Pressure-Tunable GC Columns with Electronic Pressure Control

Heather Smith and Richard Sacks*. Department of Chemistry, University of Michigan, Ann Arbor, Michigan 48109. High-precision selectivity tuning is ach...
0 downloads 0 Views 210KB Size
Anal. Chem. 1997, 69, 5159-5164

Pressure-Tunable GC Columns with Electronic Pressure Control Heather Smith and Richard Sacks*

Department of Chemistry, University of Michigan, Ann Arbor, Michigan 48109

High-precision selectivity tuning is achieved by the use of a series-coupled (tandem) ensemble of a nonpolar and a polar capillary GC column with adjustable pressure at the junction point between the columns. Changing the junction pressure results in a differential change in the holdup times for the two columns. This results in a change in the fractional contribution that each column makes to the overall separation selectivity. An electronic pressure controller is used to obtain greater control and improved tuning precision. With an available tuning range extending from less than 0.60 to greater than 0.85 for the fractional contribution of the nonpolar column, 37 unique, digitally selectable elution patterns are obtained. High-precision tuning is required since fractional contribution changes as small as 0.5% can significantly degrade resolution. With this system, resolution increments of 0.2 or less are achieved for typical critical component pairs. High-speed gas chromatography is undergoing rapid development at several academic, government, and commercial instrumentation laboratories.1-8 The introduction of highly efficient inlet systems that have very low injection time jitter and that can introduce very narrow vapor plugs into a capillary column have sparked this activity.1,4-6,8,9 With the use of these inlet systems and relatively short capillary separation columns, some simple mixtures can be completely separated in a few seconds to a few tens of seconds. However, the high probability of coelutions and the reduced peak capacity with shorter columns10 makes it much more difficult to apply these technologies to more complex mixtures. Enhanced selectivity is needed to use high-speed GC techniques for more complex mixtures.2,11,12 The most powerful approach is to tailor the selectivity for a specified set of target compounds. This can be accomplished by the use of serially linked (tandem) ensembles of capillary GC columns of very different selectivities.13,14 Adjustment of the column lengths15 or provision of a carrier gas supply with adjustable pressure16,17 at (1) Phillips, J.; Liu, Z.; Lee, R. J. Chromatogr. Sci. 1986, 24, 396. (2) Liu, Z.; Sirimanne, S.; Patterson, D., Jr.; Needham, L.; Phillips, J. Anal. Chem. 1994, 66, 3036. (3) Ehrmann, E.; Dharmasena, H.; Carney, K.; Overton, E. J. Chromatogr. Sci. 1996, 34, 533. (4) Klemp, M.; Akard, M.; Sacks, R. Anal. Chem. 1993, 65, 2516. (5) Klemp, M; Peters, A; Sacks, R. Environ. Sci. Technol. 1994, 28, 428A. (6) Sacks, R.; Klemp, M.; Akard, M. Field Anal. Chem. Technol. 1996, 1, 97. (7) Levine, S.; Berkley, R.; Ke, H. J. Air Waste Manage. Assoc. 1992, 42, 1446. (8) van Es, A.; Janssen, J.; Bally, R.; Cramers, C.; Rijks, J. J. High Resolut. Chromatogr. 1987, 10, 273. (9) Gaspar, G.; Arpino, P.; Guichon, G. J. Chromatogr. Sci. 1977, 15, 256. (10) Davis, J.; Giddings, J. Anal. Chem. 1983, 55, 418. (11) Liu, Z.; Phillips, J. J. Microcolumn Sep. 1989, 1, 249. (12) Sacks, R.; Akard, M. J. Environ. Sci. Technol. 1994, 28, 428A. S0003-2700(97)00453-8 CCC: $14.00

© 1997 American Chemical Society

the junction point between two capillary columns then is used to adjust the fractional contributions that the two stationary-phase chemistries make to the overall selectivity. While column length adjustment is simple, gas compression effects result in significant differences between the actual required column length fractions and the optimal stationary-phase fractions determined by the use of window diagram optimization procedures.18,19 The use of serially linked columns with adjustable junction pressure (tuning pressure) for high-speed GC has been described in detail.20-21 Dramatic reductions in separation time have been reported for the analysis of multifunctional mixtures. The complete separation of 15 components in under 20 s has been achieved.22 The selection of the tuning pressure that will give the best complete separation in the shortest time is based on the use of resolution window diagrams where the relative resolution21 of the most difficult to separate pair of components (critical pair) is plotted vs the tuning pressure. The tuning pressure resulting in the largest critical pair relative resolution then is used for the analysis. Despite the potential for much greater selectivity and the attendant reduction in analysis times, tandem capillary columns are not often used. There are several reasons for this. For the high-speed analysis of more complex mixtures, a column length fraction error of a few percent or less can represent the difference between adequate resolution and complete coelution of the critical pair. Differential aging of the columns in the ensemble or replacement of the columns will result in changes in the effective phase fraction and thus detuning of the tandem ensemble. Previous work using pressure-tunable column ensembles for high-speed GC20-22 has used manual pressure control for selectivity tuning. In these studies, pressure reproducibility was typically in the (1-2% range. For some critical pairs that show a high sensitivity of resolution to tuning pressure, this is inadequate. In addition, manual pressure adjustment is cumbersome and is not suitable for automated, closed-loop selectivity control. Electronic pressure control has been used recently in several applications to improve the performance of GC systems. Hermann et al.23 have used electronic inlet pressure control for inlet pressure programming in capillary GC. Overton et al.24 have applied (13) Jones, J.; Purnell, J. Anal. Chem. 1990, 62, 2300. (14) Maurer, T.; Engewald, W.; Steinborn, A. J. Chromatogr. 1990, 517, 77. (15) Purnell, J.; Jones, J.; Wattan, M. J. Chromatogr. 1987, 399, 99. (16) Deans, D.; Scott, I. Anal. Chem. 1973, 45, 1137. (17) Benicka, E.; Krupcik, J.; Kuljovsky, P.; Repka, D.; Garaj, J. Microchim. Acta 1990, 111, 1-10. (18) Purnell, J.; Williams, P. J. Chromatogr. 1984, 292, 197. (19) Ingraham, D.; Shoemaker, C.; Jennings, W. J. Chromatogr. 1982, 239, 39. (20) Akard, M.; Sacks, R. Anal. Chem. 1994, 66, 3036. (21) Akard, M.; Sacks, R. Anal. Chem. 1995, 67, 2733. (22) Akard, M.; Sacks, R. Anal. Chem. 1996, 68, 1474.

Analytical Chemistry, Vol. 69, No. 24, December 15, 1997 5159

Table 1. Compounds Used in the Test Mixture

Figure 1. Experimental system used for electronic pressure control. Key: C1, nonpolar column; C2, polar column; PC, electronic pressure controller; CG, carrier gas inlets; P1-P3, pressure measurement; R, capillary flow restrictors; FID, flame ionization detectors.

electronic inlet pressure control to field-portable GC instruments. Sippola et al.25 used electronic pressure control at the junction point between two columns to improve heart-cutting precision with multidimensional GC. This report describes the use of electronic pressure control for high-precision selectivity tuning. The objective of this work was to significantly increase the tuning resolution and repeatability of a pressure-tunable tandem column ensemble for high-speed GC. EXPERIMENTAL SECTION Apparatus. Figure 1 shows the high-speed GC instrument used in this study. The electronically controlled cryofocusing inlet system (Chromatofast Model L Cryointegrator) has been described in detail.4-6 It generates injection vapor plug band widths (σ) of ∼10 ms with a shot-to-shot injection time jitter of a few milliseconds. The inlet is interfaced to a Varian 3700 GC equipped with two flame ionization detectors (FID). Hydrogen carrier gas is provided at points labeled CG. Pressure is monitored at points P1-P3. Components labeled R are fused silica-capillary restrictors. The column ensemble consists of a pair of 0.25-mm-i.d. capillaries using 0.25-µm bonded stationary phases. The column labeled C1 is 6.0 m long and uses a nonpolar (DB-5) stationary phase. Column C2 is 5.25 m long and uses a polar (DB-Wax) stationary phase. While only one detector is needed for operation of pressure-tunable columns, both available detectors were used to obtain independent retention time and holdup time measurements from column C1. About 40% of the sample eluting from C1 is split to FID2 through a capillary restrictor. The carrier gas pressure at the junction point between the columns is controlled by a MKS 640A Series electronic pressure controller (MKS Instruments, Andover, MA). The device is an absolute pressure capacitance manometer which controls pressure in the range from 0 to 687 kPa (100 psi) with a repeatability of (0.7 kPa (0.1 psi) using a 0-5-V input signal. Column head pressure (P2) and midpoint pressure (P3) were measured with Omega DP280-P10 meters equipped with PX302-050 AV transduc(23) Hermann, B.; Freed, L.; Thompson, M.; Phillips, R.; Klein, K.; Snyder, W. J. High Resolut. Chromatogr. 1990, 13, 361. (24) Ehrmann, E.; Dharmasena, H.; Carney, K.; Overton, E. Electronic Pressure Programming for Fast Portable Gas Chromatography. Pittsburgh Conference on Analytical Chemistry and Applied Spectroscopy, Atlanta GA, 1997; Paper 705. (25) Sippola, E.; Himberg, K.; David, F.; Sandra, P. J. Chromatogr., A 1994, 683, 45.

5160 Analytical Chemistry, Vol. 69, No. 24, December 15, 1997

label

compound

bp, °C

A B C D E F G H I J K L M N

tert-butyl methyl ether tert-butyl ethyl ether isobutyl vinyl ether butyl ethyl ether methanol tetrachloromethane ethanol cis-1,2 dichloroethylene trichloroethylene trichloromethane tetrachloroethylene toluene 2-fluorotoluene isobutyl alcohol

55.2 73.1 83 96 65 76.5 78.5 60.3 87 61.7 121 110.6 114 108

ers. Carrier gas pressure was measured with a DP2000-PV meter equipped with a PX304-050A5V transducer. A small carrier gas flow is provided through a restrictor between the inlet and column C1 in order to prevent band broadening and contamination by sample diffusion into the P2 measurement line. Both detectors were interfaced to a Pentium 90-MHz computer through an electrometer/amplifier with a 5-ms time constant and a 12-bit A/D board. A sampling rate of 100 Hz was used for all measurements. Materials and Procedures. All compounds in the test mixture were reagent grade or better. The mixture used for this study is described in Table 1. This multifunctional mixture was used because the numerous coelutions that occur over the available tuning range provide a good test for the reproducibility and predictability of elution patterns from the tandem column ensemble using electronic pressure control. The carrier gas supply was passed through filters for hydrocarbons, oxygen, and water vapor. Samples were prepared in Saran gas sampling bags. Liquid samples were injected into the bag using a microsyringe, diluted with dry air, and allowed to equilibrate for 1 h. All connections in the sample flow path are made with low dead volume, all glass unions, and Y-splitters. For sample collection and focusing, the trap tube in the inlet system was cooled to a temperature of -90 °C. For injection, the trap tube was heated to ∼250 °C by the use of a capacitive discharge power supply. About 0.1 mL of sample vapor was collected in the trap. RESULTS AND DISCUSSION Hinshaw and Ettre26,27 have described the fundamental relationships and implementation of GC using tandem columns with pressure-tunable selectivity. A comprehensive bibliography of work described prior to 1986 is found in ref 26. Tunable selectivity is achieved by the use of a tandem combination of a nonpolar column np followed by a polar column p. The selectivity of the tandem ensemble is determined by the selectivities of the individual columns and the fractional contributions fnp and fp that the respective columns make to the overall separation. The fractional contributions are given by the holdup time fractions as described by eqs 1 and 2, where tmnp and tmp are the holdup times for the nonpolar and the polar columns, respectively. Overall retention time tRov for a specified component then is given (26) Hinshaw, J.; Ettre, L. Chromatographia 1986, 21, 561. (27) Hinshaw, J.; Ettre, L. Chromatographia 1986, 21, 669.

fnp ) tmnp/(tmnp + tmp)

(1)

fp ) 1 - fnp ) tmp/(tmnp + tmp)

(2)

by eq 3, where knp and kp are the retention factors for the

tRov ) tmov[fnp(knp + 1) + fp(kp + 1)]

(3)

component on the two columns, and tmov is the overall holdup time from the column ensemble. It is common practice to report retention data in terms of retention factors. For a nominal column type (specified stationary phase and phase volume ratio), retention factors are reasonably consistent, and elution order at a specified temperature is quite predictable from library values of retention factors. With tunable tandem column ensembles, however, elution patterns can be very sensitive to the values of fnp and fp. In these cases, column to column reproducibility may not be adequate, and empirically determined retention factor values based on the use of two detectors may provide more accurate tuning. In the present study, retention is expressed in terms of the overall retention factor kov as defined in eq 4.

kov ) fnpknp + fpkp

(4)

The range of selectivities attainable with a tandem column ensemble is bounded by the selectivities of the individual columns. Thus, a greater range of selectivities is available if the selectivities of the two columns are very different. The combination of polydiphenyl(5%)dimethylsiloxane and poly(ethylene glycol) has proven very useful.12,28 Because of gas compression effects, the required column length fractions for column length tuning may be significantly different from the required fraction values obtained from holdup time measurements. This makes column length tuning tedious and time consuming. Pressure tuning with electronic pressure control is far more convenient. In addition, if the column ensemble becomes detuned because of differential deterioration of the columns, retuning is very easy. However, the tuning range attainable with pressure tuning is more restricted because of gas pressure considerations. The tuning pressure (P3 in Figure 1) must be significantly less than the column head pressure (P2). As P3 approaches P2, the holdup time for the first column becomes unacceptably long, and column efficiency decreases sharply. In addition, P3 must be sufficient to ensure a carrier gas flow into the junction point between the columns. Otherwise, sample loss and contamination of the pressure control apparatus will occur. For the case of equal column lengths and phase volume ratios, the available tuning range extends from about 0.55 to 0.90 contribution from the first (nonpolar) column. If the column order is reversed, the nonpolar column contribution is adjustable in the range from about 0.10 to 0.45. Note that gas compression effects always result in a larger contribution from the first column than would be indicated from the length ratio of the columns, and equal contributions from the two columns cannot be achieved with equal column lengths and phase volume ratios. By changing the actual (28) Sandra, P.; Proot, F.; Diricks, G.; Verstappe, M.; Verzele, M. J. High Resolut. Chromatogr. 1985, 8, 782.

Figure 2. Chromatograms from the two detectors. Left, chromatogram from FID2 for the separation from the nonpolar (DB-5) column only; right, chromatogram from FID1 for the separation from the tandem column ensemble. Lines connect corresponding peaks from the two chromatograms. Letters labeling peaks are defined in Table 1.

column lengths, other tuning ranges are obtained. For the work described here, the tuning range 0.61-0.85 (nonpolar) was studied in detail. Figure 2 shows high-speed chromatograms for this mixture from FID1 (right) and FID2 (left). Note that these are high-speed separations, and the last component of the test mixture elutes from the column ensemble in less than 30 s. The tuning pressure was set at 20.0 psi. Peak labels are the same as in Table 1. Lines connect corresponding peaks in the two chromatograms. Note the large number of elution order changes between the two chromatograms. From the holdup time for the first (nonpolar) column and the overall holdup time for the tandem ensemble, the fractional contribution from the nonpolar column (eq 1) was determined to be 0.613. The useful range of tuning pressures available for the column lengths used in this study extended from 20.0 to 23.7 psi. Since the pressure controller has a resolution of 0.1 psi, 37 discrete tuning pressures are available, each obtained by a 5-mV change in the controller input. For each chromatogram, the holdup time fraction was computed. In Figure 3, the nonpolar fraction calculated from eq 1 is plotted vs the tuning pressure for all 37 chromatograms. A linear regression line is also shown. The correlation coefficient is 0.9981. Slight curvature is suggested by the data, but the largest fraction error resulting from nonconformity to the linear regression line is less that 0.005. For these 37 chromatograms, the average fraction step size is 0.0065 with a standard deviation of 0.0024. In Figure 4, overall retention factor values from the 37 chromatograms are plotted vs the nonpolar fraction for the 14 components in the test mixture. The letters used to identify the Analytical Chemistry, Vol. 69, No. 24, December 15, 1997

5161

Figure 3. Calculated nonpolar phase fraction vs tuning pressure for 37 chromatograms. The nonpolar phase fraction was computed as the holdup time fraction for the nonpolar column.

Figure 5. Extrapolation of the plots in Figure 4 to estimate the overall retention factors on the individual columns. The broken vertical lines bound the fraction range considered in Figure 4. Table 2. Measured and Calculated Retention Factors for DB-5 knp

Figure 4. Plots of overall retention factors for the components of the test mixture vs the fraction of DB-5.

plots are the same as in Table 1. Linear regression lines also are shown. The plots are all very linear with correlation coefficients in the range 0.9987-0.9138. Where two plots intersect, coelution of the corresponding peaks occurs. For this multifunctional mixture, many coelutions occur over the fraction range investigated. In these regions, reliable retention data cannot be obtained, and gaps are present in the the plots of Figure 4. Retention data can be obtained in these regions from single-component injections, but these data are not required for the method described here. Precision in retention time and holdup time measurements is limited by the 10-ms sampling interval used here. By extrapolation of the plots in Figure 4 to fraction values of 0 and 1.0, the entire tuning range can be examined. This is shown in Figure 5. The broken vertical lines correspond to the range of fractions shown in Figure 4. The retention pattern along the left vertical axis (fnp ) 0) corresponds to the chromatogram expected from the polar column only, and the pattern along the right axis (fnp ) 1.0) corresponds to the chromatogram expected for the nonpolar column only. The apparatus used here provides direct 5162

Analytical Chemistry, Vol. 69, No. 24, December 15, 1997

knp

label

regression

measured

label

regression

measured

A B C D E F G

1.25 1.41 1.56 1.77 1.05 1.60 1.08

1.26 1.41 1.55 1.76 1.04 1.55 1.08

H I J K L M N

1.37 1.81 1.40 3.10 2.47 2.65 1.43

1.41 1.80 1.41 3.03 2.42 2.59 1.41

measurement of the chromatograms produced for the nonpolar column alone. Table 2 compares the measured retention factors and the values from the extrapolations in Figure 5 for the 14 mixture components. In all cases, the values are in good agreement. In a recent report29 it was shown that retention patterns from tandem column ensembles could be predicted by the use of plots of retention factors for one column vs values for the other column. A similar approach is used in this work to determine the ability of electronic pressure tuning to achieve the retention patterns predicted by this method. Retention factors for the two columns are obtained from the end points of the extrapolations in Figure 5. That is, the extrapolated values for a phase fraction of 0 (only the polar column used) are plotted vs values for a phase fraction of 1.0 (only the nonpolar column used). The basis for the prediction of retention patterns is illustrated in Figures 6 and 7. Each compound in the test mixture occupies a single point in the retention plane. The points are labeled as in Table 1. Also shown in each figure is a phase fraction axis F, the angle of which with respect to the vertical retention axis (θ) is uniquely defined by the phase fraction ratio as described in eq 5.

tan (θ) ) fnp/(1 - fnp)

(5)

The orthogonal projections of the data points onto the phase

Figure 6. Retention factors on the polar column vs values on the nonpolar column for the components of the test mixture. The phase fraction axis F makes an angle of 57.8° with respect to the polar axis. Lines show orthogonal projections onto F. The chromatogram was obtained with the corresponding phase fraction.

Figure 7. Retention factors on the polar column vs values on the nonpolar column for the components of the test mixture. The phase fraction axis F makes an angle of 79.7° with respect to the polar axis. Lines show orthogonal projections onto F. The chromatogram was obtained with the corresponding phase fraction.

fraction axis gives the retention pattern expected from the tandem ensemble using the specified phase fractions. Figure 6 is for the case of a phase fraction (nonpolar) of 0.613. The corresponding value of θ is 57.8°. Shown along the phase fraction axis is the chromatogram obtained by adjusting the tuning pressure to 20.0 psi. This is the value indicated in Figure 3 to obtain the required phase fraction. The time axis of the chromatogram was adjusted to give the best match with the points projected onto the phase fraction axis. The point projections (29) Smith, H.; Sacks, R. Anal. Chem. 1997, 69, 145.

Figure 8. Portions of 21 chromatograms showing components M and K for tuning pressure changes of 0.1 psi. Note that the elution order is reversed on the left and right flanks of the figure.

predict the near coelutions of component pairs B-E, C-G, and D-F. This is confirmed in the chromatogram. The locations of all peaks in the chromatogram match very well with the locations of the projections on the phase fraction axis. In Figure 7, the phase fraction axis has been rotated to give a θ value of 79.7°. This corresponds to a nonpolar phase fraction of 0.846. Again, when the pressure was set at the appropriate value from Figure 3, the locations of the peaks in the resulting chromatogram were in excellent agreement with the projections on the phase fraction axis. Note that components B and E are predicted to be well separated, while in Figure 6, coelution is indicated. When Figures 6 and 7 are compared, many elution order changes are predicted, and in all cases, they are observed in the chromatograms. Note in Figures 6 and 7 that peaks marked K and M change elution order and are in an uncongested region of the chromatograms. In Figure 8, this pair of peaks is shown from 21 consecutive chromatograms, each representing a change in fraction of ∼0.0065. As the fraction is increased from left to right, the peaks move progressively closer together until they very nearly coelute. Continuing to the right, the peaks again separate but with a reversal of elution order. This is confirmed by the differences in peak heights for the components. Note the gradual decrease in peak heights observed when the left most and the right most chromatograms in Figure 8 are compared. This is the result of gradual dilution of the sample in the gas sampling bag by a small flow of carrier gas into the bag from a purge line in the cryofocusing inlet system. The degree of peak position control attainable with electronic pressure tuning is noteworthy. The difference in resolution between successive chromatograms in Figure 8 is ∼0.20 resolution unit. Since the repeatability (σ) of the phase fraction setting is ∼(0.0024, the 95% confidence interval for the phase fraction is ∼(0.0048. For the peak pair shown in Figure 8, this corresponds to a resolution uncertainty (2σ) of ∼(0.15. Analytical Chemistry, Vol. 69, No. 24, December 15, 1997

5163

The sensitivity of the resolution to changes in phase fraction is different for every critical pair. For the multifunctional mixture described here, there are over 40 different critical pairs (crossing points for the plots in Figure 5) over the entire tuning range. For the limited tuning range used in this study, there are 12 different critical pairs. The sensitivity of the resolution to changes in phase fraction is determined by the average retention for the two components and the acute angle at which the critical pair plots (Figure 4) cross. A detailed study is in progress. CONCLUSIONS The use of electronic pressure control for pressure-tunable tandem column ensembles used for high-speed GC results in significantly more precise tuning than has previously been achieved with manual methods. While the techniques described in this report were applied to relatively fast separations using a column ensemble with a total resolving power of ∼15 000 theoretical plates, the method is easily scaled up by the use of longer columns to obtain more peak capacity. Many GC separations could benefit from shorter separation times using electronically controlled tunable columns. High-speed separations using relatively short columns are more demanding regarding tuning pressure precision. This is because the total pressure along the column ensemble is smaller than for longer columns, and thus a specified pressure uncertainty corresponds to a larger range of phase fractions. Since the calculation of phase fraction in this study is based on holdup time

5164

Analytical Chemistry, Vol. 69, No. 24, December 15, 1997

measurements, accurate measurements are required. An electronically controlled inlet system with low injection time jitter is necessary for high-precision tuning of high-speed separations. With the 100-Hz sampling rate used here, fraction calculation uncertainty of ∼(0.001 is obtained. This is a major component in the overall tuning uncertainty, and a faster data system will be considered for further studies. Electronic pressure tuning also will be very useful for closedloop tuning control. With closed loop control using critical pair separation measurements, it should be possible to automate the entire tuning operation. An additional very important advantage of closed-loop tuning control is that it should provide for the development of methods to maintain accurate tuning despite differential aging of the columns in the ensemble and occasional replacement of one or both columns. ACKNOWLEDGMENT The authors greatfully acknowledge the Dow Chemical Co, Midland, MI, for use of the high-speed inlet system and the electronic pressure controller.

Received for review May 1, 1997. Accepted September 17, 1997.X AC970453V X

Abstract published in Advance ACS Abstracts, November 15, 1997.