Anal. Chem. 2006, 78, 2405-2412
A Sensitive Method for the Quantification of Acrolein and Other Volatile Carbonyls in Ambient Air Vincent Y. Seaman, M. Judith Charles, and Thomas M. Cahill*
Department of Environmental Toxicology, University of California, Davis, One Shields Avenue, Davis, California 95616
Acrolein, an unsaturated aldehyde found in both indoor and outdoor air, is considered one of the greatest noncancer health risks of all organic air pollutants. Current methods for determining acrolein often employ sorbentfilled cartridges containing a carbonyl derivatizing agent (e.g., dinitrophenylhydrazine). These methods are of limited use for unsaturated compounds due to the formation of unstable derivatives, coelution of similar compounds, long sample collection times, and ozone interferences that result in poor sensitivity, selectivity, and reproducibility. The goal of this research was to develop an analytical method for determining ppt concentrations of acrolein and other carbonyls in air with short sampling times (10 min). The method uses a mist chamber to collect carbonyls by forming water-soluble carbonylbisulfite adducts. The carbonyls are then liberated from the bisulfite, derivatized, and quantified by gas chromatography/electron capture negative ionization mass spectrometry. The method was applied to determine atmospheric acrolein concentrations at three sites in northern California reflecting hemispheric background concentrations, biogenic-dominated regions, and urban environments. The resulting acrolein concentrations were 0.056, 0.089, and 0.29 µg/m3, respectively, which are all above the EPA Reference Concentration of 0.02 µg/m3. The minimum detection limit of 0.012 µg/m3 is below that of other published methods. Methacrolein, methyl vinyl ketone, crotonaldehyde, glyoxal, methyl glyoxal, and benzaldehyde were also quantified. Acrolein, a highly reactive R,β-unsaturated aldehyde, is a pulmonary toxicant and a common constituent of both indoor and outdoor air.1,2 Acrolein is produced by the incomplete combustion of organic material as well as the oxidation of atmospheric chemicals such as 1,3-butadiene, which is a primary component of motor vehicle exhaust. Indoor sources of acrolein include heated cooking oil, cigarette smoke, incense, candles, and woodburning fireplaces.3,4 Although considered by regulatory agencies * To whom correspondence should be addressed. E-mail:
[email protected]. (1) CICAD. WHO, 2002. (2) U.S. Public Health Service In Agency for Toxic Substances and Disease Registry (ATSDR); U.S. Public Health Service, USPHS, 1990. (3) OEHHA. Chronic Toxicity Summary, Acrolein; 2000; Batch 2A. (4) Ghilarducci, D. P.; Tjeerdema, R. S. Rev. Environ. Contam. Toxicol. 1995, 144, 95-146. 10.1021/ac051947s CCC: $33.50 Published on Web 02/28/2006
© 2006 American Chemical Society
to be one of the most dangerous components of toxic air mixtures,5-7 acrolein is often omitted from studies of carbonyls in the atmosphere8-18 or is reported as “below the limit of detection”.15,19 The current EPA method of determining acrolein (method TO11A) is based on the well-documented reaction between carbonyls and dinitrophenylhydrazine (DNPH), which produces hydrazones that are separated by high-pressure liquid chromatography and detected by UV spectrophotometry (HPLC-UV).20-22 While this method is effective for many aldehydes and ketones, it has not proven reliable for acrolein and other unsaturated carbonyls. Numerous problems inherent in the methodology have been reported, including instability of the DNPH-acrolein hydrazone during collection and storage21,23-28 and poor chromatographic (5) U.S. Environmental Protection Agency. Integrated Risk Information System, Acrolein (CASRN 107-02-8); 2003. (6) Tam, B. N.; Neumann, C. M. J. Environ. Manage. 2004, 73, 131-145. (7) State of California, A. R. B.; California Air Resources Board: Sacramento, 1997; pp 1997-1911-1912. (8) van Leeuwen, S. M.; Hendriksen, L.; Karst, U. J. Chromatogr., A 2004, 1058, 107-112. (9) Pereira, E. A.; Rezende, M. O. O.; Tavares, M. F. J. Sep. Sci. 2004, 27, 28-32. (10) Pereira, E. A.; Carrilho, E.; Tavares, M. F. J. Chromatogr., A 2002, 979, 409-416. (11) Brombacher, S.; Oehme, M.; Dye, C. Anal. Bioanal. Chem. 2002, 372, 622629. (12) Pires, M.; Carvalho, L R. F. Anal. Chim. Acta 1998, 367, 223-231. (13) Grosjean, E.; Grosjean, D. Int. J. Environ. Anal. Chem. 1995, 61, 343-360. (14) Shibamoto, Y. A. J. Chromatogr., A 1994, 672, 261-266. (15) Coutrim, M. X.; Nakamura, L. A.; Collins, C. H. Chromatographia 1993, 37, 185-190. (16) Zhang, J. F.; He, Q. C.; Lioy, P. J. Environ. Sci. Technol. 1994, 28, 146152. (17) Sax, S. N.; Bennett, D. H.; Chillrud, S. N.; Kinney, P. L.; Spengler, J. D. J. Exposure Anal. Environ.Epidemiol. 2004, 14, S95-S109. (18) Bakeas, E. B.; Argyris, D. I.; Siskos, P. A. Chemosphere 2003, 52, 805813. (19) Grosjean, E.; Grosjean, D.; Fraser, M. P.; Cass, G. R. Environ. Sci. Technol. 1996, 30, 2687-2703. (20) Lipari, F.; Swarin, S. J. J. Chromatogr. 1982, 247, 297-306. (21) Tejada, S. B. Int. J. Environ. Anal. Chem. 1986, 26, 167-185. (22) Grosjean, D. Environ. Sci. Technol. 1982, 16, 254-262. (23) Kieber, R. J. a. K. M. Environ. Sci. Technol. 1990, 24, 1477-1481. (24) Goelen, E.; Lambrechts, M.; Geyskens, F. Analyst 1997, 122, 411-419. (25) Schulte-Ladbeck, R.; Lindahl, R.; Levin, J. O.; Karst, U. J. Environ. Monit. 2001, 3, 306-310. (26) Huynh, C. K.; Vu-Duc, T. Anal. Bioanal. Chem. 2002, 372, 654-657. (27) Dong, J. Z.; Moldoveanu, S. C. J. Chromatogr., A 2004, 1027, 25-35. (28) Weisel, C. P.; Zhang, J. F.; Turpin, B. J.; Morandi, M. T.; Colome, S.; Stock, T. H.; Spektor, D. M.; Korn, L.; Winer, A.; Alimokhtari, S.; Kwon, J.; Mohan, K.; Harrington, R.; Giovanetti, R.; Cui, W.; Afshar, M.; Maberti, S.; Shendell, D. J. Exposure Anal. Environ. Epidemiol. 2005, 15, 123-137.
Analytical Chemistry, Vol. 78, No. 7, April 1, 2006 2405
separation of complex carbonyl mixtures typically found in air.15,26,29,30 Since these problems can cause positive and negative biases, acrolein concentrations reported in the literature are variable and the precise values remain controversial. In addition, the long sampling times necessary when using cartridges, typically 4-12 h using flow rates of 0.1-1.0 L‚min-1, make them unsuitable for measuring short-term fluctuations and exposures. A rigorous multilaboratory study comparing various methods found the DNPH method unsuitable for acrolein.24 Progress has been made in resolving these limitations, such as using mass spectrometry instead of UV detection, but the instability of the DNPH-acrolein hydrazone under the conditions necessary for collection from air (acidified DNPH cartridge over 1-4 h) has not been overcome. OSHA method 5231 employs an XAD-2 adsorbent cartridge coated with 2-(hydroxymethyl)piperidine; however the method sensitivity of 3 ppb for an 8-h sample at 0.1 L‚min-1 is not sufficient for ambient acrolein measurements. Dansylhydrazine (DNSH) and 4-hydrazinobenzoic acid have been used to trap carbonyls in cartridges and passive samplers but have not yet provided reproducible values for ambient acrolein levels.9,10,32 The use of other carbonyl derivatizing agents, including pentafluorophenylhydrazine, o-benzylhydroxylamine, n-benzylethanolamine, cysteamine, and n-methyl-4-hydrazino-7-nitrobenzofurazan, has met with limited success due to the need for expensive equipment or reagents, inadequate sensitivity, or poor selectivity.25,30,33-35 The objective of this research was to develop and validate an analytical procedure capable of detecting acrolein and other gaseous carbonyls at low part-per-trillion concentrations with short (10 min) sample collection times. The new method employs a mist chamber, also called a Cofer scrubber, containing a sodium bisulfite solution that traps carbonyls from the air by forming stable, water-soluble sulfonates.36-42 The sulfonates are then dissociated and the free carbonyls derivatized with o-(2,3,4,5,6pentafluorobenzyl)hydroxylamine (PFBHA), forming thermally stable oxime adducts that can be analyzed by gas chromatography/electron capture negative ionization mass spectrometry (GC/ ECNI MS).33,43,44 The potential for positive and negative artifacts arising from ozone and atmospheric precursors, such as isoprene and 1,3-butadiene, was evaluated in the analytical system. The method was then validated in both the laboratory and in the field (29) Dabek-Zlotorzynska, E.; Lai, E. P. C. J. Chromatogr., A 1999, 853, 487496. (30) Otson, R.; Fellin, P.; Tran, Q.; Stoyanoff, R. Analyst 1993, 118, 1253-1259. (31) U.S. Department of Labor, O.S.H.A., 1989, http://www.osha.gov/dts/sltc/ methods/organic/org052/org052.html. (32) Zhang, I.; Zhang, L.; Fan, Z.; Ilacqua, V. Environ. Sci. Technol. 2000, 34, 2601-2607. (33) Ho, S. S. H.; Yu, J. Z. Environ. Sci. Technol. 2004, 38, 862-870. (34) Jain, V. D. T. J. Chromatogr., A 1995, 709, 387-392. (35) Yasuhara, A.; T. Shibamoto J. Chromatogr., A 1994, 672, 261-266. (36) Betterton, E. A.; Hoffmann, M. R. J. Phys. Chem. 1987, 91, 3011-3020. (37) Boyce, S. D.; Hoffmann, M. R. J. Phys. Chem. 1984, 88, 4740-4746. (38) Dufour, J. P.; Leus, M.; Baxter, A. J.; Hayman, A. R. J. Am. Soc. Brew. Chem. 1999, 57, 138-144. (39) Kaneda, H.; Osawa, T.; Koshino, S. J. Agric. Food Chem. 1994, 42, 24282432. (40) Kok, G. L.; Gitlin, S. N.; Lazrus, A. L. J. Geophys. Res.-Atmos. 1986, 91, 2801-2804. (41) Lowinsohn, D.; Bertotti, M. J. Chem. Educ. 2002, 79, 103-105. (42) Olson, T. M.; Hoffmann, M. R. J. Phys. Chem. 1988, 92, 533-540. (43) Yu, J. Z.; Jeffries, H. E.; Lelacheur, R. M. Environ. Sci. Technol. 1995, 29, 1923-1932. (44) Yu, J. Z.; Jeffries, H. E.; Sexton, K. G. Atmos. Environ. 1997, 31, 22612280.
2406 Analytical Chemistry, Vol. 78, No. 7, April 1, 2006
Figure 1. Schematic diagram for the mist chamber used for air sampling. The all-glass chambers can be stacked in series to determine collection efficiency and will allow air flow rates of up to 20 L‚min-1.
using labeled acrolein (acrolein-d4) as a matrix spike (field positive control) prior to sample collection. The sensitivity and precision of the new methodology were then determined under field conditions to give the most representative estimates of method performance. EXPERIMENTAL METHODS Reagents. Capillary GC/GC/MS grade methanol and hexane were acquired from Burdick & Jackson (Muskegon, MI). HPLC grade water and sodium sulfite were purchased from Fisher Scientific (Fairlawn, NJ). Carbonyl standards (acrolein, methacrolein, methyl vinyl ketone, crotonaldehyde, glyoxal, methyl glyoxal, benzaldehyde, benzaldehyde-d6), PFBHA, and hydrogen peroxide (30%) were purchased from Sigma/Aldrich (St. Louis, MO). Labeled acrolein (acrolein-d4) was synthesized by Cambridge Isotopes (Andover, MA). Standard solutions (10 ng/µL in methanol) were prepared for the carbonyls and two labeled compounds (acrolein-d4, benzaldehyde-d6). A mixture of the labeled carbonyls was prepared (10 ng/µL of each) for use as a matrix spike. The PFBHA stock solution was 50 mM (12.5 mg/mL in methanol). The aqueous 0.1 M sodium bisulfite was prepared by mixing 12.6 g of sodium sulfite in 1000 mL of water, adjusting the pH to 5.0 with 1.0 M H2SO4, and allowing the solution equilibrate for 7 days sealed at room temperature. Collection and Derivatization. Carbonyls were collected by drawing air through a glass mist chamber (Figure 1) containing a sodium bisulfite solution. The mist chamber is clamped to a ring stand in an upright position with a vacuum tube, connected to a small pump, attached to the top. The air enters the mist
chamber through a small orifice at the bottom, which opens to a glass tube. This tube surrounds a smaller tube connected to the bisulfite reservoir below. The air flow creates a negative pressure differential that draws the bisulfite solution up into the air stream. The result is a fine mist, which provides sufficient contact between carbonyls in the air and bisulfite ions to ensure adduct formation. A baffle in the top of the chamber keeps >95% of the mist within the chamber during a 10-min sampling period at 20 L‚min-1. An average of 9.7 ( 0.21 mL of solution was retained from an initial volume of 10.0 mL in a single mist chamber (n ) 10). Since two mist chambers are connected in series, any losses from the first chamber will enter the second chamber. A 3-µm Teflon filter is placed between the second mist chamber and vacuum source to block any mist droplets from escaping the system. Each mist chamber has a characteristic maximum flow rate, ranging from 12 to 25 L‚min-1, based on the orifice size, geometry, and vacuum strength. When used in series, each pair of mist chambers also has a characteristic maximum flow rate that ranges from 10 to 22 L‚min-1. These rates were determined for all mist chambers, both singly and in pairs, using a calibrated DryCal Lite Primary flow meter (Bios International Corp., Butler, NJ), eliminating the need for a flow controller in the field. After sample collection, hydrogen peroxide (1.04 mmol, a slight stoichiometric excess compared to bisulfite) was added to the collection solution to remove free bisulfite, and then PFBHA (1 mM final concentration) was added to the solution. The solution was acidified with 0.2 mL of 18 M H2SO4 to encourage dissociation of the carbonyl-bisulfite adducts and to ionize excess PFBHA and allowed to react at room temperature for 24-96 h. The PFBHA derivatives were then extracted twice with 5 mL of hexane, dried over sodium sulfate, and the volume reduced to 0.5 mL with nitrogen evaporation. An injection standard consisting of PFBHAderivatized acetone-d6 was added to the samples prior to analysis. Instrumental Analysis. The analysis of the PFBHA-carbonyl oximes was conducted using an Agilent 6890N gas chromatograph coupled to a 5793N quadrupole mass spectrometer. A 30-m DBXLB capillary column was used for separating the derivatives (0.25mm i.d., 0.25-µm film thickness; J&W Scientific, Folsom, CA). The initial oven temperature was set to 50 °C, which was held for 2 min and then ramped at 5 °C/min to 150 °C, 20 °C/min to 260 °C, 30 °C/min to 325 °C, and held for 5 min. The detector was operated in the negative chemical ionization mode (m/z ) 50500), which affords the highest sensitivity with fluorinated compounds. The source temperature was constant at 150 °C, and the reagent gas was methane (40%). Total ion chromatograms were used for both characterization and quantification. Quantification was performed using characteristic ions that occur in the ECNI mass spectra of PFBHAoximes. Most aldehydes and ketones, including acrolein, lose an HF fragment and have a unique, characteristic ion at [M - 20]-. For dicarbonyls and hydroxycarbonyls, the characteristic ions are [M - 181]- and [M - 51]-, respectively. A number of common pentafluorobenzyl-oxime fragments (m/z ) 178, 181, 196, 197) are also present and can be used to quickly identify the presence of an aldehyde or ketone functional group. Carbonyl-Bisulfite Adduct Formation. Previous studies demonstrated that water-soluble, nonvolatile compounds such as glyoxal and methyl glyoxal were effectively trapped in a mist
chamber containing an aqueous PFBHA solution.45 Since the reactions of PFBHA with carbonyls are relatively slow (2-24 h), more volatile compounds with Henry’s law constants less than 103 M‚atm-1 were not retained. The use of bisulfite instead of water/PFBHA in the mist chamber allowed more volatile species, including acrolein, to be trapped by forming water-soluble sulfonic acid adducts with the carbonyls. The solution was adjusted to a pH of 5.0 since the bisulfite ion (HSO3-) concentration is greatest at this pH. The reversible formation of the monosulfonate occurs rapidly on the order of seconds to minutes. Bisulfite will also add irreversibly to the double bond in unsaturated compounds, but this is a much slower reaction that only becomes important (>10% disulfonate formation) after 24 h:38
The stability of the carbonyl-bisulfite adducts was measured over a period of 30 days to determine the maximum acceptable time interval between sample collection and derivatization. PFBHA Derivatization. The optimum conditions for the carbonyl derivatization with PFBHA were determined for six aldehydes and ketones: acrolein, methacrolein, methyl vinyl ketone, crotonaldehyde, glyoxal, and methyl glyoxal. Since the presence of bisulfite severely inhibited the PFBHA derivatization of the carbonyls, hydrogen peroxide was added to oxidize the free bisulfite to sulfate. Four different molar ratios of peroxide to bisulfite were evaluated to determine the optimum peroxide concentration. In addition, the carbonyls were also mixed with peroxide (1.04 and 2.08 mmol) in the absence of bisulfite to ensure that carbonyls would not be oxidized. Last, peroxide was added to a vial containing bisulfite, isoprene, and 1,3-butadiene at concentrations that were 1000-fold greater than normally expected in the environment to ensure that peroxide did not oxidize these precursor compounds into carbonyls, thus creating positive artifacts in the sampling system. The second parameter optimized was the concentration of PFBHA used for derivatization. Four different PFBHA concentrations were evaluated for their ability to derivatize carbonyls in a 0.1 M bisulfite solution over 24 h with a slight molar excess of peroxide. The last derivatization parameter optimized was the derivatization time. Four sets of samples (n ) 3 for each condition) were prepared in a 0.1 M bisulfite solution with a slight molar excess of peroxide and were derivatized for 1, 2, 4, and 7 days. Mist Chamber Collection Efficiency. The collection efficiency (CE) of the bisulfite solution in the mist chamber was initially determined in the laboratory for the six carbonyls using a Tedlar bag containing ∼1 µg/m3 of each compound. The bag was filled by passing 200 L of zero grade air at 5 L/min-1 through a U-tube wrapped with heating tape and containing 200 ng of each carbonyl in methanol. The U-tube was held at room temperature (21 °C) for 5 min and gradually heated to 100 °C over the next 30 min. The contents of the bag were then drawn through two mist chambers in series, each containing 10 mL of 0.1 M bisulfite solution and 10 µL of the labeled matrix spike, at a rate of 20 L‚min-1 for 10 min. The resultant concentrations of the carbonyls Analytical Chemistry, Vol. 78, No. 7, April 1, 2006
2407
were determined in each mist chamber and the collection efficiency was calculated by
CE ) 1 - [B]/[A]
(1)
where [A] and [B] are the concentrations of a specific compound in mist chambers A and B, respectively. It is important to note that this formula for CE assumes that each mist chamber removes a fixed fraction of the analyte from the incoming air, regardless of the concentration. Ozone/Precursor Interferences. The presence of ozone and common atmospheric precursors, such as 1,3-butadiene and isoprene, are potential sources of positive and negative artifacts when sampling for carbonyls in air.46,47 Positive artifacts can occur when ozone oxidizes precursor species present to produce carbonyls, thus creating artificially high concentrations of the carbonyls. Negative artifacts arise from the direct oxidation of carbonyls by ozone, resulting in lower concentrations than are actually present. The use of an ozone scrubber, which is necessary in methods using hydrazine derivatizing agents (DNPH, DNSH), not only increases the cost but may also affect the collection of the sample.12,48,49 The effects of ozone and atmospheric precursors (1,3-butadiene and isoprene) on the bisulfite/mist chamber system were evaluated by passing 200 L of zero grade air or ozonated air (100 ppb) through a mist chamber containing 10 mL of 0.1 M bisulfite (blank), bisulfite + carbonyls (250 ng of each carbonyl), or bisulfite + precursors (500 µg/m3) at a rate of 20 L‚min-1 (n ) 3 for each condition). Isoprene and 1,3-butadiene were chosen as precursors since they are considered to be the most abundant naturally emitted and anthropogenic substances that react in the atmosphere to produce carbonyl species, including acrolein, methacrolein, and methyl vinyl ketone. Labeled acrolein and benzaldehyde were added to the mist chamber collection solution prior to each test. Ozone was added to the air using an ozone generator (Jelight model 600), and the ozone concentration was monitored continuously using an in-line ozone monitor (Dasibi model 1008-AH). Field Testing/Method Validation. The method was fieldtested by determining atmospheric carbonyl concentrations in three locations. The first was Salt Point, which represented clean marine-boundary layer air on the northern California coast with no terrestrial sources of carbonyls. The second site was Juniper Lake in Lassen Volcanic National Park, a remote coniferous forest area in northern California that served as the control site with biogenic sources but little or no anthropogenic influence. Longterm aerosol records indicate that Lassen Volcanic National Park is one of the cleanest areas in the 48 contiguous United States.50 (45) Spaulding, R. S.; Talbot, R. W.; Charles, M. J. Environ. Sci. Technol. 2002, 36, 1798-1808. (46) Grosjean, D.; Grosjean, E.; Williams, E. L. Environ. Sci. Technol. 1994, 28, 186-196. (47) Grosjean, D.; Williams, E. L.; Grosjean, E. Environ. Sci. Technol. 1993, 27, 830-840. (48) Rodler, D. R.; Nondek, L.; Birks, J. W. Environ. Sci. Technol. 1993, 27, 2814-2820. (49) Kleindienst, T. E.; Corse, E. W.; Blanchard, F. T.; Lonneman, W. A. Environ. Sci. Technol. 1998, 32, 124-130. (50) Malm, W. C.; Sisler, J. F.; Huffman, D.; Eldred, R. A.; Cahill, T.A. J. Geophys. Res.-Atmos. 1994, 99, 1347-1370.
2408
Analytical Chemistry, Vol. 78, No. 7, April 1, 2006
The last site was Roseville, a suburban city impacted by a major railroad depot and high motor vehicle traffic, located 15 miles northeast and normally downwind of Sacramento. Meteorological conditions for the three sites were seasonally typical during sample collection. Two or three sets of samplers, each of which consisted of two mist chambers in series, were used to collect air between 11 a.m. and 2 p.m. on separate days in August and September 2005. Ten minutes prior to collection, the mist chamber solutions (10 mL of 0.1 M bisulfite for each chamber) were spiked with 100 ng of acrolein-d4 and benzaldehyde-d6. The air was then drawn through the mist chambers for 10 min using a vacuum pump (VP0660, Medo) using either line power (Roseville) or a heavyduty deep-cycle battery with a power inverter (Salt Point, Lassen). The air flow for each set of mist chambers was determined prior to each sampling event as described earlier using identical pumps and power supplies. The flow rates for the mist chamber sets ranged from 13 to 19 L‚min-1. After the 10-min sampling period, each mist chamber solution was emptied into separate 50-mL glass screw-cap test tubes (“reaction” tubes) containing 100 µL of 30% hydrogen peroxide, 200 µL of 50 mM PFBHA, 1 mL of 1.8 M H2SO4, and 2.5 mL of hexane. The mist chambers were then rinsed twice with 3 mL of HPLC grade water and the rinses added to the reaction tubes, which were capped, gently shaken, and stored upright in a closed box at ambient temperature. Two field blanks were prepared at each site using the same process, including adding the collection solution to the mist chambers, but the vacuum pumps were not running so no air flowed through the chambers. Two reagent blanks were prepared at each site by adding 10 µL of labeled matrix spike to 10 mL of bisulfite solution, waiting 10 min, and transferring the solution to a reaction tube, which was subsequently handled and stored in the same manner as the samples. A six-point calibration curve was prepared in the field during each sampling event using ampules containing 0, 1.25, 5.0, 25.0, 50.0, and 125.0 ng of the carbonyl standards in 0.5 mL of methanol. The contents of each ampule were added to separate vials containing 10 mL of 0.1 M sodium bisulfite, which were then spiked with the labeled matrix spike, allowed to react for 10 min, and added to reaction tubes in the same manner as the samples and blanks. The calibration curve was stored with the samples; thus, it was exposed to the same conditions and time period as the samples. The samples, blanks, and calibration curve were allowed to react for 96 h, after which they were shaken for 30 s, the hexane layer was removed, and the samples were re-extracted twice with 2.5 mL of hexane. The hexane extracts were passed through a glass column containing anhydrous sodium sulfate to remove any residual water, and each extract was reduced to a volume of 0.5 mL under nitrogen evaporation. The extract concentrates were then analyzed by GC/ECNI MS. Collection efficiencies and atmospheric concentrations were determined for acrolein and selected aldehydes and ketones using acrolein-d4 as the internal standard for the unsaturated compounds and benzaldehyde-d6 as the internal standard for benzaldehyde and the dicarbonyls. The minimum detectable limit (MDL) was calculated by taking the average field blank plus 3 standard deviations of the field blanks for each site.
Figure 2. Loss (expressed as the natural logarithm of the change of concentration) of acrolein, crotonaldehyde, methyl glyoxal, and methyl vinyl ketone as a function of storage time in a 0.1 M bisulfite solution (n ) 3). The samples were stored at room temperature in the dark. The results show that unsaturated compounds eventually react irreversibly with bisulfite, but the reaction with methyl vinyl ketone occurs much faster. Methyl glyoxal, a saturated dicarbonyl, showed no decline in response during 30 days. The calculated half-lives of methyl vinyl ketone, methacrolein, crotonaldehyde and acrolein were 1.1, 12.3, 12.5, and 13.6 days, respectively.
Figure 3. Influence of peroxide concentration on the PFBHA derivatization of acrolein in water, bisulfite, and hydrogen peroxide/ bisulfite solutions (n ) 3). The presence of bisulfite caused a >90% reduction in the aqueous PFBHA derivatization yield. The addition of hydrogen peroxide (1.04:1 or 2.08:1 molar ratio of peroxide to bisulfite) resulted in derivatization yields 90-95% of the aqueous yield. Higher peroxide concentrations resulted in lower yields.
RESULTS AND DISCUSSION Carbonyl-Bisulfite Adduct Formation. The adducts of the unsaturated compounds declined over the 30 days of storage (Figure 2), which was most likely due to the irreversible formation of disulfonates. The stability of the acrolein, methacrolein, and crotonaldehyde adducts were similar (t1/2 ∼ 12.5 days) while the methyl vinyl ketone adducts disappeared rapidly (t1/2 ∼ 1.1 days). The adducts of the saturated dicarbonyls, methyl glyoxal and glyoxal, showed no losses over the 30-day period. To provide consistency and to reduce potential loses of the unsaturated species, presumably through disulfonate adduct formation, future samples were derivatized in the field immediately upon sample collection. PFBHA Derivatization. The optimal derivatization conditions were determined by testing different reagent concentrations and reaction times. The results from the peroxide optimization test
Figure 4. Effect of PFBHA concentration on carbonyl derivatization yield for a 24-h reaction period (n ) 3). Concentrations of PFBHA greater than 1 mM provided no statistically significant increases in derivatization yield (t-test, p > 0.05 for all compounds).
Figure 5. PFBHA derivatization reaction kinetics (n ) 3). The PFBHA derivatization of the unsaturated, monocarbonyls was complete after 24 h and gradually declined over the next 6 days. The saturated dicarbonyls (glyoxal and methyl glyoxal) required 2-4 days for maximum derivatization due to the need for both carbonyls to be derivatized.
(Figure 3) showed that a slight molar excess of peroxide to bisulfite (1.04:1) resulted in maximum derivatization. Molar ratios of 5:1 (peroxide/bisulfite) or more had negative effects on the analyte response, presumably through oxidation of the derivatized analytes or interference with the PFBHA derivatization reaction. The free carbonyls were unaffected by the lower concentration of peroxide since carbonyls added to the peroxide solution (not shown) gave the same responses as carbonyls in water. The addition of the precursor compounds of 1,3-butadiene and isoprene to the peroxide solution did not generate detectable carbonyl formation even at concentrations that were 1000-fold higher than expected in the environment. The minimum concentration of PFBHA required for maximum derivatization yield was determined to be 1 mM (Figure 4). Lower concentrations were less efficient in derivatizing the dicarbonyl species, while higher concentrations yielded no statistically significant increases in derivatization (t-test, p > 0.05 for all compounds) and gave larger reagent peaks in the chromatograms. Based on these results and the relatively high cost of PFBHA ($100/g), the 1 mM PFBHA solution was determined to be the optimum concentration. Analytical Chemistry, Vol. 78, No. 7, April 1, 2006
2409
Table 1. Mist Chamber Collection Efficiencies (CE) for Selected Carbonyls under Laboratory and Field Conditionsa
Figure 6. Effect of bisulfite concentration on collection efficiency (n ) 3). Collection efficiencies of the unsaturated carbonyls were minimal in water and increased with increasing bisulfite concentration. The collection efficiencies of the saturated dicarbonyls, glyoxal and methyl glyoxal, were independent of the bisulfite concentration.
The derivatization reaction was complete within 48-168 h, depending on the carbonyl species (Figure 5). There was no derivatization time that was ideal for all compounds since the unsaturated carbonyls were completely derivatized after 24 h while the dicarbonyls of glyoxal and methyl glyoxal required up to 7 days for complete derivatization. The dicarbonyls likely required longer reaction times since two carbonyl groups must be derivatized for these compounds. The unsaturated carbonyl abundances declined during a seven-day derivatization, possibly due to bisulfite addition to the double bond or reactions with residual peroxide. Due to the variable reaction time of the carbonyls, a calibration curve should be derivatized at the same time as the samples when quantifying carbonyls other than acrolein or benzaldehyde. The use of deuterated acrolein and benzaldehyde as internal standards accounts for any variations in derivatization yield for these compounds. Once the derivatives have been extracted and stored in anhydrous hexane, they appear to be stable for extended periods of time. No losses were observed when a mixture of derivatized and extracted acrolein (200 pg/µL) was stored at 4 °C and analyzed twice monthly for 6 months (average 200 ( 14, relative standard deviation (RSD) ) 7.0%, n ) 12). Collection Efficiency. The CE of the mist chamber was evaluated for water and four different bisulfite concentrations. The results showed that a bisulfite concentration of 0.1 M was necessary for maximum trapping of acrolein and other unsaturated aldehydes in the mist chamber (Figure 6). Presumably, the higher concentration of bisulfite results in more rapid adduct formation, which makes the carbonyls effectively nonvolatile and more likely to be retained in the collection solution. As expected, the CEs of the less volatile compounds, glyoxal and methyl glyoxal, were independent of the bisulfite concentration, since they were effectively collected in an aqueous PFBHA solution. With the exception of methyl vinyl ketone (CE ) 0.65), the CEs for all the compounds using 0.1 M bisulfite were greater than 0.80 (Table 1). When two mist chambers are placed in series, 96% of a carbonyl with a CE of 0.8 can be collected since 80% of the carbonyl is trapped in the first mist chamber and 80% of the remaining 20% (or 16%) is trapped in the second chamber. The collection efficiencies for the field samples were comparable to those obtained in the laboratory, even though the concentrations of the 2410 Analytical Chemistry, Vol. 78, No. 7, April 1, 2006
carbonyl
Tedlar bag (n ) 3)
Lassen (n ) 9)
Roseville (n ) 6)
acrolein methacrolein methyl vinyl ketone crotonaldehyde glyoxal methylglyoxal benzaldehyde
0.85 ( 0.04 0.82 ( 0.10 0.65 ( 0.06 0.94 0.86 ( 0.11 0.89 ( 0.05 n/a
0.72 ( 0.06 < MQL < MQL 0.79 ( 0.06 1.0 ( 0.12 0.76 ( 0.10 1.0 ( 0.17
0.70 ( 0.07 < MQL 0.67 ( 0.06 0.70 ( 0.15 < MQL 0.71 ( 0.06 0.82 ( 0.08
a If one or both mist chamber concentrations were below the minimum quantification level (< MQL), then the CE could not be determined. Collection efficiencies are not presented for Salt Point since the compounds were below the minimum quantifiable limit in the second mist chamber.
Figure 7. Effect of ozone and atmospheric precursors on mist chamber carbonyls (n ) 3). There were no differences in mist chamber carbonyl concentrations when either zero grade air or ozonated air (100 ppb) was passed through mist chambers containing bisulfite only (blank), bisulfite + 136 µg of isoprene and 100 µg of 1,3-butadiene (precursors), and bisulfite + 250 ng of carbonyls (carbonyls).
carbonyls were considerably lower. During the 0.1 M bisulfite tests (n ) 3), we recovered an average of 186 ( 31 ng of acrolein between both mist chambers, which corresponds to 93 ( 16% of the enrichment amount of 200 ng. This recovery value agrees well with the predicted collection efficiency for two mist chambers in series. Ozone and Atmospheric Precursors. The resistance of the sampling system to ozone was demonstrated by passing zero grade air and ozonated air (100 ppb) through mist chambers containing bisulfite only (blank), bisulfite + precursors, and bisulfite + carbonyls. There were no carbonyls generated in the bisulfite-only or the bisulfite + precursor chambers, thus demonstrating that ozone produced no positive artifacts from interactions with precursor compounds. In addition, no apparent differences were seen between using zero grade air or ozonated air in the chambers containing carbonyls, thus showing that ozone did not degrade the carbonyls present in solution during sample collection (Figure 7). The bisulfite/mist chamber system is inherently resistant to interferences from ozone for two reasons: (1) bisulfite is rapidly
Table 2. Selected Carbonyl Concentrations and Minimum Detection Levels (MDLs) in Ambient Air from Pristine and Urban Areasa
a
carbonyl
MDL (range)
Salt Point (n ) 6)
Lassen (n ) 9)
Roseville (n ) 6)
acrolein methacrolein methyl vinyl ketone crotonaldehyde glyoxal methylglyoxal benzaldehyde
0.012-0.035 0.007-0.033 0.038-0.050 0.009-0.048 0.050-0.216 0.029-0.038 0.020-0.044
0.056 ( 0.011
