Cryofocusing inlet with reverse flow sample collection for gas

Pressure-Tunable GC Columns with Electronic Pressure Control. Heather Smith and Richard Sacks. Analytical Chemistry 1997 69 (24), 5159-5164...
1 downloads 0 Views 697KB Size
Anal. Chem. 1993, 65, 2516-2521

2516

Cryofocusing Inlet with Reverse Flow Sample Collection for Gas Chromatography Mark A. Klemp,?Michael L. Akard, and Richard D. Sacks’ Department of Chemistry, University of Michigan, Ann Arbor, Michigan 48109

An inlet system for high-speed gas chromatography is characterized. Sample is cryofocused in a bare metal capillary tube and released by a fast temperature increase caused by an electric current. The gas flow directionthrough the capillary tube is reversed after sample collection so that the sample injection onto the capillary column for separation occurs from the downstreamend of the metal tube. The system offers narrower initial band widths, elimination of memory effects, and reduced thermal decompositionduring reinjection relative to previously describedcryofocusinginlet systems. Sample vapor can be collected and cryofocused from any source pressure greater than -20 Torr. A configuration for direct air sampling is described. INTRODUCTION High-speed gas chromatography (GC) is gaining popularity for applications in chemical process monitoring, environmental monitoring, and screening of organic solvents. Highspeed separations are achieved by the use of relatively short capillary columns operated a t unusually high carrier gas flow rates.1-12 The successful operation of a high-speed capillary GC system requires the minimization of extracolumn band broadening. Often, the injection technique is a major source of these extracolumn effects. A variety of injection techniques have been developed for high-speed applications. These include fluid logic gates,213 a small orifice in the side of a moving capillary tube,- low dead volume sample valves? and electrically heated cold traps.Sl2 Previous studies have described an electrically heated cold trap inlet system for high-speed analysis.gJ0 This system is based on condensation and cryointegration of sample vapor from a conventional injection port onto a C d N i alloy capillary tube. After sample collection, the cold trap tube is resistively

* Author to whom correspondence should be addressed.

+ Present address: Chromatofast, Inc., 912 N. Main St., Suite 14, Ann Arbor, MI 48104. (1) Giddings, J. C.; Seager, S. L.; Stucki, L. R.; Stewart, G. H. Anal. Chem. 1960,32,867-870. (2) Wade, R. L.; Cram, S. P. Anal. Chem. 1983,44, 131-139. (3) Annino, R.; Leone, J. J . Chromatogr. Sci. 1982,20, 19-26. (4) Tijssen, R.; Van den Hoed, N.; Van Kreveld, M. E. Anal. Chem. 1987,59, 1007-1015. (5) Peters, A.; Sacks, R. J. Chromatogr. Sci. 1991,29,403-409. (6) Peters, A.; Sacks, R. J. Chromatogr. Sci. 1992,30, 187-191. (7) Van Es, A.; Janssen, J.; Bally,R.; Cramers, C.; Rijks, J. A. HRC h CC, J. High Resolut. Chromatogr. Chromatogr. Commun. 1987,10,273279. (8) Hopkins, B.; Pretorius, V. J. Chromatogr. 1978, 158, 465-469. (9) Ewels, B.; Sacks, R. Anal. Chem. 1985,57, 2774-2779. (10) Lanning,L.;Sacks, R.; Mouradian, R.; Levine, S.; Foulke, J. Anal. Chem. 1988, 60,1994-1996. (11) Klemp, M.; Sacks, R. J. Chromatogr. Sci. 1991, 29, 114-121. (12) Van Es, A,; Janssen, J.; Cramers, C.; Rijks, J. HRC & CC, J.High Resolut. Chromatogr. Chromatogr. Commun. 1988,11, 852-855. 0003-2700/93/0365-2516$04.00/0

heated with a capacitive discharge power supply.’O This serves to flash vaporize the sample and deliver it to the column as a narrow vapor plug. This cold trap inlet system was used with a vacuum-pump-operated back-flush and sample recycle system for the analysis of high-purity solvents.llJ”16 By venting most of the solvent through the flame ionization detector (FID) and then using a reverse gas flow direction to retrap the sample components remaining on the column, improved precision and accuracy were achieved for components which elute on the solvent tail. While the electrically heated cold trap system proved to be analytically useful, certain operating features either limited its range of use or required extra procedures for satisfactory operation. These limitations include thermal decomposition of some sample components during trap heating and injection port memory effects. In addition, with these previous designs, it is necessary to introduce sample vapor into the trap tube a t the elevated pressure needed to drive the carrier gas through the trap tube and column. In the study reported here, a new configuration of the electrically heated cold trap is investigated. This system provides minimal sample decomposition during the injection process and eliminates memory effects from the injection port. In addition, direct introduction of sample vapor can be achieved from any source pressure greater than -2.7 kPa (20 Torr). A significantly narrower injection bandwidth also is obtained.

EXPERIMENTAL SECTION Apparatus. Figure 1 shows a schematic of the system

illustrating flow directions during three operating modes. Component specifications are found in Table I. Part a shows the sample collection mode; (b) shows the sample injection and separation mode; (c) shows the column back-flushmode. Valves VI and Vz are pneumatically controlled onloff valves. The capillary separation column is represented by C and the electrically heated capillary cold trap by T. The trap tube is cooled by a continuous flow of cold gas. The cold trap, the cooling system, and the capacitive discharge heating circuit have been described in detail.BJ0 Components labeled R1,&, Ra,and R4 are lengths of deactivated fused-silica capillary tubing used to provide flow restrictions. Points labeled PH,Ps, PL,y d PRare the pressures of the carrier gas supply, the sample inlet, the vacuum pump, and the detector, respectively. Arrows show gas flow directions. Pressure PL from the vacuum pump represents the lowest pressure of the system at -1.3 kPa. The detector pressure at the FID is 1 atm. The carrier gas pressure PH is the highest pressure, and it is adjusted to provide the appropriate carrier gas flow through the column. The sampling port pressure PS can range from just below PHto just above PL. Figure l a shows the sample collection and cryofocusing mode. In this mode valve VI is open, Vz is closed, and the cold trap is maintained at a low trapping temperature. With the vacuum pump in the system, points PH,Ps, and PRall have flows to PL. (13) Klemp, M.; Sacks, R. J. Chromatogr. Sci. 1991, 29, 243-247. (14) Klemp, M.; Sacks, R. J . Chromatogr. Sci. 1991, 29, 248-252. (15) Rankin, C.; Sacks, R. J. High Resolut. Chromatogr. 1990,13,674678. 0 1993 American Chemlcal Soclety

ANALYTICAL CHEMISTRY, VOL. 65, NO. 18, SEPTEMBER 15, 1993

Figure 1. Schematicdiagram of the reverse fbw samplingcryofocuslng Instrumentfor hlgh-speed OC: (a)sample cdlectlon mode; (b) reln)ectkn and separatlon mode; (c)column back-flush mode. PH,PL,Ps, and PR represent the pressures of the carrler gas, vacuum pump, sample source, and detector, respectlvely. V1 and V2 are pneumatically operated gas valves, R1-R4 are capillary flow restrlctors, T Is the cryofocuslng tube, and C Is the Separation column.

Thus, sample vapor introduced at PS will flow through the restrictors R2 and Rs and then into the cold trap, where it will be cryofocused. This reverse flowmeans the sample is introduced into the trap end closest to the separation column. Notice that, during the collect and trap mode, the column is back-flushed with flame gases from the FID.l6 Note also that the sample does not come into contact with any valve surfaces. This is important for eliminating possible contamination from valves and sample alteration or condensation on the valve surfaces.16 Figure l b shows the injection and separation mode. Valve V1 is closed and Vz is open, causing the flow through the cold trap to be from left to right. The trap is then resistively heated with a capacitive discharge power s ~ p p l y . ~This J ~ serves to rapidly vaporize the sample, delivering it as a narrow vapor plug to the analytical column. The carrier gas flow is initially split between R1 and valve V2. Gas flows from the trap and Ra are then combined after the cold trap. The inlet restrictor Rz is purged with the remaining carrier gas flow from Rd during this operation. The restrictor Rd is chosen to supply only the flow necessary to purge R2 while adding only minimal flow into Ra. The back-flush mode is shown in Figure IC. In this mode, both valves are open, and flows through the column and the cold trap are from right to left. Therefore, any components which have not yet eluted from the column are back-flushed toward the cold trap. If the trap is cooled to trapping temperatures, the back-flushed components will be cryofocused again for a subsequentinjection. This givea the systemthe capabilityof selective component recycling.11Js16 Alternatively, the trap can be maintained at an elevated temperature with a sustaining current from the power supply. This will flush the remaining sample through the trap and out to vent through the vacuum pump at PL. During the back-flush mode the sample inlet restrictor Rz is still being purgedwith carrier gas. This prevents the collection of a new sample and eliminates contamination from residual sample in &. Materials and Procedures. The components were installed in a modified Varian 3700 GC. The split injector and FID from the Varian were used without change. For direct air monitoring, (16) J o w o n , J. A.;Vejrosta, J.; Novak, J. J. Chromtogr. 1982,236, 307-312.

2517

sampling was done directly into the open end of restrictor Rz. The FID was connected to a high-speed electrometer/amplifier which was constructed in-house. A filter with a 5-ma time constant was used to reduce 60-Hz interference. Fused-silica columns 385cm long,0.25-mm i.d. with a 0.25-pm film of methyl silicone-bonded stationary phase, were used for most studies. Hydrogen carrier gas average linear velocity through the separation column ranged from 120 to 150 cm/s (3.5 to 4.4 mL/ min). These average velocitieswere measured by the procedure outlined by K1emp.l' Column temperature was 60 OC unless otherwise noted. The Varian split injector was maintained at 200 OC. The FID was held at 200 OC. Trapping temperature was -80 OC, unless otherwise noted. Typical injections were 1.0 pL. Sample Components were obtained from Aldrich and were at least 98% pure. The air mixture was obtained from Scott Specialty Gases. The component concentrations were all close to 100 ppm in dry air. The air sampling was done directly out of a Tedlar bag from Chromatography Research Supplies. The end of fused-silica restrictor tube R2 was passed through a 22-gaugeneedle piercing a septum in the bag. The needle was then withdrawn leaving the tube embedded in the septum. Split flow ratios were determined at room temperature. A Hastings Model ST mass flowmeter with a PR-4AJ four-channel monitor was inserted after the injector and before Rzto determine the flow from the injector. Another digital mass flow meter measured the split flow. The split ratio was 5501. A typical run started with the sampling mode for 8 a. Manual injection into the Varian injection port was done directly after the start of the sampling mode. Immediately following the sampling mode, a brief back-flush mode lasting 2.0 s was used to deliver to the trap tube any sample still left in restrictor Rs. Next, valve V1 was closed, while valve V2 was left open. The capacitive-discharge trap heater circuit was then fired after a short flow stabilization pause. This allowed the flow through the trap to return to the direction of the inject and separate mode as shown in Figure lb. The data acquisition for the chromatogram was started simultaneously with the heating of the trap tube.

RESULTS AND DISCUSSION Timing Considerations. The duration of the back-flush that follows the sampling mode can be varied. The purpose of the back-flush is to trap all the remaining sample in the restrictor RB. The effect of increasing this time beyond 1s was not significant. For times under 1a, not all the sample was trapped and the peak area decreased. This back-flush mode ends when valve VI is closed (Figure lb) and the flow changes from the reverse direction to the normal direction. Varying the time between closing valve VI and heating the trap tube did have a significant effect on the chromatogram. Figure 2 shows octane peaks with different retention times due to varying this delay. The time is in seconds after the trap heating pulse is applied. Peaks A-F correspond to delays of 0,0.5,1,2, 4, and 10 s, respectively. Note that the retention times for peaks with delays of 4 and 10s (peaks E and F) are approximately the same. Therefore, it appears that the normal flow direction resumes within -4 8.

Delays as long as 30 s did not significantly change the retention times. This would indicate that there was no appreciable creep of the trapped component along the length of the trap during normal flow. Retention time reproducibility for all delay times was in the range from 0.20% to 0.06% relative standard deviation. Injection Bandwidths. A necessary consideration in the development of a high-speed GC system is the minimization (17)Klemp, M. A.; Sacks,R.D.J. Chromatogr. Sci. 1991,29,507-510. (18) Gasper, G.; Arpino, P.; Cuiochon, G . J. J. Chromatogr. Sci. 1977, 15,256-261. (19)Graydon, J.; Grob, K.J. Chromatogr. 1983,254, 266269.

2518

ANALYTICAL CHEMISTRY, VOL. 65, NO. 18,

SEPTEMBER 15, 1993

Table I. Instrumentation Specifications for the Reverse Flow Sampling Inlet System and the High-speed GC

Gas Chromatograph Varian Model 3700 with inlet splitter, SGE Model BPR-30 back-pressure regulator, J&W Scientific Model SA202-3(3)-1 gas flow controller, and flame ionization detector Cold Trap System quartz tube, 25-mm i.d., 28-mm o.d., 100 mm long Cu (70%)/Ni(30%),0.508-mm i.d., 0.305-mm o.d., 25 cm long iron/constantan, 36 gauge, type J; copper/constantan, 36 gauge, type T SGE SS microtubing union for 0.4-mm 0.d. ends of trap tube wrapped with 2.5 cm X 12.5 cm heating tape, 25 W each Vacuum Back-Flush Central Scientific Hyvac 7, two stage SGE micropneumatic on/off valve in 'L" configuration with 50-mm stem Valcor solenoid valve Model H55P18DlA Chromfit all-glasssplitter 0.10.25-0.25 mm, 25-cm-long, 0.25-mm4.d. fused-silicatube to vacuum pump deactivated fused-silicatubing, 0.1-mm i.d. R1,14 cm long; Rz, 11cm long; %,40 cm long; & 15 cm long Data Collection Metrabyta DAS-8PGA 12-bit A/D board Dell Systems 200 AT-286

cooling sheath trap tube thermocouples connectors transfer lines vacuum pump pneumatic valves electric valve connector restrictom A/D computer

E.F

't

I

6

2l

4 1

I

1

0 1

B

0 "

5.9

"

6.1

I

6.3

"

I

"

6.5

6.7

'

6.9

"

7.1

Time (S)

Flgurr 2. *Octane peaks for different delay times prior to cold trap heating. Delay times for peaks A-F are 0, 0.5, 1, 2, 4, and 10 s.

of extracolumn band broadening.19 Of the described techniques to minimize injection band broadening,'-l2 the electrically heated cold trap has the advantage that it can cryofocus a sample from a dilute and disperse source and then rapidly vaporize the sample as a narrow vapor plug. One problem associated with the cold trap is that the trap tube must be long enough to cryofocus components of the highest volatility that are of interest in the sample. Studies have suggested that capillary cold traps longer than 20 cm should be used for efficient operation.'E Often the majority of the sample will be cryofocused along the temperature gradient at the upstream end of the tube, causing the rest of the tubing to act as a dead volume during reinjection. This dead volume has the largest efficiency effect on components with small capacity ratios, where high efficiency is most needed. The effect of this dead volume is minimized by initially placing the sample at the downstream end of the cold trap (Figure lb). Figure 3 demonstrates this improvement with pentane peaks from the normal flow inlet system in trace A and the reverse flow sampling inlet system in trace B. For both cases, 0.11 nL of sample was collected in the cold trap. The flow rate for the peak in trace A was increased slightly to give similar retention times for both peaks. Notice that with the reverse flow sampling inlet system,the pentane peak is significantlynarrower and taller. The number of theoretical plates generated for peak B was 22 400, compared to only 9400 for peak A. Injection bandwidths with the cold trap were evaluated on the reverse sampling inlet system. The system in Figure 1

3.5

3.6

3.7

3.8

3.9

Time (S)

Flgure 3. Pentane peaks for the normal flow sampllng inlet system (A) and the reverse flow sampling system (B).

was used, but the analytical column C was replaced by a series combination of a 10.0-cm piece of 0.1-mm i.d. and a 25-cm piece of 0.25-mm i.d. fused-silica deactivated tubing. The contribution to band broadening from longitudinal diffusion was thus minimized, and other sources of on-column band broadening were nearly eliminated. Samples of 0.09 nL of n-hexane were injected from the cold trap. The sample was cryofocused at a trap temperature of -95 OC and injected at a trap tube temperature of 150 OC. The oven temperature was 90 o c . Injection bandwidths from the cold trap were -6 ms (standard deviation). Note that band broadening due to longitudinal diffusion was insignificant compared to the injection bandwidth. Bandwidths as low as 3 me were achieved with careful adjustment of flow rates and restrictor lengths. It was assumed for this study that the FID did not contribute significantly to band broadening. The location and width of the trapped components should be controlled by the sample gas velocity through the trap tube, the diffusion time to the trap wall, and the accommodation coefficient (stickingprobability). For the low trapping temperatures used in this study, the accommodation coefficient should be close to unity. The diffusion time from the center to the wall of the trap tube is inversely proportional to the diffusion coefficient. For the conditions used in this study root mean square (rms) diffusion times are typically in the range from 0.1 to 1ms. At a sample gas velocity of 100 cmh, the sample would travel an average distance of 0.1-1 mm along the trap axis before reaching the tube wall. Assuming that 10r m s diffusion times are needed for quantitative sample collection, the length of

ANALYTICAL CHEMISTRY, VOL. 65, NO. 18, SEPTEMBER 15, 1993

2519

B E %

-120

-100

-80

-80

-40

a

"

"

"

"

F

"

-20

Trap Temperature ('C) Figure 4. Trapping efficiency plots for +pentane (A), +hexane (B), +heptane (C), and ROctane (D). See text for injection conditions.

the condensed-phasesample plug should be in the range from 1to 10mm. This is consistent with the reinjection bandwidth of -10 ms observed in previous studies at gas velocities of 100 cm/s.m In the present configuration, the pressure in the trap tube is 20 Torr during sample collection. Diffusion coefficients increase linearly with decreasing gas density, but local gas velocities in the trap tube also increase linearly with decreasing density. These two effects should cancel with the result that the length of the condensed sample plug should be nearly independent of the pressure in the trap tube. TrappingEfficiency. In order to determine the trapping efficiency with respect to temperature, samples were cryofocused for 8 s at various temperatures. Peak areas from the chromatogram following subsequent sample injection are plotted vs temperature in Figure 4. Plots A-D are for n-pentane, n-hexane, n-heptane, and n-octane, respectively. Sample injection size was constant at 0.2 p L with a split ratio of 5501. Eachcompound was tested separately. Reinjection temperatures of approximately 90,100,120,and 160 "C were used for the four compounds,respectively. It was noted that significantly lower temperatures resulted in quantitative injection compared with previous cold trap designs. The plots show sharp temperature cutoffs for quantitative trapping. Pentane requires lower temperatures, as expected due to higher volatility, and is quantitatively trapped at about -80 OC. This temperature may be misleading because it was measured at one point near the center of the trap tube. Different components probably are trapped at different locations along the trap. With the large temperature gradient present at the end of the trap tube, the actual trapping temperature could be significantly different from the values reported in Figure 4. The boiling point for pentane at atmospheric pressure is 36 OC and the melting point is -137 OC. The actual pressure during trapping is -2.7 kPa. At this pressure, the reported boiling point is -50 OC.21 This would indicate that pentane is trapped as a liquid on the metal surface. Octane is quantitatively trapped at relatively mild conditions of about -30 "C. The boiling point of octane at 2.7 kPa is -19 OC. Thermal Decomposition. One concern associated with the use of bare metal traps is the possibility of thermal decomposition of the sample during the reinjection process. The use of stationary phase-coated tubes or metal-coated

-

~~~~

-

~

~

(20) Rankin, C. Automated High-speed Repetitive GC Systems for Vapor Analysis. Thesis, University of Michigan, Chemistry Dept., 1991; pp 147-153. (21) Lide,D. R. Handbook of ChemistryandPhysics;ChemicalRubber Publishing Co.: Boca Raton, FL, 1991; pp 6-73477.

l:/40

150

250

350

450

50

Injection Temperature C) Figure 5. Sample recovery vs InJeCtkmtemperature for the normal flow sampling system (a) and the reverse flow sampling system (b): (A)toluene; (8) noctane;(C)pxylene;(D) ~ o n a n e(E) ; P-chkrotoluene; (F) 1,3dichlorobenrene.

fused-silica tubes may decrease sample decomposition, but the possibility of decomposing the stationary phase and the slower heat transfer by using fused-silica inserts are unfavorable trade-offs. Earlier studies with bare metal tubes have shown favorable operation with minimal decomposition for some samples.m@ In a previous study using a back-flush recycle system with a Cu/Ni alloy trap tube, chromatograms showed significant reduction in sample decomposition when the sample was positioned at the downstream end of the trap tube.22 Figure 5 show a quantitative comparison of samplerecovery between the earlier system with the samplecryofocused at the upstream end of the trap (a) and the reverse flow sampling system design with the sample cryofocused at the downstream end (b). Plots A-F show recovery for 0.29-nL injections of a mixture containing toluene, n-octane, p-xylene, n-nonane, 2-chlorotoluene, and l,&dichlorobenzene,respectively. For each peak in the chromatograms, the ratio of the peak area to the total area of all peaks in a given chromatogramis plotted against trap injection temperature. In part a, with the normal flow configuration, injection temperatures less than 150 OC show reduced relative peak areas for the two highest boiling point components,n-nonane and l,&dichlorobenzene. This indicates incomplete vaporization from the trap tube. All other components show larger relative areas because the overall area of the chromatogram isreduced. Nodecompositionwas deted i n (a) for injection temperatures less than 150 OC. For trap temperatures between 150 and 300 OC, all components show relatively constant peak areas. This suggests that, for this temperature range, nearly quantitative injection occurs with minimal decomposition. Although some decomposition was detected at 300 OC, the amount is insignificant. At higher injection temperatures for part a, all components show significant decomposition with the exceptionof toluene.

-

(22) Klemp, M.; Sacks,R. J. HighResolut. Chromatogr. 1991,14,236240.

2520

ANALYTICAL CHEMISTRY, VOL. 65, NO. 18, SEPTEMBER 15, 1993 13 r

10

ob 0

A

C

B

10

01

50

250 350 450 Injection Temperature ('C) Figure 6. Sample recovery vs injection temperature for the normal flow sampling system (a) and the reverse flow sampling system (b): (A) 2-fluorotoluene; (B) 2-chlorotoluene; (C) 2-bromotoluene. 150

The toluene peak increases in size because it is a principal decomposition product of 2-chlorotoluene. At these temperatures, light hydrocarbons, probably methane and ethane, become dominant features in the chromatograms. The data are dramatically different with the reverse flow sampling inlet configuration in part b. Notice that the peak areas for all components are relatively constant over the entire injection temperature range. Since the areas for peaks D and F remain constant even at the lowest injection temperature of 75 OC, there appears to be quantitative injection even a t these low temperatures. As the injection temperatures increase, all components resist decomposition. Even at the highest injection temperature of 450 OC, no decomposition products are detected. Figure 6 gives quantitative information on injection efficiency and degree of decomposition for a more temperature sensitive mixture of halogenated aromatic hydrocarbons. Again, each trace in plots a and b represents the ratio of the peak area of the component of interest to the total area of all peaks in a given chromatogram plotted against the injection temperature. Traces A-C correspond to 2-fluorotoluene, 2-chlorotoluene, and 2-bromotoluene, respectively. Plots a are data obtained with the normal flow configuration, with the sample trapped a t the upstream end of the trap tube, and plots b are data obtained with the reverse flow sampling inlet system, where the sample is trapped a t the downstream end of the tube. About 0.16 nL of the sample mixture was injected in each system. In (a) with the earlier inlet system configuration, injection temperatures of less than 150 O C show incomplete reinjection for the highest boiling point component, 2-bromotoluene.This again results in higher relative peak areas for the other two components because of a decrease in the overall peak area of the chromatogram. Note that trace C does not show any significant range of temperatures for quantitative reinjection without decomposition. Even at the temperatures giving maximum peak area (-175 "C), decomposition products are detected. A t the highest injection temperature, almost no

l

J

2

1

'

4

l

l

6

I

I

B

@

I

10

I

12

I

I

14

1

"

18

l

'

18

'

20

Time (S)

Figure 7. Memory effect study for the normal flow sampling inlet system (a) and the reverse flow sampling system (b). The second component of the large solvent peak at 6 s for case a is the memory effect artifact from the normal flow sampllng system: (A) mctane; (B) pxylene; (C) *xylene; (D) rrnonane.

-

2-bromotoluene survives the injection process. Again for (b), where the sample was cryofocused at the downstream end of the trap tube, there is both quantitative reinjection at the low injection temperatures and resistance to decomposition at the highest injection temperatures. Memory Effects. Earlier studies with cryofocusing inlet system reported a memory effect artifact associated with the cryofocusing of large samples.ll A study was done to determine the extent of any memory effects with the reverse flow sampling inlet system. A variety of solvents with sample size ranging from 0.1 to 50 nL were collected in the cold trap from the Varian injection port. The solvents were then injected onto the column from the cold trap. After a few seconds to allow the cold trap to recool to -75 OC, the power supply was again discharged. Any solvent remaining in the cold trap was then detected with the FID. For all the solvents studied, there were no detectable memory effects from the second firing of the cold trap heater circuit. Since there is a purging flow through the restrictor Rz and inlet PSbetween trap heating pulses, the lack of memory effect with this new inlet configuration indicates that the source of memory effect from the earlier inlet system may have been from cryofocusing residual sample vapor remaining in the Varian 3700 injection port. Notice in Figure ICthat there is also a purge flow out of the system through the injection port for the back-flush mode. Therefore, for back-flush recycle operations, only the sample remaining in the separation column is returned to the trap tube and not any material remaining in the injection port. Figure 7 illustrates this point. Samples of 2.25 nL of isooctane were injected into the system and then back-flushed and refocused after the majority of the solvent was flushed through the detector. The chromatograms shown are the reinjected portions of the recycled components in the mixture. Part a is from the normal flow inlet system, and part b is from the reverse sampling system design using the injection port purge flow. Chromatogram a was obtained on a 4-m-long, 0.25mm-i.d. column with a 0.5-pm stationary-phase film. Chromatogram b used the same conditions except for a stationaryphase film thickness of 0.25 pm. The oven temperature was adjusted to give approximately the same retention time for the solvent. Peaks A-D are spiked components of n-octane, p-xylene, o-xylene,and nonane, respectively,and components X, Y, and Z are unidentified impurity peaks in the isooctane solvent. Note that concentrations of these trace components were different for the two chromatograms.

ANALYTICAL CHEMISTRY, VOL. 65, NO. 18, SEPTEMBER 15, 1993

0

4

8

le

16

20

Time (S) Flgure 8. Chromatogram from an air sample containing +pentane (A), n-hexane (B), benzene (C),mheptane (D),toluene (E), and +octane (F). Cryofocusing was done at -90 "C.

With the earlier system there is a double solvent peak between 4 and 6 s which is not present with the reverse flow sampling inlet system. The first-eluting component of this double peak was from the recycled isooctane that was retrapped at the downstream end of the cold trap tube. The second component was a memory effect from the Varian injection port. It is this memory effect component that limits the utility of the technique for components which elute immediately after the solvent. Direct Air Monitoring. When the system was used as a direct air monitor, the Varian injector was removed. Sampling was done directly through the capillary restrictor Rz from a Tedlar bag. Note that the sample was therefore slightly diluted by the purging carrier gas during the re-inject mode and during the back-flush mode. In arealair monitoring (23) Rankin, C.; Sacks, R. LC-GC 1991,9,42&434.

2521

application there would be no dilution effect. Figure 8 shows a chromatogram of an air sample containing n-pentane (A), n-hexane (B),benzene (C), n-heptane (D), toluene (E), and n-octane (F). Concentrations ranged from 50 to 105 pL/L of dry air. Sample collection and cryofocusing were done at -90 "C. Sample injection from the cold trap tube was accomplished at 190 "C. The separation was accomplished with an 8.0-m length of 0.25-mm-i.d. nonpolar methyl silicone column having a 0.25-pm stationary-phase film. The oven temperature was 40 "C. Some advantages over other air sampling methods include the absence of valves in the sample flow path, the ability to control the amount of air sampled, and the removal of most of the oxygen from the injected sample. The amount of air sampled can be varied by changing the length of time in the sampling mode (Figure la). Changing the restriction of Rz also will change the amount of sample collected. Sampling time can be easily changed with a software adjustment, avoiding alteration of the physical system. Further study is needed to characterize the direct air monitoring configuration. A previous study using the cryofocusing inlet system in the normal flow sampling configuration and a precision vapor generator resulted in detection limits in the low ppb (v/v) range for organic compounds in air samples.23 Most permanent gases are not trapped in bare metal tubes a t the temperatures used in this study. Delaying the return to the normal flow direction after sampling, as shown in Figure IC,will allow the trap and connecting tubes and restrictors to be swept free of air. Thus, no nitrogen or oxygen should be introduced into the analytical column. The removal of oxygen will reduce oxidation problems with sensitive columns at elevated temperatures. RECEIVED for review November 18, 1992. Accepted June 5, 1993.