Anal. Chem. 1996, 68, 2874-2878
A Comparison of Cryofocusing Injectors for Gas Sampling and Analysis in Fast GC Anthony J. Borgerding† and Charles W. Wilkerson, Jr.*
Chemical Science and Technology Division, Los Alamos National Laboratory, Los Alamos, New Mexico 87545
The performance of two cryofocusing injectors for fast gas chromatography has been studied. The first system traps analytes onto bare metal tubes and rapidly vaporizes them upon ballistically heating the tube using a capacitance discharge. The second is a microloop injector in which analytes are cryotrapped onto short lengths of narrow-bore fused silica tubing with various coatings. The ballistically heated injector is capable of sampling and injecting compounds from air faster than the microloop system, because the metal tube can be heated and cooled more rapidly. Both systems are capable of cryotrapping compounds as volatile as butane at -90 °C, and the microloop system can trap ethane when a section of a porous layer open-tubular (PLOT) column is used as the sample loop. In addition, the microloop injector can be used without cryointegration to analyze compounds regardless of their volatility, as long as they are present in the samples at detectable concentrations. Because the ballistically heated injector is flushed prior to injection, it can introduce only compounds that are adsorbed onto its metal trap. Comparison of chromatograms obtained using the two injectors show similar chromatographic resolution. Both traps are susceptible to freezing during the cryotrapping step, but the use of an inline Nafion dryer allows air saturated with water vapor to be sampled using both systems for 3 min without plugging the trap. Thermal decomposition during the injection step can occur for labile species in the ballistically heated trap, but even the highly unstable compound ethyl diazoacetate may be injected without breakdown in the microloop system. The rapid separation of compounds using fast gas chromatography (fast GC) is becoming more popular, and many of these experiments may be routinely accomplished on time scales ranging from a few seconds to 1 min. The theory behind the reduction in analysis times, including dependence on injection band duration, flow rates, and physical parameters of the column, has been studied extensively,1-5 and a variety of fast GC systems, both research and commerical, have been reported. Sacks and co-workers have developed a cryofocusing injector for fast GC † Present address: Department of Chemistry, University of North Dakota, Grand Forks, ND 58202. (1) van Es, A.; Janssen, J.; Bally, R.; Cramers, C.; Rijks, J. J. High Resolut. Chromatogr. 1987, 10, 273-279. (2) Schutjes, C.; Vermeer, E.; Rijks, J.; Cramers, C. J. Chromatogr. 1982, 253, 1-16. (3) Tijssen, R.; van den Hoed, N.; van Kreveld, M. Anal. Chem. 1987, 59, 10071015. (4) Schutjes, C.; Vermeer, E.; Scherpenzeel, G.; Bally, R.; Cramers, C. J. Chromatogr. 1984, 289, 157-162. (5) Desty, D. In Advances in Chromatography; Giddings, J., Keller, R., Eds.; Marcel Dekker: New York, 1965; Vol. 1, pp 199-228.
2874 Analytical Chemistry, Vol. 68, No. 17, September 1, 1996
that they have applied to a variety of analytical problems.6-8 Their system utilizes a ballistic heating step to rapidly vaporize and inject analytes into the column after cryofocusing them on a metal trap. This approach is especially advantageous for the analysis of gases, since it can trap analytes from relatively large sample volumes, thereby increasing the concentration sensitivity of the measurement to allow detection at part per billion by volume (ppbv) levels.9 Volatile compounds from any gas-phase sample should benefit from this approach. Recently, our laboratory introduced an injection system that also provides the advantages of cryofocusing but uses a microvolume loop injection system to introduce the analytes to the column.10 In this paper, we compare these two injection systems with regard to the time required for sampling and injection and the effect of each on chromatographic resolution. In addition, experiments have been performed to test the ability of each injector to handle a range of analyte volatilities, samples with high moisture content, and analytes that are prone to thermal decomposition. EXPERIMENTAL SECTION Ballistically Heated Cryoinjector. A Varian Model 3400 GC (Walnut Creek, CA) was fitted at the factory with a commercial version of the Sacks cryofocusing injector (Cryointegrator L, Chromatofast, Ann Arbor, MI). Prior to the experiments described below, several modifications were made to the instrument. Since the commercial system was built to trap and inject samples introduced via syringe through a standard split injection port, changes were made to the plumbing for gas sampling. No changes were made to the cryogenic cooling system or to the ballistic heater. Figure 1 shows the final configuration, which is based on the system used for gas sampling by Akard and Sacks.9 A manually actuated two-way valve (Valco Instruments, Houston, TX) controls carrier flow, sending it either through the trap or to a vent port. A four-port valve (Valco) controls flow to and from the sampling port and the detector. For some experiments, samples pass through a Nafion dryer (Perma Pure, Toms River, NJ), which is operated with a 25 mL/min counter flow of dry air. An Edwards Model E2M5 vacuum pump is used to pull sample through the trap. A small utility pump continuously moves fresh sample to the four-port valve prior to trapping. Electrically actuated valves (V1 and V2) control the vacuum provided by these pumps. During the sample step (Figure 1, top), the carrier is vented to prevent competitive pumping of this gas. In this way, sample flow rates are increased. When the sampling time is completed, the valve to the vacuum pump (V1) closes, and the (6) Ewels, B.; Sacks, R. Anal. Chem. 1985, 57, 2774-2779. (7) Klemp, M.; Sacks, R. J. Chromatogr. Sci. 1991, 29, 243-247. (8) Klemp, M.; Peters, A.; Sacks, R. Environ. Sci. Technol. 1994, 28, 369A. (9) Akard, M.; Sacks, R. J. Chromatogr. Sci. 1994, 32, 499-505. (10) Borgerding, A.; Wilkerson, C., Jr. Anal. Chem. 1996, 68, 701-707. S0003-2700(96)00187-4 CCC: $12.00
© 1996 American Chemical Society
Figure 2. Diagram of microloop injection system.
Figure 1. Configuration of ballistically heated injector showing flow patterns during sampling, purging, and injection steps.
carrier gas path is switched to direct flow through the trap to purge the sample line (Figure 1, middle). The duration of the purge for most of the experiments is 18 s. After this time, the four-port valve is switched, and carrier gas flows to the column and detector (Figure 1, bottom). A flow stabilization time of 9 s elapses before the trap is ballistically heated to inject the analytes into the column. The valve to the utility vaccum pump (V2) is closed during the sampling and purge steps to prevent water formed in the flame ionization detector from being pulled through the column, but it opens when carrier flow is switched to the column, allowing fresh sample to be taken through the four-port valve in preparation for the next sampling step. Microloop Cryoinjector. The microloop system, which is part of a modified Hewlett Packard 5890 GC (Palo Alto, CA), has been described previously.10 A schematic of the system is shown in Figure 2. A Valco C6WP six-port valve enclosed in a 4 in. × 6 in. valve oven controls sample and carrier flow across the microloop/trap. The microloop consists of a 20 cm length of a SB-Phenyl-5 column (Dionex, Sunnyvale, CA), having an inner diameter of 100 µm and a film thickness of 0.5 µm. Alternatively, a 20 cm section of 320 µm i.d. GasPro GSC porous layer opentubular (PLOT) column was used (Astec, Whippany, NJ). A tee which is supplied with liquid nitrogen from a cryovalve (SGE, Austin, TX) covers a portion of the loop. A thermocouple wire is placed inside the tee along with the sample loop, and temperature is measured using a Fluke Model 52 digital thermometer (Everett, WA). During sampling (valve positioned as shown by the dotted lines in Figure 2), the loop is cryogenically cooled while sample
is pulled through it. The valve is then turned half-way, isolating the loop and stopping all flow. The liquid nitrogen is turned off, and the loop is warmed by the valve heater, vaporizing the trapped analytes. The valve is then turned such that it is positioned as shown by the solid lines in Figure 2, putting the microloop in the carrier stream and injecting its volume into the column. For some experiments, an inline Nafion dryer is placed between the sample source and the injector, as described above for the ballistically heated system. Gas Chromatography. Within each set of comparative experiments, the column, oven conditions, and carrier gas flow rate were held constant. Details are provided in the Results and Discussion section for each measurement. Flame ionization detectors were used for all analyses. Signals for the microloop system were recorded by a Hewlett Packard Model HP3395 integrator. Experiments using the ballistically heated injector were controlled by a personal computer using Varian Star chromatography software (version 4.0). Reagents. Cyclopentane and toluene were obtained from EM Science (Gibbstown, NJ), and hexane, heptane, and octane from JT Baker (Phillipsburgh, NJ). These compounds were reported by the suppliers to be >99% pure. Ethyl diazoacetate was obtained from Aldrich (Milwaukee, WI) with >90% purity. All compounds were used without further purification. Bag samples of these chemicals were prepared by injecting liquid phase compounds into a Tedlar bag with a microliter syringe and diluting with zerograde air. The specifics of this process have been previously described.9,10 The 15 ppmv gaseous mixtures of C1-C6 n-alkanes and C2-C4 alkynes were purchased from Scott Specialty Gases (Plumbsteadville, PA). Analytical Chemistry, Vol. 68, No. 17, September 1, 1996
2875
Figure 4. Replicate injections of a 15 ppmv mixture of acetylene, propyne, 1-butyne, and 2-butyne using the microloop system without cryointegration (no trapping).
Figure 3. Analysis of n-hexane, benzene, n-heptane, toluene, and n-octane using the ballistically heated injection system with purge times of 6, 12, and 18 s.
RESULTS AND DISCUSSION Time Required per Injection. Because the separation of analytes proceeds very quickly, the limiting factor in total analysis time when using fast GC is often sample collection and injection. For each system, maximum sample flow rates are similar10,11 (about 1 mL/min), which infers that the time required to achieve cryofocusing and vaporizing conditions is important. The ballistically heated system is advantageous in this regard, since it is capable of heating the tube to inject sample and then returning to the cold state in as little as 1 s. In contrast, the microloop system requires approximately 1 min to cool and heat during the sampling and injection processes. The microloop system has an advantage, however, in that its configuration separates the sampling line from the GC carrier gas stream. Because of this, purging of the sample line is not required. In the ballistically heated system, the sampling lines needed to be purged for 18 s to completely flush analytes that were not cryotrapped. The effect of purge time on the results obtained using the ballistically heated system was studied by cryotrapping a mixture of n-hexane, benzene, n-heptane, toluene, and n-octane for 30 s at -90 °C. After injection, these components were separated using a 3 m, 250 µm i.d. DB-1 column with a 0.25 µm film thickness. The oven temperature was 30 °C, and the flow rate 3.8 mL/min. Figure 3 shows the results obtained using different purge times. Note that the start time on these chromatograms corresponds to the switching of carrier flow to the column at the end of the purge step and that “firing” of the trap (injection) occurs 9 s after that event. With a 6 s purge, wide peaks appear not only prior to the elution of the analytes that were trapped and injected via the ballistic heating step but also prior to firing the trap. These extraneous peaks correspond to analyte in the sample line between the four-port valve and the cryotrap, implying an incomplete flushing of this zone prior to switching the carrier flow to the column. With a 12 s purge time, most but not all of the untrapped (11) Peters, A.; Klemp, M.; Puig, L.; Rankin, C.; Sacks, R. Analyst 1991, 116, 1313-1320.
2876 Analytical Chemistry, Vol. 68, No. 17, September 1, 1996
analyte has been removed, as shown by the reduced signal of the wide peaks. An 18 s purge time (Figure 3, bottom) removes all traces of untrapped analyte. Notice that the signal for all analytes decreases with increasing purge time. This is perhaps due to trapped analytes vaporizing from partially cooled areas of the trapping tube prior to firing the trap. As the trapping tube is not totally contained within the cryocooled region, a temperature gradient exists along the metal tube adjacent to the cooled zone. Analytes may adsorb to this section of the tube but are more easily removed during the purge step. Octane, since it is the least volatile, will remain trapped farther away from the cooled zone than the other analytes. This is shown in Figure 3 by the fact that the area of the octane peak remains essentially constant for both the 12 and 18 s purge times. In contrast, the more volatile components of the mixture, such as hexane, continue to lose peak area as the purge time is increased. Even with an 18 s purge time, the ballistically heated system takes over 40 s less time for cryofocusing and injection of a sample compared to the microloop injector. However, for samples that can be detected without cryointegration, the microloop system can inject samples as soon as the previous separation is complete, or at any time during the run. Figure 4 shows replicate injections of acetylene, propyne, 1-butyne, and 2-butyne using the microloop system without cryointegration. A 4 m, 250 µm i.d. DB-1 column with a 1.0 µm film thickness was used at an oven temperature of 25 °C and a flow rate of 3.3 mL/min. To maximize the number of separations that can be completed in a given time, injection of the loop contents into the carrier stream occurred before the last compound of the previous separation was eluted. In this way, five analyses were completed in less than 1 min. The reproducibility of the measurements is excellent due to the constant volume of the injection loop (e.g., 2% RSD for 2-butyne). The ballistically heated system in this configuration cannot make this rapid measurement because the sample line needs to be flushed prior to each injection. Under conditions where the gas sample stream is at positive pressure, ballistically heated cryotraps have been configured to inject subsequent samples immediately after the elution of the final component in a run.11 In this configuration, however, samples must be at positive pressure, precluding the direct analysis of ambient air samples. Furthermore, measurements of compounds as volatile as acetylene are not possible using the ballistically heated system in any configuration because acetylene is not adsorbed onto the metal trap. Thus, while the ballistically heated injector allows more analyses in a given time
Figure 5. Analysis of C1-C6 alkanes trapped at -90 °C. (a) Ballistically heated injector with metal trap. (b) Microloop injector with liquid stationary phase-coated fused silica trap. (c) Microloop injector with solid stationary phase-coated fused silica trap.
than the microloop system in most cases, the versatility of the microloop system makes it advantageous for many samples, particularly those containing very volatile compounds at relatively high concentrations. Trapping Characteristics. An important characteristic of a cryofocusing injector is its ability to effectively trap very volatile compounds. This is of special significance in fast GC if the technique is to be used for analyzing trace gases from atmospheric samples for air pollution research.12 To date, most work using cryofocusing injectors has been done on compounds with boiling points greater than or approximately equal to that of pentane (35 °C). We have previously shown that trapping onto tubes with organic coatings allows C5-C8 compounds to be trapped at relatively higher temperatures compared to trapping onto bare silica tubes.10 We have hypothesized that such coatings might also allow trapping of more volatile compounds. To study this experimentally, a 10 ppm mixture of C1-C5 n-alkanes was trapped at -90 °C for 6 s using each system and injected onto a 25 m × 320 µm i.d. DB-624 column with a film thickness of 1.8 µm. The oven temperature was 45 °C, and the flow rate at the column outlet was 5.2 mL/min. The chromatographic results are shown in Figure 5 for the bare metal trap of the ballistically heated system (Figure 5a) and the microloop system using two different traps: a 100 µm i.d. tube coated with a 0.5 µm liquid film of 5% phenyl, 95% methyl polysiloxane material (SB-Phenyl-5 column, Figure 5b), and a 320 µm i.d. tube with a solid (PLOT) stationary phase coating (GasPro column, Figure 5c). In all cases, butane and pentane are completely sampled, as evidenced by their signals not rising as compared to signals from experiments done at lower sampling temperatures (data not shown). Contrary to expectations, only a small enhancement of propane signal was observed when the nonpolar liquid film-coated tubes were used compared to the bare metal trap. However, when the tube with the solid stationary phase was used (Figure 5c), ethane and propane were (12) Lai, J.; Matisova, E.; He, D.; Singer, E.; Niki, H. J. Chromatogr. 1993, 643, 77-90.
completely sampled. The ability to trap compounds as volatile as ethane suggests that the microloop fast GC injector with solid loop coatings should be applicable for studies of non-methane hydrocarbons in atmospheric samples. Recent advances using a ballistic heating system in a new configuration by Gorsuch and Klemp have also shown the ability to trap compounds as volatile as ethane.13 The peak widths, and therefore the chromatographic resolution, of the separations using the ballistically heated system and the microloop system with the 100 µm i.d. loop (Figure 5a and b) are similar. The estimated baseline band duration of the injection impulse using this microloop system, based on an empirical model and previous measurements,10 is 20 ms. While reported values for the injection band duration in ballistically heated systems are somewhat less,6 the difference is negligible in view of the other contributions to peak broadening in this separation. The predicted injection band duration of the microloop system using the 320 µm i.d. loop is over 100 ms, and, accordingly, the peaks in the separation using that system (Figure 5c) are broader than those in the other separations. Another difference in the chromatograms is that signals from methane and ethane appear when using the microloop systems, while no such peaks appear when using the ballistic heating system. As discussed earlier, this is because the microloop sampling system does not require a flushing step. Therefore, unretained sample in the gas phase remains in the trapping tube with the analytes that have been cryotrapped. In the ballistically heated system, nonretained analytes are swept out of the trap and the sampling lines during the purge step. This result is significant, since it again shows that the microloop system can be used for fast GC analysis of even those compounds that are too volatile to be trapped, provided they are at a detectable concentration. The ballistically heated system cannot be used in this manner. Susceptibility to Freezing during Sampling. Injection systems that utilize cryofocusing often have problems with plugging that result from moisture in the sample freezing inside the injector.14-16 For atmospheric gases and moist samples from chemical process streams, freezing of the sampling lines is of particular concern. To study the effect of moisture on the trapping capabilities of each system, an approximately 1 mL/min flow of air that had been humidified by bubbling it through a water reservoir was passed through each trap, both of which were cooled to -90 °C. Plots of the flow rate through the injector versus time are shown in Figure 6. Due to its larger diameter (300 vs 100 µm), the ballistically heated injector remained unfrozen for a longer period than the microloop system, but both systems failed to sustain a moist air sample flow for more than 4 s. By eliminating the moisture from the sample, this problem can be mitigated. We have employed inline Nafion drying tubes to effectively desiccate the sample flows. The device consists of an inner Nafion membrane tube through which the sample passes, while a counter flow of dry air flows through an outer tube. Moisture in the sample stream diffuses through the membrane and reaches equilibrium with dry counter flow. Use of these dryers effectively prevented plugging of the traps and allowed (13) Gorsuch, A.; Klemp, M. Presented at Pittcon ‘96, Chicago, IL, March, 1996; Paper 107. (14) Pankow, J. Environ. Sci. Technol. 1991, 25, 123-126. (15) Burford, M.; Hawthorne, S.; Miller, D. J. Chromatogr. A 1994, 685, 95111. (16) Lepine, L.; Archambault, J. Anal. Chem. 1992, 64, 810-815.
Analytical Chemistry, Vol. 68, No. 17, September 1, 1996
2877
Figure 6. Flow of water-saturated air through each cryotrapping injector versus time.
sampling of humidified gas samples for 3 min or more, as shown in Figure 6. Because the 100 µm i.d. microloop has a higher resistance to flow than the 300 µm i.d. trap in the ballistically heated system, a higher pressure is required in order to obtain a flow similar to that in the ballistically heated system. The loop restriction may also contribute to the relatively unstable flow recorded for the microloop system, which may lead to greater uncertainty in the sample volumes collected, particularly for short sampling times. Nafion dryers can remove some very polar organic molecules such as alcohols and amines from samples,14 but most relatively nonpolar volatile organic compounds are unaffected by this drying process.17 Analysis of a mixture containing cyclopentane, nhexane, benzene, n-heptane, toluene, and n-octane in air using each system with and without the dryer on-line gave similar results, indicating that these compounds were unaffected by the drying process (data not shown). Decomposition during Injection. The ballistic heating system has previously been shown to decompose thermally labile compounds upon injection.18 Design changes, such as lowering injection temperature and cryofocusing analytes on the side of the trap nearest the column to minimize the exposure to potentially catalytic metals during the heating stage, have reduced this problem.9,19 The ballistic trap heating nevertheless is likely to cause degradation of very sensitive compounds. The heating step in the microloop injection system is much milder, with slower heating and usually a lower ultimate temperature, making it less susceptible to problems with degradation. Figure 7 shows the analysis of ethyl diazoacetate using both systems. This compound is very sensitive to heating, and two degradation products form when the ballistic heating system is used, as shown in Figure 7 (bottom). While the composition of these products has not been conclusively determined, GC/MS analysis of ethyl diazoacetate using standard split injection at 200 °C showed the most prominent decomposition product to be loss of N2 to form ethyl acetate. Ethyl diazoacetate also undergoes catalytic reactions in the presence (17) Janicki, W.; Wolska, L.; Wardencki, W.; Namiesnik, J. J. Chromatogr. A 1993, 654, 279-285. (18) Klemp, M.; Sacks, R. J. High Resolut. Chromatogr. 1991, 14, 235-240. (19) Klemp, M.; Akard, M.; Sacks, R. Anal. Chem. 1993, 65, 2516-2521. (20) Doyle, M. P. Acc. Chem. Res. 1986, 19, 348-363.
2878 Analytical Chemistry, Vol. 68, No. 17, September 1, 1996
Figure 7. Analysis of ethyldiazoacetate using microloop injection system (top) and ballistically heated injection system (bottom). Conditions: 3 m, 0.25 mm i.d. DB-5 column, 30 °C oven temperature, 3 mL/min flow rate.
of transition metals.20 Catalytic decomposition may therefore be occurring on the metal trap, which is a Cu-Ni alloy, despite attempts to minimize exposure to these metals, as indicated earlier. Using the microloop system, and reinjecting at 40 °C, no indication of decomposition was seen (Figure 7, top). This indicates that, for thermally labile compounds that decompose during vaporization in the ballistically heated injector, using a microloop injector may be advantageous. CONCLUSIONS For most samples, the ballistically heated cryotrapping injector may be more beneficial for sampling and injection of compounds from air compared to the cryotrapping microloop injector. The primary advantage of the ballistically heated injector is its ability to rapidly cycle between cold and hot states, whereas a significantly longer time is required for these activities using the microloop system. Therefore, more samples can be run in a given time using the ballistically heated system. However, the microloop system has benefits for samples that are problematic for the ballistically heated system. For example, very volatile compounds can be detected using the microloop injection system, due to the enhanced trapping capabilities of coated sample loops, and because untrapped compounds are not purged from the sample inlet. In addition, the gentle vaporization step in the microloop system permits the analysis of very thermally labile compounds, which decompose when analyzed using the ballistically heated system. ACKNOWLEDGMENT This work was supported by the Department of Energy Office of Environmental Management through the DOE Methods Compendium Program. Received for review February 27, 1996. Accepted June 6, 1996.X AC9601876 X
Abstract published in Advance ACS Abstracts, July 15, 1996.
