Dynamic Microwave-Assisted Extraction Coupled ... - ACS Publications

Feb 27, 2003 - large-volume injection gas chromatography. The Injection interface was a programmable temperature vaporizer. The system performance tes...
1 downloads 0 Views 117KB Size
Anal. Chem. 2003, 75, 1713-1719

Dynamic Microwave-Assisted Extraction Coupled On-Line with Solid-Phase Extraction and Large-Volume Injection Gas Chromatography: Determination of Organophosphate Esters in Air Samples Magnus Ericsson and Anders Colmsjo 1*

Department of Analytical Chemistry, Stockholm University, 106 91 Stockholm, Sweden

An on-line method was developed for the extraction, cleanup, and analysis of airborne organophosphate esters collected on glass fiber filters. The extraction and cleanup step was performed by conducting the dynamic microwaveassisted extraction (DMAE) coupled to solid-phase extraction (SPE). This system was further connected to include large-volume injection gas chromatography. The Injection interface was a programmable temperature vaporizer. The system performance test was investigated using spiked glass fiber filters. The DMAE-SPE recovery of the organophosphate esters was found to be greater than 97%. The repeatability of the uncorrected peak areas and the retention times was determined to be 4.2-8.0 and 0.03% relative standard deviation, respectively, and limits of detection were in the range 61-186.2 pg/m3. The method was tested in a newly restored office, in which several of the targeted organophosphate esters were detected. The total sampling and analysis time was less than 1.5 h. The use of microwave-assisted extraction (MAE) has grown in frequency and scope over the past decade.1 It is widely recognized now as a versatile extraction technique for solid samples. In several comparative studies, MAE has proven to be equal or superior to other techniques such as Soxhlet extraction, pressurized liquid extraction (PLE), and supercritical fluid extraction (SFE).2-4 The main advantages of MAE are usually the rate of extraction due to the fast heating and the elevated temperatures, but also the ease of instrument operation. Elevated temperatures and the high associated mass-transfer rates can be essential when the goal is quantitative and reproducible extraction. A drawback with MAE is that the heating is limited to dielectric matter. The two primary mechanisms for energy absorption using microwave power are ionic conduction and rotation of dipoles. Ionic conduction heating is due to the electrophoretic migration * Corresponding author. Tel: +46 8 16 71 97. Fax: +46 8 15 63 91. [email protected]. (1) Sparr Eskilsson, C.; Bjorklund, E. J. Chromatogr., A 2000, 902, 227-250. (2) Frost, S. P.; Dean, J. R.; Evans, K. P.; Harradine, K.; Cary, C.; Comber, M. H. I. Analyst (Cambridge, U. K.) 1997, 122, 895-898. (3) Lopez-Avila, V.; Young, R.; Teplitsky, N. J. AOAC Int. 1996, 79, 142-56. (4) Zuloaga, O.; Etxebarria, N.; Fernandez, L. A.; Madariaga, J. M. Trends Anal. Chem. 1998, 17, 642-647. 10.1021/ac026287v CCC: $25.00 Published on Web 02/27/2003

© 2003 American Chemical Society

of ions when a microwave field is applied. The resistance of the matter to this flow will generate heat as a consequence of friction. Dipolar molecules couple electrostatically to the microwaveinduced electric field and tend to align themselves with it. Since the microwave field is alternating in time, the dipoles will attempt to realign as the field reverses and so are in a constant state of oscillation at the microwave frequency. Frictional forces cause heat to be developed due to the motion of the dipoles.5 The traditional form of MAE is static; i.e., it does not involve continuous transport of extracted analytes out of the extraction vessels. Thus, this kind of method is not amenable to on-line coupling since continuous flow is a prerequisite for coupling to a cleanup step and a final gas chromatographic (GC) separation. Therefore, the authors have recently developed a dynamic microwave-assisted extraction 6 technique and coupled it to solidphase extraction (SPE).7 Automated analytical systems have many obvious benefits, such as speed of analysis, together with reductions in manual handling, risk of contamination, loss of analytes, and sample consumption. A drawback, however, is the increased complexity of on-line systems. Several papers have been published on the topic of the extraction of a solid sample for further transfer of the received extract on-line, to a GC. Typically, the extraction and cleanup step in GC-hyphenated systems involves SFE 8 or PLE 9,10 and highperformance liquid chromatography (HPLC) or SPE. In a recent publication, Shimmo et al. have shown the potential of SFE-LCGC to extract and analyze organic acids.8 An application was made with aerosol samples collected on Teflon filters. The instrument included on-line derivatization of the acids. Benzoic acid and butyric acid were identified at the level of nanograms per cubic meter. Pressurized hot water extraction coupled to LC-GC has also been described in two recent papers by Hyo¨etyla¨inen et al. (5) Meredith, R. Engineers’ Handbook of Industrial Microwave Heating; Institution of Electrical Engineers: London, 1998. (6) Ericsson, M.; Colmsjo, A. J. Chromatogr., A 2000, 877, 141-151. (7) Ericsson, M.; Colmsjo, A. J. Chromatog., A 2002, 964, 11-20. (8) Shimmo, M.; Hyo ¨etyla¨inen, T.; Hartonen, K.; Riekkola, M.-L. J. Microcolumn Sep. 2001, 13, 202-210. (9) Hyo ¨etyla¨inen, T.; Andersson, T.; Hartonen, K.; Kuosmanen, K.; Riekkola, M. L. Anal. Chem. 2000, 72, 3070-3076. (10) Kuosmanen, K.; Hyo ¨etyla¨inen, T.; Hartonen, K.; Riekkola, M.-L. J. Chromatogr., A 2002, 943, 113-122.

Analytical Chemistry, Vol. 75, No. 7, April 1, 2003 1713

9,10

in which polyaromatic hydrocarbons and brominated flame retardants in sediment samples were determined. The novel cleanup system, including a Tenax trap and a normal-phase HPLC column, allowed the use of flame ionization detection. Limits of detection were improved compared to traditional methods, and in addition, the amounts sampled could be as low as 10-100 mg of sediment. The methods in the papers 8-10 required large-volume injections of the final effluent, which were performed with the retention gap and early solvent vapor exit technique.11 In this study, a programmable temperature vaporizer (PTV) was utilized to interface the sample pretreatment and the GC column. To our knowledge, the only applications of solid sample extraction coupled on-line to a PTV-GC to date have been made by means of SFE.12-13 However, these systems did not involve large-volume injections since the effluent was gaseous carbon dioxide. In this study, organophosphate esters are studied as target molecules. These compounds are used on a large scale as additive flame-retarding agents, plasticiziers, or both in diverse products. Plastic material, lubricants, and electronic goods are some common fields of applications. A wide range of biological effects resulting from exposure to organophosphate esters has been reported. For instance, it has been shown that triphenyl phosphate has contact allergenic properties,14 and it is a potent inhibitor of human blood monocyte carboxyl esterase.15 The effects of some chlorinated organophosphate esters have been investigated on mice and rats. Both tris(2-chloropropyl) phosphate and tris(2chloroethyl) phosphate has been shown to have gonadotoxic effects16,17 and tris(2-chloroethyl) phosphate has shown neurotoxic and carcinogenic effects.18 In this study, the dynamic microwave-assisted extraction (DMAE) connected on-line with SPE, outlined in a recent publication,7 is further developed to include on-line transfer of the DMAE-SPE effluent to a GC equipped with a PTV injector. An application is made with organophosphate esters in indoor air collected on glass fiber filters.

EXPERIMENTAL SECTION Chemicals. The DMAE extraction solvent was methanol supplied by BDH Laboratory Supplies (Poole, U.K.). The water added to retain the organophosphate esters was precleaned by a UHQII system (Elga, High Wycombe, U.K.). The SPE cartridges were packed with polymer laboratories reversed phase (PLRP-S, 15-20 µm, 100 Å) and were purchased from SPARK Holland (AJ Emmen, The Netherlands). Methyl tert-butyl ether (MtBE) (Rathburn Chemicals, Walkerburn, Scotland) was used as the SPE (11) Grob, K. On-Column Injection in Capillary Gas Chromatography: Basic Technique, Retention Gaps, Solvent Effects; A. Huthig: Heidelberg, Germany, 1987. (12) Xianwen, L.; Janssen, H.-G.; Cramers, C. A. J. Chromatogr., A 1996, 750, 215-226. (13) Houben, R. J.; Janssen, H. G. M.; Leclercq, P. A.; Rijks, J. A.; Cramers, C. A. J. High Resolut. Chromatogr. 1990, 13, 669-73. (14) Environmental Health Criteria 111. Triphenyl phosphate. In International Program on Chemical Safety; World Health Organization: Geneva, 1991. (15) Emmet, E. A.; Tanaka, G. L., F.; Bleecker, M.; Fox, R.; Darlington, A. C.; Synkowski, D. R.; Dannenberg, A. M. J.; Taylor, W. J.; Levine, M. S. J. Occup. Med. 1985, 27, 905. (16) Shepel’skaya, N. R.; Dyshinevich, N. E. Gig. Sanit. 1980, 2, 85. (17) Shepel’skaya, N. R.; Dyshinevich, N. E. Prom-st. Ser: Toksikol. Sanit. Khim. Plastmass 1979, 4, 39. (18) Matthews, H. B.; Eustis, S. L.; Haseman, J. Fundam. Appl. Toxicol. 1993, 20, 477.

1714

Analytical Chemistry, Vol. 75, No. 7, April 1, 2003

eluent and as solvent for the external and internal standards. All solvents were of analytical reagent grade. The concentration of each organophosphate ester in the reference solution was ∼1 ng/ µL. The internal standard spiked onto the glass fiber filter samples was tri(n-pentyl) phosphate at a concentration of 2.79 ng/µL. All standard substances were purchased from Larodan Fine Chemicals (Malmo¨, Sweden). Dynamic Microwave-Assisted Extraction Combined with Solid-Phase Extraction and Large-Volume Injection Gas Chromatography. The system (Figure 1A) was assembled in our laboratory. It consisted of a solvent delivery system (Merck 655A12; Figure 1A, P1) that pumped extraction solvent through a preheating column and into the extraction vessel (both from Jour Research, Onsala, Sweden; Figure 1A.2 and 1A.3, respectively). Both the column and the vessel were heated by a microwaveassisted oven (EMM2361, Electrolux, Stockholm, Sweden; Figure 1A.1). The temperature of the solvent was measured using a grounded thermocouple, type K (Pentronic; Figure 1A.5) mounted in a modified tee (Jour Research; Figure 1A.4) and regulated by a temperature set-point controller (Eurotherm; Figure 1A.6). To retain pressure over the DMAE and to keep the methanol in a liquid state, the flow of extract was restricted, with a piece of narrow-bore tubing, 50 mm × 0.064 mm i.d. (Jour Research; Figure 1A.7), placed prior to the SPE system. The SPE eluent and the water were delivered by two separate HPLC pumps (Figure 1A, P2 and P3; models LC-6A, Shimadzu, and Prostar 220, Varian, Palo Alto, CA, respectively). The flows of methanol and water were combined in a mixing tee (Jour Research). The solidphase extraction was performed using 10 × 2 mm disposable cartridges installed in a cartridge holder (both from SPARK Holland; Figure 1A.8). Nitrogen gas that had been dried and purified by passage through a moisture filter, and a charcoal filter (Chrompack, Varian) was passed through the cartridge in order to dry the solid phase prior to GC analysis. Four-, 6-, and 10-port valves were used (VICI AG, Valco International; Figure 1A, V1V4) to switch solvents and gas streams, as appropriate. The injection valve (V4) was outfitted with a fused-silica leak 300 mm × 0.05 mm i.d., to purge the solvent and solute remaining in the transfer line in the reverse direction when the injection was complete (Figure 1A.10). A PTV-equipped GC with a nitrogen phosphor detector (NPD) from Agilent Technologies (Wilmington, DE; Figure 1A.9) was used to analyze the organophosphate esters. The PTV was set at solvent vent mode, with a solvent vent flow of 200 mL/min and solvent vent time of 5 min. The temperature program was 60 °C (5.1 min) followed by a 700 °C/min gradient to 400 °C (held for 10 min), and the purge flow was 33.1 mL/min for 5.9 min. The PTV liner was packed by the manufacturer (Agilent Technologies) with deactivated glass fiber wool. The GC oven was programmed as follows: 50 °C for 6 min, followed by gradients of 50 °C/min to 175 °C, then 15 °C/min to 320 °C, and held for 1 min. The column used was a DB5-MS supplied by J & W (Folsom, CA; length, 30 m; i.d., 0.25 mm; phase, 0.10 µm), the flow rate of nitrogen carrier gas was 40 cm/s, and the NPD was set at a temperature of 325 °C. The air, hydrogen, and nitrogen makeup gas flow rates were 60, 5, and 4 mL/min, respectively. The NPD used an external current supply (DETector Engineering & Technology, Inc, Walnut Creek, CA) set at 2.7 A

Figure 1. Schematic diagram of the DMAE-SPE-LVI-GC system and different modes of operation: (1) microwave oven, (2) preheater, (3) extraction vessel, (4) mixing tee, (5) thermocouple, (6) temperature regulator, (7) restrictor, (8) SPE cartridge, (9) PTV-GC-NPD, and (10) fused-silica leak. V1-V4, valves. P1-P3, pumps. Working modes: (A) extraction and trapping; (B) system cleanup and drying with nitrogen; (C) transfer and GC analysis.

with a bias voltage set at 5. The black ceramic bead assembly supplied by Agilent was chosen for this application since it offers better peak shape for phosphorus-containing solutes, although it has lower nitrogen sensitivity compared to the white bead assembly. The preheater, extraction vessel, all tubing, and finger-tight fittings were made of polyetheretherketone (PEEK). The extraction vessel had an internal volume of 57 µL. The tubes were of HPLC standard, with o.d. of 1.59 mm. All tubing had an internal diameter of 0.25 mm. A personal computer-based laboratory data system (ELDS Pro, Chromatography Data Systems, Svartsjo¨, Sweden) was used to program a relay card for switching the valves, starting the GC, and controlling on/off mode of the DMAE. HP-chemstation software (Agilent Technologies) was used for registering the detector signal and to control all GC parameters. Experimental Design. Two variables influencing the DMAESPE were investigated by means of experimental design. The variables were the total volume of MtBE required for complete elution from the SPE (VE-SPE ) 300-900 µL) and the flow rate of the added water (Φw ) 1000-2000 µL/min). A total of six experiments, including two central points, were performed using a full factorial design.19 The recoveries (%) of three different organophosphate esters detected in the extraction of spiked filters were used as responses, namely, tri-n-butyl, tris(2-chloropropyl), and triphenyl phosphate. The results were analyzed by multiple linear regression (MLR) performed using

Modde 4.0 software (Umetri AB, Umeå, Sweden) installed on a personal computer. The effects of varying the duration of extraction and the extraction temperature were not subjected to any study, but those variables were kept at the same levels as in a previous study.7 Air Sampling. The organophosphate esters were collected on 25-mm binder-free borosilicate glass fiber filters (Gelman Science Inc. Ann Arbor, MI). Prior to sampling, the filters had to be thoroughly cleaned due to the ubiquity of the organophosphate esters in indoor air. A cleaning routine including sonication of the filters for 20 min each in methanol, acetone, and dichloromethane was therefore applied. The filters were placed in a sampler holder made from anodized aluminum.20 Air was pumped through the sampler using a battery-operated personal sampler pump (224-PCXR7, SKC Inc, Eighty Four, PA). The flow rate was 2.5 L/min, and the sampling time for the office with new furniture and computer was 25 min. This yields a total sampled air volume of 62.5 L. Either the samples were analyzed immediately after sampling or the filters were wrapped in foil and stored in a freezer at -18 °C until analysis. Analysis Procedure for DMAE-SPE-LVI-GC. After air sampling, the first step (Figure 1A) in the analysis procedure was to cut the glass fiber filter into thin strips, which were carefully placed in the extraction vessel with a pair of tweezers. One filter filled the larger part of the vessel. A 1.0-µL portion of the internal standard was added using a 5-µL GC syringe (Agilent Technolo-

(19) Carlson, R. Design and optimization in organic synthesis; Elsevier: New York, 1992.

(20) Oestman, C.; Carlsson, H.; Bemgaard, A.; Colmsjoe, A. Polycyclic Arom. Compd. 1993, 3, 485-92.

Analytical Chemistry, Vol. 75, No. 7, April 1, 2003

1715

Table 1. Factorial Design Settings and Results recoveryb (at various level settingsc) phosphate estera

[300/1000]

[900/1000]

[300/2000]

[900/2000]

[600/1500]

[600/1500]

tri-n-butyl tris(2-chloropropyl) triphenyl

79 32 77

86 31 82

81 95 81

102 107 101

92 93 88

91 93 89

a

Phosphate substituent. b Expressed as percent of spiked amount (%). c Variable settings, [elution volume (µL)/water flow rate (µL/min)].

gies). Prior to extraction, a new SPE cartridge was loaded. The solid phase was first cleaned with 7.5 mL of methanol and then activated with 9.0 mL of water. The methanol and water pumps (P1, P3) were stopped while the extraction vessel was installed with finger-tight connections in the system. The pumps (P1, P3) and the relay program were started simultaneously (Figure 1A). The microwave heating, set at 110 °C, was started after a short delay, to ensure that the extraction vessel was properly filled with extraction solvent. The water flow rate was set according to the screening design, i.e., 2000 µL/min, the methanol flow rate was 500 µL/min, and the duration of microwave-assisted extraction was set to 10 min. The pressure over the extraction vessel during the extraction was ∼30 bar, with a slight oscillation due to the bistable (on/off) nature of the microwave heating. In the next step (Figure 1B), valves V1 and V3 were switched. The SPE cartridge was dried for 15 min with N2 at a flow rate of 100 mL/min. Simultaneously, the rest of the system was precleaned with MtBE delivered by P2 at a flow rate of 200 µL/min. In the final step (Figure 1C), valves V2 and V4 were switched to back-flush the SPE-trapped analytes into the PTV injector. The GC was started simultaneously. The fraction transferred to the PTV was 840 µL. Linearity. Method linearity was tested according to good laboratory praxis (GLP) on six levels of concentration. Coefficient of determination is usually not sufficient for the judgment of the curvature of a line. For this purpose, a curve according to

y ) kxR + m

was vertically adapted to the measured points by means of the method of least squares. An exponent close to unity is a strong indication on linearity, but usually a value in the range 0.9 < R < 1.1 can be considered as acceptable.21 RESULTS AND DISCUSSION In the development of the DMAE-SPE-LVI-GC method, the DMAE-SPE step was optimized separately off-line, but the on-line coupling to the GC, i.e., the transfer of the eluted fraction and the solvent elimination in the PTV, was investigated while the entire system was operating. The complete method was further evaluated by an overall system performance test, in which LOD, linearity and dynamic range, retention time, and absolute area repeatability were investigated. Following this, the complete DMAE-SPE-LVI-GC method was applied to air samples collected in an office. (21) ISO 6143: 2001; p 6.

1716

Analytical Chemistry, Vol. 75, No. 7, April 1, 2003

DMAE-SPE. The DMAE-SPE extraction vessel was miniaturized to 57 µL, a 10-fold reduction in volume compared to the previous vessel.7 The small vessel volume facilitates the on-line coupling to a GC, thus substantially lowering the limit of quantification of target compounds. System miniaturization also enhances the extraction performance since it enables more vessel volumes of extractant to be flushed through the vessel without increasing the volume of extraction solvent used. However, a disadvantage with miniaturization is that it increases the risk of wall effects since it raises the surface area to solvent volume ratio. The risk of adsorption is largely governed by the solvent, solute, and wall properties. With the duration of extraction set to 10 min, and with the methanol flow rate set to 500 µL/min, the required volume of methanol for the extraction of one air sample was 5 mL, compared to the 20 mL of chlorinated solvent required by a previous offline method.22 The temperature of the microwave-assisted extraction can have a strong influence on the rate of mass transfer. Extraction kinetics are usually accelerated with increases in temperature, up to the point at which target analytes are degraded. In this study, the extraction temperature was set to 110 °C, close to the highest possible temperature allowed by the selected apparatus. Higher temperatures would have required the oven to be operated at maximum power throughout the extraction, which would have caused potentially problematic heating of the oven walls since the oven interior contained relatively little microwave-absorbing material. A solid phase with a highly cross-linked polymer made of poly(styrene-divinylbenzene) was used in the SPE cartridge to retain extracted compounds. The surface area of this phase is roughly 1000 m2/g and the main interaction is due to hydrophobic forces, although charge transfer and π-π interactions may also be involved. Since the target compounds are able to form both hydrophobic and π-π bounds, this phase was also the first choice when experimental conditions were to be discovered for the organophosphate esters. An experimental design was performed in order to establish how efficiently the chosen SPE phase was able to trap and release the extracted compounds, focusing on water flow rate and the volume of MtBE required to elute the organophosphate esters from the SPE. In Table 1, the investigated levels and the result are presented. In Table 2, the model coefficients and their confidence intervals are presented. The design was implemented with six extractions at all investigated points, with the twin goals (22) Carlsson, H.; Nilsson, U.; Becker, G.; Oestman, C. Environ. Sci. Technol. 1997, 31, 2931-2936.

Table 2. Factorial Design Results model coefficients for main variables and their interaction effectb phosphate estera

elution vol β1

water flow rate β2

interaction β12

tri-n-butyl tris(2-chloropropyl) triphenyl

7.1 (5.6) 2.8 (32.1) 6.4 (4.3)

4.3 (5.6) 35 (32.1) 5.9 (4.3)

3.2 (5.6) 3.4 (32.1) 3.5 (4.3)

a Phosphate substituent. b Levels of the MLR (recovery ) β + 0 β1x1 + β2x2 + β12x1x2) calculated model coefficients, figures in parentheses are the confidence intervals at the 95% level. Significant model coefficients are shown in boldface type.

of minimizing both the SPE elution volume and water consumption. The MLR calculations showed that the elution volume had significant impact on the models for the arylated and the alkylated organophosphate esters but not for the chlorinated analytes. The water flow rate showed significant impact solely for the chlorinated analytes. The interaction effect of the variables did not yield any significant influence on the model for any of the selected responses. At the low level, with 1000 µL/min of water added to the 300 µL/min flow of MtBE, giving a volume/volume ratio of 67:33, substantial losses of chlorinated compounds were observed (recovery ∼32%). This was solely due to breakthrough in the SPE cartridge and, hence, not to incomplete desorption. There were also losses of both arylated and alkylated organophosphate esters (recovery ∼80%), due in these cases to both the water flow rate and SPE elution volume being too low; i.e., both breakthrough and incomplete desorption occurred. At the central point, the added flow rate of water was 1500 µL/min (75%), which was sufficient to trap all of the investigated analytes fully, but the SPE elution volume at this setting (600 µL) was insufficient to allow quantitative extraction of the selected components. At the high level, the water flow rate was 2000 µL/ min (80%) and the SPE elution volume was 900 µL. With these settings, the extraction was quantitative for all investigated compounds. However, the volume of MtBE could be reduced to 800 µL while recoveries were maintained. Consequently, these levels were chosen as the experimental conditions for further coupling. Since the chosen SPE was effective for both trapping and releasing the solutes, no other phases were investigated. Consideration of differences in the octanol-water partition coefficients (log Kow) of the selected compounds can give further insight into mechanisms influencing the results. For example, the log Kow values for triphenyl phosphate and trichloropropyl phosphate are 4.7 and 2.5, respectively, while the alkylated phosphates have higher log Kow values of ∼9.23 Thus, a high water content is needed to trap the chlorinated organophosphate esters, but less is needed for the alkylated and arylated compounds. Consequently, more MtBE is needed to desorb the alkylated and arylated compounds compared to the chlorinated compounds. Since the presence of water might demolish the chromatographic analysis and activate the liner and the column, the solid phase by passing nitrogen through it prior to elution and injection into the GC is essential. Fifteen minutes was previously established to be a suitable drying time.7 (23) Physical Properties Database. Syracuse Research Corp., Syracuse, NY.

Transfer of the Eluate and the Large-Volume Injection. When the eluted SPE fraction was transferred to the PTV-GC, a small carryover (∼3-4%) from the previous run was detected for all the investigated compounds due to adsorption onto the rotor of the injection valve (Figure 1, V4) and the fused-silica transfer line. The silica transfer line and the channel of the valve rotor were not flushed with fresh solvent prior to each extraction, since this flow path could not be flushed without performing an injection. To reduce the carryover effect, 40 µL of MtBE in addition to the 800 µL needed to release the trapped organophosphate esters was injected each time. This reduced the carryover to a level of 1-2%. Approximately 1 mL of solvent was needed to fully eliminate the carryover, but increasing the injection volume to 1 mL would have reduced the repeatability of the peak areas and the useful lifetime of the liner. Therefore, the small carryover was accepted in further experiments. The maximum transfer rate was set at 200 µL/min. At higher transfer rates, flooding of the PTV became a problem. Flooding occurs when the speed of vaporization of the solvent of the injected fraction is much lower than the flow rate of eluent into the injector. Consequently, some of the analytes will be flushed out through the ventilation flow line, resulting in lower yields and unpredictable results. The compound tri-n-butyl phosphate was subject to some discrimination in the PTV, roughly ∼10% being lost at the temperature and flow rate used. This discrimination was mainly dependent on the injector temperature; lower temperatures caused flooding while higher temperatures increased the discrimination. The vent flow rate was a robust parameter and did not have the same dramatic effects on the discrimination when minor adjustments were made. At all investigated rates, the results were shown to be reproducible (Table 3). System Performance. The performance of the system was investigated with respect to the extraction of spiked filters. The recovery of the organophosphates for the DMAE-SPE step was found to be higher than 97% (Table 3) for all the selected compounds. As shown above, ∼10% of tri-n-butyl phosphate was lost in the PTV. Losses other than those due to discrimination of analytes in the PTV are mainly due to the sample handling prior to extraction and the small carryover, as discussed above. The method LOD, recovery, peak area, and retention time reproducibility are shown in Table 3. The first column of Table 3 lists the lowest detectable levels of each compound trapped on the filters. These levels were found to be in the range of 4.6-14 pg. The LOD for the office air samples is presented in column two. The lowest detectable levels ranged from 61 pg/m3 for tri-n-butyl phosphate to 186 pg/m3 for tris(2ethylhexyl) phosphate. These results were obtained with a sampling time of 30 min and a pump flow rate of 2.5 L/min. At lower levels, prolonged sampling is needed. Reproducibility parameters (relative standard deviations, RSDs) for the retention time and peak area of each compound are listed in columns 5 and 6 of Table 3. These experiments (five extractions over a period of 3 days) were done with filters spiked at the 1 ng /filter level. The RSDs of the uncorrected peak areas were low, in the range of 4.1-8%. The RSDs for the retention times were also low and reproducible, in the range of 0.03-0.04%. This is advantageous as reliable retention times are essential for identifying phosphate esters with this system. Analytical Chemistry, Vol. 75, No. 7, April 1, 2003

1717

Table 3. System Performance organophosphate ester

LODa (pg/filter)

LODb (pg/m3)

DMAE-SPE recoveryc (%)

peak area RSDd (%)

retention time RSDe (%)

tri-n-butyl phosphate tris(2-chloroethyl) phosphate tris(2-chloropropyl) phosphate* tris(2-butoxyethyl) phosphate triphenyl phosphate tris(2-ethylhexyl) phosphate

4,6 8,2 5,4 9,2 8,6 14

60,9 109,3 72,7 122,3 114,1 186,2

98.5 101.2 97.0 103.4 97.5 99.3

4,2 8 5,5 4,1 6,3 5,4

0,03 0,03 0,04 0,03 0,03 0,03

a The lowest amount needed to be trapped on the filters for detection in the DMAE-SPE-LVI-GC method. b LOD with 30 min of sampling at a flow rate of 2.5 L/min. c Recovery achieved with the DMAE-SPE offline method with spiked filters. d Absolute peak area determined with spiked filters at the 500 pg level, n ) 5. e n ) 5. *Integrated sum of three isomers.

Table 4. Method Linearity organophosphate ester

r2 coeff of determinationa

curvatureb

linearity range (ng/filter)

tri-n-butyl phosphate tris(2-chloroethyl) phosphate tris(2-chloropropyl) phosphatec tris(2-butoxyethyl) phosphate triphenyl phosphate tris(2-ethylhexyl) phosphate

0.999 0.999 0.996 0.999 0.989 0.992

y ) 225.4x0.99 - 1.2 y ) 223.4x1.07 - 0.2 y ) 168.7x1.11 + 24.1 y ) 195.3x1.06 - 14.3 y ) 144.3x0.98 - 0.4 y ) 171.0x0.90 - 10.4

0.13-1.5 0.13-5 0.14-9 0.14-4 0.14-8 0.14-7

a Linearity was investigated at six levels with triplicates for each point. b Should be in the range of 0.9-1.1 for good linearity. c Integrated as the sum of three isomers.

For the linearity studies, filters were spiked with a standard mixture at six different levels with triplicates at each level, Table 4. For the range 130 pg to 9 ng, all compounds had a factor of curvature within the range of 0.90-1.11, which is very nearly within the limits (0.9-1.1) commonly used to accept linearity. Linearity was also investigated by plotting the response factors versus the spiking level. In this case, the linearity was evaluated by investigating the deviation from the zero slope (k ) 0). Low deviations were discovered for all the compounds (0-8%) except for tris(2-chloroethyl) phosphate, which in the worst case had 20% deviation. Application of the Method: Determination of Organophosphate Esters in Indoor Air Samples. To apply the entire method to real samples, the organophosphate ester status of the air in an office was investigated. The room was equipped with new furniture and a new personal computer and monitor, used for a week prior to the first sampling occasion. Air samples were collected in the breathing zone of a user sitting in front of a computer monitor. Samples were collected on two occasions, 10 days apart. As shown in Table 5, several phosphate esters were discovered. A typical chromatogram from a sample taken in the office is shown in Figure 2. A few and well-defined peaks can be seen for tri-n-butyl, tris(2-chloropropyl), tris(2-butoxyethyl), and triphenyl phosphate. The most abundant phosphate ester in the investigated room was tris(2-chloropropyl) phosphate. This compound consists of three isomers that appear as three separated peaks in the chromatogram (Figure 2) forming a pattern that is easy to recognize. The isomers were integrated and quantified as one compound. The levels of concentration for these compounds were registered to be ∼60 ng/m3 at the two sampling occasions (RSD ) 10.6 and 18.5%, n ) 3). Tris(2-chloropropyl) phosphate is known 1718 Analytical Chemistry, Vol. 75, No. 7, April 1, 2003

Table 5. Levels of Organophosphate Esters Found in Office Air Samples office concn (ng/m3) (SD) n ) 3

organophosphate ester

new computer and furniture

10 days later, computer switched off

tri-n-butyl phosphate tris(2-chloroethyl) phosphate tris(2-chloropropyl) phosphatea tris(2-butoxyethyl) phosphate triphenyl phosphate tris(2-ethylhexyl) phosphate

2.5 (0.9) 1.6 (0.1) 60.3 (6.4) 7.6 (0.7) 20.1 (5.4) ndb

4.2 (0.7) 0.6 (0.1) 59.5 (11) 41(4) 10.4 (3.6) nd

a

Integrated as the sum of three isomers. b nd, not detected.

to be used as a flame-retarding and plastizing additives in glue. Thus, probable sources include the glues used in the office chair upholstery, carpets, laminated chipboards, etc. Confirmed sources were found to be the glue in the laminated chipboards used in the new furniture and the upholstery of the office chairs. This was established by analyzing small amounts of the material, typically a few milligrams. Further studies will be performed in order to identify all sources of this compound. The second chlorinated compound investigated, tris(2-chloroethyl) phosphate, was also discovered at levels of 1.6-0.6 ng/ m3. The concentration did not appear to vary significantly with respect to time, indicating that this compound did not originate from the computer equipment or the new furniture. Triphenyl phosphate was also discovered in the air samples. This compound is likely to originate from the plastic material of the computer monitor, which can contain up to 10% of noncovalently added triphenyl phosphate. This was also confirmed by

The concentrations found for tri-n-butyl phosphate were at the levels of 3-4 ng/m3. No significant trend with respect to time in these levels was detected, indicating that this compound did not emanate from the computer equipment or the new furniture.

Figure 2. Chromatogram from an office air sample. The substituent names of the investigated organophosphate esters are given for each peak.

extracting a few milligrams of plastic from the cover. The level found at the first sampling was 20.1 ng/m3 (RSD ) 26.9%, n ) 3). At the second occasion, this level had decreased to 10.4 ng/m3 (RSD ) 34.6%, n ) 3), indicating that air levels of triphenyl phosphate were declining gradually with use. This supports an earlier report by Carlsson et al.24 Two alkylated phosphates, tris(2-butoxyethyl) phosphate and tri-n-butyl phosphate, were also discovered in the office. The level of tris(2-butoxyethyl) phosphate found at the first sample occasion was 7.6 ng/m3 (RSD ) 9.2%, n ) 3) while at the second the level had increased to 41 ng/m3 (RSD ) 9.8%, n ) 3). The origin of this compound was not established and will be investigated in further studies. (24) Carlsson, H.; Nilsson, U.; Oestman, C. Environ. Sci. Technol. 2000, 34, 3885-3889.

CONCLUSIONS Dynamic microwave-assisted extraction coupled to SPE and large-volume injection gas chromatography was shown to be a powerful tool for the analysis of organophosphate esters in indoor air. Its efficiency is due to the effective extraction (>97%), quantitative preconcentration, and introduction of the whole sample into the GC. The limit of detection that the DMAE-SPELVI-GC method yield allows short sampling times to be used or, with extended sampling times, the detection of low concentrations of analytes. Thus, it clearly has great potential for samples collected spotwise and for personal monitoring. The technique can also be considered for screening large numbers of samples since the sampling and analysis time is short (1.5 h). The instrument is also highly flexible since the analytical conditions of the system, the solid phase, the extractant, and the eluent solvent, can all be readily be changed. Diverse compounds from air samples or other matrixes can thus be recovered and analyzed. ACKNOWLEDGMENT Anders Christensen and Petter Tollba¨ck are acknowledged for their help and valuable discussions on the topic of PTV. Ove Jonsson is acknowledged for his help with the phosphate standards. Received for review November 6, 2002. Accepted January 28, 2003. AC026287V

Analytical Chemistry, Vol. 75, No. 7, April 1, 2003

1719