Anal. Chem. 2005, 77, 1853-1858
Determination of Monosaccharide Anhydrides in Atmospheric Aerosols by Use of High-Performance Liquid Chromatography Combined with High-Resolution Mass Spectrometry Christian Dye* and Karl Espen Yttri
Norwegian Institute for Air Research, Instituttveien 18, P.O. Box 100, N-2027 Kjeller, Norway
A novel method for determination of the monosaccharide anhydrides (MAs) levoglucosan, mannosan, and galactosan in atmospheric aerosols has been developed. The method is based on solvent extraction of aerosol filter samples and chemical analysis performed with highperformance liquid chromatography (HPLC) combined with time-of-flight high-resolution mass spectrometry. The MAs were separated on two series connected 2.1 mm × 150 mm reversed-phase HPLC columns and identified by negative electrospray ionization mass spectrometry using the m/z 113, 129, and 161 as monitoring ions. The limit of quantification (LOQ) at S/N 10 ranges between 20 and 40 pg for the three isomers. The LOQ can easily be improved. Samples from Elverum (urban background) and Oslo (urban) in Norway were used in validation experiments. Tetrahydrofuran was found to be the most efficient extraction solvent. The choice of solvent is of crucial importance to minimize adsorption to the glassware. Interactions between the MAs and active sites on the quartz filter surface are observed. The content of organic aerosols in the atmosphere affects the global climate through their modification of the solar radiation energy reaching the earth-atmosphere system (IPCC, 2001). In this context, it is important to know whether the organic aerosol sources are of anthropogenic or natural origin in order to quantify the impact of human activity on the global climate. A common method in source apportionment studies is to combine quantitative chemical analysis of source-specific tracers with receptor model calculations.1,2 During the past decade, the monosaccharide anhydrides (MAs) levoglucosan (1,6-anhydro-β-D-glucopyranose), mannosan (1,6-anhydro-β-D-mannopyranose), and galactosan (1,6anhydro-β-D-galactopyranose) have received increased attention due to their properties as specific tracers for biomass burning3,4 (Figure 1). MAs are generated during combustion and pyrolysis * To whom correspondence should be addressed. E-mail:
[email protected]. (1) Schauer, J. J.; Rogge, W. F.; Hildemann, L. M.; Mazurek, M. A.; Cass, G. R.; Simoneit, B. R. T. Atmos. Environ. 1996, 30, 3837-3855. (2) Manchester-Neesvig, J. B.; Schauer J. J.; Cass, G. R. J. Air Waste Manage. Assoc. 2003, 53, 1065-1079. (3) Fraser, M. P.; Lakshmanan, K. Environ. Sci. Technol. 2000, 34, 45604564. (4) Simoneit, B. R. T. Appl. Geochem. 2002, 17, 129-162. 10.1021/ac049461j CCC: $30.25 Published on Web 02/04/2005
© 2005 American Chemical Society
Figure 1. Chemical structure of the target monosaccharide anhydrides.
of cellulose and hemicellulose5 and are an important class of compounds in the organic aerosol subfraction denoted watersoluble organic compounds (WSOCs).6,7 Even though the MA content of organic aerosols has been the topic in many studies, a limited diversity is seen in the methodology of the chemical analysis.4,8-10 The most frequently used principles are based on extraction of the filter media with dichloromethane, which in some studies is modified with a more polar solvent, followed by derivatization with various silylating agents. After derivatization, the sample is analyzed with GC/MS. The benefit of this methodology is that polar compounds, which normally are not compatible with GC/MS analysis due to poor chromatographic behavior and low volatility, easily can be analyzed due to the increased volatility and improved chromatographic properties of their corresponding derivatives. Several studies using this methodology have revealed that the WSOC fraction is dominated by oxygenated polar multifunctional compounds containing carboxyl, carbonyl, hydroxyl, amide, and amino groups.9,11,12 (5) Simoneit, B. R. T.; Schauer, J. J.; Nolte, C. G.; Oros, D. R.; Elias, V. O.; Fraser, M. P.; Rogge, W. F.; Cass, G. R. Atmos. Eviron. 1999, 33, 173182. (6) Zappoli, S.; Andracchio, A.; Fuzzi, S.; Facchini, M. C.; Gelencse´r, A.; Kiss, G.; Kriva´csy, Z.; Molnar, A.; Me´sza´ros, E.; Hansson, H.-C.; Rosman, K.; Zebu ¨ hr, Y. Atmos. Environ. 1999, 33, 2733-2743. (7) Yttri, K. E.; Dye, C.; Slørdal, L. H.; Braathen, O. A. Submitted (8) Fabbri, D.; Chiavari, G.; Prati, S.; Vassura, I.; Vangelista, M. Rapid Commun. Mass Spectrom. 2002, 16, 2349-2355. (9) Graham, B.; Mayol-Bracero, O. L.; Guyon, P.; Roberts, G. C.; Decesari, S.; Facchini, M. C.; Artaxo, P. J.; Maenhaut, W.; Ko ¨ll, P.; Andrae, M. O. J. Geophy. Res. 2002, 107, D20, 8047, (10) Pashynska, V.; Vermeylen, R.; Vas, G.; Maenhaut, M.; Claeys, M. J. Mass Spectrom. 2002, 37, 1249-1257. (11) Nolte, C. G.; Schauer, J. J.; Cass, G. R.; Simoneit, B. R. T. Environ. Sci. Technol. 2001, 35, 1912-1919. (12) Schauer J. J.; Kleeman, M. J.; Cass, G. R.; Simoneit, B. R. T. Environ. Sci. Technol. 2001, 35, 1716-1728.
Analytical Chemistry, Vol. 77, No. 6, March 15, 2005 1853
One basic question regarding the MAs is whether it is necessary to use derivatization in order to perform a reliable chemical analysis. Nonderivatization sample preparation combined with GC analyses of MAs have been applied.13,14 However, these studies did not include a detailed methodology description, and other studies have reported poor chromatographic behavior15 without derivatization. The objective of the present work was to develop a highperformance liquid chromatography (HPLC)/high-resolution mass spectrometry (HRMS) method targeted at MAs. The method is based on solvent extraction of aerosol filter samples and subsequent chemical analysis with HPLC/HRMS. Several advantages were expected to be obtained compared to current aerosol science status: (1) An HPLC/HRMS method is expected to be less labor and less time-consuming, as the sample preparation is less extensive due to the omitted derivatization step. (2) An HPLC/HRMS method will increase the diversity of chemical methods available for determination of MAs. It is of great scientific value to compare the results from analytical methods obtained with different approaches. (3) An HPLC/HRMS method is complementary to the frequently used GC/MS methods implying a potential to extend the knowledge of the organic aerosol composition. This can be obtained by accurate mass measurements and deduction of the elemental composition of new compounds.18,19 Although an HPLC/HRMS method will provide new advantages, one should note that LC/MS instruments generally are more expensive than that of GC/MS. It is also possible to run one type of ion source, i.e., negative electrospray ionization (ES), with different types of mass spectrometers such as ion trap, quadrupole, and time of flight (TOF), which all will provide different performances. In addition, within one instrument type, higher instrument-to-instrument variability is normally found for LC/MS than for GC/MS. In two recent studies, flow injection analysis combined with negative ES ion trap mass spectrometry has been used to study the WSOC fraction.16,17 However, these methods were not developed for quantifying MAs in atmospheric aerosols. To our knowledge, this is the first time an HPLC/HRMS method for determination of MAs has been reported. EXPERIMENTAL SECTION Chemicals. HPLC gradient grade acetonitrile (ACN), methanol (MeOH), and tetrahydrofuran (THF) were obtained from Merck (Darmstadt, Germany). Absolute 2-propanol and ethanol were obtained from Arcus, and dichloromethane (DCM) (Supra(13) Frazer, M. P.; Lakshmanan, K. Environ. Sci. Technol. 2000, 34, 45604564. (14) Fine, P. M.; Cass, G. R.; Simoneit, B. R. T. Environ. Sci. Technol. 2001, 35, 2665-2675. (15) Zdra´hal, Z.; Oliveira, J.; Vermeylen, R.; Claeys, M.; Maenhaut, W. Environ. Sci. Technol. 2002, 36, 747-753. (16) Gao, S.; Hegg, D. A.; Hobbs, P. V.; Kirchstetter, T. W.; Magi, B. I.; Sadilek, M. J. Geophys. Res. 2003, 108 (SAF 27), 1-16. (17) Cappiello, A.; De Simoni, E.; Fiorucci, C.; Mangani, F.; Palma, P.; Trufelli, H.; Decesari, S.; Facchini, M. C.; Fuzzi, S. Environ. Sci. Technol. 2003, 37, 1229-1240. (18) Hogenboom, A. C.; Niessen, W. M. A.; Little, D.; Brinkman, U. A. Th. Rapid Commun. Mass Spectrom. 1999, 13, 123-133. (19) Eckers, C.; Wolff, J.-C.; Haskins, N. J.; Sage, A. B.; Giles, K.; Bateman, R. Anal. Chem. 2000, 72, 3683-3688.
1854
Analytical Chemistry, Vol. 77, No. 6, March 15, 2005
Solv) and acetic acid were obtained from Merck. Standards of mannosan (1,6-anhydro-β-D-mannopyranose) and galactosan (1,6anhydro-β-D-galactopyranose) and ammonium acetate were purchased from Sigma. Levoglucosan (1,6-anhydro-β-D-glucopyranose) was obtained from Fluka. The water used for HPLC analysis and sample preparation was purified by a Waters MilliQ system. The chemicals used for TOF mass calibration were analytical grade. The glassware at the laboratory was cleaned by soaking it in a RBS-25 cleaning solution (R. Borghgraef s.a.- n.v., Brussels, Belgium) for 24 h with a subsequent MilliQ water rinse procedure. Aerosol Sampling. Aerosol samples were collected on nonprefired quartz fiber filters (Whatman QM-A, 8 in. × 10 in.) using a high-volume sampler (flow rate 1.1 m3 min-1) at an urban site (Oslo) during November-December 2001 and at an urban background site (Elverum) during February-March 2002. At the Elverum site, aerosol samples were collected on Zefluor Teflon filters (2-µm pore, 47 mm, Gelman) as well, using a low-volume sampler (10 L min-1). Both samplers were equipped with a PM10 inlet providing a 50% cutoff for particles with an equivalent aerodynamic diameter of 10 µm. At Elverum, the two aerosol samplers were operated in parallel. The sampling time was 23 h. Apparatus. Liquid chromatography was performed with an Agilent 1100 liquid chromatography system (Agilent Technologies, Waldbronn, Germany), equipped with an autosampler, a quaternary pump, an on-line degassing system, and a diode array detector (UV). The compound separation was performed with two equal series-connected reversed-phase C18 columns (Atlantis dC18, 2.1 mm i.d. × 150 mm length, 3 µm, Waters). A stainless steel inlet filter (Supelco, 0.8 µm) was used in front of a precolumn with the same stationary phase as the separation columns. Gradient elution was performed with water as solvent A and acetonitrile as solvent B. The binary gradient had a flow rate of 0.2 mL min-1 and started isocratic with 100% A the first 5 min. Solvent B was introduced linearly up to 100% at 20 min and kept isocratic until 28 min. At 28.3 min, the setting was 100% A and the column was equilibrated up to a run time of 60 min. Electrospray Ionization. The analytical detector was a Micromass LCT orthogonal-TOF MS equipped with a Z-spray electrospray ion source and a 4-GHz time to digital converter (Micromass Ltd., Wythenshawe, Manchester, U.K.). The instrument was operated in negative mode and calibrated for accurate mass measurements of selected samples. The lenses may be tuned for optimal sensitivity or optimal resolution. The rf dc offset value 1, rf dc offset value 2, and aperture were tuned for a signal resolution of 6500 at half-height for accurate mass measurements. When high sensitivity was needed, the lenses were tuned for a resolution of 4500, which increased the sensitivity by a factor of 4. The electrospray source parameters were optimized to the following values: sample cone 31 (m/z 113 and 129) and 20 V (m/z 161), capillary voltage 2.5 kV, extraction cone 3 V, source temperature 125 °C, desolvation temperature 350 °C, cone gas flow 24 L h-1, and desolvation gas flow 600 L h-1. The pusher frequency was operated in automatic mode with a typical value of 20 MHz. The spectrum integration time was 0.9 s with a interscan delay of 0.1 s. The instrument was operated in continuum mode in the m/z range 50-600. The data processing and instrument (HPLC/HRMS) control were performed by the MassLynx software.
Figure 2. Mass spectra of the monosaccharide anhydrides at cone voltage 31 V.
Accurate Mass Measurements. A verification of the compound identity was obtained by accurate mass measurements of selected samples and standards. Lock mass reference compound, i.e., sulfadimethoxime m/z [M - H]- 309.0658 and leucine enkephalin m/z [M - H]- 554.2615 in 50% acetonitrile in water, were introduced into the instrument by using a Harvard 22 syringe pump (Harvard Apparatus inc., South Natick, MA) and a T-piece connection between the HPLC column and the ion source. Sample Preparation. Two punches (1 cm2) of the exposed filters (quartz fiber and Teflon) were extracted in 2 mL of THF by ultrasonic bath agitation. The filters were extracted twice, and the extracts were pooled. Undissolved sample material and filter debris were removed from the sample solution by filtration (Alltech, syringe filter, 0.45 µm, Part No. 2047). The pooled extract was evaporated to 1 mL by a gentle stream of N2. The elution strength of the sample extract was adjusted with water (0.8 mL) before injection. The standards for quantitative chemical analysis were prepared by dissolving known amounts of the individual MA compounds in 50% THF in MilliQ water. As a part of the method validation, DCM and 50% THF in water were tested as sample extraction solvents. The samples were extracted twice in 2 mL of the specified solvent by ultrasonic bath agitation. After extraction, the sample solutions were filtered through a syringe filter. The DCM solution was evaporated to dryness by a gentle stream of N2 in room temperature and redissolved in 50% THF in water. An MA stock solution in water was used in the solvent compatibility studies. The experiments were performed by mixing an aliquot of the stock standard solution with the specified solvents to obtain a mixture of 40% solvent in water. RESULTS AND DISCUSSION Monitoring Ions. The mass spectra obtained by the single MAs at a cone voltage of 31 V are given in Figure 2. Due to the chromatographic overlapping peaks of mannosan and levoglu-
cosan (m/z 161 in Figure 3) it is necessary to use the less abundant ions m/z 129 and 113, respectively. These ions provide baseline separation for the target compounds, which is crucial when working with quantitative analysis. Less ion source fragmentation of the MA compounds can be obtained by reducing the cone voltage to 20 V, which in turn increases the sensitivity significantly of the m/z 161 for all of the target MA compounds. In the MassLynx software, it is possible to cycle the cone voltage such that it alternates between 20 and 31 V in the same run. This provides selectivity and optimum sensitivity in the same run and was used for routine analysis. Typical data of the HPLC/HRMS instrument performance are found in Table 1. The data in Table 1 were obtained with unit resolution in order to be comparable with ion trap and quadrupole instruments. However, in routine analysis, ions were profiled with a peak width of ∼30 mDa, which doubles the signal-to-noise ratio compared to unit resolution and hence reduces the limit of quantification (LOQ) correspondingly. The elemental composition was inferred for the MA standards and some selected samples from Elverum and Oslo in the first part of the method development. The measured accurate masses of the monitoring ions are given in Table 2. The deviations from the theoretical monoisotopic values are typically less than 2 mDa. These findings were valid for the selected samples that were analyzed as well. Accurate mass measurements were found not to be needed on a routine basis in order to obtain reliable chemical analysis of the target MAs in our sample set. However, this may change if samples are collected from new sites or if the sample matrix is more complex. HPLC Column Selection. Six commercially available HPLC columns were tested for MA selectivity and signal tailing (see Table 3). The Atlantis dC18 column and the carbohydrate column had comparable chromatographic performances. The other columns did not have sufficient MA selectivity. Due to extensive column bleeding, the carbohydrate column was not compatible with the mass spectrometer ion source. In addition to MA Analytical Chemistry, Vol. 77, No. 6, March 15, 2005
1855
Figure 3. Mass chromatograms of the monosaccharide anhydrides. The injection volume is 5 µL, and the standard concentrations are galactosan 176 µg/L, mannosan 165 µg/L, and levoglucosan 77 µg/L. Table 1. Instrumental Performance galac mann levo galacto manno levo m/z 113 m/z 129 m/z 113 m/z 161 m/z 161 m/z 161 precision (RSD),a % LOQb
4
14
6
9
8
4
300
375
255
33
22
46
a Four standard injections, 100 µg/L of each MA, and 7 h between each injection in a sequence. b Absolute amount injected in picogram at S/N 10.
Table 3. Columns That Were Tested for the MA Application manufacturer Waters ACE ACE ACE Waters Waters
stationary phase Xterra MS C18, 3.5 µm C18, 3 µm -A6296 CN, 3 µm -A6297 phenyl, 3 µm -A6298 carbohydrate (amino) 5 µm WATO 44355 Atlantis dC18, 3 µm
dimension
accepted performance?
2.1 mm × 150 mm 2.1 mm × 150 mm 2.1 mm × 150 mm 2.1 mm × 150 mm 4.6 mm × 250 mm
no no no no no
2 x (2.1 mm × 150 mm)
yes
Table 2. Suggested Elemental Composition of Monitoring Ionsa monitoring m/z
measured mass (Da)
suggested elemental comp
deviation from theor mass (Da)
113 129 161
113.0239 129.0177 161.0455
C5H5O3 C5H5O4 C6H9O5
0.0008 -0.0011 0.0005
a
The data in the table are obtained by use of standards.
separation, the Atlantis dC18 column has a great potential for separation of other polar oxygenated multifunctional compounds. The column-generated background noise was low. Solvent Compatibility. During the early stages of method development, unexpected results and day-to-day variations of the instrument response were observed for samples extracted with water and MA compounds dissolved in water. Two possible hypotheses were postulated to explain this: (1) adsorption of MAs to the glassware and (2) undesirable chemical reactions between the MAs and the solvents used. A simple experiment was conducted to test whether mixtures of water and solvents would be less variable than water alone. To 3 mL of a quantification standard containing 400 µg/L of the three MAs in water was added 1856 Analytical Chemistry, Vol. 77, No. 6, March 15, 2005
Figure 4. Results from the solvent compatibility experiments expressed as relative HPLC/HRMS instrument response of a monosaccharide anhydride standard dissolved in 40% of the specified solvent in water. The response of 40% THF in water is defined as 100%. ***, poor chromatography. Variables in order from left to right are m/z 161 galactosan, m/z 129 mannosan, and m/z 113 levoglucosan.
2 mL of the specified solvent. The elution strength was adjusted with water. The results are presented in Figure 4. The total ion current chromatograms obtained in ES(-) do not indicate that new byproducts in the mass range 50-600 Da is the prevailing effect. Total ion chromatograms in ES(+) and atmospheric pressure chemical ionization were not carried out. Thus, focus should be kept on adsorption effects and how to avoid
Table 4. Results of Sample Extraction Experimentsa Performed on a Quartz-Fiber Filter (Q) and a Teflon Filter (T) Exposed at the Same Site and at the Same Time galacb manno levo galacto manno levo m/z 113 m/z 129 m/z 113 m/z 161 m/z 161 m/z 161 Q-THF RSD, % Q-THF:H2O RSD, % Q-DCM RSD, % T-THF RSD, % T-THF:H2O RSD, % T-DCM RSD, %
nd nd nd nd nd nd
124 14 65 13 77 7 184 26 149 21 107 9
721 2 284 13 432 25 879 11 548 23 545 26
9 33 3 21 3 18 38 6 38 17 42 27
130 8 70 0 92 8 177 13 132 16 135 18
686 2 275 6 423 12 896 12 535 9 587 12
a Mean of three parallels (ng m-3). b Below detection limit. Instrument resolution ∼6500.
them. A solvent mixture with 40% THF in water seems to be the most compatible solvent for the MA compounds whereas ordinary HPLC solvents such as MeOH and ACN should be avoided. The consequence of using pure water combined with MAs is severe adsorption to the glassware. This is of special importance with respect to the working definition of WSOCs, of which MAs are expected to be among the most abundant individual identified constituents. Real sample matrix constituents may inhibit some of the adsorption or may even amplify this effect. The net effect is rather unpredictable without comprehensive knowledge about the sample matrix. Extraction Experiments. The extraction procedure was tested on one Teflon (T) filter and one quartz (Q) fiber filter that were exposed at the same site and at the same time. Three parallel samples were prepared for each filter type and solvent combination by using one punch (1 cm2) in each parallel and by extracting the samples as previously described. Based on the solvent compatibility experiments, pure THF and 50% THF in water were selected as solvents. In addition, DCM was selected for comparison with established analytical methods using this solvent. The results are given in Table 4 as the mean of three parallels. The repeatability is good for all three compounds, and the quantitative agreement between the monitoring ions is good as well. Teflon filters provide systematically higher extraction yields for the three isomers than obtained with quartz fiber filters. (See Figure 5.) A comparison of the extraction performed with Q-THF and T-THF indicates that the quartz fiber filter retain about 76, 32, and 18% for galactosan, mannosan, and levoglucosan, respectively, relative to the Teflon filter. A possible explanation may be strong interactions between the MAs and active sites on the quartz fiber filter. A comparison of T-THF versus T-DCM and Q-THF versus Q-DCM indicates that THF is the most efficient extraction solvent. In both cases, the extraction yield of levoglucosan obtained with DCM is only 60% of the amount obtained with THF. The recovery from the extraction procedure was tested by consecutive extraction of an exposed filter three times as described previously. The results from the exposed samples are presented in Figure 6, showing the relative amount extracted in each of three extraction steps. The least abundant isomers,
Figure 5. Results from filter extraction experiments expressed as extraction efficiency relative to T-THF (100%). Q-THF: quartz fiber filter extracted in THF. Q-THF:H2O: quartz fiber filter extracted in 50% THF in water. Q-DCM: quartz fiber filter extracted in DCM. T-THF: Teflon filter extracted in THF. T-THF:H2O: Teflon filter extracted in 50% THF in water. T-DCM: Teflon filter extracted in DCM. The filter media has been exposed at the Elverum site. The data have been normalized. Variables in order from left to right are m/z 161 galactosan, m/z 129 mannosan, and m/z 113 levoglucosan.
Figure 6. Recovery experiments expressed as the relative recovery obtained by use of tetrahydrofuran as extraction solvent in three consecutive extraction steps performed on two filter punches (a and b) from the same quartz fiber filter sampled at the Elverum site. Variables in order from left to right are m/z 161 galactosan, m/z 129 mannosan, and m/z 113 levoglucosan.
galactosan and mannosan, are almost completely removed in the first step whereas the corresponding percentage for levoglucosan is 66%. The large difference between the three extraction steps indicates high extraction efficiency of extractable MAs; however, due to the nonextractable MAs (Figure 5), it is difficult to determine the exact recovery percentage. Two blank filter punches (1 cm2) were spiked with 20 µL each of a THF/water (1:1) solution containing 10 mg/L of the individual MAs. The spiked punches were extracted twice as previously described. The results obtained with spiked blank filters were 99% after extracting the filter twice with THF. The higher recovery obtained with spiked samples at the second extraction step can possibly be explained by a looser adsorption to the filter of spiked MA compounds compared with the case of real sampling and more complex matrix-bonded MA Analytical Chemistry, Vol. 77, No. 6, March 15, 2005
1857
Table 5. Methodological Limit of Quantificationa (in ng m-3) Obtained with the Specified Ion Chromatogram at a Signal/Noise Ratio of 10 filter media high-vol quartz filter (414 cm2); sample vol 828 m3 a
galac mann levo galac mann levo m/z 113 m/z 129 m/z 113 m/z 161 m/z 161 m/z 161 30.0
38.0
26.0
3.0
2.0
5.0
Instrument resolution ∼4500.
compounds. Internal standards compatible with HPLC/HRMS and that are adequate for this application, i.e., isotope-labeled MAs,20 are to our knowledge not commercial available. Hexanetriol15 has been tried in our experiments without success. Based on the findings above, it is recommended to extract the quartz fiber filters in at least two steps to obtain satisfactory recovery of the extractable MAs. The methodological LOQ based on the preceding work description is summarized in Table 5. This limit is sufficient for the most common MA sampling strategies and can easily be improved by evaporation of solvent from the sample extracts or by increasing the sample volume. (20) Fine, P. M.; Chakrabarti, B.; Krudysz, M.; Schauer, J. J.; Sioutas, C. Environ. Sci. Technol. 2004, 38, 1296-1304.
1858
Analytical Chemistry, Vol. 77, No. 6, March 15, 2005
CONCLUSIONS HPLC/HRMS has been proven to be a powerful tool for quantifying MAs in atmospheric aerosols. A 20-50-pg sample of the MAs is required to reach the LOQ, which is sufficient for the most common MA sampling strategies. This LOQ can easily be improved. THF should be preferred as solvent in order to minimize adsorption of MAs to the laboratory glassware and quartz fiber filter surface. THF is also found to be the best extraction solvent. Based on the Elverum samples, the extraction yield of levoglucosan obtained with DCM is only 60% of that obtained with THF regardless of filter type. In the present study, the use of Teflon filters provided higher concentrations of the MAs than the use of quartz fiber filters. An advantage of the method is that sample preparation prior to chemical analysis is limited. It offers a new valuable analytical approach for studies using MAs as tracers for biomass burning. ACKNOWLEDGMENT This work was supported by the Norwegian research council (NFR), VISTA (The Norwegian Academy of Science and Letters and Statoil) project 6143, and the Norwegian Institute for Air research (NILU). The authors are grateful for the reviewers’ constructive and competent feedback on the manuscript. Received for review April 8, 2004. Accepted August 24, 2004. AC049461J
