Anal. Chem. 2001, 73, 279-285
Pulsed Flow Modulation for High-Speed GC Using a Pressure-Tunable Column Ensemble Tincuta Veriotti, Megan McGuigan, and Richard Sacks*
Department of Chemistry, University of Michigan, Ann Arbor, Michigan 48109
A computer-driven pressure controller is used to deliver pressure pulses to the junction point of two series-coupled columns using different stationary-phase chemistries. The column ensemble consists of a trifluoropropylmethyl polysiloxane column followed by a dimethyl polysiloxane column. Each pressure pulse causes a differential change in the carrier gas velocities in the two columns, which lasts for the duration of the pulse. A pressure pulse is used to selectively increase the separation of a component pair that is separated by the first column but coelutes from the series-coupled ensemble. If both components are on the same column when the pulse is applied, a small change in the ensemble separation occurs. If one component of the pair is on the first column and the other component is on the second column, a pressure pulse can result in a much larger change in the ensemble separation for the component pair. A model with a spreadsheet algorithm is used to predict the effects of a pressure pulse on the trajectories of component bands on the column ensemble. The effect of the initiation time of a pressure pulse is investigated for a two-component mixture that coelutes from the column ensemble. For the case where the entire pressure pulse occurs when one of the components is on the first column and the other component is on the second column, the peak separation from the ensemble increases nearly linearly with the product of the pressure pulse amplitude and the pulse duration. Peak shape artifacts are observed if the pressure pulse occurs when a solute band is migrating across the column junction point. The introduction of the fused-silica, wall-coated open tubular (capillary) GC column in 19791 represents a paradigm shift in the analysis of volatile and semivolatile organic compounds. Up to this time, the principal emphasis in GC methods development often involved the preparation of a stationary phase with high selectivity for a particular set of compounds. Here selectivity refers to the pattern of peaks eluted from the column and the values of the separation factor R for all component pairs. Hundreds of different stationary-phase chemistries were described.2-4 With the introduction of the fused-silica capillary column, the emphasis (1) Dandeau, R.; Bente, P.; Rooney, T.; Hiskes, R. Am. Lab. 1979, 11, 61. (2) Rotzsche, H. Stationary Phases in Gas Chromatography; Elsevier: New York, 1991. (3) 3.Guide to stationary phases for gas chromatography; Analabs: North Haven, CT, 1969. (4) Baiulescu, G.; Ilie, V. A. Stationary Phases in Gas Chromatography; Pergamon Press: New York, 1975. 10.1021/ac000665j CCC: $20.00 Published on Web 12/14/2000
© 2001 American Chemical Society
gradually shifted toward the development of more universal columns with great resolving power.5,6 The much greater resolving power obtained from long capillary columns allowed for the complete separation of more complex mixtures with analysis times comparable to those achieved with packed columns. Here, selectivity is sacrificed for resolving power. Shorter capillary columns (5-20 m) operated at relatively high carrier gas flow rates have been used to obtain much faster separations but with substantially reduced resolving power.7-10 If these techniques for high-speed GC (HSGC) are to prove useful for more complex mixtures, issues of selectivity will need to be addressed.11,12 A practical approach to achieving enhanced selectivity for HSGC with capillary columns involves the series coupling of two capillary columns using different stationary-phase chemistries. By combining the two chemistries in different proportions, a variety of selectivities (peak elution patterns) can be obtained.13-16 A very flexible approach to enhanced selectivity is the use of a series-coupled column ensemble with an adjustable-pressure source of carrier gas at the junction point between the columns.15-19 Adjustment of this pressure results in a differential change in the carrier gas velocities in the two columns. An increase in the junction point pressure results in reduced carrier gas velocity in the first column and increased velocity in the second column. If the holdup time of the first column increases and that of the second column decreases (increase in junction point pressure), all mixture components have increased residence time in the first column and decreased residence time in the second column; this increases the influence of the stationary-phase chemistry of the first column and decreases the influence of the second column. A reduction in the junction point pressure has the opposite effect. (5) Lipari, F. J Chromatogr. 1990, 503, 51. (6) Berger, T. A. J Chromatogr. 1996, 42, 63. (7) Gaspar, G. J. Chromatogr. 1991, 566, 331. (8) Annino, R. J. High Resolut. Chromatogr. 1996, 19, 285. (9) Sacks, R.; Smith, H.; Nowak, M. Anal. Chem. 1998, 70, 29A. (10) Cramers, C. A.; Janssen, H. G.; van Deursen, M. M.; Leclercq, P. A. J. Chromatogr., A 1999, 856, 315. (11) Sacks, R.; Coutant, C.; Grall, A. Anal. Chem., in press. (12) Sacks, R.; Coutant, C.; Veriotti, T.; Grall, A. J. High Resolut. Chromatogr. 2000, 23 (3), 225. (13) Purnell, J. H.; Williams, P. S. J. Chromatogr. 1984, 292, 197. (14) Ingraham, D. F.; Shoemaker, C. F.; Jennings, W. J. J. Chromatogr. 1982, 239, 39. (15) Sandra, P.; David, F.; Prood, M.; Diricks, G.; Verstappe, M.; Verzele, M. J High Resolut. Chromatogr., Chromatogr. Commun. 1985, 8, 782. (16) Hinshaw, J. V.; Ettre, L. S. Chromatographia 1986, 21, 561. (17) Deans, D. R.; Scott, I. Anal. Chem. 1973, 45, 1137. (18) Akard, M.; Sacks, R. Anal. Chem. 1994, 66, 3036. (19) Akard, M.; Sacks, R. Anal. Chem. 1995, 67, 2733.
Analytical Chemistry, Vol. 73, No. 2, January 15, 2001 279
If the junction point pressure is controlled electronically, a variety of software-selectable ensemble retention patterns can be obtained for a specified set of target compounds, each one representing a different composite stationary-phase chemistry. Electronically controlled instruments have been described that have 50 to over 200 unique junction point pressure set points.9,12,16,20 This is equivalent to having available a large number of single columns with a relatively fine gradation of stationary-phase chemistry. As mixture complexity increases under conditions of fixed analysis time and fixed ensemble resolving power, the probability of coelutions increases rapidly, and no available combination of stationary-phase chemistries may have adequate selectivity for a complete separation. Recently, programmable junction point pressure has been successfully used for cases where no single pressure resulted in a complete separation.21-25 With pressureprogrammable selectivity, the ensemble junction point pressure initially is set to provide adequate separation of the early-eluting components. After these components have all migrated across the junction and into the second column, the pressure can be changed without affecting the ensemble elution pattern for these components. An on-the-fly pressure change then is used to establish carrier gas velocity conditions more favorable for the complete separation of the next group of components to elute from the column ensemble. This procedure can be repeated several times, if necessary, to achieve a more complete separation of the entire mixture.21,23 A significant limitation of these methods is the fact that a change in the junction point pressure used to increase the separation of a particular component pair usually results in reduced separation of one or more other component pairs. Thus, the selection of a junction point pressure for a specified set of target compounds always represents a compromise condition. This report describes the use of a relatively short pressure pulse to increase the separation of a specified component pair without significantly affecting the peak pattern and separation of other components in the mixture. A previously described solute band trajectory model22,25 is used to predict the effects of a pressure pulse on ensemble retention times and peak separation. Experimental results are presented that validate the model. EXPERIMENTAL SECTION Apparatus. The pressure-controlled, dual-column ensemble has been described.22 The system uses a Varian 3500 capillary GC (Varian Instruments, Walnut Creek, CA) as an experimental platform. The split inlet and the two flame ionization detectors (FID) from the GC are used without change. The column ensemble consists of a trifluoropropylmethyl polysiloxane column (Rtx-200, Restek, Bellafonte, PA) followed by a dimethyl polysiloxane column (DB-1, J&W Scientific, Folsom, CA). Each column is 10 m long, 0.25-mm i.d. and uses a 0.25-µm-thick bonded stationary phase film. The column junction point is connected to a computer-driven pressure controller (model 640A, MKS Instruments, Andover, MA) with pressure step size of 0.1 psi and pressure repeatability (20) Smith, S.; Sacks, R. Anal. Chem. 1997, 69, 5159. (21) Smith, H.; Sacks, R. Anal. Chem. 1998, 70, 4960. (22) Leonard, C.; Sacks, R. Anal. Chem. 1999, 71, 5501. (23) Grall, A.; Zellers, E. T.; Sacks, R. Environ. Sci. Technol. 2000, in press. (24) Coutant, C.; Sacks, R. Anal. Chem., in press. (25) Grall, A.; Sacks, R. Anal. Chem., in press.
280
Analytical Chemistry, Vol. 73, No. 2, January 15, 2001
of (0.01 psi. This device is used to apply pressure pulses to the column junction point. A 1.0-m-long, 0.25-mm-i.d. uncoated fusedsilica tube connected between the column junction point and the pressure controller and vented to the atmosphere is used to reduce the pressure equilibration time for downward pressure changes and to prevent contamination of the controller by injected samples.22 One of the Varian FIDs (FID1) is used to monitor a portion of the effluent from the first column, and the other FID (FID2) is used to monitor the output from the column ensemble. A 1.0-mlong, 0.10-mm-i.d. deactivated fused-silica tube is used to transport ∼10% of the effluent from the first column to FID1. The chromatogram obtained from this detector is used to determine the time at which to initiate the pressure pulse. The detector currents are monitored with electrometer/amplifier circuits built in-house and having time constants less than 10 ms. The electrometers are interfaced to a Gateway 2000 PC (PC-75, Gateway 2000, Sioux Falls, SD) using a 16-bit A/D board (model C10DAS-1602, Computer Boards Inc, Mansfield, MA). Instrument control and data acquisition operations are implemented with LABTECH notebook software (Laboratory Technologies, Inc., Wilmington, VA). Chromatogram processing is accomplished with Grams/32 software (Galactic Industries, Salem, NH). Materials and Procedures. Hydrogen, after purification with filters for oxygen, water vapor, and hydrocarbons, was used as carrier gas. This gas source is connected to both the Varian split inlet and the junction point pressure controller. For every experiment, a 10-µL headspace sample was injected from an equalvolume mixture of reagent-grade n-octane and 1-pentanol. The injector split ratio was 5:1. The inlet and detector temperatures were 250 °C. The inlet pressure was 35.0 psig. All chromatograms were generated isothermally at 36 °C. To obtain accurate holdup time and retention time values for the first column, the gas transport time from the column junction point to the detector was subtracted from the measured retention times obtained with FID1. The transport time was computed using standard equations for gas flow in capillary tubes.26 Note that this transport time varies with column junction point pressure. Retention factors for the first column were computed from the corrected retention time and holdup time values. Holdup time and retention time values for the second column were obtained by subtracting the corresponding values for the first column from the ensemble values obtained from FID2. From these values, retention factors for the second column were computed. The retention factor values for the individual columns as well as the column dimensions and the viscosity of the carrier gas at the column operating temperature were used as inputs to a band trajectory model that was previously developed and described in refs 22 and 25. This model uses spreadsheet calculations and takes into account the carrier gas flow velocity profile along the column axis, the change in retention for each mixture component as it crosses the junction from the first to the second column, and programmed changes in the junction point pressure. The junction point pressure in the absence of the pressure pulse (quiescent pressure) was chosen to be equal to the pressure that would occur at the junction point in the absence of any external connections to the junction point (natural pressure). This (26) Grant, D. Capillary Gas Chromatography; Wiley: New York, 1996.
Figure 1. Upward (a) and downward (b) pressure pulses applied to the column junction point. Broken lines show nominal pulse shape.
(natural) pressure results in an ensemble holdup time that is near the minimum value achievable with the dual-column ensemble. RESULTS AND DISCUSSION Pressure Pulse Shape. The pressure controller used for these studies was not designed for implementing rapid pressure changes. When the set point pressure is changed to a lower value, the gas trapped in the controller dead volume (>7 cm3) must bleed out through the second column and the added vent line before pressure equilibration occurs. The equilibration time depends on the pressure-step size and on the target pressure. Figure 1 shows the pressure versus time profiles used for upward pressure pulses (a) and downward pressure pulses (b). Broken lines show the nominal pulse shape, and solid lines show the actual pressure as monitored by the pressure transducer in the controller output. In both cases, the nominal pulse width ∆t was 3.0 s, and the quiescent set point pressure (pressure in the absence of a pulse) was 36.0 psia. The pressure change ∆P during a pulse was either plus or minus 8.0 psi. Feedback damping for the pressure controller was adjusted to minimize ringing and overshoot in the output pressure profile. For the case of the upward pressure pulse, the leading edge is very sharp and ringing is rapidly damped. The trailing edge, however, shows much slower pressure equilibration, and nearly 2 s is required for the quiescent pressure to be reestablished. For the downward pressure pulse, the leading edge is a transition to lower pressure, and the pressure equilibration time is about
Figure 2. Solute band position vs time plots from spreadsheet calculations for n-octane (1) and 1-pentanol (2) for a 3-s-wide upward pressure pulse initiated 3 (a), 13 (b), and 18 s (c) after sample injection. Solid-line plots are for the case with the pressure pulse, and broken-line plots are for the case without the pressure pulse. Plots with alternate dots and dashes are for an unretained component. Portions of the plots in broken-line boxes are shown on an expanded scale to the right of each set of plots. Peaks shown are for chromatograms obtained with the indicated conditions.
equal to the 3.0-s pulse width. The longer equilibration time for the leading edge of the downward pressure pulse relative to the trailing edge of the upward pulse is the result of the lower target pressure (28.0 psia) in the downward pulse. The trailing edge of the downward pressure pulse is very sharp, but ringing continues for ∼1 s. Band Trajectory Model Results and Testing. A pair of mixture components can coelute from a tandem-column ensemble if both components have very similar retention factor values on both columns or if one of the components has a larger retention factor on one of the columns and the other component has a larger retention factor on the other column. In the former case, adjustment of the column junction point pressure may not separate the two components, and a longer column ensemble (greater resolving power) or an ensemble with greater selectivity for the component pair may provide the only solutions. In the latter case, a change in the junction point pressure often will result in adequate separation of the component pair. However, a change in the junction point pressure may result in the coelution of other component pairs that were previously separated. Figure 2 shows how the application of a pressure pulse to the column junction point at an appropriate time after injection can Analytical Chemistry, Vol. 73, No. 2, January 15, 2001
281
be used to separate a particular component pair without significantly affecting the pattern of peaks from other mixture components. For each of the three cases shown in Figure 2, injection occurs at zero time in the lower left corner of the plots. The horizontal line in the center of each part of the figure corresponds to the column-ensemble junction point. Elution from the ensemble occurs along the top horizontal line at 20 m. The pair of vertical lines in each of the three cases shows the time interval corresponding to the application of a 3-s-wide (nominal) pressure pulse from a quiescent value of 36.0-44.0 psia. The band trajectory plots labeled 1 and 2 are for n-octane and 1-pentanol, respectively. The solid-line plots are for the case with the pressure pulse, and the broken-line plots are for the case without the pressure pulse. The band trajectories represented by lines with alternate dots and dashes are for an unretained component and describe the carrier gas velocity profile along the column-ensemble axis for the case with the pressure pulse. The plots for the unretained component show no discontinuous slope change at the junction point since the natural junction point pressure was used in the absence of a pulse. To the right of each set of plots, the portion of the band trajectory plots in the broken-line box is shown on an expanded time scale, and the actual chromatograms obtained with the corresponding conditions are shown above the termination of the band trajectory plots. The first column in the ensemble has the polar trifluoropropylmethyl polysiloxane stationary phase, and the polar 1-pentanol (plot 2) migrates more slowly. In the second, nonpolar column, the nonpolar n-octane migrates more slowly. The result is that, in the absence of a pressure pulse, the band trajectory plots for these compounds cross just prior to elution from the ensemble. The peak-apex separation is less than 0.5 s, and only a single peak is observed in the chromatogram. For case a, the pressure pulse is applied while both components are still on the first column. Since the pulse results in an increase in junction point pressure, the pressure drop along the first column is reduced, and the local migration velocities of both components are sharply reduced. This results in greater residence times on the first column, and the ensemble retention times are shifted to greater values, but the peak-apex separation is not significantly changed. For case b, the pressure pulse is applied after the n-octane has crossed the junction but before the 1-pentanol has reached the junction. The larger junction point pressure during the pulse results in an increase in local carrier gas velocity in the second column, and component 1 accelerates. The result is a decrease in ensemble retention time for component 1. Since component 2 is still on the first column when the pulse is applied, its migration rate decreases appreciably during the pressure pulse, and the band arrives at the junction considerably later than without the pressure pulse. This results in an increase in ensemble retention time. The overall result is a differential shift in retention times for the two components, and a large separation is observed in the chromatogram. For case c, the pressure pulse is applied after both components have migrated across the junction, and both components show increased migration velocities during the pulse. The result is decreased ensemble retention time for both components but no significant change in peak-apex separation. These results show that ensemble values of peak-apex separation can be significantly 282 Analytical Chemistry, Vol. 73, No. 2, January 15, 2001
Figure 3. Chromatograms for the two-component mixture with a 3.0-s-wide (nominal) upward pressure pulse initiated at the times (in s) indicated to the left of each chromatogram. The broken vertical line shows the peak-apex retention time for the single peak observed without the pressure pulse.
increased for components that are on opposite sides of the junction point when the pressure pulse is applied. For components that are on the same side of the junction during the pulse, significant retention time shifts occur, but peak-apex separations and thus the pattern of peaks eluting from the ensemble are not significantly changed. Figures 3 and 4 show chromatograms obtained with upward and downward pressure pulses, respectively, initiated at various times after sample injection. The numbers to the left of the chromatograms give the pulse initiation time (s) after injection. Without the pressure pulse, the peak-apex retention time for the single peak containing both the n-octane and the 1-pentanol is 29.5 s. This time is indicated by the broken vertical lines. The nominal pulse width was 3.0 s. See Figure 1 for the actual pulse shapes. Note that an upward pulse (to higher junction point pressure) results in lower local carrier gas velocity in the first column and higher velocity in the second column. The opposite situation occurs for a downward pressure pulse. In Figure 3, a pulse initiation time of 10.0 s results in the entire pulse occurring while both compounds are on the first column. This shifts the peak-apex retention time to 31.8 s but with very little change in peak shape. When the pulse initiation time is delayed until 11.0 s after injection, both compounds are on the first column at the beginning of the pulse, but part of the n-octane crosses the column junction before the end of the pulse. The result is a broad, distorted double peak. With a pulse initiation time of 13.0 s, the n-octane band has completely crossed the junction prior to the pulse, and the
Figure 4. Chromatograms for the two-component mixture with a 3.0-s-wide (nominal) downward pressure pulse initiated at the times (in s) indicated to the left of each chromatogram. The broken vertical line shows the peak-apex retention time for the single peak observed without the pressure pulse.
1-pentanol remains on the first column during the entire duration of the pulse. The result is a large shift to lower ensemble retention time for the n-octane and no significant change in retention time for the 1-pentanol relative to the 10.0-s pulse initiation time. The peak-apex separation for the 13.0-s delay case is 3.8 s. The peak tailing observed for the 1-pentanol is often seen for alcohols and is observed when only this compound is injected and no pressure pulse is used. For a 15.5-s pulse initiation time, the 1-pentanol reaches the column junction before the pulse is complete, and a very broad, low-amplitude feature is observed in the chromatogram. For the 18.0-s pulse initiation time, both compound bands have migrated across the junction prior to pulse initiation, and only a single peak is seen in the chromatogram. However, the peak apex retention time has shifted by 1.5 s relative to the case with no pressure pulse. The same general trends are observed for the downward pressure pulse shown in Figure 4, but the retention time shifts relative to the no-pressure-pulse case (broken vertical line) are in the opposite direction. In addition, the maximum peak separation is only 1.8 s. For the no-pressure-pulse case, the first-column elution times for the two compounds differ by ∼4.9 s. For an upward pressure pulse initiated after the first component (noctane) crosses the column junction, the local carrier gas velocity in the first column decreases sharply. This substantially increases the residence time on the first column of the 1-pentanol. This is
Figure 5. Ensemble retention time (a, b) and peak separation (c, d) vs pulse initiation time for n-octane (1) and 1-pentanol (2) for a 3-s-wide upward pressure pulse. (a, c) spreadsheet calculations; (b, d) measurements from chromatograms. Error bars show standard deviations for three injections. The broken horizontal lines in (a) show the calculated ensemble retention times for the case of no pressure pulse.
clearly seen in the band trajectory plots in Figure 2b. The increased residence time allows the pressure pulse to be completed before 1-pentanol reaches the junction, and a complete separation can be obtained from the column ensemble. For a downward pressure pulse, the local carrier gas velocity on the first column increases during the pulse, and the pulse is not completed prior to the leading edge of the 1-pentanol band reaching the junction. The result is a significantly poorer separation and loss of some of the 1-pentanol through the vent line. This is seen in the reduced size of the 1-pentanol peak in Figure 4 for the 14.0-s pulse initiation time. Some sample loss is inevitable for the downward pressure pulse if the junction point pressure falls below the natural pressure during the pulse and if the solute band reaches the junction point prior to the end of the pulse. This is a clear disadvantage of the downward pressure pulse. Pressure Pulse Properties and Peak Separation. Figure 5 shows plots of ensemble retention time (a, b) and peak separation (c, d) vs pulse initiation time for n-octane1 and 1-pentanol2 for an upward pressure pulse with a 3.0-s nominal pulse width. Plots a and c are from spreadsheet calculations using the solute band trajectory model and the retention factor values for the two compounds on the individual columns. Plots b and d are from measurements on actual chromatograms. Error bars show standard deviations for three injections. The broken horizontal lines in (a) show the calculated ensemble retention times for the case Analytical Chemistry, Vol. 73, No. 2, January 15, 2001
283
of no pressure pulse. These are the same as the retention times observed for injections of the individual components without a pressure pulse. Peak separation with no pressure pulse is ∼0.4 s. For the upward pressure pulse, pulse initiation times less than 9 s result in a nearly constant shift in retention times of ∼1.6 s for both compounds since the pulse is complete before either compound reaches the junction. The small increase in retention times observed with increasing initiation time during this interval is the result of carrier gas acceleration along the column. For pulse initiation times in the range 9-11 s, component 1 reaches the junction point before completion of the pulse, and the retention time for this component sharply decreases. For pulse initiation times in the range 11-14 s, the entire pulse occurs while component 1 is on the second column and component 2 is on the first column. Here, a separation of 2.3-2.4 s is observed with the higher values occurring near the end of this time interval because of carrier gas acceleration. If the pulse initiation time is 16-24 s, both components are on the second column during the entire pulse, and the peaks show nearly equal shifts to shorter retention times. As the pulse initiation time is further increased, one or both peaks elute from the ensemble before the pulse is completed, and the retention times return to their no-pulse values. Note that, in Figure 5a, the plots for the two compounds cross for a pulse initiation time of ∼10 s. This is explained by reference to the band migration trajectory plots in Figure 2. Without a pressure pulse, the band trajectories cross just prior to elution from the ensemble with the result that component 2 elutes from the column ensemble ∼0.4 s before component 1. With an upward pressure pulse beginning 10 s after injection, component 1 reaches the junction before the end of the pressure pulse and its ensemble retention time is shifted lower by ∼0.4 s. This results in complete coelution of the two components. Further delay in the start of the pressure pulse by 1 s results in a larger shift for compound 1. This results in an elution order change and a relatively large separation. The qualitative features of plots a and b are very similar. For pulses that occur when both compounds are on the same column, only a single peak apex is observed in the chromatograms, and the ensemble elution time of the peak apex is plotted in Figure 5b. The nearly flat regions of the peak separation plots in panels c and d of Figure 5 are expected from the retention time plots in panels a and b of Figure 5. The maximum peak separation for the upward pressure pulse shown in Figure 5d is ∼3.8 s compared with the maximum predicted value of 2.4 s from Figure 5c. In part this is the result of the nonideal pulse shape shown in Figure 1. This will be discussed. Figure 6 is similar to Figure 5 except the data are for the case of a downward pressure pulse with a nominal width of 3.0 s. For the downward pressure pulse with initiation times less than 4 s, the entire pulse is complete before either of the compounds reaches the column junction point, and the retention times decrease only slightly in this interval with increasing delay in the pulse initiation. For pulse initiation times in the range 4-10 s, component 1 reaches the junction prior to completion of the pulse, and its ensemble retention time increases with increasing initiation time due to the lower carrier gas velocity on the second column during the pulse. Component 2 reaches the junction point in ∼10 s, and for pulse initiation times greater than ∼10 s, the ensemble 284 Analytical Chemistry, Vol. 73, No. 2, January 15, 2001
Figure 6. Ensemble retention time (a, b) and peak separation (c, d) vs pulse initiation time for n-octane (1) and 1-pentanol (2) for a 3-s-wide downward pressure pulse. (a, c) spreadsheet calculations; (b, d) measurements from chromatograms. Error bars show standard deviations for three injections. The broken horizontal lines in (a) show the calculated ensemble retention times for the case of no pressure pulse.
retention time for this compound increases rapidly with increasing pulse initiation time until ∼16 s when component 2 crosses the junction. Further increases in pulse initiation time have only a minor effect on retention times until ∼26 s when the compounds begin to elute from the ensemble before completion of the pulse. For pulse initiation times greater than ∼30 s, both components elute prior to the start of the pulse and the retention times are the same as the no-pulse values. No crossing of the plots for the two compounds is observed in Figure 6a for the case of the downward pressure pulse. A downward pressure pulse causes the retention time for component 1 to shift to larger values for pulse initiation times greater than 5 s. There is no change in elution order and thus the plots in Figure 6a do not cross. Note that the shift in ensemble retention times for component 1 occurs for initiation times greater than 8 s for the upward pressure pulse and for times greater than 5 s for the downward pulse. This is the result of the lower carrier gas velocity on the first column during an upward pulse and the higher velocity during the downward pulse. The more triangular peak separation versus pulse initiation time plots for the downward pressure pulse also are the result of the increased carrier gas velocity in the first column during the pulse. This causes component 2 to begin shifting to larger retention time values for pulse initiation times greater than 10 s, while compound 1 has not completely shifted to higher values for initiation times less than 11 s. For the downward pressure
Figure 5a. Since the peak separation is proportional to both the pressure pulse amplitude and width, the peak separation should be proportional to the pulse area (∆P∆t). Data from spreadsheet calculations used a nominal pressure pulse area of 24 psi‚s (∆P ) 8.0 psi, ∆t ) 3.0 s). Pulse area measurement from Figure 1 gives an actual area of ∼31.2 psi‚s for the upward pressure pulse, and when the calculated value of peak separation is multiplied by the ratio 31.2/24, the calculated separation is 3.1 s, which is in reasonable agreement with the actual peak separation of 3.8 s. The actual area of the downward pulse is ∼13.6 psi‚s, and when the calculated value of peak separation is multiplied by the ratio 13.6/24, the calculated separation is 1.8 s, which is in good agreement with the actual peak separation of 1.7 s.
Figure 7. Plots from spreadsheet calculation of peak separation vs pressure pulse width ∆t (a) and amplitude ∆P (b) for an upward pressure pulse beginning 12.0 s after sample injection. For plot a, the pulse amplitude was 8.0 psi. For plot b, the pulse width was 3.0 s.
pulse, the maximum peak separation from Figure 6d is 1.7 s, while the predicted value from Figure 6c is 3.3 s. These differences between the predicted and observed maximum peak separation values are due in part to the pulse shapes shown in Figure 1. Figure 7 shows results from spreadsheet calculations of the effects on peak separation caused by changes in the pressure pulse width ∆t (a) and amplitude ∆P (b) for the case of an upward pressure pulse. For plot a, ∆P is 8.0 psi, and for plot b, ∆t is 3.0 s. For both plots, the pulse initiation time is 12 s so that component 1 has crossed the junction but component 2 is still on the first column at the start of the pulse. For pulse widths up to 9 s, the peak separation is nearly linear with pulse width. For this portion of the plot in Figure 7a, the linear regression correlation coefficient is 0.9999. For pulse widths greater than 9 s, component 2 crosses the junction prior to the end of the pulse, and further increases in pulse width have little effect. Note that, without the pressure pulse, component 2 crosses the junction 16.3 s after injection. With the upward pressure pulse, the lower carrier gas velocity in the first column during the pulse increases the junction crossing time to 21 s if the pulse continues to at least this time. This explains why the linear portion of the plot in Figure 7a extends to a pulse width of 9 s. The plot of peak separation versus pulse amplitude is linear with a linear regression correlation coefficient of 0.9994 for the amplitude range 2-10 psi. The intercepts for both plots in Figure 7 are nonzero because of the elution order change described in (27) Veriotti, T.; Sacks, R. Anal. Chem., submitted.
CONCLUSIONS This preliminary study shows the utility of using relatively narrow pressure pulses to control the separation of specific component pairs eluting from a pressure-controlled column ensemble. For mixture components that are on the same column during the pulse, little change in peak pattern and peak-pair separation occurs in the ensemble chromatogram. Relatively large peak separation increases are observed only when one of the mixture components is on the second column and the other component is still on the first column during at least a portion of the pressure pulse. This should be very useful for more complex mixtures where most components are adequately separated and the goal is to obtain enhanced separations for a relatively few component pairs or small groups without loosing separation quality for the rest of the mixture components. This approach also may be more useful than pressure (selectivity) tuning and programming since for most of the duration of the analysis the quiescent junction point pressure can be adjusted to give minimum ensemble holdup time. Upward pressure pulses may be more useful than downward pulses for several reasons. First, an upward pressure pulse results in a decrease in the carrier gas velocity in the first column, and this increases the available time for the pulse to be completed before the second component reaches the junction. Second, a downward pressure pulse is more likely to result in sample loss through the vent line if the pressure during the pulse falls below the pressure that would exist at the column junction point in the absence of the pressure controller. Third, the pressure equilibration time is smaller for an upward pulse for any specified quiescent-pressure value. However, downward pressure pulses do have some useful applications, and downward pressure pulses from a valve-operated vent line to atmospheric pressure recently have been used to eliminate the need for a pressure controller.27 It is clear that a better method for generating pressure pulses is needed. While shorter pressure equilibration time can be obtained by the use of a smaller pneumatic restriction in the vent line connected to the outlet of the pressure controller, a price is paid in increased carrier gas consumption, and eventually, a point is reached where pressure control can no longer be obtained. Work in progress with a pneumatically actuated micro gas valve indicates that relatively rectangular pressure pulses with widths of less than 1 s can be obtained. Received for review June 9, 2000. Accepted November 5, 2000. AC000665J Analytical Chemistry, Vol. 73, No. 2, January 15, 2001
285
