Helical Sorbent Microtrap for Continuous Sampling by a Membrane

and Ecole des Mines d'Albi, Campus Jarlard, F 81013 Albi Cedex 09, France ... trap interface for on-line gas chromatographic monitoring of organic...
1 downloads 0 Views 101KB Size
Anal. Chem. 2003, 75, 736-741

Helical Sorbent Microtrap for Continuous Sampling by a Membrane and Trap Interface for On-Line Gas Chromatographic Monitoring of Volatile Organic Compounds Ionel Ciucanu,*,† Adrian Caprita,‡ Adrian Chiriac,† and Radu Barna§

Department of Chemistry, West University of Timisoara, Strada Pestalozzi 16, RO-1900 Timisoara, Romania, Department of Organic Chemistry, University of Agriculture Sciences, Calea Aradului 118, RO-1900 Timisoara, Romania, and Ecole des Mines d’Albi, Campus Jarlard, F 81013 Albi Cedex 09, France

A helical sorbent microtrap consisting of a helical sorbent fixed inside a silicosteel capillary tube is presented. The main parameters that affect the safe sampling time of the helical sorbent microtrap in continuous sampling by a membrane and trap interface for on-line gas chromatographic monitoring of organic volatiles in gaseous samples are examined, taking into account the helical configuration of the sorbent, the presence of the membrane in system, and the properties of the analytes. Thermal desorption of analytes from the helical sorbent trap was also examined having regard to the influence of the turbulent flow generated by the helical sorbent in the heat transfer and the effect of thermal backward flow on the peak shape. The practical application of the helical sorbent microtrap in a membrane and trap interface was demonstrated by on-line gas chromatographic monitoring of four volatile organic compounds in the fume hood air and of volatile organic compounds from a diesel engine exhaust. The limit of detection was in the picogram per milliliter range, depending on the time of trapping and the parameters that affect the permeation through the membrane. The concentration of volatile organic compounds (VOCs) in air can fluctuate in both time and space. Portable devices for continuous monitoring can almost immediately detect these fluctuations, and a quick decision can be taken to keep the emission sources under control. Much of the impetus for air and water analysis comes from the practice of industrial or occupational hygiene. Open path Fourier transform infrared spectrometers,1 portable mass spectrometers,2,3 and portable gas chromatographs4,5 are * Corresponding author: [email protected]. † West University of Timisoara. ‡ University of Agriculture Sciences. § Ecole des Mines d’Albi. (1) Xiao, H. K.; Levine, S. P.; Herget, W. F.; Darcy, J. B.; Spear, R.; Pritchett, T. Am. Ind Hyg. Assoc. J. 1991, 52, 449-457. (2) Hart, K. J.; Wise, M. B.; Griest, W. H.; Lammert, S. A. Field Anal. Chem. Technol. 2000, 4, 93-110. (3) Syage, J. A.; Nies, B. J.; Evans, M. D.; Hanold, K. A. J. Am. Soc. Mass Spectrom. 2001, 12, 648-655. (4) Tuan, H. P.; Janssen, H. G.; Cramers, C. A.; Mussche, P.; Lips, J.; Wilson, N.; Handley, A. J. Chromatogr., A 1997, 791, 187-195.

736 Analytical Chemistry, Vol. 75, No. 4, February 15, 2003

used in near real-time monitoring of VOCs in the field. Progress in gas chromatography (GC) makes it an excellent technique for near real-time monitoring of VOCs from air and water since the analytes are separated prior to their detection, allowing a flexible qualitative and quantitative analysis of these complex mixtures. Since the VOCs have a low concentration in air and water, direct injection of large volumes of sample in GC6,7 causes excessive band broadening with losses of chromatographic resolution. Moreover, a large amount of water or air injected in the chromatographic column degrades the stationary phase. Hence, diluted environmental samples need to be preconcentrated before injection into GC a column. Many preconcentration methods for VOCs have been tried in order to trap the analytes in a suitable medium with subsequent elution and injection in GC. Usually, the preconcentration was done either by cryogenic trapping8-15 with or without a packing material or by ambient trapping using a sorbent material.16-24 Since cryogenic trapping gives bulky and expensive equipment, and troubles by trapping water from sample, concentration on sorbent material such as black carbon derivatives, porous polymers, silicon polymers, and molecular sieves has the highest applicability. However, packed traps also have serious disadvantages due to incomplete sample recovery, high (5) Ciucanu, I.; Chiriac, A. J. Sep. Sci. 2002, 25, 447-452. (6) Fujii, T. J. Chromatogr. 1977, 139, 297-302. (7) Teply, J.; Dressler, M. J.J. Chromatogr. 1980, 191, 211-223. (8) Hopkins, B. J.; Pretorius, V. Anal. Chem. 1978, 158, 465-469. (9) Graydon, J. W.; Grob, K. J. Chromatogr. 1983, 254, 265-269. (10) Werkhoff, P.; Bretschneider, W. J. Chromatogr. 1987, 405, 87-98 (11) Ju ¨ ttner, F. J. Chromatogr. 1988, 442, 157-163. (12) Hagman, A.; Jacobsson, S. J. Chromatogr. 1988, 448, 117-126. (13) Springston, S. R. J. Chromatogr. 1990, 517, 67-75. (14) Helming, D.; Greenberg, J. P. J. Chromatogr., A 1994, 677, 123-132. (15) De la Calle-Guntinas, M. B.; Ceulemans, M.; Witte, C.; Lobinski, R.; Adams, F. C. Microchim. Acta 1995, 120, 73-82. (16) Nunez, A. J.; Gonzalez, L. F.; Janak, J. J. Chromatogr. 1984, 300, 127-162. (17) Kaiser, R. E.; Rieder, R. J. Chromatogr. 1989. 477, 49-52. (18) Burger, B. V.; Le Roux, M.; Murno, Z. M.; Wilken, M. E. J. Chromatogr. 1991, 552, 137-151. (19) Krieger, M. S.; Hitd, R. A. Environ. Sci. Technol. 1992, 26, 1551-1555. (20) Mitra. S.; Yun, C. J. Chromatogr. 1993, 648, 415-421. (21) Thomas, C. L. P. LC-GC Int. 1993, 6, 8-12. (22) Helmig, D.; Vierling, L. Anal. Chem. 1995, 67, 4380-4386. (23) Tuan, P. H.; Janssen, H. G.; Cramers, C. A. J. Chromatogr., A 1997, 791, 177-185. (24) Baltussen, E.; David, F.; Sandra, P.; Janssen, H. G.; Cramers, C. Anal. Chem. 1999, 71, 5193-5198. 10.1021/ac026238i CCC: $25.00

© 2003 American Chemical Society Published on Web 01/22/2003

background contamination, and formation of artifacts. 22,25,26 To avoid some of these disadvantages, capillary columns coated with polysiloxane chromatographic stationary phases were used. The low capacity of the capillary column to retain analytes was improved using cold traps,13,27,28 multichange traps,19 ultra-thickfilm traps,18,29 and long traps.30 The trap has two important functions. First, the trap retains the analytes for their concentration. The sorbent can retain a little amount of analytes for a limited time, and then breakthrough occurs because the analytes are eluted by the carrier gas or the sorbent material is overloaded by the analytes. The appearance of the sampled molecules in the outlet stream is characterized by the breakthrough volume31,32 and the breakthrough time,25,32,34 which were roughly approximated from the GC retention parameters by the injecting of analyte into the trap. However, determination of the breakthrough volume is unsuitable for mixtures containing competing or high concentrations of analytes. Second, the analytes retained on the sorbent trap are desorbed for introduction into a GC capillary column. The desorption of analytes from the sorbent material could be performed by solvent extraction or by thermal desorption. The disadvantages of solvent extraction such as artifact reactions, loss of analytes, and broadening of solvent peaks were overcome by thermal desorption. The thermal desorption method has a long history,26 and can be performed off-line35 or on-line.5,23,28,36,37 In continuous on-line sampling,5,36,37 the analytes are continuously collected and enriched by a membrane and a trap interface (MTI) and transferred directly by thermal desorption from the trap into the GC column. The errors associated with sample collection, transport, storage, and laboratory are avoided and an automatic and sensitive system for monitoring of analytes is provided. Recently,38 a helical sorbent for solid-phase extraction has been introduced for collecting and concentrating of analytes prior to GC analysis. This paper describes the helical sorbent trap of a portable device used for continuous sampling by a membrane and trap interface and on-line gas chromatographic monitoring of volatile organic compounds from air. The main parameters that affect safe sampling time of the sorbent and the thermal desorption of analytes were examined. EXPERIMENTAL SECTION Reagents and Materials. Analytical grade benzene, toluene, trichloroethylene, and tetrachloroethylene were purchased from (25) Vahdat, N.; Sweaarengen, P. M.; Jhonson, J. S.; Priante, S.; Mathews, K.; Neidhardt, A. Am. Ind. Hyg. Assoc. J. 1995, 56, 32-38. (26) Harper, M. J. Chromatogr., A 2000, 885, 129-151. (27) Grob, K.; Habich, A. J. Chromatogr. 1985, 321, 45-58. (28) Lai, J. Y. K.; Matisova, E.; He, D.; Singer, E.; Niki, H. J. Chromatogr. 1993, 643, 77-90. (29) Roeraade, J.; Blomberg, S. J. High Resolut. Chromatogr. 1989, 12, 138141. (30) Bicchi, C.; D’Amato, A.; David, F.; Sandra, P. J. High Resolut. Chromatogr. 1989, 12, 316-321. (31) Raymond , A.; Guiochon, G. J. Chromatogr. Sci. 1975, 13, 173-177. (32) Brown, R. H.; Purnell, C. J. J. Chromatogr. 1979, 178, 79-90. (33) Wood, G. O.; Moyer, E. S. Am. Ind. Hyg. Assoc. J. 1989, 50, 400-407. (34) Yoon, Y. H.; Nelson, J. H. Am. Ind. Hyg. Assoc. J. 1984, 45, 509-516. (35) Matz, G.; Kibelka, G.; Dahl, J.; Lennemann, F. J. Chromatogr., A 1999, 830, 365-376. (36) Luo, Y. Z.; Yang, M. J. J. Pawliszyn, J. High Resolut. Chromatogr. 1995, 18, 727-731. (37) Mitra, S.; Zhu, N.; Zhang, X.; Kebekus, B. J. Chromatogr., A 1996, 736, 165-173. (38) Ciucanu, I. Anal. Chem. 2002, 74, 5501-5506.

Figure 1. Schematic diagram of the MTI-GC experimental system: (1) hydrogen cylinder: (2) nitrogen cylinder: (3) fan; (4) injector porthole; (5) sampling chamber; (6) membrane module; (7) microtrap; (8) gas chromatograph; (9) computer; (10) printer; (11) capacitive discharge power supply; (12) permeation cell; (13) permeation tube; (14) aluminum block.

Merck (Darmstadt, Germany). Hydrogen, nitrogen, and compressed air were from Linde (Timisoara, Romania). The poly(dimethylsiloxane) (PDMS) sorbent used for the helical trap was a commercially available SE-54 stationary phase (Alltex, Hoogeveen, The Netherlands). Teflon tubes used as transfer lines for carrier gas were of 0.25mm i.d. and 1.5-mm o.d. and were purchesed from Western Analytical Products (Murrieta, CA). Fused-silica capillaries (0.2mm i.d., 0.35-mm o.d..) with a nonpolar deactivation coating used as transfer lines for analytes were from Supelco (Bellefonte, PA). All carrier gas connections were made using stainless steel Valco zero dead-volume fittings (Alltex GmbH, Unterhaching, Germany) and graphite-Vespel ferrules (Supelco). Silicosteel (stainless steel deactivated with silica) tubes for traps were from Restek (Bellefonte, PA). The aqueous samples were prepared by injecting different volumes of standard solutions into 50 mL of water. Three standard solutions were prepared by adding 1, 10, and 100 µL of analyte, respectively, in 10 mL of methanol. The aqueous samples were mixed with a magnetic stirrer to generate quickly homogeneous solutions. The concentration of analytes was in the 10-6-10-9 mol/L range and for the headspace was calculated taking into account Henry’s law and dimensionless Henry’s constants.39,40 Apparatus and Methods. A schematic diagram of the MTIGC system5,41 used in this study is presented in Figure 1.The main components of this system are the membrane module, the helical sorbent trap, and the gas chromatograph. The VOCs were continuously permeated through the membrane. The carrier gas passed through the membrane module and carried the permeated analytes into the sorbent trap. The analytes were concentrated by trapping at room temperature. After trapping, the analytes were desorbed at fixed intervals of time by pulse flash electrical heating as a concentrated band of sample and were injected into a GC column for separation. The appearance of analytes in the outlet stream of the trap was monitored with a flame ionization detector which was calibrated by direct liquid injection. The trap was connected to the chromatographic column through a deactivated transfer line. The membrane was a flat sheet of PDMS or bisphenol A polycarbonate-poly(dimethylsiloxane) (BPAPC-PDMS) from (39) Gossett, J. M. Environ. Sci. Technol. 1987, 21, 203-208. (40) Ashworth, R. A.; Howe, G. B.; Mullins, M. E.; Rogers, T. N. J. Hazard. Mater. 1988, 18, 25-36. (41) Ciucanu, I.; Chiriac, A.; Captita, A. J. Chromatogr., A 2002, 964, 1-9.

Analytical Chemistry, Vol. 75, No. 4, February 15, 2003

737

Figure 2. Schematic diagram of the helical sorbent microtrap.

Specialty Silicone Products Inc. (Ballston Spa, NY). It had a thickness of 0.025 mm and an outlet surface in membrane module of 163 mm2. Since the flat membrane was not self-supported, it was supported in a membrane module.5,41 The pressure of carrier gas lifted the membrane and allowed it to carry the permeated analytes to the trap. The membrane module was set in the middle of the sampling chamber, which was a stainless steel canister or the headspace of a sample vial. The sampling chamber was equipped with a fan inside that agitated the air outside of the membrane surface. The vial was sealed with a silicon rubber, and the solution of the analytes was stirred with a magnetic stirring bar at 1200 rpm. The sample was injected with a syringe or was delivered from a standard gas generator.41 The tests of the sorbent trap without the membrane were performed using a six-way injection valve, Rheodyne model 58826 (Supelco), to connect the trap to the sampling chamber. The carrier gas was directly introduced into the sampling chamber for experiments without membrane. The sampling chamber was thermostated, and the temperature in the membrane module was maintained at 24 °C for all experiments. The trap was periodically flash heated by passing electrical current directly through the wall of the metal tube of the trap by discharging a capacitor from time to time. The cycle of trapping and heating was repeated automatically from a few seconds to hours using a timer. The temperatures during trapping and thermal desorption were measured with fine J-K thermocouple wires (Omega, Engineering, Inc.) connected to an amplifier and a recorder. The tests of sorption in the trap at lower temperature were performed with a 2 × 2 cm one-stage Peltier cooler (Melcor Materials Electronic Products Corp., Trenton, NJ). The hot surface of the cooler was glued to an aluminum heat sink to facilitate heat release with the assistance of a fan. A dc variable power supply manufactured in the laboratory was used to modify the temperature of the cooler. The cool surface and the trap tubing were isolated with glass wool. All the analyses were performed using an HP 6890 gas chromatograph (Hewlett-Packard, Avondale, PA) equipped with a flame ionization detector. The temperature in the oven was 60 °C. The carrier gas was hydrogen at 5 mL/min. GC separations was performed on a MXT-1 silicosteel capillary column (3 m × 0.32 mm i.d., 3-µm PDMS thickness film) from Restek. A slight modification of the gas chromatographic injector was required for this experiment. The carrier gas line from the mass flow controller was disconnected from the injector and was reconnected to the gas inlet of the membrane module using a piece of 30 cm Teflon tubing. Construction of the Traps. Figure 2 shows a schematic diagram of a helical sorbent fixed inside of a silicosteel tube (80 mm long, 0.75 mm i.d. × 0.95 mm o.d.).The preparation of the 738

Analytical Chemistry, Vol. 75, No. 4, February 15, 2003

helical sorbent has already been presented.38,42 The helical support for the microtrap was made by wrapping with a constant pitch a 0.07-mm-diameter chromium-aluminum alloy wire (AB Kanthal, Hallstahammer, Sweden) around a 1.0-mm-diameter straight wire. After cleaning with water, acetone, and chloroform, the surface of the helical support was coated with several layers of PDMS by immersion of the helical support in a solution of 75 and 50% PDMS in dichloromethane. PDMS cross-linking was accomplished by adding 0.8% dicumyl peroxide to the coating solution. After each coating, the helical support was heated in a GC oven for 10 min and then was weighed. A 260-mm length of wire wrapped in helical configuration was uniformly coated with ∼19.6-mm3 PDMS with a surface area of 441.08 mm2. The thickness of the final film of PDMS on the helical sorbent was ∼0.05 mm and depended on the concentration of the PDMS solution and the number of coatings. The helical sorbent was introduced into a silicosteel tube and fixed inside at both ends with two rings. The helical sorbent trap was connected to a GC injector with a nitrogen flow of 3 mL/ min, heated from 150 to 200 °C at 0.3 °C/min for final crosslinking, and then heated to 270 °C for 24 h for conditioning. At both ends, the trap tube was connected in circuit with two Valco zero dead-volume fitting. The bore of the fitting was 0.25 mm to the end of the trap connected to the carrier gas and 0.75 mm to the end between the trap and capillary column. A thick-film (0.05 mm) silicone rubber capillary trap was carried out by the dynamic coating of a polysiloxane solution (20% SE-30 in dichloromethane with 0.8% dicumyl peroxide18,43) in a silicosteel capillary tube (0.75 mm i.d. × 0.95 mm o.d.). The polymer was cross-linked using the above technique. The final thickness was obtained by performing three consecutive coatings. The length of the thick-film trap was 85 mm. RESULTS AND DISCUSSION Safe Sampling Time of the Helical Sorbent Trap. The safe sampling time was limited in continuous on-line sampling by the penetration of the sorbent trap with analytes, which was evaluated for air sampling by breakthrough time.25,33,34,44 Numerous equations have been proposed for predicting the breakthrough time (tb) of a sorbent trap. Most of them33,34 were developed by starting from the mass balance between the gas entering into the sorbent trap and the sum of the gas retained by trap and the gas penetrating through the trap; breakthrough time is expressed as

tb )

[

(

)]

C0 - C Wc FQ Ws ln C0Q Kv C

(1)

where Ws is the weight of sorbent (g), Wc is the sorption capacity (mg/g), Q is the volumetric flow (mL/min), F is the density (g/ mL), Kv is the sorption rate constant min-1), Co is the inlet concentration (mg/mL), and C is the breakthrough concentration mg/mL). Practically, the breakthrough time was measured from the time when the analyte was introduced into the trap until the time that 1% of the inlet concentration was observed in the outlet (42) Ciucanu, I. Patent Application PCT/RO 00209, 2001. (43) Blomberg, S. Roeraade, J. J. High Resolut. Chromatogr. 1990, 13, 509512. (44) Mitra, S.; Hu, Y. H.; Chen, W.; Lai, A. J. Chromatogr., A 1996, 727, 111118.

Figure 3. Influence of the carrier gas flow rate on breakthrough time for benzene (b), toluene (9), trichloroethylene (2), and tetrachloroethylene (O). The concentration of each analyte was 1.5 × 10-6 mol/L in the headspace; the carrier gas was hydrogen.

stream,25 e.g., in our experiments. until the detector signal reached the limit of detection (signal-to-noise ratio equal to 3). As expected from eq 1, Figure 3 shows that the breakthrough time is inversely proportional to the flow rate. It can be seen that the breakthrough time has the highest values at low flow rates if other conditions remain the same. Between 2 and 5 mL/min, the breakthrough time decreases very fast for the following reasons. First, the amount of analyte sampled by the membrane interface was increased significantly in this range of flow rates.41 Second, the amount of analyte that reached the sorbent trap in a given time increased at high flow rates. Third, an increase in the flow rate generated an increase in the velocity of the carrier gas and an increase of the energy transferred to the adsorbed molecules. Consequently, the number of analyte molecules stripped from the surface of the sorbent will be higher and breakthrough time lower. The trapping test with thick-film capillary trap resulted in a similar variation of the breakthrough time. However, the breakthrough times in the thick-film capillary trap were ∼10 times shorter than in the helical trap. The amount of analyte retained by the traps for the same sorbent and in the same experimental conditions depends only on the amount of sorbent.38 The amount and the thickness of the PDMS sorbent were approximately the same for both traps. Only the trapping surface is double for the helical sorbent. The only explanation of this difference in breakthrough time could be found in the improvement of the mass transfer of analytes in the boundary layer of the helical sorbent. The boundary layer is practically a static layer generated as a result of the friction between the flowing sample fluid and the surface of the sorbent and where the mass transport is performed by molecular diffusion. With a helical sorbent, the flow had a rotational motion around the helix,45 and as a function of the velocity of the fluid and the pitch of the helix,46 turbulence was generated on the surface of the sorbent. In a transition and turbulent flow, the thickness of the boundary layer was reduced, which improved the mass transfer of the analytes from the carrier gas to the sorbent and increased the sorption rate of the analytes by sorbent. In a thick-film capillary, the flow was laminar and the flow path lines were parallel with the wall of the tube, generating (45) Grambill, W. R.; Bundy, R. D.; Wasbrough, R. W. Chem. Eng. Prog. Symp. Ser. 1961, 57, 127-137. (46) Kreith, F.; Margolis, D. Appl. Sci. Res. Sect. A 1959, 8, 457-473.

a thick boundary layer. The breakthrough time at room temperature with a thick-film trap was not due to overloading or saturation of the trap but to the low diffusion of the analytes from the carrier gas to the sorbent in laminar flow. In this case, the analytes could go through the short capillary traps without interacting with the sorbent film. It can be seen from Figure 3 that the breakthrough time depended on the physical properties of the analytes. Benzene, with low molecular weight, volatility, and boiling point, had the lowest breakthrough time. Also very important was the affinity of the analyte for a certain sorbent. A mixture of sorbents could be used to enhance the breakthrough time for a larger range of analytes. The humidity did not affect the breakthrough time of the tested analytes on PDMS because this sorbent material and the analytes are hydrophobic. Anyhow, the presence of water in the system was limited by its low permeation through the silicone membrane. According to eq 1, the effect of the analyte concentration on the breakthrough time is more complicated. Practically, however, for progressively increased analyte concentration, the breakthrough time decreased, following a trend such as shown in Figure 3. The test was performed with and without membrane in the system, using the standard gas generator.41 Without the inlet membrane, the breakthrough time curve was diminished by a factor that depended on the permeability of the analyte in the membrane. For example, for benzene without the inlet membrane, the breakthrough time was 7 times lower than with the membrane. The effect of the variations in ambient temperature on the trapping process was tested between 10 and 30 °C. A linear variation of the breakthrough time (tb) with temperature (T) was noticed for each analyte (for benzene: tb ) - 4.54 T + 184.51; R2 ) 0.9984). A variation of the trap temperature by 5 °C around room temperature (22 °C) modified the breakthrough time by ∼25%. Thermal Desorption of the Analytes from the Helical Sorbent Trap. The thermal desorption of the analytes from the helical sorbent is the reverse process of sorption. The desorption was performed by the hot carrier gas, which was heated electrically through the wall of the silicosteel capillary tube. The heat was transferred to the sorbent by forced convection,47 because there was no direct contact between the sorbent and the wall. The maximum temperature of the wall was limited by the thermal stability of the analytes and of the sorbent. The desorption process must be extremely rapid in order to avoid excessive band broadening of the analytes before GC injection. The thermal desorption process can be enhanced by a fast heating and by improving the heat transfer from the carrier gas to the sorbent and the mass transfer of analytes in the boundary and sorbent layers. The diffusion of analytes in the sorbent layer was faster when the temperature of the sorbent was higher because the diffusion coefficients in PDMS increased with an increase in temperature.38 The rate-controlling step of this process was the transfer through the boundary layer between the hot carrier gas and the sorbent. The turbulent rotational flow on the surface of the sorbent, as the result of the helical configuration of the sorbent, generated a thinner boundary layer between sorbent and gaseous sample and a faster heat and mass transfer. In the thick(47) Bejan, A. Heat Transfer; John Wiley & Sons: New York, 1993.

Analytical Chemistry, Vol. 75, No. 4, February 15, 2003

739

Figure 4. Profile of the electric potential applied to the ends of the trap. (A) one capacitor; (B) three capacitors.

Figure 5. Influence of voltage applied to the ends of the trap on the area of the desorbed benzene peak. The carrier gas flow rate was 5 mL/min; the trapping time was 8 s.

film capillary trap, the flow was laminar and the heat and mass transfer was slow in this boundary layer, generating broad peaks. The trap was flash heated by applying a voltage at the ends of the metal tube of the trap by discharging an electric capacitor for a few milliseconds. Figure 4 shows the time of discharge of the capacitor and the profile of the electrical potential at the ends of the trap. Profile A was obtained by applying the electric potential of one capacitor of 17 mF. In this case, the potential is very high in the first milliseconds and then decreases exponentially. The profile of the electrical current was similar to that of the electrical potential. Profile B was obtained by applying the potential of three capacitors of 5 mF, one by one, at very short intervals with an electronic scheme48 consisting of three capacitors and two inductors arranged in series. Profile A resulted in a high temperature in the first milliseconds and then the temperature decreased similar to the electrical potential. For the same amount of transmitted heat, profile B resulted in a lower temperature than profile A. The heating time was longer, and thermal decomposition was avoided. Since the trap was heated only for milliseconds, it was very difficult to accurately measure the exact heating rate. A measurement using a thermocouple showed that the temperature in profile A, at the moment of discharge, was ∼260 °C for 1-2 ms. In profile B, the temperature of the wall of the trap was 200 °C and was maintained constant for ∼10 ms. In these conditions, the thermal desorption blanks showed no peak. Figure 5 shows that, by increasing the voltage on the trap, the area of the desorbed benzene peak increased until it reached a steady value. A higher voltage resulted only in the degradation of sorbent polymer. At the highest optimal voltage, for 1-min trapping, and by bypassing the membrane with a six-way valve, the recovery of benzene by one flash heating was ∼95 ( 3%. The carryover could not be avoided in a continuous sampling mode. The peak area also depended on the amount of analyte extracted by the membrane. The highest peak area for benzene was (48) Lanning, L. A.; Sacks, R. D.; Mouradian, R. F.; Levine, S. P.; Foulke, J. A. Anal. Chem. 1988, 60, 1994-1996.

740 Analytical Chemistry, Vol. 75, No. 4, February 15, 2003

Figure 6. Consecutive gas chromatograms of on-line monitoring by MTI-GC of the four volatile organic compounds in air. Peaks: (1) benzene; (2) trichloroethylene; (3) toluene; (4) tetrachloroethylene. The concentration of each analyte was 0.7 × 10-6 mol/L from the gas generator; the trapping time was 1 min; the carrier gas flow rate was 5 mL/min. For other chromatographic conditions, see the Experimental Section.

obtained using PDMS membrane. The peak area with the BPAPC-PDMS membrane was 10% lower due to a lower mass transport of benzene in this material. The flash heating of the trap produces an expansion of the volume of carrier gas and analytes in the trap tubing and spreading of the sample band. The pressure is suddenly increased, and the analytes are pushed backward and forward relative to the sorbent. Practically, the sample is divided into two parts, and as a result, at low flow rates of carrier gas, splitting and tailing of the peaks will occur. The splitting was avoided and the tailing was reduced by diminishing the backward flow at carrier gas flow rates above 2 mL/min, by reducing the free volume behind the sorbent in the trap and by using a flow restrictor. The best restrictions were obtained using a Valco fitting with a internal diameter of 0.25 mm at the inlet and 0.75 mm at the outlet of the trap. The tailing of the peaks was also reduced by the fast thermal desorption that occurred as a result of improvements in heat and mass transfer in the sorbent and in the boundary layer between the sorbent and the gaseous sample and as a result of the helical configuration of the sorbent and the flash heating of the trap. The desorption profile of the peaks is illustrated by consecutive gas chromatograms (Figure 6) obtained in a continuous sampling mode of a mixture of four volatile organic compounds (pollutants) with the aid of the MTI, followed by their pulse introduction into the GC column. The sample was generated at a constant concentration in a fume hood by a standard gas generator. The thermal stability of the sorbent trap extended beyond 4 monthes of continuous work at optimal voltage. The chromatograms were reproducible for prolonged times, and the shape of the peaks was very good. The peaks produced by the helical sorbent trap injection were sharper, narrower, and higher than ones produced with a thick-film capillary trap. As a consequence, the resolution and detection limit were improved. The limit of detection also depended on the time of trapping and the parameters that affected the permeation through the inlet membrane and was in the nanogram per milliliter range under the above experimental

monitoring of this complex mixture, giving a relatively simple gas chromatogram with the possibility to reduce the analysis time. The helical sorbent microtrap could also be used for continuous sampling of analytes from a liquid sample followed by their on-line introduction into a liquid chromatograph for analysis.42

Figure 7. Consecutive gas chromatograms of on-line monitoring by MTI-GC of the volatile organic compounds from a diesel engine exhaust. Conditions: the temperature of the GC oven was 100 °C; the carrier gas flow rate was 2.5 mL/min; the trapping time was 3 min; other conditions as in Figure 6.

conditions. The decrease of the analytes concentration in the air to 10-9 mol/L increased the breakthrough time and allowed for longer trapping time, with detection limits up into picogram per milliliter range. Figure 7 shows the consecutive gas chromatograms of a diesel engine exhaust. Car exhaust, especially from the diesel engine, is an important source of city pollution. The sampling was performed from the exhaust pipe of a diesel car at idling speed. No sampling chamber was used. Diesel exhaust was delivered from the stainless steel pipe directly on the surface of the membrane. The membrane allowed only the permeation of the VOCs, but the aerosols and the compounds with high boiling points could not permeate41 to reach the helical sorbent trap. This selectivity of the membrane was important for fast on-line

CONCLUSIONS This study shows that a helical sorbent microtrap is a very good solution for continuous sampling of the volatile organic compounds by a membrane and trap interface for performing automatically on-line preconcentration and GC analysis with the aid of a portable device. The results demonstrate that safe sampling of the helical sorbent microtrap is increased by the helical configuration of the sorbent that generates a turbulent rotational flow on the surface of the sorbent and by using low carrier gas flow rates, low concentration of analytes, and low trapping temperatures. The generation of the sharp GC peaks and reduced band broadening are a result of fast thermal desorption of the sample from the sorbent by flash heating of the trap, of a fast heat and mass transfer in the boundary layer between the sorbent and gaseous sample due to the turbulent flow generated by the helical configuration of the sorbent material, and of a diminishing of the backward flow in the trap. The helical sorbent microtrap exhibited good reproducibility and long time stability. ACKNOWLEDGMENT The authors thank to Dr. Iulia Lazar (Northeastern University, Boston) for reading the manuscript. Received for review November 15, 2002.

October

15,

2002.

Accepted

AC026238I

Analytical Chemistry, Vol. 75, No. 4, February 15, 2003

741