to the total plate height in chromatography. ⇒ Hdiff describes the contribution from longitudinal diffusion ⇒ Hconv that from convective mixing ⇒ Hex,m that stemming from the kinetics of mass exchange from the mobile phase to the interface between mobile and stationary phase ⇒ Hex,s that from the kinetics of mass exchange from the stationary phase. Consequently the total plate height is the sum of the four contributions: 7 Ernst Kenndler: Gas Chromatography H = H diff + H conv + H ex ,m + H ex ,s (12) (ad Hdiff ) The contribution of longitudinal diffusion, that in direction of zone migration, is caused by the concentration gradient occurring between the sample and its surrounding in this direction. According to the Einstein equation the resulting peak variance in the space domain is given by (13) σ z2 = 2 Dmi t Broadening of the peak with time according to eq. 13 is shown in Figure 4. The equivalent expression for the relation between plate height and variance is (14) σ z2 = Hz 1,2 concentration 1,0 δ function t0 lim ∆ z = 0 t1 0,8 concentration 10 0 1 50 200 z 0,6 t2 0,4 0,2 t3 0,0 z Figure 4 Development of peak broadening with increasing time due to diffusion. The concentration zone was infinitely narrow at time 0 (δ - or Dirac function) This contribution is as more pronounced as larger the diffusion coefficient, Dm,i , of the analyte in the mobile phase is, and as longer the time is available for diffusion. Consequently, 8 Ernst Kenndler: Gas Chromatography this increment increases with decreasing velocity of the mobile phase; it is proportional to 1/v. It follows that the plate height contribution due to longitudinal diffusion is H diff = 2 Dm,i (15) v It should be mentioned that longitudinal diffusion in the stationary phase does not significantly contribute to peak broadening, because the diffusion coefficients in the stationary phase are 5 to 6 orders of magnitude smaller than those in the gas phase. (ad Hconv ) The radial velocity profile of the mobile phase flowing through the column introduces an effect to peak broadening due to convective mixing. In case of an open tube (as it is in capillary GC) the flow profile can be analytically described in a simple way, in contrast to packed beds. It has a parabolic shape with zero velocity at the capillary wall, and maximum velocity at the centre of the tube with radius rC. For a non-retained component the contribution to the plate height is described by the so-called Taylor-dispersion H conv = rc2 v 24 Dm ,i (16) radius, r v ma x 30 00 0 25 00 0 2000 0 15 000 10 00 0 5 00 0 0 -5 000 velocity - 10 00 0 Figure 5 Profile of the velocity, v, of the mobile phase in an open tube as function of radius, r The resulting dispersion due to the parabolic flow profile and the radial diffusion would give the following peak for a non-retained component 9 Ernst Kenndler: Gas Chromatography 2 ,4 2 ,2 2 ,0 concen tration 1 ,8 1 ,6 1 ,4 1 ,2 parabolic 1 ,0 flow profile 0 ,8 0 ,6 flo w profile 0 ,4 ra dial diffusion 0 ,2 0 ,0 -0,2 20 10 0 z -10 -20 Figure 6 Peak profile resulting from parabolic flow and radial diffusion (ad Hex,m ) Both mass exchange terms have their origin in the finite rate of mass transport from the inner part of the mobile or stationary phase, respectively, to the interface between these two phases. Roughly it can be said that at the front of the peak in the mobile phase a higher concentration exists than the equilibrium concentration. At the rear side the opposite situation occurs. From these kinetic reasons peak dispersion occurs, which increases with increasing velocity of the mobile phase. The effect in the mobile phase is connected to the flow situation. Therefore the contribution of the finite mass transfer in the mobile phase is combined with that stemming from the flow profile, which enlarges eq. 16. The combination of both effects leads to Hconv + ex ,m = (1 + 6k i, + 11k i,2 ) rc2 (1 + k i, ) 2 24 Dm,i v (17) (ad Hex,s ) The term which stems from the finite kinetics of mass transport in the stationary phase is given by H ex ,s d 2f k i, 2 = v 3 (1 + k i, ) 2 Ds,i (18) 10 Ernst Kenndler: Gas Chromatography where df is the thickness of the stationary phase layer, and Ds,i is the diffusion coefficient in the stationary phase. 1.2.1 Golay equation The Golay equation describes the total plate height, given by the sum of the particular increments that contribute to peak dispersion: H = 2 Dm,i v + (1 + 6k i, + 11k i,2 ) rc2 (1 + k i, ) 2 24 Dm,i v+ d 2f k i, 2 v 3 (1 + k i, ) 2 Ds,i (19) This equation shows the dependence of the particular increments, and thus the total plate height, as a function of the velocity of the mobile phase, apparently one of the most important experimental variables to influence peak dispersion. Eq. 19 can be rewritten in a simplified form as H= B B + (Cm + C s ). v = + Cv v v (20) 1,0 0,8 H 0,6 0,4 C.v 0,2 0,0 B/v 0 20 40 60 v 80 100 Figure 7 Plot of the plate height as a function of the mobile phase velocity (Golay equation). The depiction of the plate height in relation to the mobile phase velocity is given in Figure 7. It results in the summation of the hyperbolic B-term of eq. 15, that depends on 1/v, with the C-term, which increases linearly with v. From Figure 7 it can be seen that a singular velocity exists where the plate height exhibits a minimum value. Here peak dispersion is smallest. If minimum dispersion (highest efficiency) is necessary for sufficient separation, this particular velocity has to be selected. On the left-hand side of the H vs. v curve the plate height steeply 11 Ernst Kenndler: Gas Chromatography increases due to pronounced diffusion and the time of analysis increases as well. Insofar there is no cause to select such velocities in practice. On the right hand side of the minimum, in contrast, the plate height increases considerably slowly, and approaches the C-term asymptotically at high velocity. In this range the working conditions are selected favourably when the separation is large enough due to sufficient selectivity. In this case the time of analysis, which is often an important analysis parameter, will be reduced with not too high loss in efficiency. 1.2.1.1 Gas as compressible fluidum It was pointed out that in GC there is an average linear velocity, <v>, which differs from the local velocity, v, at a certain position at the longitudinal coordinate of the column due to the compressibility of the gaseous mobile phase. The particular velocity depends on the pressure drop across the column. v p p0 Figure 8 Schematic drawing of the local flow velocity, v, (indicated by arrows) of a compressible fluidum across the capillary. p pressure, p0 pressure at the column outlet The deviation of the volume flow velocity can be derived by the correction factor according to Martin and Synge 3 ( p p0 ) − 1 j= 2 ( p p0 )3 − 1 2 (21) where p and p0 are the pressures at the top and the end of the capillary. Consequently the minimum plate height is observable only in a small part of the separation capillary, where the particular mobile phase velocity is optimal. All other velocities 12 Ernst Kenndler: Gas Chromatography deviate in principle from this value for minimum plate height. However, under usual conditions in capillary GC (for columns not too narrow and long) p does not exceed p0 by more than 1.5 fold. In this case the difference between the average linear velocity (which is measured by <v> = L/tR0) and the actual velocity at any section inside the capillary is negligible, and the plate height does not change significantly across the capillary. For this reason the variation of the velocity due to the compressibility of the gaseous mobile phase normally is not a matter for consideration. 1.2.2 Plate height vs. retention factor. Another important fact can be seen from the Golay equation as well: the plate height is dependent on the capacity factor. It follows that for given instrumental conditions H may change drastically for different components of a sample mixture, given that these components have considerably different k´ values. For a capillary column, e.g., with a very thin film of stationary phase, for which the third term in eq. 19 can be neglected, the dependence of the plate height from the k` values of the sample components can be expressed by Figure 9. 10 8 , ,2 (1+6k +11k )/(1+k ) , 2 12 6 4 2 0 0 2 4 k , 6 8 10 Figure 9 Dependence of the second term of the Golay equation (eq. 19) on the capacity factor. This term reaches values between 1 (for k` of zero; here peak broadening is given by Taylor dispersion) and approximately 11 at large values for k`. This means that for otherwise equal experimental conditions the plate height of the peaks within one chromatogram can vary by as much as one order of magnitude. If the stationary phase of the capillary column has not such a thin film, the Cs-term may play a role as well. For such cases the dependence of this term on the k` of the analytes is shown in Figure 10. 13 Ernst Kenndler: Gas Chromatography 0,30 k`/(1+k`) 2 0,25 0,20 0,15 0,10 0,05 0,00 0 2 4 6 8 10 k` Figure 10 Cs-term as function of the retention factor (eqs. 18 and 19) A closer analysis of the dependence of the plate height on the capacity factor of the solutes enables conclusions on the properties of the separation column concerning the significance of the particular terms. According to the weight of the two terms different relations of H as function of k` are found for the individual columns. The theory of chromatography in capillary column delivers the instrument to evaluate the occurring effects. Note that it is the plate number, N, rather than the plate height, which is decisive for the separation of two analytes: N= L H (22) The plate number of the particular components of a sample can simply be calculated from their retention time, and the corresponding standard deviation σ (or the half peak width w), all parameters taken in the time domain, and in the same units by 2 2 t R ,i t = R ,i * 5.54 N i = w σ t ,i t ,i (23) 14 Ernst Kenndler: Gas Chromatography PRACTISE OF GAS CHROMATOGRAPHY 2 A gas chromatograph consists of several parts, which are described in the following in more detail (cf. e.g. refs. 1-13 ) and can indeed often be handled as modules in instrumental practice. They are schematically shown in Figure 11. Roughly, a chromatograph is composed from the carrier gas supply, the sample inlet, the column, positioned in a column oven, the detector(s) and a device for data collection, acquisition and processing. Carrier Gas Supply Sample Inlet Column Oven Detector Data Collect. Acquis. Process. Figure 11 Scheme of a gas chromatograph 2.1 Carrier gas As mobile phase an inert gas is used, which is delivered by a gas generator, or a gas cylinder. The most common carrier gases are H2, N2, He. They must be of very high purity, because traces of water or oxygen may decompose the stationary phase, which leads to column bleeding and finally destruction of the column. Therefore special devices for gas purification are installed often prior to the sample inlet. The choice of the carrier gas depends on several demands, e.g. the appropriate operation of the detector (for the combination of GC with MS e.g., He is needed), on safety reasons (H2 is explosive), or on the price (N2 is the cheapest gas), but also on demands on separation efficiency and speed. Due to its lowest viscosity of all gases, H2 e.g. allows to operate the column with the highest mobile phase velocity - and therefore lowest analysis time - at comparable efficiency (see below). Summary of the demands on the carrier gas: Chemically inert High purity (water, oxygen) 15 Ernst Kenndler: Gas Chromatography Detector compatibility Economic / Safety reasons Efficiency / Speed Concerning efficiency and speed depending on the kind of carrier gas we ask for the condition of plate height minimum. This is dH =0 du (24) With the simplified Golay equation (eq. 20) we obtain H min = 2 BC ≈ rc (25) 3 It can be seen that this plate height is independent on the kind of mobile phase (see Figure 12). It is depending on the radius of the capillary only. This plate height is reached at the mobile phase velocity u min = B C (26) ≈ 7 Dmi rc The velocity where minimum plate height is reached depends on the gas used as mobile phase, because the diffusion coefficient is related to the gas viscosity. Therefore hydrogen as mobile phase reaches minimum plate height at higher mobile phase velocity than e.g. nitrogen. Faster analyses can be achieved therefore. 10 plate height, H 8 6 H2 4 2 N2 0 0 5 10 15 20 mobile p hase velocity, v Figure 12 Peak height vs. mobile phase velocity for hydrogen and nitrogen as carrier gases 16 Ernst Kenndler: Gas Chromatography 2.2 Sample inlet In GC the sample is normally brought into the separation system in liquid solution (sampling techniques for vapours, e.g. head space or adsorption / thermodesorption injection, or for solid samples, e.g. pyrolysis injection, are not discussed here). The sample is dissolved in an organic solvent, which is normally injected into the carrier gas flow by the aid of a syringe or a valve. Indeed the quantitative and non-discriminating introduction of the sample into the column is the most critical part of all experimental steps in capillary gas chromatography. Although GC is a well-developed and established method, sample introduction in capillary GC is still a non-trivial task due to practical limitation. This results in a number of different injector types, which should be selected in practice according to the nature of the sample and the demands in accuracy and reproducibility of the analysis. In contrast to capillary GC, sample introduction in packed column GC is not a problem. This is due to the fact that in packed bed GC the column volume is large, and the phase ratio is considerably large, too. Here the sample, dissolved in a volatile organic solvent, is simply injected into a heated and thermostatted injector cell, where it is evaporated, and the vapour is transported by the carrier gas into the column. Note that 1 µl of liquid sample delivers several hundred µL vapour after evaporation. It is clear that such a large volume would overflow the entire column in case of a capillary column. As an example: a capillary with 0.2 mm I.D. and 25 m length has a total volume of less than 800 µL. It is obvious that it is not possible to introduce the entire evaporated sample directly into the column in capillary GC. Therefore mainly two possibilities are proposed to overcome this problem in practise: (i) In case of the “evaporating” injectors the sample is inserted by a syringe into the heated injector and evaporated. Either only a part of the evaporated sample is allowed to enter the capillary – this is realised with the split injector; or the main part of the solvent (and a small part of the sample as well) is separated in the injector from the sample components. In this splitless mode the sample components normally are recondensed at the top of the column either by “cold trapping” or by “ solvent trapping”. (ii) the total liquid volume is brought into the cold injector by the aid of a syringe, and solely the solvent is evaporated carefully first and usually recondensed either at the top of the column, or in the “retention gap”. Due to careful selection of the temperature conditions the sample remains at the top of the separation system. Then the sample is evaporated, too, and introduced into the separation capillary. 17 Ernst Kenndler: Gas Chromatography 2.2.1 Split and splitless injector In the split mode (see Figure 13A) the sample is rapidly injected and evaporated in the liner of the heated injector. The gas flow (vaporised sample mixed homogeneously with the carrier gas) is divided at the top of the column by the aid of a needle valve (which enables to adjust different split ratios). The main part of the gas mixture is flushed out, and only a small part is allowed to stream into the separation capillary. The injector has the advantage that the injected zone is narrow, and the small sample aliquot entering the capillary avoids overloading of the column. Although very flexible in practice, this injector has a number of disadvantages. In many cases mass discrimination of sample components is observed, especially when the range of their volatility differs much. Therefore systematic errors for quantitative analysis may occur. Another disadvantage, especially in trace analysis, is the fact that only a part of the analytes is transferred into the capillary, and reaches the detector. The main part of the sample is deleted via the split and therefore lost for detection. To overcome such problems the injector can be operated in the splitless mode. Here the capillary is first run in the split mode. Directly prior to injection the split is closed. The sample is then slowly injected into the heated injector, and sample and solvent are evaporated. It is most important that in this mode the column is kept at relatively low temperature, lower than the boiling point of the solvent. Therefore the volatile solvent condenses at the top of the column, and forms a kind of a stationary phase here. Volatile sample components, which evaporate in the injector, too, are dissolved again in the liquid formed, and are therefore refocused (“solvent trapping”). Less volatile sample components, which were also evaporated in the hot injector, are recondensed on the top of the colder column, and focused, too (“cold trapping”). After these two processes have taken place, the split is opened (after about 30 – 90 s) and the rest of the solvent is flushed via the split valve. The solvent that initially forms a liquid sheath at the top of the column evaporates gradually, with progressive evaporation from the injector to the detector side. This effect supports the refocusing of the sample components. Finally the sample is evaporated by the application of a temperature program, which is obligatory for this injection technique. The entire procedure avoids the large tailing of the solvent peak observed otherwise, and allows the transfer of the main part of the sample components into the column and, finally, into the detector. It is therefore a favourable technique for the insertion of diluted samples in trace analysis. 18 Ernst Kenndler: Gas Chromatography A) B) Figure 13 (A) Schematic drawing of a split-splitless injector (from ref. 10 with permission). (B) Schematic drawing of an on-column injector. 1 Carrier gas inlet, 2 Sealing; 3 Capillary column, 4 Cooling gas (from ref. 14 with permission). The recondensation of the solvent at the top of the capillary can be critical by damaging the stationary phase (because it might be partially dissolved by the condensed liquid), and therefore only columns with chemically bonded phases should be used. Another limitation is the necessary wettability of the stationary phase by the condensed solvent, otherwise droplets are formed rather than a liquid layer. These problems may be overcome by the use of an empty (widebore) capillary without stationary phase, mounted between injector and separation column. In this capillary indeed both refocusing processes – solvent trapping and cold trapping – as described above can be established. This construction is named “retention gap”, and is used in the splitless mode and, more common, with the on-column injection technique. 19 Ernst Kenndler: Gas Chromatography It should be mentioned that the discriminating evaporation of volatile sample components (according to their boiling points) when injected into the hot injector can be diminished by a modification called “programmed temperature vaporiser” (PTV injector). The PTV device can be used for the split and the splitless mode as well. Here injection is carried out into the cooled injector. When operated initially at a temperature slightly above the boiling point of the solvent (but below those of the most volatile sample components), the main part of the solvent can be separated via the open split (“solvent purge”). After flushing the solvent the PTV injector is heated up rapidly, and the sample is transferred to the top of the column as a more or less narrow band. Summary of the advantages and disadvantages of split and splitless injectors: SPLIT INJECTOR Advantage Injected zone narrow Small sample aliquot avoids overloading Limitations Mass discrimination of sample components (different range of volatility) Systematic errors for quantitative analysis In trace analysis: only part of analytes to detector SLITLESS INJECTOR Advantage Avoids large tailing of solvent peak Allows transfer of main part of sample components into detector Trace analysis: favourable technique for insertion of diluted samples Limitations Recondensation of solvent at top of capillary: possible damaging stationary phase Only columns with chemically bonded phases Necessary wettability of stationary phase for condensed solvent (droplets) 2.2.2 On-column injection With the on-column technique the sample solution is directly inserted into the column with the aid of a syringe with a long, narrow needle, whereby the injector (Figure 13B) is maintained at low temperature. Due to the restricted mechanical stability of the thin syringe needle a normal septum cannot be used, and is replaced by special sealing. In most cases a piece of a wide bore capillary (retention gap) is connected to the thinner separation capillary to avoid “column flooding” by the large volume of sample vapour. Retention gap: empty capillary - fused silica, widebore (0.5 mm i.d.), 20 –200 cm L without stationary phase mounted between injector and separation column 20 Ernst Kenndler: Gas Chromatography The capillary for the retention gap is mounted in the column oven, whose temperature must be adjusted to the boiling point of the solvent. If the temperature is below the boiling point, solvent trapping takes place. If it is selected slightly above the boiling point, cold trapping of the sample components occurs. In both cases the chromatogram must be developed with an adequate temperature program of the column, which is an essential step when using this injector type. Advantage of on-column injector: avoids mass discrimination effects trace analysis: enables quantitative insertion of sample into column (and detector) labile components not stressed thermally 2.3 Columns In gas chromatography packed bed columns and capillary column are used. Packed columns are tubes made of glass or metal with 2-4mm I.D. and 1-6 m length. They are filled with porous particles, which act as support of the stationary liquid phase, which is coated on the porous material. Capillary columns are open tubes with 0.1 to 0.5 mm I.D. and 5 to 100 m lengths. Most common dimension, however, are 0.3 mm I.D. and 25 m length. Originally the capillaries were made from metal or glass; in the last decade fused silica replaced all other materials. Fused silica has the advantage of a very inactive inner surface, which avoids adsorptive interactions between analytes (especially when they are polar) and adsorption centres, leading otherwise to tailing peaks or even loss of material due to irreversible adsorption (see note below on Grob`s test mixture). It has the further advantage of extremely high mechanical stability that reduces breakage of the columns. The stationary phase is coated as a thin layer (with 0.1 to 5 µm film thickness) onto the inner wall of the open tube. Normally this phase is a liquid. Due to modern column technology, which enables cross-linking of the polymer molecules of the liquid, and even attachment of the phase at the silica surface due to chemical bonding, the initially liquid phase might behave as large, single, polymeric molecule. Interestingly these cross-linked phases thermodynamically behave very similar to the initial liquid. 21 Ernst Kenndler: Gas Chromatography PACKED COLUMN CAPILLARY COLUMN Stationary phase Packing Figure 14 Schematic drawing of packed and capillary columns It should be mentioned that a mixture of solutes, which allow conclusions about specific adsorptive sites, could test the adsorptivity of columns. The most common is the so-called Grob test mixture, which consists of octan-2-on, octanol-1, 2,6-dimethyl phenol, 2,4-dimethyl anilin, naphthalin, tridecan and tetradecan. 2.3.1 Stationary phases Stationary phases must cover a wide range of “polarity as indicated in the following Table. Apolar phase vapour pressure Polar phase intermolecular forces dispersion polarisation dipole-dipole hydrogen bonding Besides being the source for separation selectivity in GC, the demand on stationary phases is thermal stability. It is clear that polymeric compounds best fulfil the latter restrictions. Especially siloxane polymers have a high thermal stability, caused by the Si-O backbone of the silicon chain. The type of the substituents attached at this chain implements selectivity. CH3 groups as substituents give the lowest polarity, resulting in a polymer that is nearly as 22 Ernst Kenndler: Gas Chromatography A) O O Si O Si O Si R1 H3C CH R1 R2 R2 3 CH3 H3C B) O Si O O n O O Figure 15 Structural formulae of stationary phases: A) Polysiloxane-based phases. Polydimethyl siloxane: R1, R2: CH3 Methylphenyl polysiloxanes: different ratio R2/R1; R1, R2 = Or R1, CH3, R2 = CH3 Methyl cyanopropylphenyl polysiloxanes: R1, R2 = CH3 Or R1, R2= or C N B) Polyethyleneglycol High Temperature Phases a) Carboran modified polysiloxane b) Silarylen polymer 23 Ernst Kenndler: Gas Chromatography apolar as a hydrocarbon. Gradual substitution of the CH3 groups by polarizable phenyl moieties increasingly changes polarity, and enables therefore separation of moderately polar analytes. Introduction of cyanopropyl substituents partially replacing phenyl groups in the silicon chain leads to a phase with highest polarity amongst the siloxanes. Polyethylenglycol, finally, allows interactions based on hydrogen bonds, and is therefore best suited for the separation of analytes that are H-donors, e.g. alcohols. These phases are depicted in Figure 15. Two examples for high temperature phases are also given. The one consists of dimethyl polysiloxane polymers, which determine the selectivity. These polymeric chains are connected with carborane (carbon-boron compounds) anchors, which are responsible for the usability at temperatures as high as about 400°C. The second phase in the example given introduces temperature stability by implementation of phenyl groups directly into the polymer chain. 2.3.2 Rohrschneider / McReynolds index A concept to characterise the polarity of stationary phases was introduced by Rohrschneider and McReynolds. It is based on the finding that the interaction of polar functional groups of a solute is reflected by the difference of its retention indices on a polar and a nonpolar stationary phase, respectively. Taken a very apolar phase as a reference (for this purpose squalane is used, a branched C30 alkane, hexamethyl tetracosane), the retention index difference on a certain phase relative to squalane is a measure for the “polarity” of this phase. If e.g. the solute is an alcohol, it can be formally divided into the alkyl rest (which interacts only by weak dispersion forces with all phases), and the OH group that is able to donate hydrogen bonds. Interaction with stationary phase molecules, able also for hydrogen bonding, will lead to stronger retention on the polar phase than on an apolar phase, and therefore a large increase in retention index , ∆IR, will result. According to the concept of Rohrschneider / McReynolds, the increase of the retention index for the alcohol on the polar stationary phase is given by a certain constant, characteristic for the given stationary phase. Measured for butanol-1 as reference solute it is squalane y , = I bupolar tan ol −1 − I bu tan ol −1 (27) To characterise the stationary phase polarity in a more general manner, the concept uses 5 special solutes, which are considered to represent typical chemical interactions. For each of them, the constants x´, y´, z´, u´ and s´ are defined accordingly; they are shown in Table 2. 24 Ernst Kenndler: Gas Chromatography The polarity of the stationary phase is expressed by the sum of the constants, the Rohrschneider / McReynolds index, Σ Σ = x´+ y´+ z´+u´+ s´ (28) Examples for stationary phases most common in practice are given in Table 3. If two phases do not differ in their indices by more than about 200, their polarity can be considered as nearly equal. In general it is not meaningful to use two such phases. This does not mean, however, that probably their selectivities concerning one special type of interaction are negligible. Table 2 Rohrschneider / McReynolds index, Σ 5 special solutes represent typical chemical interactions Σ = x´+ y´+ z´+u´+ s´ Reference solute Rohrschneider/ McReynolds constant Type of interactions Typical for Dipole π-complex H-bond benzene x´ - donor - butanol-1 y´ - donor 2-pentanon z´ acceptor - 1-nitropropane pyridine u´ s´ acceptor donor - olefines, aromatic compounds alcohols, phenols, acids, amides aldehydes, ketones, esters, ethers nitro-, nitrilo compounds Amines, aromatics Table 3 Rohrschneider / McReynolds constants and indices for characterisation of stationary phase polarity Stationary phase Squalane Dimethyl silicon Phenyl methyl silicon Cyanopropylphenyl (cpph) dimethy silicon Composition 100 % methyl 5 % phenyl 50 % phenyl 75 % phenyl 6 % cpph 50 % cpph 100 % cpph PEG 20 M x´ 0 17 32 119 178 50 y´ 0 57 72 158 204 115 z´ 0 45 65 162 208 107 u´ 0 67 98 243 305 164 s´ 0 43 67 202 208 103 Index Σ 0 229 334 884 1103 539 227 523 322 373 757 536 336 659 368 489 942 572 398 801 510 1823 3682 2308 25 Ernst Kenndler: Gas Chromatography 2.4 Column oven The column oven has the function to adjust the column temperature to an accurate and reproducible value. It is prerequisite in practice to establish these temperature conditions for the column over the entire length, not only where the temperature sensor is located in the oven. The column oven should not only meet these demands in the isocratic manner; it must also enable the application of appropriate gradients, either a single linear or ballistic, or multiple gradients by temperature programming. It should be noted that after such programming a very important but often underestimated aspect is the establishment of the exact initial column temperature after cooling. The inappropriate re-establishment often is the source of systematic measuring errors. 2.5 Special Detectors For GC a number of detectors have been developed. For trace analysis, however, not all of them have increased importance. Most important detectors in this area are (beside the mass spectrometer) • Flame ionisation detector (FID) • Electron capture detector (ECD) • Alkali flame ionisation detector (NPD) These three detectors will be discussed in more detail in the following. 2.5.1 Flame ionisation detector The FID is a mass flow sensitive detector. It is based on the measurement of the electric charges, which are produced in a small hydrogen flame. In the absence of organic molecules in the carrier gas, this flame is very poor in charged particles, because the combustion of hydrogen with oxygen delivers only a very small number of ions or electrons. Indeed the residual current is in the range of only 10-12 A (under normal working conditions, i.e. when a voltage of about 200-300 V is applied between the flame and the collector electrode). This extremely small residual current is amplified and represents the background signal (the blank). Organic molecules, which possess CH-groups, form CHn•-radicals (n=0-3) at the periphery of the hydrogen flame, where also excited O2* and OH* molecules are formed. 26 Ernst Kenndler: Gas Chromatography Reaction of the radicals with the excited molecules leads to the formation of positively charged ions and electrons, e.g. according to (29) C • + OH ∗ = CHO + + e − Consequently the positively charged molecular ions (e.g. CHO+) and the electrons formed increase the current when organic molecules are entering the detector flame, and the analytes are detected in this way (note that there is not full combustion to CO2). Figure 16 Schematic drawing of a flame ionisation detector (from ref. 15 with permission). The more CH-groups a molecule contains, the larger is the detector response. Heteroatoms in the molecule lead to a smaller response. Molecules without CH-groups do not deliver a signal, except due to overloading effects of the detector. The FID is therefore not universal, as e.g. water, CO2, NOx , CS2, CCl4, etc. are not or only poorly detected. The same is the case for the lower alcohols or substances with many heteroatoms. Beside the analyte properties, the detector response is also dependent on its geometry, and on the flow rate of the burning gases. Favourable flow rates for H2 are in the range of 2530 mL/min, those for the air are by a factor of 10 higher. Especially the H2 flow must be selected carefully, because its optimal range is narrow (see Figure 16). The use of so-called make-up gas is normally advisable in capillary GC for the improvement of the detector response. It is also favourable to avoid loss in separation efficiency due to extra-column 27 Ernst Kenndler: Gas Chromatography effects caused by the dead volume of the detector. Because the FID is mass flow dependent, the make-up gas does not negatively influence the detector sensitivity. Figure 17 Dependence of the sensitivity of the FID on the flow rates of H2 and N2, respectively. N2 is added as make-up gas (from ref. 14 with permission). Performance data FID sensitivity ~ 0.015 As/g limit of detection ~ 10-11 to 10-12 g carbon/s linearity 7 orders of magnitude Time constant ~2 ms 2.5.2 Electron capture detector In principle the ECD is an ionisation chamber. It has a very high specifity and sensitivity for compounds that have atoms in their molecules with high electron affinity. Especially halogens exhibit this property, but also for oxygen containing groups or nitro-groups this detector responds very well (see the following Table). It will be, however, discussed below, that it is not sufficient to take only the electron affinity into account to interpret the detector response. X H C O CN NO2 ElectronAffinity [eV] 0.72 1.2 2.34 3 3.9 X J Br Cl F Electron Affinity [eV] 3.12 3.52 3.78 4.1 28 Ernst Kenndler: Gas Chromatography Roughly, the detector principle is based on the property of analytes, AX, containing such atoms or groups, X, to attract electrons according to (30) AX + e − = AX − + energy Electrons are generated by a radioactive β-source like 63 Ni. However, these “primary” electrons exhibit a too high energy (and therefore too high speed) to be captured from the affine groups in the analytes. They interact first with carrier gas, which is present in very large excess compared to the trace analytes; thus the chance for interaction with the β-particles is much greater for the former than for the latter molecules. As carrier gases (cg) those with large mass are used, e.g. N2 or Ar (mixed with 5% CH4). According to the reaction scheme given in eq. 31 the β-particles generate thermal, “secondary” electrons by interaction with the carrier gas molecules (these electrons have much lower, namely only thermal energy). β− (31) + − (therm ) cg → cg + e The current produced from these low energy electrons in the detector cell is measured and amplified. As long as there is no analyte with electron affine groups in the carrier gas, the background current (the blank) with about 10-8 A is delivered. When analytes with e.g. halogen substitution are eluted from the column and enter the detector, electrons from the basic current are captured according to eq. 30, and the current is decreased, detecting the analytes in this way. In principle the number of charged particles is not changed in this reaction, because instead of an electron a negatively charged sample ion is generated. However, the velocity of these molecular ions (10 cm/s) is many orders of magnitude smaller than that of the electrons (105 cm/s). For this reason the ions do not reach the collector electrode; they recombine faster with positively charged carrier gas molecules under formation of the neutral molecules according to AX − + cg + = AX + cg (32) However, under such conditions no signal would be obtained. For favourable measuring conditions a pulsed voltage is applied, as indicated in Figure 18. For a closer insight into the signal generation of the ECD we have to differentiate two processes: the one is the simple addition of an electron, as expressed by eq. 30. The more common process, on the other hand, includes a dissociative reaction of the product formed after electron addition according to AX + e − = A• + X − (33) 29 Ernst Kenndler: Gas Chromatography Figure 18 Schematic drawing of the ECD with pulsed polarisation voltage. b pulse width, e.g. 3µs; D pulse distance, e.g. 10-200 µs; h pulse height, e.g. 50 V (from ref. 15 with permission). Therefore not only the electron affinity, but also the binding energy between the carbon atom and the electron affine atom or group decides over the extent of reaction. This fact explains the result of the relative sensitivity, rel. Si, of different compounds as given in the following Table. Detector Response and Sensitivity Substance Rel. Si Binding energy [kJ/mol] Fluorobenzene Chlorobenzene Bromobenzene Iodobenzene 1 100 600 37 000 C-F C-Cl C-Br C-I 538 391 281 210 Electron Affinity [eV] F 4.1 Cl 3.78 Br 3.52 I 3.12 Performance data ECD several 10 fg of analytes (e.g. lindan) per injection linearity 3 to 4 orders of magnitude 30 Ernst Kenndler: Gas Chromatography 2.5.3 Alkali flame ionisation detector The alkali flame ionisation detector, also named thermionic detector, belongs to the group of ionisation detectors in which thermal energy is used as source for ionisation. From the construction point of view it is a modification of the FID, with a pearl of alkali salt (Rb, Cs) located between the flame and the collector electrode. Heating of the alkali salt pearl leads to the emission of alkali atoms. The detector is responding especially to analytes containing nitrogen or phosphorus atoms. According to the special conditions the detector can be run either in the P- or in the N- and P- specific mode. The detailed mechanism of detection is still a matter of question. Here we follow the discussion given by Kolb et al. (cf. e.g. ref. 15). 2.5.3.1 P mode At elevated temperature alkali ions A+ in the (silicate) pearl are neutralised by electrons delivered from the electrical source applied (A+ + e- = A). The neutral alkali atoms, A, evaporate. These atoms are thermally excited giving A*. In the hydrogen flame H• and OH• radicals are formed. However, these two radicals can only recombine to H2O when a partner is present that is able to overtake their energy of formation. This may occur with a triple collision, with excited A* as partner. As a result A* is ionised according to the reaction scheme H • + OH • + A∗ = H 2 O + A+ + e − (34) The alkali cation A+ formed is collected at the negatively charged pearl, whereas the electron is migrating to the collector electrode. In this way the background current of the detector is formed in the P mode. Phosphorus containing molecules, when present, are transformed to phosphorus oxide radicals, R•, in the flame of the detector. These radicals react in a double collision (which is more probable than a triple collision with H• and OH•) with excited A* (35) R • + A∗ = R − + A+ e.g. O = P = O • + A∗ = [O = P = O ] + A+ − (36) The P-containing anion R- reacts further with an OH• radical to R − + OH • = ROH + e − (37) e.g. 31 Ernst Kenndler: Gas Chromatography [O = P = O ] − (38) + OH • = HPO3 + e − This reaction delivers the electron for the signal current specific for P-containing analytes. Figure 19 Schematic presentation of the thermionic detector in the P mode (from ref. 15 with permission). Note that the flame jet is grounded, and has therefore a positive potential. For this reason the electrons formed by reaction of CH- containing molecules due to the same processes as with the FID are not interfering the detector signal, because they are conducted to ground, and do not reach the collector anode. 2.5.3.2 NP mode The N analogous oxides formed from N-containing molecules would decompose rapidly in the flame of the detector, and would therefore not react with alkali. For this reason the detector is run in the NP mode under reducing conditions, leading to CN radicals instead. These conditions are established by decreasing the flow rates of hydrogen to about 1-3 mL/min, and that of air to less than 100 mL/min. The flame goes out then, but the remaining free hydrogen is ignited at the electrically heated pearl. It forms a kind of plasma around the pearl, where a CN radical formed adds an electron taken from excited A*, forming a cyanide ion and alkali cation (see P-mode, eq. 35) CN • + A∗ = CN − + A+ (39) 32 Ernst Kenndler: Gas Chromatography The CN anion formed finally reacts either with H• or OH• radicals to HCN or HCNO, respectively. Figure 20 Schematic representation of the thermionic detector in the NP mode (from ref. 15 with permission). This reaction delivers the electron for the signal current analogous to eq. 37. As the formation of CN radicals is essential, such a structure must be present initially in the analytes to enable detection. Therefore e.g. organic nitro compounds are detectable, but not nitrogen oxides, or nitrate esters. In the following table some data for the performance of the detector in both modes are given. Typical performance data for the NPD in P- and NP- mode, respectively. Parameter P mode NP mode P N Sensitivity 1 C/g P 5 C/g P 0.5 C/g N Limit of detection 5.10-14 g P/s 10-14 g P/s 10-13 g N/s Selectivity (vs. carbon) 106 105 104 Linearity 105 105 105 33 Ernst Kenndler: Gas Chromatography 3 FURTHER READINGS (1) Scott, R. P. W. Introduction to Analytical Gas Chromatography; 2nd ed.; Marcel Dekker, 1998. (2) Jennings, W. G.; Mittlefehldt, E.; Stremple, P. Analytical Gas Chromatography; 2nd ed.; Academic Press, 1997. (3) McNair, H. M.; Miller, J. M. Basic Gas Chromatography; Wiley, 1997. (4) Grant, D. W. Capillary Gas Chromatography; Wiley, 1996. (5) Fowlis, I. Gas Chromatography; 2nd ed.; Wiley, 1995. (6) Scott, R. P. W. Techniques and Practices of Chromatography; 2nd ed.; Marcel Dekker, 1995. (7) Grob, R. L. Modern Practice of Gas Chromatography; 3rd ed.; Wiley, 1995. (8) Baugh, P. E. Gas Chromatography: A Practical Approach; Oxford, 1994. (9) Hinshaw, J. V.; Ettre, L. S. Introduction to Open Tubular Column Gas Chromatography; Advanstar, 1994. (10) Grob, K. Split and Splitless Injection in Capillary Gas Chromatography; 3rd ed.; Hüthig, 1993. (11) Hill, H. H.; McMinn, D. G. Detectors for Capillary Chromatography; Wiley, 1992. (12) Grob, K. On-Column Injection in Capillary Gas Chromatography; 2nd ed.; Hüthig, 1991. (13) Poole, C. F.; Poole, S. K. Chromatography Today; Elsevier, 1991. (14) Baars, B.; Schaller, H. Fehlersuche in der Gaschromatographie; VCH, 1994. (15) Kolb, B. Gaschromatographie in Bildern; Wiley-VCH, New York, 1999. (16) Kenndler, E.; Huber, J. F. K. In Analytiker Taschenbuch; Springer, 1989. 34