Programmable Control of Column Selectivity for Temperature

A computer-driven pressure controller connected to the junction point of a series-coupled ensemble of two capillary GC columns having different ...
3 downloads 0 Views 121KB Size
Anal. Chem. 2000, 72, 5450-5458

Programmable Control of Column Selectivity for Temperature-Programmed GC Carrie Coutant and Richard Sacks*

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

A computer-driven pressure controller connected to the junction point of a series-coupled ensemble of two capillary GC columns having different stationary-phase selectivity is used to obtain on-the-fly (programmable) changes in ensemble selectivity. Changes in the junction-point pressure result in differential changes in the local carrier gas velocity in the two columns, and this results in changes in the pattern of peaks eluting from the ensemble. When used with relatively fast temperature programming (30 °C/min), the pattern of eluting peaks can be very sensitive to the time at which a selectivity (junction-point pressure) change is implemented. These elution pattern changes are described for a set of six PCB congeners that elute with a small range of retention times. The components are considered as a group, and changes in their elution pattern are described for a single junction-point pressure change, which is implemented at various times after sample injection. If the pressure change is implemented after the components have migrated across the junction point, the final pressure has relatively little impact on the ensemble retention pattern. Pressure changes implemented prior to the components reaching the junction can have a large effect and usually result in a pattern of peaks similar to the pattern obtained when the final pressure is used for the entire separation. For pressure changes made when the group of components is near the junction point, the observed peak pattern may be very sensitive to the time of the pressure change. The time at which the junction-point pressure change occurs is varied in 1.0-s intervals. Artifacts such as peak doubling and peak focusing or broadening are observed if a migrating band is crossing the column junction point at the time of the programmed pressure change. Since the introduction of the fused-silica wall-coated open tubular (capillary) GC column,1 the number of frequently used stationary phases has dramatically decreased. This is the result of the extraordinary resolving power of long capillary columns. In many cases, methods development in capillary GC has been reduced to the relatively straightforward task of selecting a sufficiently long column with a stationary phase selected from relatively few choices and an appropriate temperature or temperature programming rate to obtain adequate separation of the most difficult to separate component pair (critical pair). Since narrow(1) Dandeneau, R.; Bente, P.; Rooney, T.; Hiskes, R. Am. Lab. 1979, 11, 61.

5450 Analytical Chemistry, Vol. 72, No. 21, November 1, 2000

bore capillary columns operate with high efficiency at relatively high average carrier gas velocities,2 high resolving power often can be achieved with analysis times comparable to values associated with low-resolution packed-column separations. In recent years, more emphasis has been placed on the development of high-speed separation techniques due to the need for higher sample throughput and lower analysis costs.3-6 Faster analyses are achieved with shorter capillary columns and higher average carrier gas velocities. The result is reduced resolving power and the need to reexamine issues related to column selectivity and methods development. With high-speed GC (HSGC) techniques resolving power is significantly reduced, and the onesize-fits-all approach to column selection often is inadequate. To meet this challenge, columns and separation strategies with enhanced selectivity are required. Enhanced selectivity has been achieved with mixed stationary phases, where the volume ratio of the phases is tailored for use with a specified set of target compounds.7-11 Window-diagram methods have been developed for the selection of the optimal volume ratio in the stationary-phase mixture.10,11 Once a mixedphase column has been designed, it may be very useful for the separation of the target compound mixture, but any changes in the target compound list may render the column completely inadequate. Series-coupled ensembles of two capillary columns using different stationary phases can achieve the same selectivity as a mixed-phase column but with greater convenience and flexibility.12-15 If an additional carrier gas supply with adjustable pressure is connected to the junction point of the series-coupled column ensemble, continuously adjustable selectivity can be (2) Desty, D. H.; Goldup, A.; Swanton, W. T. Performance of Coated Capillary Columns. In Gas Chromatography; Brenner, N., Callen, J. E., Weiss, M. D., Eds.; Academic Press: New York, 1962; pp 105-135. (3) Gaspar, G. J. Chromatogr. 1991, 566, 331. (4) Annino, R. J. High Resolut. Chromatogr. 1996, 19, 285. (5) Sacks, R.; Smith, H.; Nowak, M. Anal. Chem. 1998, 70, 29A. (6) Cramers, C. A.; Janssen, H. G.; van Deursen, M. M.; Leclercq, P. A. J. Chromatogr., A 1999, 856, 315. (7) Freeman, R. R.; Kukla, D. J. Chromatogr. Sci. 1986, 24, 392. (8) Chien, C. F.; Laub, R. J., Kopecni, M. M. Anal. Chem. 1980, 52, 1402. (9) Pilgrim, G. W.; Keller, R. A. J. Chromatogr. Sci. 1973, 11, 206. (10) Laub, R. J.; Purnell, J. H. J. Chromatogr. 1975, 112, 71. (11) Laub, R. J.; Purnell, J. H. Anal. Chem. 1976, 48, 799. (12) Purnell, J. H.; Williams, P. S.; J. Chromatogr. 1984, 292, 197. (13) Ingraham, D. F.; Shoemaker, C. F.; Jennings, W. J. Chromatogr. 1982, 239, 39. (14) Sandra, P.; David, F.; Prood, M.; Diricks, G.; Verstappe, M.; Verzele, M. J High Resolut. Chromatogr., Chromatogr. Commun. 1985, 8, 782. (15) Hinshaw, J. V.; Ettre, L. S. Chromatographia 1986, 21, 561. 10.1021/ac000610h CCC: $19.00

© 2000 American Chemical Society Published on Web 09/28/2000

obtained.14-18 Further, if adjustable junction-point pressure is obtained with a high-precision, computer-driven pressure control device, a range of software-selectable selectivities can be achieved.5,19,20 Recently, programmable selectivity, where on-the-fly selectivity (junction-point pressure) changes are made during an analysis, has been used to obtain enhanced HSGC separations of more complex mixtures.21-23 With programmable selectivity, the junction-point pressure of a tunable, dual-column ensemble is initially set to give the desired separation quality for the early-eluting components, and sometime after sample injection, the junctionpoint pressure is changed in order to improve the separation quality of a later-eluting group of components. If necessary, this procedure can be repeated. Essentially, this divides the mixture components into groups, and selectivity (junction-point pressure) optimization methods are applied separately to the solute groups. This procedure can result in significantly faster separations and may result in a complete separation in cases where no single junction-point pressure will work. Selectivity tuning has been applied to temperature-programmed GC,24 and an optimization algorithm has been proposed for the selection of the junction-point pressure that will give the most complete high-speed separation. Using this technique, a mixture containing 12 common pesticides often found in human samples was completely separated in ∼140 s using a 30 °C/min temperature-programming rate. Using a two-step selectivity program and a 30 °C/min temperature program, a 31-component mixture of purgeable organic compounds was separated in ∼140 s.21 None of this previous work considered in detail how the target pressure and the time at which an on-the-fly junction-point pressure change is implemented effect retention patterns with temperature-programmed GC. The situation is significantly more complex than for isothermal separations due to the very rapid acceleration of solute bands just prior to elution when relatively high temperature-programming rates are used. In this report, a model is developed that describes the effects on ensemble retention times of single, on-the-fly pressure changes during an isothermal analysis. Results from this model then are used to interpret the effects on ensemble retention times of the same pressure changes for temperature-programmed GC. Results are described for a test mixture containing six PCB congeners that elute in the same general vicinity of the temperature-programmed chromatogram. EXPERIMENTAL SECTION Apparatus. The HSGC instrument used for these studies has been described in detail.22 A Varian 3500 capillary GC (Varian, Walnut Creek, CA) is used as an experimental platform. The pressure-programmable column ensemble consists of a 10-m-long trifluoropropylmethyl polysiloxane column (Rtx-200, Restek, Bel(16) Deans, D. R.; Scott, I. Anal. Chem. 1973, 45, 1137. (17) Akard, M.; Sacks, R. Anal. Chem. 1994, 66, 3036. (18) Akard, M.; Sacks, R. Anal. Chem. 1995, 67, 2733. (19) Smith, H.; Sacks, R. Anal. Chem. 1997, 69, 5159. (20) Sacks, R.; Coutant, C.; Veriotti, T.; Grall, A. J. High Resolut. Chromatogr. 2000, 23, 225. (21) Smith, H.; Sacks, R. Anal. Chem. 1998, 70, 4960. (22) Leonard, C.; Sacks, R. Anal. Chem. 1999, 71, 5501. (23) Grall, A. J.; Zellers, E. T.; Sacks, R. D. Environ. Sci. Technol., submitted. (24) Sacks, R.; Coutant, C.; Grall, A. Anal. Chem. 2000, 72, 524A.

lafonte, PA) (first column) and a 10-m-long dimethylpolysiloxane column (DB-1, J&W Scientific, Folsom, CA). Both columns are 0.25-mm i.d. and use 0.25-µm-thick bonded phases. The Varian inlet splitter and flame ionization detector (FID) are used for all studies. A computer-driven, high-precision absolute pressure controller (MKS model 640A, MKS Instruments, Andover, MA) is connected to the column junction point in order to obtain on-the-fly selectivity programming. This device can control the pressure in the 0-100 psia range in 0.1 psi steps with a repeatability of (0.01 psi. A vent line consisting of a 1.0-m-long, 0.25-mm-i.d. uncoated fused-silica tube is connected between the pressure controller output and the column junction point. The other end of the tube vents to atmospheric pressure. The vent line is used to obtain more rapid pressure equilibration after pressure set-point changes and to extend the useful pressure range without the risk of controller contamination by injected samples.22 With the vent line, pressure equilibration time for upward pressure changes is less than 1.0 s and for downward pressure changes is typically 1-6 s, depending on the size of the pressure step and the value of the target pressure. The FID is interfaced to a PC (P5-75, Gateway 2000, Sioux Falls, SD) by means of an electrometer/amplifier (built in house) and a 16-bit A/D board (CIODAS-1602, Computer Boards, Inc., Mansfield MA). The A/D board also provides the interface for the pressure controller. The controller requires a 0-5.0-V input, and a 5-mV change in the input voltage produces a set-point pressure change of 0.1 psi. The interface board is controlled by means of Labtech Notebook software (Laboratory Technologies, Inc., Wilmington, VA). Chromatogram processing is accomplished with Grams/32 software (Galactic Industries, Salem, NH). Excel spreadsheet calculations are used to predict the effects of junctionpoint pressure changes on isothermal separations. Materials and Procedures. Hydrogen is the carrier gas and is purified for oxygen, water vapor, and hydrocarbons by filters prior to the injection port. Six individual PCB congeners as 100 µg/mL solutions in hexane (Ultra Scientific, North Kingstown, RI) were mixed in equal volumes to obtain a test sample containing ∼16.7 µg/mL of each congener. The sample components and their IUPAC congener numbers are listed in Table 1. These congeners were chosen because they elute relatively close together and their elution pattern is quite sensitive to changes in the ensemble junction-point pressure. For each experiment, 1.0 µL of the target test mixture was injected into the Varian split inlet using a 10:1 nominal split ratio and a 250 °C injection port temperature. All chromatograms were obtained using a starting column temperature of 150 °C and a liner temperature-programming rate of 30 °C/min, beginning at the time of sample injection. In addition, retention factor data from previous studies21-23 for the two columns in the ensemble were used for spreadsheet calculations to demonstrate the effects on retention time of junction-point pressure changes for an isothermal separation of volatile compounds. These compounds also are listed in Table 1. RESULTS AND DISCUSSION Programmable Selectivity for Isothermal GC. To develop a conceptual framework for the temperature-programmed studies reported here, a model was developed for the simpler case of a single junction-point pressure step applied during an isothermal Analytical Chemistry, Vol. 72, No. 21, November 1, 2000

5451

Table 1. Compound Lists for PCB Test Mixture and for Isothermal Modeling of Volatile Compounds PCB Mixture

l1 ) tp/[tm1,s(k1 + 1)]

(3)

Substitution of eq 3 into eq 2 and rearranging gives eq 4. Equation

peak no.

IUPAC no.

compound

1 2 3 4 5 6

110 77 151 149 118 186

2,3,3′,4′,6-pentachlorobiphenyl 3,3′,4,4′-tetrachlorobiphenyl 2,2′,3,5,5′,6-hexachlorobiphenyl 2,2′,3,4′,5′6-hexachlorobiphenyl 2,3′,4,4′,5-pentachlorobiphenyl 2,2′,3,4,5,6,6′-heptachlorobiphenyl

peak no.

Volatile Component Mixture compound

bp (°C)

1 2 3 4 5

n-nonane ethylbenzene m-xylene o-xylene 2-hexanol

151 136 138-139 143-145 136

tR ) tm1,f(k1 + 1) + tp[1 - (tm1,f/tm1,s)] + tm2,f(k2 + 1) (4) 4 assumes that the entire solute band migration in the second column occurs at the final junction-point pressure. If the pressure change occurs when the solute band is in the second column, then the fractional length of the second column l2 traversed by the solute band at the time of the pressure change is given by eq 5.

l2 ) [tp - tm1,s(k1 + 1)]/[tm2,s(k2 + 1)]

(5)

Substitution of eq 5 into eq 2 and rearranging gives eq 6. Equation separation. In this case, retention factors are constant on each column during the course of the separation. The goal is the prediction of ensemble retention times as functions of the time tp after sample injection at which the pressure change occurs. Equation 1 gives the ensemble retention time tR as the sum of

tR ) tG1,s + tG1,f + tG2,s + tG2,f

(1)

the solute-band migration times tG in the first (1) and second (2) columns and for the starting junction-point pressure (s) and the final junction-point pressure (f). If the pressure change occurs when the solute band is in the first column, then the entire migration time on the second column occurs with the final junction-point pressure, and the third term on the right side of eq 1 is zero. It the pressure change occurs after the solute band has migrated across the junction and is in the second column, then the entire migration time on the first column occurs with the starting junction-point pressure, and the second term on the right side of eq 1 is zero. If gas compression effects are neglected, then the carrier gas velocity is constant along each of the columns for a specified valve of junction-point pressure. In this case, eq 1 can be recast in terms of the retention factors k1 and k2 on the two columns and the holdup times tm on the two columns at the two valves of junctionpoint pressure.

tR ) tm1,s(k1 + 1)l1 + tm1,f(k1 + 1)(1 - l1) + tm2,s(k2 + 1)l2 + tm2,f(k2 + 1)(1 - l2) (2) where l1 is the fraction of the total first column length that the solute band has traversed at the time of the pressure change and l2 is the fraction of the total second column length that the solute band has traversed at the time of the pressure change. Note that one of the terms on the right side of eq 2 is always zero. Since gas compression effects are neglected, the length fraction values l1 and l2 can be expressed in terms of the pressure change time tp and the band elution times for the respective columns. For the case where the pressure change occurs while the solute band is still in the first column, 5452

Analytical Chemistry, Vol. 72, No. 21, November 1, 2000

tR ) tm1,s(k1 + 1) + [tp - tm1,s(k1 + 1)][1 - (tm2,f/tm2,s)] + tm2,f(k2 + 1) (6) 6 assumes that the entire solute band migration in the first column occurs at the starting junction-point pressure. From eqs 4 and 6, plots of ensemble retention time versus the junction-point pressure change time should be straight lines with slopes of 1 - (tm1,f/tm1,s) if the pressure change occurs while the solute band is in the first column and 1 - (tm2,f/tm2,s) if the pressure change occurs while the solute band is in the second column. For a junction-point pressure change to a higher pressure value, tm1,f/tm1,s is greater than 1.0, and tm2,f/tm2,s is less than 1.0. In this case, the plots of tR vs tp have a negative slope for pressure changes occurring while the solute band is on the first column and a positive slope for pressure changes occurring while the solute band is in the second column. The opposite situation occurs for a junction-point pressure change to a lower pressure value. Note that these slopes are the same for all components in a mixture. Figure 1 shows plots of ensemble retention time versus the time of the junction-point pressure change for the five-component mixture of volatile compounds listed in Table 1 using junctionpoint pressure changes from 22 to 31 psia (a) and from 31 to 22 psia (b). Numbers near the plots correspond to the compound numbers in Table 1. These plots were generated from spreadsheet calculations using a previously described algorithm.22 In the algorithm, gas compression effects are included, and local carrier gas velocity is calculated for 1.0-cm intervals along the column ensemble axis. Retention factors then are used to calculate the solute migration times for each 1.0-cm interval. Integration over the ensemble length gives the ensemble retention times. Each plot has two branches. The left branch corresponds to the pressure change occurring while the solute is in the first column, and the right branch corresponds to the pressure change occurring while the solute is in the second column. The flat portion of each plot for large pressure change times corresponds to the pressure change occurring after the component has eluted from the column ensemble, and thus, the entire separation is conducted with the starting junction-point pressure. For every plot, the

Figure 1. Retention times vs junction-point pressure change initiation times for an on-the-fly presssure change from 22 to 31 psia (a) and from 31 to 22 psia (b) during an isothermal separation. Retention factor data (40 °C) were used for the list of volatile compounds in Table 1. Plot numbers correspond to compound numbers in the table.

intersection of the left and right branches corresponds to the elution time from the first column of the corresponding mixture component. Curvature of these plots is the result of carrier gas compression effects. Linear regression fits to the left and right branches of the plots give slope values that are within 1-2% of the values calculated from eqs 4 and 6. An important feature observed in the plots of Figure 1 is that a pressure change occurring when components are on the same column can significantly change retention times but has little effect on the pattern of peaks from the column ensemble. However, pressure changes occurring when components are on different columns can produce a large change in the ensemble peak pattern. This provides a useful means for adjustment of the ensemble selectivity by adjustment of the time of the junction-point pressure change. Retention Characteristics for a PCB Test Mixture. Chromatograms of the six-component PCB mixture were obtained for junction-point pressures in the 20.0-32.0 psia range in 1.0 psi intervals. In all cases, a 30 °C/min temperature program was initiated at the time of injection. Values of retention time are plotted versus junction-point pressure in Figure 2a. No on-the-fly junction-point pressure change was used during the separation. For each mixture component, a minimum retention time value is observed for a junction-point pressure of ∼26 psia. This value is close to the pressure that would exist at the column junction point

Figure 2. Retention time vs junction-point pressure for the six PCB congeners (a) and plots of nearest-neighbor peak apex separation vs junction-point pressure for the six PCB congeners (b). Numbers identifying the retention time plots correspond to the compound numbers in Table 1. Letters identifying the peak separation plots correspond to the nearest-neighbor peak pair with (A) for the first two peaks to elute and (B) for the second and third peaks. Broken vertical lines indicate the junction-point pressure values used for subsequent study.

in the absence of any additional connections. As the controller set-point pressure deviates from this value, a steady increase in ensemble holdup time occurs. At the high and low ends of the pressure range shown in the figure, the average linear carrier gas velocity in the first and second columns, respectively, decreases rapidly, and the component retention times increase accordingly. Note that the plots for components 2 and 3 cross at a junctionpoint pressure of ∼25 psia, and the use of this pressure will result in the coelution of these mixture components. A similar situation occurs for components 5 and 6 at a junction-point pressure of 28 psia. At the high-pressure end (31-33 psia), the plots for component pairs 1/3 and 2/4 coalesce, and these component pairs are not adequately separated in this pressure range. From the retention time versus junction-point pressure plots in Figure 2a, the nearest-neighbor separation plots shown in Figure 2b were constructed. Plot A gives the separation in seconds of the first two peaks to elute irrespective of which components are involved. In the pressure range from 20 to 25 psia, components 1 and 2 are the first pair to elute, and their separation increases with increasing junction-point pressure. At a pressure of 25 psia, components 2 and 3 coelute, and their elution order changes at higher pressures. The result is that, at pressures greater than 25 Analytical Chemistry, Vol. 72, No. 21, November 1, 2000

5453

Figure 3. Chromatograms of the six-component PCB mixture using junction-point pressure values of 31 (a) and 22 psia (b). Peak numbers correspond to the compound numbers in Table 1.

Figure 4. Chromatograms of the six-component PCB mixture using an initial junction-point pressure of 22 psia with a change to 31 psia occurring 0 (a), 40 (b), 80 (c), 120 (d), and 180 s (e) after injection.

psia, components 1 and 3 are the first-eluting peak pair. The separation of this pair then decreases with increasing junctionpoint pressure until they coelute at 32 psia. Plot B gives the separation of the second and third peaks to elute. This peak pair corresponds to components 2 and 3 over the entire pressure range with a coelution and change in elution order occurring at a junction-point pressure of 25 psia. This method of using nearestneighbor peak separations for the purpose of determining the optimal junction-point pressure for temperature-programmed GC is useful because peak widths in temperature-programmed GC can be relatively independent of retention time over a limited retention time range. From the plots in Figure 2b, junction-point pressure values of 22 and 31 psia were chosen for studies using an on-the-fly pressure change during the analysis. These pressures are indicted by the broken vertical lines in Figure 2b and are the same pressures used for the isothermal retention time plots in Figure 1. At 31 psia, components 1 and 3 should nearly coelute, and the separation of 2 and 4 (plot C in Figure 2b) should be ∼1 s. This is confirmed by the chromatogram shown in Figure 3a. At 22 psia, component pair 1/2 (plot A in Figure 2b), component pair 2/3 (plot B), and component pair 4/5 (plot D) should all have separations of ∼2 s. This is confirmed by the chromatogram shown in Figure 3b. Selectivity Programming for Temperature-Programmed GC. All selectivity programming studies were performed using either a pressure step from 22 to 31 psia or a pressure step from 31 to 22 psia at varying times after sample injection. Pressure steps

from the low pressure to the high pressure are rapid, with the measured time needed for the transition from 10 to 90% of the target value less than 1.0 s. Because the large internal volume of the pressure controller (>7 cm3) must be vented before pressure equilibration is complete, downward pressure transitions take significantly longer. For the pressure step used in this study, the time needed for the transition from 10 to 90% of the target value is 5.8 s. Figure 4 shows chromatograms obtained with a starting pressure of 22 psia and an on-the-fly pressure change to 31 psia occurring 0 (a), 40 (b), 80 (c), 120 (d), and 180 s (e) after sample injection. Note that chromatogram a corresponds to the entire separation occurring with a junction-point pressure value of 31 psia, and not surprisingly, the chromatogram is very similar to the one shown in Figure 3a. Since the final junction-point pressure is greater than the starting pressure, the ensemble retention times should decrease when the pressure change occurs while the mixture components are on the first column. This is seen in chromatograms b and c in Figure 4. When the pressure change occurs 40 s after sample injection (chromatogram b), the same peak pattern is observed as for the case where the entire separation was performed at this pressure (tp ) 0), but all retention times are significantly reduced, and peaks are significantly broader. When tp is increased to 80 s (chromatogram c), retention times are further reduced and much broader peaks are observed. However, peak apex separations also

5454 Analytical Chemistry, Vol. 72, No. 21, November 1, 2000

Figure 5. Retention time vs the time after injection of a junctionpoint pressure change from 22 to 31 psia (a) and from 31 to 22 psia (b). The numbers identifying the plots correspond to the component numbers in Table 1.

have increased, and the resolution of peaks 2 and 4 is significantly greater than in chromatogram a. When tp is 120 s (chroamtogram d), the mixture components are on the second column, and the peak pattern is similar to that observed when the entire separation is conducted with a junctionpoint pressure of 22 psia (starting pressure), except the retention times are substantially shorter, and the peaks are narrower. When tp is 180 s (e), all mixture components elute prior to the pressure change, and component retention times and peak widths are very similar to those in chromatogram b in Figure 3, which was obtained with a junction-point pressure of 22 psia. While the peaks are broader in chromatogram e relative to d, peak separations also are greater with the result that the resolution of adjacent peak pairs is comparable for the two chromatograms. The conditions used for chromatogram d would be beneficial because of substantially shorter analysis time. Figure 5a shows component retention times for a starting junction-point pressure of 22 psia and an on-the-fly pressure change to 31 psia implemented with various tp values using time intervals of 10 s. Figure 5b shows component retention times for a starting junction-point pressure of 31 psia and an on-the-fly pressure change to 22 psia implemented with various tp values, again using time intervals of 10 s. Error bars corresponding to one standard deviation for three injections are shown for each point, but in most cases, they are barely discernible. Under temperature-programmed conditions, component residence times on the first column are greater than on the second column for the case of comparable holdup times for the two

columns. This is due to the acceleration of the solute bands as the column temperature increases. For the plots in Figure 5a, the general decrease in ensemble retention times with increasing tp if the components are in the first column and then increase in retention times if the components are in the second column is comparable to the isothermal data in Figure 1a, but the right and left branches of the plots for each component show much more curvature because of the steady decrease in retention factors as the column ensemble is heated. For the conditions used in Figure 5a, components cross the junction point in the time range from about 80 to 100 s after injection if no pressure change occurs prior to this time. When tp is greater than ∼100 s, all mixture components migrate through the first column entirely under the influence of the 22 psia junction-point pressure, and their elution times on the first column are not affected by latter pressure changes. Elution times on the second column, however, increase steadily with increasing tp beyond ∼100 s due to the lower average carrier gas velocity on the second column with a junction-point pressure of 22 psia relative to 31 psia. If tp is greater than ∼170 s, all components have eluted from the column ensemble, and the entire separation is performed with a junction-point pressure value of 22 psia. This case is indicated by the flat portion of the plots for tp greater than ∼170 s. The plots in Figure 5b for the case of a junction-point pressure change from 31 to 22 psia are comparable with the isothermal plots in Figure 1b in that the ensemble retention times are shifted to larger values with increasing tp if the pressure change occurs while the mixture components are on the first column, but the shapes of the left and right branches are very curved and skewed because of the steady reduction of retention factors during the temperature-programmed separation. The largest retention time values in Figure 5b occur when tp is in the range 140-150 s. This corresponds to the elution times for the mixture components from the first column when the junction-point pressure is 31 psia. For this case, the ensemble retention times are in the range 185-195 s, and the mixture components spend ∼75% of their total migration time on the first column. If the pressure change is implemented after the mixture components have migrated across the column junction, the ensemble retention times decrease rapidly with increasing tp. If tp is more than ∼175 s, all components have eluted from the ensemble, and the junction-point pressure change has no effect. This corresponds to conducting the entire separation with a junction-point pressure of 31 psia. Note that, for this case, over 80% of the ensemble retention time is spent on the first column. Again, this is the result of the large holdup time for the first column for a junction-point pressure of 31 psia and use of a relatively fast temperature program. Note that the elution order changes in Figure 5, indicated by the crossing points in the corresponding plots, all occur near the minimum retention times for Figure 5a and near the maximum retention times for Figure 5b. These regions are shown in greater detail in parts a and b of Figure 6, respectively. These data were taken with tp intervals of 1.0 s. Error bars show standard deviations for three injections. For each plot there is an abrupt slope change, which indicates that the pressure change is occurring at the time that the Analytical Chemistry, Vol. 72, No. 21, November 1, 2000

5455

Figure 6. Retention time vs the time after injection of a junctionpoint pressure change from 22 to 31 psia (a) and from 31 to 22 psia (b) shown on an expanded time scale. The numbers identifying the plots correspond to the component numbers in Table 1.

corresponding component migrates across the column junction point. Note that the elution order changes are very abrupt, and a difference in tp of 1 or 2 s can result in a reversal of elution order for a particular component pair. This provides exciting possibilities for the control of retention patterns with temperature-programmed GC. As long as a junctionpoint pressure change is implemented while the mixture components of interest are still on the first column, the effects of that change will be reflected in the ensemble elution pattern. If the pressure change is implemented after the components of interest have crossed the column junction point, the pressure change will not affect the order of peaks eluting from the ensemble. Changes in pressure that occur much before or much after the migration of the components into the second column have very little effect on the elution pattern. This suggests that the elution patterns for small groups of components can be manipulated quite independently, and the ensemble retention pattern for a component group depends primarily on the junction-point pressure at the time when the group migrates across the column junction point. Other peak patterns can be obtained if the pressure change occurs when some of the component bands are in the first column and some in the second column when the pressure change occurs. For example, components 1 and 3 in Figure 6a reach the junction point at ∼84 s if no pressure change occurs prior to this time. In this case, they migrate across the junction when the junction-point pressure is 22 psia and the peak separation upon 5456 Analytical Chemistry, Vol. 72, No. 21, November 1, 2000

Figure 7. Chromatograms of the six-component PCB mixture using a junction-point pressure change from 22 to 31 psia occurring at various times after injection. The numbers to the left of the chromatograms give the time (s) after injection of the junction-point pressure change. Peak identification numbers correspond to the component numbers in Table 1.

elution from the ensemble is considerably greater than for the case where the pressure change is made earlier than 84 s, and these components migrate across the junction when the pressure is 31 psia. Without a junction-point pressure change, component 2 reaches the junction ∼89 s after injection. This is the pressure change time in Figure 6a for which the slope change for this component occurs. If the junction-point pressure change is delayed more than 89 s, components 2 and 3 show a reversal of elution order. Figures 7 and 8 show chromatograms for the conditions described in parts a and b of Figure 6, respectively. The numbers to the left of the chromatograms give the tp values in seconds. For Figure 7, the tp values increase by 1.0 s from top to bottom for each consecutive chromatogram. For Figure 8, tp values increase by 2.0 s from top to bottom for each consecutive chromatogram. For the top chromatogram in Figure 7 (tp ) 82 s), the pressure change occurs while all six components are still on the first column. Thus, all components migrate across the junction with the junction-point pressure set at 31 psia. This is a relatively high pressure, and the local carrier gas velocity in the first column is correspondingly low. This results in the mixture components eluting from the first column as relatively broad peaks with relatively large temporal spacing between the peaks. The chromatogram appears as if the time axis had been stretched relative

Figure 8. Chromatograms of the six-component PCB mixture using a junction-point pressure change from 31 to 22 psia occurring at various times after injection. The numbers to the left of the chromatograms give the time (s) after injection of the junction-point pressure change. Peak identification numbers correspond to the component numbers in Table 1.

to the case where the local carrier gas velocity at the outlet of the first column is lower at the time that the mixture components cross the junction. When tp is increased to 84 s, components 1 and 3 are partially across the junction when the pressure change occurs. When the pressure change occurs, the carrier gas velocity in the first column decreases and the velocity in the second column increases. The result is that the portions of these component peaks that have crossed the junction speed up while the portions remaining on the first column slow. This results in peak doubling, which is seen in the chromatogram. When tp is 87 s, components 2, 4, and 6 are all crossing the junction, and peak doubling is observed for these components as well. When tp is 89 s, all components except 5 have crossed the junction. Note that component 5 (the last peak in the chromatogram) is broader than the other peaks. This is because the other components crossed the junction when the junction-point pressure was 22 psia and thus eluted from the first column under conditions of relatively high carrier gas velocity. For tp values of 90 and 91 s, component 5 is crossing the junction and the elution order of components 5 and 6 changes with the result that a single broad and distorted feature is observed in the ensemble chromatogram for these two components. When tp is 92 s or more, all components are on the second column at the time of the pressure change,

and the ensemble elution pattern is similar to the case where the entire separation is conducted with a junction-point pressure of 22 psia. The chromatograms in Figure 8 illustrate the ensemble retention pattern changes that occur for a junction-point pressure change from 31 psia at the time of injection to a value of 22 psia. If the pressure change occurs prior to the components crossing the column junction, the carrier gas velocity at the outlet of the first column will be relatively high, and the peaks eluting from the first column are relatively narrow and closely spaced. If this downward pressure change occurs while a component is migrating across the junction, the portion of the component band on the first column will speed up and the portion on the second column will slow. This can result in distorted peaks from the column ensemble but not in peak doubling. If tp is 130 s, all components are still on the first column, and the ensemble retention pattern is similar to the case where the entire separation is conduced with a junction-point pressure of 22 psia. For tp values in the range from 132 to 138 s, a change in ensemble elution order for components 2 and 3 is observed. This change in pattern is less abrupt with respect to changes in tp values than for the case of an increase in junction-point pressure shown in Figure 7. This is due to the longer pressure equilibration time for the downward pressure change, which is caused by the dead volume in the pressure controller and its associate plumbing.22 If tp is 142 s, components 2 and 3 have crossed the junction and appear as a single, broad peak. When these components cross the junction, the junction-point pressure is still 31 psia, and thus, the carrier gas velocity at the outlet end of the first column is relatively low. This results in broader injection plugs for the second column relative to the remainder of the components, which cross the junction with a junction-point pressure of 22 psia. When tp is 148 s, only component 5 is still on the first column and only this component shows a relatively narrower peak in the ensemble chromatogram. If tp is more than 150 s, all components are on the second column, and the pressure change has no effect on the elution order. CONCLUSIONS The work reported here lays the foundation for understanding the effects of a single upward or downward junction-point pressure change for a pressure-programmable ensemble of two capillary columns with temperature-programmed operation. Any upward pressure change will result in a reduction in ensemble retention times for all mixture components still on the column ensemble, and the reduction will be greater for components that are closer to the junction point at the time of the pressure change. Any downward pressure change will result in an increase in ensemble retention times, and the increase will be greater for components that are closer to the junction point at the time of the pressure change. The ensemble retention pattern depends primarily on the junction-point pressure at the time that the mixture components cross the junction. The function of the first column is to disperse the solute bands from the sample and sequentially inject them into the second column. If a pressure change occurs when sample bands are near the junction, significant stretching or pinching of the chromatogram occurs. If the junction-point pressure is increased just before a band is injected into the second column, the carrier gas velocity Analytical Chemistry, Vol. 72, No. 21, November 1, 2000

5457

in the first column will be reduced and the band will be injected more slowly into the second column. This will result in larger ensemble peak widths and greater peak apex separation, but the resolution will be affected primarily by how the pressure change alters the efficiency of the second column. The chromatogram will appear stretched relative to the case where the entire separation is conducted at the higher pressure. If the junction-point pressure is decreased just before a band is injected into the second column, the carrier gas velocity in the first column will increase, and the band will be injected more quickly into the second column. This will result in smaller ensemble peak widths and smaller peak apex separations. The chromatogram will appear compressed relative to the case where the entire separation is conducted at the lower pressure. If a pressure change occurs during the passage of a component through the junction point, peak doubling or peak shape distortion is likely. However, the effects of a pressure change on peak shape for a downward pressure change are complicated by the relatively long pressure equilibration time for the junction-point-pressure control system used in this work.

5458

Analytical Chemistry, Vol. 72, No. 21, November 1, 2000

Programmable column selectivity with temperature programming provides precise control of retention patterns over a certain range for groups of compounds in a chromatogram. Future work will include investigating the effect of multiple pressure steps as well as pressure pulses on the chromatography. In addition, the application of these techniques to more complex mixtures will be investigated. ACKNOWLEDGMENT The authors gratefully acknowledge Varian Instruments, Walnut Creek, CA, for equipment gifts, the Centers for Disease Control and Prevention (CDCP) through the National Environmental Laboratory for financial support of this work, and Tincuta Veriotti, Department of Chemistry, University of Michigan, for the spreadsheet modeling work.

Received for review May 30, 2000. Accepted August 24, 2000. AC000610H