Anal. Chem. 1998, 70, 4960-4966
Column Selectivity Programming and Fast Temperature Programming for High-Speed GC Analysis of Purgeable Organic Compounds Heather Smith and Richard D. Sacks*
Department of Chemistry, University of Michigan, Ann Arbor, Michigan 48109
High-speed gas chromatograms are obtained by the use of relatively short lengths of capillary column operated at relatively large carrier gas flow rates. This approach is difficult for more complex mixtures because of the reduced peak capacity available with shorter columns. A solution to this problem is the use of tunable column ensembles consisting of the series (tandem) combination of a polar and a nonpolar column. By adjusting the pressure at the junction point between the columns, the selectivity of the ensemble can be adjusted within the limits imposed by the individual columns. For mixtures representing a relatively large boiling point range and containing more than ∼20 components, high-speed, isothermal separations are less effective. These limitations are significantly reduced by combining fast temperature programming with selectivity programming. Selectivity programming is obtained by changing the pressure at the column junction point one or more times during the course of an analysis. In the work described here, the column ensemble temperature and the junction pressure are initially set to give a high-quality separation of the earliest eluting components. After these components have eluted, a linear temperature ramp of ∼35 °C/min is initiated. As the temperature increases, the pressure is adjusted to change the selectivity and thus facilitate the separation of groups of components as they migrate through the column ensemble. Using this approach, a mixture of 30 purgeable organic compounds is separated in less than 2.5 min. High-speed gas chromatography (HSGC) can dramatically reduce analysis times for volatile and semivolatile organic compounds (VOCs).1-5 This should result in increased sample throughput and reduced analysis costs for many environmental applications. For process monitoring, HSGC should result in better process control and thus improved product quality. Highspeed separations are obtained by using relatively short capillary columns operated at higher than usual carrier gas flow rates. (1) Klemp, M.; Peters, A.; Sacks, R. J. Environ. Sci. Technol. 1994, 28, 369A. (2) Sacks, R.; Klemp, M.; Akard, M. Field Anal. Chem. Technol. 1997, 1, 97. (3) Van Es, A. High-Speed Narrow Bore Capillary Gas Chromatography; Huthig Buch Verlag: Heidelberg, 1992. (4) Ehrmann, E. U.; Dharmasena, H. P.; Carney, K.; Overton, E. B. J. Chromatogr. Sci. 1996, 34, 533. (5) Akard, M.; Sacks, R. J. Chromatogr. Sci. 1994, 32, 499.
4960 Analytical Chemistry, Vol. 70, No. 23, December 1, 1998
However, these conditions result in reduced peak capacity and thus increased probability of peak overlap.6 This makes it difficult to apply HSGC methods to mixtures of more than a few components. Series-coupled (tandem) ensembles of two column types of very different selectivity have been successfully applied to highspeed separations of mixtures containing 10-20 components.7-9 By adjusting the carrier gas pressure at the junction point between the columns, the selectivity of the ensemble can be changed. Optimization procedures10-13 and models14,15 of multiphase separations have been developed to aid in the determination of the junction point pressure and column temperature needed to minimize the time required for the complete separation of a specified set of target compounds. The majority of HSGC studies have used isothermal operation. Algorithms have been reported for the simultaneous optimization of column selectivity and isothermal temperature.16,17 Using these techniques, mixtures containing 15-20 components have been completely separated in a time frame of 30-60 s.18 For more complex mixtures, two significant problems have limited the application of these techniques. First, as the number of components in the mixture is increased, the probability that a singlecolumn junction pressure will result in a complete separation of all components is decreased.9 This problem is exacerbated as the peak saturation6 (ratio of number of components to peak capacity) increases. Second, for mixtures covering a relatively wide boiling point range, isothermal operation results in longer analysis times and reduced peak detectability, as is the case for conventional isothermal GC. (6) Davis, J. M.; Giddings, J. C. Anal. Chem. 1983, 55, 418. (7) Akard. M.; Sacks, R. Anal. Chem. 1994, 66, 3036. (8) Akard, M.; Sacks, R. Anal. Chem. 1995, 67, 2733. (9) Sacks, R.; Akard, M. J. Environ. Sci. Technol. 1994, 28, 428A. (10) Purnell, J. H.; Williams, P. S. J. Chromatogr. 1984, 292, 197. (11) Purnell, J. H.; Williams, P. S. J. Chromatogr. 1985, 321, 249. (12) Othman, M. Y. B.; Purnell, J. H.; Wainwright, P.; Williams, P. S. J. Chromatogr. 1984, 289, 1. (13) Hevesi, T.; Krupcik, J.; Chretien, J. R. Fresenius J. Anal. Chem. 1995, 352, 643. (14) Smith, H.; Sacks, R. Anal. Chem. 1997, 69, 145. (15) Sacks, R.; Smith H.; Nowak, M. Anal. Chem. 1998. 70, A29. (16) Kaiser, R. E.; Rieder, R. I. HRC & CC, High Resolut. Chromatogr. Chromatogr. 1979, 2, 416. (17) Matisova, E.; Kovacicova, E.; Garaj, J.; Kraus, G. Chromatographia 1989, 27, 494. (18) Akard, M.; Sacks, R. Anal. Chem. 1996, 68, 1474. 10.1021/ac980463b CCC: $15.00
© 1998 American Chemical Society Published on Web 10/28/1998
EXPERIMENTAL SECTION Apparatus. A simplified drawing of the apparatus is shown in Figure 1. A Varian model 3700 GC equipped with two flame ionization detectors is used as a platform for the complete instrument. The inlet is a high-speed cryofocusing device (Cryointegrator model L, Chromatofast, Ann Arbor, MI). Injection plug widths (σ values) typically are on the order of 10 ms. With this inlet system, injection is computer controlled with a shot-toshot injection time jitter of less than 10 ms. The column ensemble consists of a 5-m-long, 0.25-mm-i.d. DB-5 (nonpolar) column C1 (J&W Scientific) followed by a 5-m-long, 0.25-mm-i.d. StableWax (polar) column C2 (Restek Corp.). Both columns use 0.25-µm stationary-phase film.
The Varian detectors were used without change. Detector FID1 shown in Figure 1 monitors the final output from the tandem column ensemble. Detector FID2 is connected to the column junction point by means of a 25-cm-long, 0.1-mm-i.d. deactivated fused-silica tube R. About 40% of the effluent from the first (nonpolar) column is split to this detector. This detector is used to measure the holdup time for the first column. Hydrogen carrier gas is introduced at points labeled CG. The pressure at the junction point between the columns is controlled by an MKS model 640A electronic pressure controller PC (MKS Instruments, Andover, MA). This device controls the absolute pressure in the range 0-100 psi with a resolution of 0.1 psi. The column head pressure and the junction pressure were measured with Omega DP280-P10 meters equipped with PX302-050AV transducers. Carrier gas pressure was measured with a DP2000PV meter equipped with a PX304-050A5V transducer. The GC oven temperature was monitored by the use of a thermocouple located inside the oven. An Omega model 115JC meter was used. The meter sensitivity is 1.0 mV/°C. A Pentium-90 computer with a 16-bit A/D board (National Instruments PCIMD16XE-50) was used to control the pressure program and data acquisition from the two detectors as well as sample collection and injection by the cryofocusing inlet system. The resolution of the analog output voltage applied to the pressure controller was ∼0.08 mV. LabView software was used to control the instrument. An electrometer/amplifier, built in-house and having a 5-ms time constant, was used to interface the detectors to the A/D board. The temperature program was initiated manually. Procedures and Materials. Mixtures of VOCs were prepared by mixing approximately equal volumes of the neat compounds. Aliquots of ∼15 µL of the mixtures were injected into 12 in. × 12 in. Saran gas sampling bags and diluted with dry nitrogen. At least 1 h of equilibration time was allowed before use. Individual components were obtained from Aldrich. Table 1 lists the VOCs and their boiling points. The cryofocusing inlet device uses a vacuum pump to sample directly the atmospheric pressure vapor in the bag. For each analysis, the VOCs from ∼0.4 mL of sample gas were collected. The trapping temperature was typically -90 °C, and the injection temperature was ∼200 °C. Hydrogen carrier gas was purified with filters for water vapor, oxygen, and hydrocarbons. Average carrier gas velocities in the two columns vary with the junction-point pressure. All calculations for window diagrams were done in an Excel spreadsheet. For each junction-point pressure investigated, overall retention times for the column combination were logged in for all peaks along with overall holdup time and the holdup time for the first column only. Unambiguous peak identification was obtained by spiking mixtures with individual components. Data were combined for several column temperatures. A window diagram procedure was used to select the operating conditions for the 12 least-retained compounds, which were separated under both isothermal and isobaric (constant junction-point pressure) conditions.
(19) Smith, H.; Sacks, R. Anal. Chem. 1997, 69, 5159. (20) Hinshaw, J. V.; Ettre, L. S. Chromatographia 1986, 21, 669. (21) Harris, W. E.; Habgood, H. W. Programmed Temperature Gas Chromatography; Wiley: New York 1966; pp 201-2.
RESULTS AND DISCUSION Selectivity Tuning. The use of tunable column ensembles with electronic pressure control at the column junction is a
Figure 1. Simplified diagram of the HSGC instrument using electronic control of column junction-point pressure: inlet, cryofocusing inlet system; C1, nonpolar column; C2, polar column; PC, electronic pressure controller; FID, flame ionization detectors; R, fused-silica capillary restrictor; CG, carrier gas inlets.
These limitations can be reduced by a combination of two operational strategies, programmable selectivity and fast temperature programming. If the column junction pressure is changed during the course of an analysis, selectivity programming is achieved. Here, the pressure can be set initially to give a good separation of the earliest eluting components, and after they elute, the pressure is changed to facilitate the separation of the next eluting group of components. This process can be repeated as many times as necessary in order to achieve a high-speed separation. Using electronic pressure control, the junction-point pressure can be set very accurately and reproducibly.15,19,20 For the relatively short capillary columns used for HSGC (typically 5-10 m), relatively high-temperature programming rates should be used, since faster separations are achieved without significant loss of resolution.21 In this report, a three-step pressure program is combined with a 35 °C/min temperature ramp initiated after a short isothermal run in order to separate a mixture of 30 purgeable VOCs in less than 2.5 min. A window diagram procedure8,10,18 was used to determine the optimal column junction pressure and isothermal temperature for the separation of the first 12 components. An empirical approach then was used to establish the selectivity program and the temperature program needed for completion of the analysis.
Analytical Chemistry, Vol. 70, No. 23, December 1, 1998
4961
Table 1. Mixture Components and Boiling Points label
compound
bp (°C)
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31
1,1-dichloroethene methanol ethanol methylene chloride 1,1,1-trichloroethane carbon tetrachloride cis-1,2-dichloroethene benzene chloroform trichloroethene 1,2-dichloroethane 1,2-dichloropropane toluene bromodichloromethane 2-fluorotoluene trans-1,3-dichloropropane tetrachloroethene 1,1,2-trichloroethane ethylbenzene p-xylene chlorobenzene o-xylene 3-ethyltoluene 1,3,5-trimethylbenzene 3-chlorotoluene 2-ethyltoluene 1,2,4-trimethylbenzene bromoform 1,2,3-trimethylbenzene 1,3-dichlorobenzene 1,2-dichlorobenzene
31.7 64.6 78.5 40 75 76 60.1 80 61 87 83.5 96.8 110.6 87 114 112 121 113.5 136.2 138.4 132 144.4 162.5 164.7 162 162 171.5 149.5 176.1 172 182
powerful approach for the high-speed separation of multifunctional mixtures. By adjusting the pressure at the column junction, the relative contributions that the two columns make to the overall separation selectivity is changed. With the system described here, which uses two 5-m-long, 0.25-mm-i.d. columns, the fractional contribution of the first (nonpolar column) can be adjusted in the range 0.55-0.95. The corresponding junction point pressure ranged from 20 to 26 psia. Using electronic pressure control with a resolution of 0.1 psi, this 6.0 psi tuning range can generate 60 unique retention patterns. Figure 2 shows portions of a series of chromatograms obtained with tuning pressure values changing by 0.2 psi for each consecutive analysis. The peak numbers correspond to the numbers in Table 1. All chromatograms were obtained at a column temperature of 70 °C. Chromatogram a was obtained with a tuning pressure of 22.3 psia, which corresponds to a nonpolar column contribution of 0.727. For chromatogram b, the pressure was increased to 22.5 psia, resulting in a change of nonpolar fraction to 0.741. Note that peak 17 is completely buried under peak 19. This is also indicated by the increase in height of peak 19. For chromatogram c, the pressure was increased to 22.7 psia, and peak 17 is causing some distortion of the left-hand edge of peak 19. For (d) (22.9 psia), peak 17 is under the right-hand edge of peak 18. The process continues, and for chromatogram f (23.3 psia), peak 17 reappears but to the left of peak 18. Finally, chromatogram h (23.7 psia) provides the best separation of the three compounds. Further increases in pressure results in reduced resolution of peaks 18 and 19. Figure 2 clearly illustrates the utility and the degree of relative peak position control obtained using electronic pressure control. Note that, for a 0.1 psi pressure change, which is the resolution of the pressure control device, the fraction change is only ∼0.007. If isothermal temperature is also varied, many additional unique 4962 Analytical Chemistry, Vol. 70, No. 23, December 1, 1998
Figure 2. Portions of chromatograms obtained at different junctionpoint pressures. See Table 1 for peak identification. Pressure values increased in 0.2 psi steps from 22.3 psia for (a) to 23.7 psia for (h).
retention patterns are available. The goal of the optimization strategy is the identification of the temperature and junction-point pressure values that will give the fastest complete separation. With these two-dimensional optimization procedures, mixtures containing 15-20 components have been separated in a time frame of 30-60 s.7-9,18 Window-Diagram Optimization. Optimization procedures usually use a critical pair analysis, where the column junctionpoint pressure is adjusted to give the best separation of the most difficult to separate component pair (critical pair). For an n-component mixture, there are (n2 - n)/2 unique peak pairs, all of which must be considered as the possible critical pair. For a complex, multifunctional mixture, many different peak pairs will be the critical pair for different combinations of column temperature and junction-point pressure. For a 15-component mixture, there are 105 possible coelution pairs. For a 30-component mixture, the number jumps to 435, and the probability that a pressure-temperature combination exists that will result in adequate separation of all mixture components drops accordingly. For the column combination, an overall retention factor k can be described based on overall retention times and the overall holdup time for the column combination. These overall retention factors are linear functions of the fractional contribution that either column makes to the overall selectively.22,23 The fractional contributions Fp and Fnp for the polar and nonpolar columns, respectively, are found from the holdup times, tmp and tmnp for the individual columns, respectively.
Fnp ) tmnp/(tmnp + tmp)
(1)
Fp ) tmp/(tmnp + tmp)
(2)
With the instrument used in this study, the overall holdup time (tmnp + tmp) and tmnp can be measured for every experiment. (22) Mayer, H.; Karpathy, O. J. Chromatogr. 1962, 8, 308. (23) Purnell, J.; Jones, J.; Wattan, M. J. Chromatogr. 1987, 399, 99.
Figure 3. Overall capacity factor valves k versus the fractional contribution of the nonpolar column Fnp for the 31 components of the test mixture. Oven temperature was 70 °C.
Figure 3 shows plots of overall retention factors versus Fnp values for all peaks from ∼180 chromatograms. Overlapping data were obtained over several days. Small day-to-day changes in atmospheric pressure as well as column head pressure result in small changes in Fnp values for a given junction-point pressure value. These changes typically are less than one instrumental pressure step. The column temperature was 70 °C. The linear patterns of points, which identify the mixture components, are obvious even without the linear regression lines. Slopes of the lines are governed by the polarity of the compounds relative to the pair of stationary phases used in the column combination. Wherever lines cross, the corresponding components coelute and are considered the critical pair. Data are not plotted in these regions since accurate retention factor values cannot be obtained for individual components from coeluting peaks. From the data in Figure 3, a window diagram was constructed considering all mixture components. This procedure involves the calculation of the relative resolution8 of all nonredundant component pairs. The relative resolution R for any component pair is described in eq 3, where ∆k is the difference in overall retention
R ) ∆k/(kav + 1)
(3)
factors for the two compounds and kav is the average value. The relative resolution of a peak pair depends on the fractional contributions of the two columns but is independent of the total resolving power of the column combination. Once the optimal values of the fractional contributions are determined, the total column length can be adjusted to achieve the required actual resolution of the critical pair. Tuning of the junction-point pressure then is used to reestablish the required values of fractional contribution.
Figure 4. Window diagram (a) and chromatogram (b) for an isothermal separation at 70 °C. The broken vertical line in (a) indicates the conditions used for the chromatogram. The inset in (b) shows the early portion of the chromatogram on an expanded time scale.
From the regression lines in Figure 3, R values were computed for incremental changes in Fnp and the minimum value determined for each Fnp value. Figure 4 shows the window diagram (a) and a chromatogram (b) for the complete mixture at a column temperature of 70 °C. In the window diagram, the minimum R values (critical pair values) are plotted versus the Fnp values. Every zero-resolution point on the plot corresponds to the coelution of a different critical pair. Slope changes in the plot correspond to different component pairs becoming the critical pair as the value of Fnp is changed. The chromatogram shown in part b of Figure 4 was obtained using an Fnp value of 0.70, as indicated by the broken vertical line in the window diagram of part a. The window diagram indicates that this fraction value provides the best separation. Experience with tunable column ensembles of the resolving power used here suggest that relative resolution values in the range 0.03-0.04 are needed to obtain an actual resolution greater than 1.0. Thus, the peak value of ∼0.018 is inadequate. This is confirmed in the chromatogram. While the chromatogram is complete is ∼85 s, the early-eluting components (shown in the inset using an expanded time scale) show numerous coelutions. Initial Isothermal Operation. To improve the separation of the early-eluting components, the column temperature was lowered and window diagrams were plotted for different isothermal temperatures. The poly(ethylene glycol) polar column used here has a minimum recommended operating temperature of 40 °C. However, inadequate retention of the most volatile components at this temperature still resulted in inadequate resolution. A column temperature of 30 °C worked very well with no significant loss in column efficiency or retention reproducibility. Analytical Chemistry, Vol. 70, No. 23, December 1, 1998
4963
Figure 6. Chromatogram obtained with 30 °C isothermal operation for 37 s after injection followed by a 35 °C/min temperature ramp to 90 °C. The junction-point pressure was the same as for Figure 5.
Figure 5. Window diagram (a) and chromatogram (b) for an isothermal separation at 30 °C of the first 12 components of the test mixture. The broken vertical line in (a) indicates the conditions used for the chromatogram.
Figure 5a shows a window diagram for the optimization of the first 12 mixture components to elute from the column ensemble at 30 °C. Peaks are identified in Table 1. Note that, even at 30 °C, components 1 and 2 (1,1-dichloroethene and methyl alcohol, respectively) show inadequate retention and coelute over the entire available tuning range. For further studies, they were treated as a single component. The chromatogram is complete in ∼37 s. Components 6 and 7 are the critical pair. A slight decrease in the Fnp value, accomplished by a decrease in the set point pressure of the electronic pressure controller, results in components 7 and 8 becoming the critical pair. Note that by reducing the number of components considered in the windowdiagram optimization from 31 (Figure 4) to 12, a simpler diagram is obtained with several windows having peak relative resolution values of ∼0.040. Fast Temperature Programming. If the conditions used for Figure 5b are continued, the retention factor for the last-eluting compound (1,2-dichlorobenzene) is nearly 70, and the isothermal analysis of the complete mixture will take ∼15 min. In addition, the detectability of the later-eluting peaks would be reduced significantly. The use of fast temperature programming with linear ramp rates greater than ∼20 °C/min is very attractive for HSGC analysis of mixtures covering a relatively wide boiling point range. The maximum programming rates (assuming a linear temperature ramp) that can be used without seriously compromising peak capacity vary inversely with column holdup time. For the relatively small holdup time values used in this study, programming rates greater than ∼25 °C/min were found most useful. 4964 Analytical Chemistry, Vol. 70, No. 23, December 1, 1998
For the mixture of purgeable VOCs used in this study, the analysis time can be reduced to less than 2.5 min by using a temperature program with a ramp rate of 35 °C/min starting ∼37 s after injection. Using an initial temperature of 30 °C, the chromatogram from the first 12 components is the same as in Figure 5. The complete chromatogram is shown in Figure 6. The tuning pressure was held constant (isobaric separation). The measured temperature profile also is shown. The jaggedness in the temperature profile is the result of inadequate amplification. While the measured temperature ramp is quite linear over most of the ramp duration, significant departures occur at the beginning and end of the ramp. For the lower programming rates and longer ramp durations associated with conventional temperature programming, these departures from linearity are not apparent. The features observed in the temperature profile are very reproducible and have no adverse impact on the final separation quality. The first 12 components elute before the start of the temperature program, and the chromatogram of these components is identical to the one in Figure 5b. While the separation is complete in just under 140 s, several coelutions are prominent. These include components 14 and 15 as well as 18, 20, and 21. Components 24 and 25 also are inadequately separated. A variety of temperature programming rates were investigated, but none resulted in a complete separation is less than 2.5 min. Selectivity Programming. After elution of the first 12 components under isothermal and isobaric conditions, several pressure change and temperature ramp rate regiments were investigated. An example is shown in Figure 7a. Note that a different sample was used for this study, and thus peak heights cannot be compared to those in Figures 4-6. Part b of Figure 7 shows the measured temperature (T) and pressure (P) profiles. For this experiment, an initial isothermal, isobaric separation was used for the first 37 s after injection. These are the same starting conditions as used for Figures 5 and 6. At 37 s, a 60 °C/min temperature ramp was initiated to achieve a final temperature of 90 °C. Isothermal conditions were then maintained until the end of the separation. Also at 37 s, the column junctionpoint pressure was increased from 23.6 to 24.6 psia and then held constant until the end of the separation.
Figure 7. Chromatogram obtained with 30 °C isothermal operation for 37 s after injection followed by a 60 °C/min temperature ramp to 90 °C. The initial junction-point pressure was the same as for Figure 5. The pressure set point was increased to 24.7 psia 37 s after injection.
Figure 8. Chromatogram obtained with 30 °C isothermal operation for 37 s after injection followed by a 35 °C/min temperature ramp to 90 °C. The initial junction-point pressure was the same as for Figure 5. The pressure was increased to 26.3 psia 37 s after injection and reduced to 22.5 psia 56 s after injection.
The departures from linearity in the temperature profile are quite prominent. A nominal programming rate of 60 °C/min clearly exceeds the maximum linear programming rate for the GC instrument used here. However, the temperature profile is very reproducible, and the departures from linearity did not effect the outcome of the experiments described here. Note that the pressure transition is very sharp. The pressure profile is shown on an expanded vertical scale, and the absolute pressure change during the pressure jump is only ∼4.3%. The hump in the chromatogram baseline, which begins ∼40 s after injection, appears to be a result of a carrier gas flow perturbation caused by the abrupt pressure change at the columnjunction point. After ∼80 s, the baseline becomes more or less stable, but at a slightly lower level than prior to the pressure change. Note that the chromatogram in Figure 7 was obtained with a more sensitive electrometer/amplifier setting than for the case of Figure 6. At the setting used for Figure 6, the hump in the baseline is barely noticeable. The chromatogram shown in Figure 7a is completed in about the same time as the one in Figure 6. Note that peaks 13-18 are completely separated in Figure 7a, while in Figure 6, without the pressure change, peaks 14, 15, and 18 all had coelutions. However, several coelutions occur latter in the chromatogram of Figure 7a. For a variety of pressure step values resulting in a final pressure greater than ∼24.5 psia, and temperature ramp rates greater than ∼30 °C/min, complete separation of components 1318 was achieved. However, none of the conditions investigated resulted in complete separation of components 19-31.
Figure 8a shows a chromatogram in which a complete separation was achieved in less than 2.5 min. Part b of Figure 8 shows the measured temperature and junction-point pressure profiles. In this case, a 35 °C/min temperature ramp was initiated 37 s after injection, and two pressure changes were employed. For the first pressure change, which was initiated 37 s after injection, the junction pressure was increased from 23.6 to 26.3 psia. At 56 s after injection, the pressure set point was changed to 22.5 psia. However, this transition to lower pressure is quite slow, and a constant pressure is not observed until ∼90 s after injection. This relatively slow pressure change is probably associated with excessive dead volume in the pressure controller and the connecting line. When the junction-point pressure is decreased, the carrier gas trapped in this volume must bleed through the pneumatic restriction of the second column in the system. Note that the pressure profile is very reproducible and did not adversely effect the empirical optimization procedure used in this study. Note that peaks 15-18 are broader than the later-eluting peaks. These components migrate through most of the combined column length under conditions of relatively high junction-point pressure. Under these conditions, the efficiency of the column combination is relatively low. The efficiency of pressure-tunable column ensembles varies significantly with the junction-point pressure.24 This is the result of the effects of gas velocity changes on efficiency. For the studies reported here, the fractional contribuAnalytical Chemistry, Vol. 70, No. 23, December 1, 1998
4965
tion of the first column ranged from 0.612 to 0.956. For the former value, the average carrier gas velocity in the first column was 0.96 m/s and in the second column was 1.51 m/s. These values are in the range commonly used for HSGC. For the latter value, the average gas velocities were 0.21 and 4.6 m/s, respectively. These values are far from optimal, and considerable loss in column efficiency occurs. CONCLUSIONS This report illustrates the power of the combination of column selectivity programming and fast temperature programming for the high-speed separation of more complex mixtures. An empirical approach to optimization, in which the column selectively is adjusted for groups of compounds as they elute from the column, appears to be effective. For still more complex mixtures spanning a wider boiling point range, the temperature ramp can be continued, and additional changes in the column junction-point pressure can be implemented as needed. For the column lengths and carrier gas velocities used in this study, the required temperature-programming rates can be obtained from most laboratory GC instruments, and modification (24) Purnell, J. H.; Williams, P. S. J. Chromatogr. 1985, 325, 1.
4966 Analytical Chemistry, Vol. 70, No. 23, December 1, 1998
for pressure-programmable selectivity is straightforward. The use of an inlet system that is capable of generating narrow injection plugs and relatively fast electrometer/amplifier and data acquisition electronics is necessary for the efficient utilization of relatively short columns. The reduction of dead volume in the pressure controller and its interface to the column system should result in more rapid transitions to lower junction-point pressures. The procedures used to select the temperature ramp rate and the pressure profile were largely empirical. It is very possible that other conditions could result in a still faster complete separation of the test mixture. Since retention patterns can be changed incrementally and very reproducibly by the use of electronic pressure control, more efficient optimization algorithms may be developed. Pressure profiles with sharp downward as well as upward transitions should facilitate this development. The use of continuous junction-point pressure changes may further enhance the efficiency of pressure-programmable column selectivity. Received for review April 27, 1998. Accepted September 24, 1998. AC980463B
