Characterization of Secondary Aerosol from the Photooxidation of

Environ. Sci. Technol. 2001, 35, 3626-3639

Characterization of Secondary Aerosol from the Photooxidation of Toluene in the Presence of NOx and 1-Propene MYOSEON JANG* AND RICHARD M. KAMENS Department of Environmental Sciences and Engineering, CB# 7400, Rosenau Hall, The University of North Carolina at Chapel Hill, Chapel Hill, North Carolina 27599

Secondary organic aerosol (SOA) from the photooxidation of toluene in a hydrocarbon-NOx mixture was generated in a 190 m3 outdoor Teflon chamber. The photooxidation reaction of toluene in the gas phase leads to substituted aromatics (TOL-AR), nonaromatic ring retaining (TOL-R), and ring opening products (TOL-RO). In this work, the following ring opening oxycarboxylic acids were newly identified: glyoxylic acid, methylglyoxylic acid, 4-oxo-2-butenoic acid, oxo-C5-alkenoic acids, dioxopentenoic acids, oxo-C7alkadienoic acids, dioxo-C6-alkenoic acids, hydroxydioxoC7-alkenoic acids, and hydroxytrioxo-C6-alkanoic acids. The newly characterized TOL-R and TOL-RO products included methylcyclohexenetriones, hydroxymethylcyclohexentriones, 2-hydroxy-3-penten-1,5-dial, hydroxyoxo-C6-alkenals, hydroxy-C5-triones, hydroxydioxo-C7-alkenals, and hydroxyC6-tetranones. Products in both the gas and aerosol phases were derivatized with O-(2,3,4,5,6-pentafluorobenzyl)hydroxylamine hydrochloride (PFBHA) for carbonyls and pentafluorobenzyl bromide (PFBBr) for carboxylic acid and phenol groups and analyzed using a gas chromatograph/ mass spectrometry (GC/MS) in an electron impact mode (EI) and a gas chromatograph/ion trap mass spectrometry (GC/ ITMS) in both chemical impact and EI modes. To confirm different isomers, the PFBHA-derivatives of products were rederivatized by silylation using N,O-bis(trimethylsilyl)trifluoroacetamide (BSTFA). The Fourier transform infrared spectroscope (FTIR) was used to obtain additional functional group information for SOA products impacted on a zinc selenide FTIR disk. The major SOA products under the high NOx conditions of the above experiment included methylnitrophenols, methyldinitrophenols, methylbenzoquinones, methylcyclohexenetriones, 4-oxo-2-butenoic acid, oxo-C5-alkenoic acids, hydroxy-C3-diones, hydroxyoxo-C5alkenals, hydroxyoxo-C6-alkenals, and hydroxydioxo-C7alkenals. Of the major SOA products, the experimental partitioning coefficients (iKp) of aldehyde products were much higher and deviated more from predicted iKp values. This is an extremely important result, because it shows that aldehyde products can further react through heterogeneous processes, which may be a very significant SOA generation mechanism from the oxidation of aromatics in the atmosphere. 3626

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 18, 2001

Introduction Interest in secondary organic aerosol (SOA) formation in the atmosphere has been renewed because of its possible impacts on the radiative balance associated with climate change (1, 2), visibility degradation (3, 4), and health effects (5). Possible candidates for SOA formation include the terpene system from terrestrial vegetation (6) and aromatics from anthropogenic sources. The atmospheric oxidation reactions of these organic compounds can produce either nitrated (710) or nonnitrated oxygenated organic compounds which can result in SOA formation through either a self-nucleation process or the gas/particle partitioning on preexisting particulate matter (11-17). Volatile aromatic compounds comprise up to 44% of the urban volatile hydrocarbon mixture in various parts of the world (18). Of the aromatics, benzene, toluene, xylenes, ethylbenzene, and 1,2,4-trimethylbenzene make up 60-75% of this load, with toluene being one of the most significant compounds. While the primary reaction of aromatics with OH radicals has been addressed in many studies (9, 19-23), the further reaction mechanisms that lead to multigeneration products are ambiguous and need to be established. Only a few studies have attempted to identify aromatic SOA composition. Yu et al. (24, 25) identified the gas-phase carbonyl products from the oxidation of alkylbenzenes including toluene in the presence of nitrogen oxides (NOx) using the O-(2,3,4,5,6-pentafluorobenzyl) hydroxylamine hydrochloride (PFBHA) derivatization method. Smith et al. (26, 27) also examined the OH-initiated degradation products of toluene in the presence of NOx. Forstner et al. (10) generated SOA from alkylbenzenes-NOx systems in an outdoor smog chamber and reported the molecular compositions of SOA sampled on quartz fiber filters. For the toluene-NOx system, 3-methyl-2,5-furandione and dihydro2,5-furandione composed 46% of the identified SOA composition (10). If one estimates the vapor pressures and associated partitioning of these compounds, less than 0.02% of these compounds would be expected in the aerosol phase at 300 µg/m3 of fine TSP at 298 K, which seems to contradict Forstner’s data. This discrepancy may be explained by positive gas-phase sampling artifacts on quartz filters (28). In another more recent aromatic chamber oxidation study, Kleindienst et al. (29) reported many of the same gas-phase products observed in the above study. The important message is that either there is a lack of information on particle phase products that result from the oxidation of aromatics or some of the SOA composition information appears contradictory. A few studies have recently attempted to model SOA formation yields from the oxidation or photooxidation of terpenes or aromatics (11-16, 30-32). What was clearly lacking in those studies were detailed speciations of the products that result in SOA formation and time series concentration in the gas and aerosol phases. Without this information, it is very difficult to further develop realistic predictive models with experimental data. In this study, a characterization of the gas and particle photooxidation products from a toluene/NOx/1-propene photochemical reaction was undertaken. Toluene was selected, because it is one of the most ubiquitous aromatics and is regarded as a significant source to the anthropogenic SOA formation (33). The objectives of this study were to (1) describe as many new reaction products as possible from the photooxidation of * Corresponding author phone: (919)966-3861; fax: (919)966-7911; e-mail: [email protected]. 10.1021/es010676+ CCC: $20.00

 2001 American Chemical Society Published on Web 08/18/2001

TABLE 1. Conditions for the Outdoor 190 m3 Smog Chamber Experiments To Produce SOA from the Toluene/NOx/1-Propene Photochemical Reaction System sampling system

date mm/dd/ yyyy

reaction timea (h)

T (K)

RH (%)

duration (min)

sampling vol (m3)

∆Tolb (ppm)

TSP (µg m-3)

fomc

SOA yieldd (%)

filter-filter-denuder filter-filter-denuder filter-filter-denudere filter-filter-denuder impaction on glass platee impaction on ZeSn FTIR disk

05/26/2000 05/26/2000 05/26/2000 05/26/2000 05/26/2000 05/26/2000

1.78 2.63 3.26 4.10 2.55 2.55

302.0 303.6 303.4 304.0 303.0 303.0

38 34 34 34 34 34

20 20 20 20 100 100

0.45 0.45 0.45 0.45 1.54 1.74

2.82 3.47 3.55 3.62

417 1830 1670 1530

0.61 0.91 0.88 0.91

1.72 9.14 7.92 7.26

a Sampling mid-time after injection of toluene gas into the outdoor chamber. b ∆Tol ) amounts of reacted toluene (corrected by gas leaking rate using SF6 tracer gas in the outdoor chamber). c Extractable organic fraction of the front filter sample. d SOA formation yield % ) 100 (SOA total mass)/(amounts of reacted toluene). e These samples were analyzed for the SOA products.

toluene in the presence of NOx, (2) permit the development of a kinetic mechanism and model to explain the formation of these reaction products, and (3) obtain information on the gas/particle phase distribution of the products.

Experimental Section Materials and Instruments. PFBHA, pentafluorobenzylbromide (PFBBr), decafluorobibenzyl (internal standard for derivatization), and N,O-bis(trimethylsilyl)trifluoroacetamide (BSTFA) were purchased from Aldrich (Milwaukee, WI). Semivolatile carbonyl products from the photooxidation reaction of a toluene/NOx/1-propene mixture were derivatized with PFBHA (17, 25, 25, 34), and carboxylic acids and phenol groups were derivatized with PFBBr. BSTFA was also used as a derivatizing agent (17, 35) for carboxylic acids and hydroxyl groups. To identify molecular structures of derivatives, a Varian Saturn II gas chromatograph/ion trap mass spectrometry (GC/ITMS) was used. The GC separation of products was performed on a 30 m × 0.25 mm i.d. J&W DB-5 fused-silica capillary column (0.25 µm film thickness). The GC temperature program was 60-280 °C at 8 °C/min and held for 0.5 min at 280 °C. ITMS employed both CH4-chemical impact (CH4-CI) and electron impact (EI) modes over a mass of 50 to 650 amu, with a scan time of 3 scan/s. The products which were not derivatized by PFBHA, PFBBr, and BSTFA were detected with a Hewlett-Packard 5890 gas chromatograph (30 m, 0.25 mm i.d., J&W DB-5 with a 0.25 µm film thickness) interfaced to a 5971 mass selective detector. The temperature program was 60 °C for 1 min, 60-290 °C at 10 °C/min, and held for 1 min at 290 °C. The additional functional group analysis for SOA products was performed using a Fourier transform infrared spectroscope (FTIR) (Nicolet Magma 560, Nicolet) with a deuterated triglycine sulfate (DTGS) detector. SOAs were collected directly on an ungreased zinc selenide (ZnSe) disk (25 mm in diameter) by impaction. The scan number and resolution for FTIR were 8 and 2 cm-1. Outdoor Chamber Sampling. The gas- and particle-phase samples for this study were generated in a large outdoor 190 m3 Teflon film chamber (15, 16, 36) under sunlight on May 26, 2000. The experimental conditions (temperature, T, and relative humidity, RH) and SOA yields during the course of the reactions are shown in Table 1, along with the total suspended particulate matter (TSP), the mass fraction of the absorptive liquidlike material (fom) which is accounted from extractable organic mass fraction by organic solvent (dichloromethane), and reacted toluene amounts (∆Tol). The initial concentrations of toluene, 1-propene, and NOx were 14.8, 5.5, and 4.4 ppm, respectively. 1-Propene was added to increase the overall reaction rate. The NO, NO2, and O3 concentrations for the duration of the outdoor chamber experiment are shown in Figure I in Supporting Information. Gas and particle-phase samples were simultaneously col-

FIGURE 1. FTIR spectrum for secondary aerosols impacted on the ungreased ZeSe FTIR window. lected using a sampling train that consisted of two (front and back) 47 mm Teflon inpregnated glass fiber filters (type T60A20, Pallflex Products Corp., Putnam, CT), followed by a downstream 5-channel annular denuder. The detailed sampling, workup, and quantitative analysis procedures have been reported previously (16, 36, 37). Prior to the addition of NOx and hydrocarbons, ammonium sulfate [(NH4)2SO4] seed aerosol was added to the chamber. It was generated using a nebulizer (Model 3075 Atomizer, TSI, MN) to aspirate aqueous salt solution (0.0067 M) into the chamber. Approximately 72 µg/m3 of seed aerosol was added to the chamber, and 98% of the initial seed particle number, as calculated from the experimentally measured particle distribution, was associated with particles less than 0.1 µm in diameter. Ozone was measured by a Bendix chemiluminescent ozone meter (Model 8002, Roncerverte, WV). NO and NOx were measured using a chemiluminescent analyzer (Monitor Labs, model 8440E, San Diego, CA). The precision associated with an individual NOx and O3 calibration was ( 5%. Gas-phase concentrations of toluene were monitored using a gas chromatograph (Shimadzu Model 8A, stainless steel column packed with Supelco 5% Bentanone 34, 1.5 m, 3.2 mm i.d) associated with a flame ionization detector and calibrated with a commercially prepared NIST traceable gas mixture. Product Characterization. Twenty microliters of decafluorobibenzyl solution (1 mg/mL in dichloromethane), as an internal standard for the derivatization procedure, was added to 1 mL GC vials containing 200 µL of gas and particle sample extracts in dichloromethane. The vials were then diluted with a mixture of hexane, dichloromethane, and acetonitrile (2:4:4 in vol./vol/vol.). The contents of this vial were then split into three parts: half was used for the carbonyl analysis, a quarter for carboxylic acid analysis, and the remaining quarter was archived. For carbonyl analysis, vials containing extracts from the front and back filter samples and impactor samples were VOL. 35, NO. 18, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

3627

TABLE 2. Molecular Structures of Aromatic Ring Retaining Products from Photooxidation of Toluene in the Presence of NOx/1-Propenea

a G: gas-phase sample. P1: front filter sample. P2: back filter sample. I: impactor sample. m: major component. s: small amounts as a minor component. t: trace amounts. -: not found.

gently evaporated under a dry nitrogen stream to dryness and reconstituted with 0.5 mL of acetonitrile followed by 50 µL of PFBHA (8.57 mg/mL in water). Denuder samples were also treated in the same way but were diluted with 200 µL of PFBHA. The vials were then held at room temperature for 18 h. The detailed workup procedure for PFBHA derivatives has been reported previously (24, 36). The vials with PFBHAcarbonyl derivatives were then split into two injection vials; one was for the GC/MS analysis of the PFBHA derivatives and the other was used for further analysis of carboxylic acids and hydroxyl groups. This was done by directly adding 10 µL of BSTFA and sonicating the vials at 60 °C for 20 min. The PFBBr derivatization method is used for structural identification of carboxylic acids and phenolic hydroxyl groups. The BSTFA derivatization combined with PFBHA was also used to clarify structural isomer products retaining hydroxy and carboxylic acids. A quarter (250 µL) of each sample for PFBBr derivatization of carboxylic acids was diluted with 1 mL of acetonitrile. One hundred microliters of PFBBr (12.14 mg/mL in dichloromethane) for gas-phase samples and 25 µL for particle samples (front and back filters) were added to the vials along with 50 mg of powdered potassium carbonate (K2CO3). Further details on this procedure can be found in Chien and Jeffries (38).

Results and Discussion Identification and Reaction Pathways for Photooxidation Products. It is well-known that the OH radical addition into the aromatic ring under atmospheric conditions is a predominant reaction compared to hydrogen abstraction (32, 39). Atkinson and Ashmann (40) suggest that the rate constant of toluene with the OH radical is about 6.0 × 1012 cm3 molecule-1 s-1 at room temperature along with a branching ratio of 0.07-0.12 for the H-abstraction on a methyl group. The mechanism of aromatic photooxidation following OH 3628

9


radical addition to the aromatic ring is still not completely certain, although theoretical calculations have been employed to determine the most energetically favored intermediate (41, 42). Tentatively identified products from photooxidation of toluene in the presence of NOx/1-propene are listed in Table 2-6 along with their molecular structures. Toluene photooxidation products were categorized into three classes in this study: aromatic retaining (TOL-AR series in Table 2), nonaromatic ring retaining (TOL-R series in Table 3), and ring opening (or ring cleavage) products (TOL-RO series in Table 4-6). Their characteristic fragment ion peaks are listed in Appendixes 1-3 and GC chromatograms for PHBHA, PHBHA/BSTFA, and PHBBr-derivatives are shown in Figure II-V in the Supporting Information. Some ring opening products retaining alkenyl double bonds can produce geometric isomers consisting of cis and trans forms due to restricted rotations on double bonds. Alkenes conjugated with carbonyls and carboxylic acid groups can also be photochemically isomerized from cis to trans forms and vice versa. The structural isomers resulting from an initial OH attack on an aromatic ring or other radical reactions on different ring positions (o, m, and p) should be considered for aromatic and nonaromatic ring products. Some carbonyl compounds retaining a hydroxyl group in the R-position of carbonyls can be tautomerized between keto and enol forms: e.g., three tautomers for TOL-RO-C3 (2-hydroxy-1,3-propandial, 3-hydroxy-2-oxo-propanal, and 2,3-dihydroxy-2-propenal). For convenience, only one or some of isomers, mostly based on favored pathways, were used for product molecular structures and the product formation mechanisms. Furthermore, diverse structural isomers, which have either the same molecular weight or the same functional groups, but different arrangement of functional groups, were seen in the ring opening reaction products

TABLE 3. Molecular Structures of Nonaromatic Ring Reserved Products from Photooxidation of Toluene in the Presence of NOx/1-Propenea


TABLE 4. Molecular Structures of Ring-Opening Carbonyl Products from Photooxidation of Toluene in the Presence of NOx/1-Propenea


VOL. 35, NO. 18, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

3629

SCHEME 1. Reaction Pathways for Aromatic Retaining Products

as well. It was necessary to utilize different derivatization methods for differentiating the molecular structures of reaction products, which had the same molecular weight but different structures and functional groups. To do this, a combined interpretation using the three different derivatization methods (PFBHA, PFBBr, and PFBHA/BSTFA) was used. The derivatives of multifunctional oxy-products were readily identified by analyzing the reconstructed chromatograms for a characteristic intense peak. For example, the 181 ion results from a pentafluorobenzyl ion fragment (C6F5CH2+) of either the PFBBr-derivatized carboxylic acids or the PFBBH-derivatized carbonyls, M-197 is also a representative fragmentation peak (C6F5CH2O+) for both PFBBr and PFBHA derivatives, and M-225 amu is another representative fragmentation peak (C6F5CH2OCO +) for PFBBr-derivatives of carboxylic acids. For BSTFA derivatization, the mass fragmentations at 73 [(CH3)3Si+] and 75 [(CH3)2SiOH+] amu are typical ion peaks for carboxylic acids and hydroxyl groups. M - 15, M - 73, and M - 89 [(CH3)3SiO+] are also 3630

9


characteristic fragmentation ion peaks for BSTFA-derivatives (35). M-117 and 117 amu [(CH3)3SiOCO+] are fragment ion peaks for BSTFA-derivatives of carboxylic acids. The detection of M + 181 mass fragments by reaction with a pentafluorobenzonium ion is unique to either PFBBr or PFBHA-derivatives in the CH4-CI mass spectrum (24, 25, 36, 38). M + 73 mass fragments associated with a trimethyl silyl ion are commonly found in BSTFA derivatives in the CH4-CI mass spectrum. Aromatic Compounds. Table 2 presents tentatively identified TOL-AR products by photooxidation of toluene in the presence of NOx/1-propene. PFBHA and PFBBr derivatives for TOL-AR products were analyzed using a GC/ITMS in CH4-CI mode, and PFBHA/BSTFA-derivatives were by GC/ITMS in EI mode (Appendix 1 in Supporting Information). PFBHA and PFBBr derivatives can be readily interpreted by typical fragmentation patterns: M/M + 1, 181, 197, M - 181, M - 197, M - 225, and M + 181 amu. PFBHA/BSTFAderivatives were also characterized by representative fragment peaks described above. For PFBHA/BSTFA-derivatives,

SCHEME 2. Reaction Pathways for the Products Retaining Nonaromatic-Six-Membered Rings

M - 15 ion peaks are often much more intensive than molecular ion peaks (M). Derivatives of nitrophenol class products showed a fragment peak at M - 46 by loss of a nitro group (NO2). Either PFBBr- or PFBHA/BSTFA-derivatives for methylnitrophenol (TOL-AR10, MW ) 153) generated particular ion peaks associated with the reaction of methyl nitroaryllium [(CH3)(NO2)C6H3+]. The fragment ion at 467 amu (M + 134) in the PFBBr derivatization for product TOLAR10 corresponds to a reaction of a molecular ion (M ) 333) with a methyl nitroaryllium ion. For PFBHA/BSTFA derivatives, the fragment ion at 435 (M + 210) amu is from a reaction of a molecular ion (M ) 225) with a fragment ion at M - 15 [CH3NO2C6H3OdSi(CH3)2+]. Scheme 1 shows the pathway for forming TOL-AR products. Products with more than two substituents on the aromatic ring often show structural isomers. The most probable intermediate results from OH addition to the orthoposition of the toluene ring (pathway 1a in Scheme 1) (43). The further reaction pathways in Scheme 1 are described only for a major intermediate. The reaction intermediate initiated by OH radicals reacts with NO2 and produces nitroaromatics (9). The phenoxy radicals can be created through a hydrogen abstraction of phenolic hydrogens by an OH radical (44) and further react with NO2, resulting in nitrophenols and dinitro phenol isomers. About 7-12% of the toluene-OH radical reaction is processed by H-abstraction from the methyl group, leading to benzaldehyde (illustrated in pathway 1b of Scheme 1). Benzaldehyde further reacts with the OH radical, forms a

benzoylperoxy radical in the presence of O2 and NO, and leads to benzoic acid by a reaction of a benzoylperoxy radical with HO2. The benzylic-hydrogen abstraction reaction also occurs with other reaction products such as methylphenol or methylnitrophenol compounds and produces substituted aromatic aldehydes and acids as shown in Scheme 1. Nonaromatic Ring Retaining Products. The tentatively identified molecular structures for TOL-R products are shown in Table 3, and fragment ions obtained from PFBHA- and PFBHA/BSTFA-derivatives are represented in Appendix 2 (Supporting Information). To characterize some products that are not derivatized by the above methods, such as furans and 2,5-furandiones, underivatized samples were directly injected to a GC/MS in a EI mode (Hewlett-Packard 5890 GC interfaced to a 5971 MSD). While small amounts of 2,5furandione and 3-methyl-2,5-furandione were tentatively identified only in gas-phase samples (Table 3), furan and methylfuran were not observed because they were too close to the solvent peak and lost in the ion source solvent delay time of the MS program. The identification of 2,5-furandione and 3-methyl-2,5-furandione here is consistent with the studies of Forstner et al. (10), but these compounds were major product components in Forstner et al., composing 46% of total identified organic compositions (10). The high particle phase concentration of furandiones and hydrofurandiones in the Forstner et al. study may have resulted from sampling artifacts by filter adsorption from the gas phase. VOL. 35, NO. 18, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

3631

SCHEME 3. Reaction Pathways for the Products Retaining Five-Membered Rings

The TOL-R products (nonaromatic-six-membered rings, furans, and furandiones) can be produced via several possible pathways as shown in Schemes 2 and 3. For nonaromatic six-membered ring oxy-products (Scheme 2), it has been reported that intermediates initiated by the OH radical addition to toluene lead to epoxides by the reaction with oxygen (10, 25). Then, the epoxy ring can either be retained (path 2a) (24, 25) or open and lead to methylbenzoquinons (path 2b and 2c). Methylbenzoquinones can be attacked by OH radicals, further react with oxygen in the air, and lead to newly identified products, TOL-R8 (methylcyclohexenetriones) and TOL-R10 (hydroxymethylcyclohexenetriones) isomers. In this work, however, epoxy compounds were not observed. Analyses for PFBHA derivatives in this study suggest that a product TOL-R8 (MW ) 138), which was previously identified as an epoxybenzoquinone, may be a methylcyclohexene tricarbonyl. Additionally, we did not observe epoxide product TOL-R9. It is also possible that an epoxide group may be hydrolyzed or react with PFBHA in the acidic aqueous solution during PFBHA workup procedure. The mechanism for forming furans and furandiones from the photooxidation reaction of toluene in hydrocarbon-NOx mixture is shown in Scheme 3. Furanoid products can be produced through a bridged oxide intermediate on a bicyclo ring (path 3a) (45). Although furan and methylfurans were not observed here due to instrumental limitations, other carbonyl-retaining furanoids (product TOL-R3 and product TOL-R5) were tentatively identified using PFBHA-carbonyl derivatization. A reaction mechanism for forming furandiones has been described previously (10, 45, 46). The conjugated dialdehyde with an intra carbon double bond as a ring opening product loses one of its aldehydic hydrogen atoms via either photolysis or attack by OH radicals and forms an acyl radical. The intra-aldehyde group attacks an acyl radical and leads to furandiones through cyclization and loss of a hydrogen atom, as shown in path 3b of Scheme 3. Ring Opening Products. Tables 4-6 present molecular structures of ring opening products (TOL-RO) tentatively identified from the photooxidation reaction of toluene/NOx/ 1-propene system. The reaction pathways for TOL-RO 3632

9


product formation are shown in Schemes 4 and 5. The hydroxymethylphenoxy radical intermediates lead to 6-oxo2,4-heptadienal and methyl-6-oxo-2,4-hexadienal isomers (product TOL-RO-A7, MW ) 124) via a ring opening reaction (Scheme 4). Product TOL-RO-A7 is oxidized to product TOLRO-B6 (MW ) 140) through the reaction of peroxide radical intermediates with HO2 radicals. OH radicals can also attack the alkenyl double bonds of TOL-RO-A7 and mainly form stable allylic radicals (-CdC-C‚) as shown in Scheme 4. These allylic radicals undergo isomerization and lead to either multifunctional oxy-products or oxygenated smaller molecular weight compounds via carbon-carbon cleavage (Scheme 4). Products TOL-RO-C6 and TOL-RO-C8 can be regarded as major ring-opening compounds of SOAs for the toluene/ NOx/1-propene photochemical reaction system (Scheme 4). Methylbenzoquinones (product TOL-R7) can react with OH radicals again, and lead to highly oxidized multifunctional products, TOL-RO-B7 (MW ) 142), TOL-RO-B8 (MW ) 172, trace), and TOL-RO-B9 (MW ) 174, trace) in the presence of NOx, as shown in Scheme 5. Molecular structures were tentatively characterized by a combination of three different derivatization methods. Of the photooxidation products, some compounds have the same molecular weight but a different functional group: these include product TOL-RO-B4 vs product TOL-RO-C5 (MW ) 114) and product TOL-RO-B5 vs product TOL-RO-C6 (MW ) 128). One set of compounds has a carboxylic acid and the other an alcohol group. The PFBHA mono- or di- derivatives and PFBHA/BSFTA derivatives of the iso-mass products were not distinguished from each other. Therefore, the PFBBr derivatization for a carboxylic acid group was simultaneously implemented with PFBHA and PFBHA/BSFTA derivatization methods in this study to correctly interpret the identified molecular structures. For products with more than three carbonyl groups, the molecular structure assignment was conducted based on PFBHA-mono/di-derivatives because the molecular weight of PFBHA-tri-derivative is out of the scan range (650 amu). Appendix 3 in the Supporting Information shows the typical fragmentation patterns obtained from GC/ITMS in either CI or EI mode for PFBHA, PFBBr, PFBHA/BSTFA derivatives of ring opening products.

TABLE 5. Molecular Structures of Ring-Opening Oxocarboxylic Acid Products from Photooxidation of Toluene in the Presence of NOx/1-Propene


In Appendix 3, mostly the di-derivatives are included with the exclusion of mono-derivatives for the carbonyl products which can have mono- and di-derivatives. Nitrogen Retaining Products: Peroxyacylnitrates and Alkylnitrates. The existence of alkyl nitrates (RONO2) or peroxyacyl nitrates [RC(O)OONO2] was clarified using FTIR spectroscopy. The OH or NO3 radicals (mainly OH radicals during daytime) attack the alkene double bond products [R1R2CdCR3R4] of TOL-R and TOL-RO series in Tables 2 and 3 and form alkyl radicals [R1R2C(OH)CR3R4]. Then, alkyl radicals form alkylperoxy radicals [R1R2C(OH)C(OO‚)R3R4] by the reaction with oxygen and furthermore alkoxy radicals [R1R2C(OH)C(O‚)R3R4] in the presence of NOx. The alkyl nitrates are mostly obtained from either the reaction of alkoxy radicals with NO2 or alkylperoxy radicals with NO. The substituted benzaldehydes in TOL-AR series and multifunctional aldehydes in TOL-R and TOL-RO series lose an aldehydic hydrogen by OH radical or photochemical reactions, form peroxyacyl radicals [RC(O)O‚] by reaction with oxygen, and lead to peroxyacyl nitrates by reaction with NO2.

Figure 1 represents the characteristic FTIR spectrum for secondary aerosols impacted on the ungreased ZnSe FTIR window. The O-H stretching of the hydrogen-bonded alcoholic hydroxyl group is seen at 3500-3100 cm-1, and the O-H stretching of hydrogen bonded COOH groups at 33002500 cm-1. The CdO stretching of carboxylic acid, ketones, and aldehydes is shown at 1640-1780 cm-1. The polar nitro group shows two strong bands at 1559 and 1340 cm-1. The strong peak at 1645 cm-1 is from NO3 stretching of alkyl nitrates or peroxyacyl nitrates (8), strongly verifying that alkyl nitrates or peroxyacyl nitrates contribute to secondary aerosol formation. Ozone Reactions. The ozone produced from the NO/ NO2/O3 photoreaction cycle further reacts with both alkene and phenolic compounds of secondary organic products in the gas phase and results in further oxidized secondary organic compounds retaining more carboxylic acid and carbonyl functional groups. A detailed description of possible secondary ozone reaction mechanisms was not included in this study, because additional work is needed to fully VOL. 35, NO. 18, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

3633

TABLE 6. Molecular Structures of Ring-Opening Hydroxy-Carbonyl Products from Photooxidation of Toluene in the Presence of NOx/1-Propenea


understand the significance and contribution of these reactions to SOA products. The NO/NO2 in the reaction system may also be bound as alkyl nitrates or peroxyacyl nitrates. The NO, NO2, and O3 time series show that chemiluminescent NO2 concentrations declined slowly, and O3 concentrations remained almost constant after 1:00 pm EDT in the afternoon (see Figure I of Supporting Information). Thus, the NO2 signal at this point in the reaction is almost entirely composed of nitrate compounds and not NO2, since NO2 would photolyze. This result strongly suggests that nitrate formation from substituted alkenes, substituted benzaldehydes, and nonaromatic aldehydes created from toluene ring opening reactions is considerable. Major Products in the Gas and the Particle Phases. Sampling for photooxidation products was accomplished with two different sampling systems: a filter-filter-denuder system and an impaction system. A significant issue related to the analysis of particle-phase concentration (iCp) and the gas-phase concentration (iCg) for a secondary organic product (i) is the artifacts associated with sampling. A filter-filterdenuder sampling train has been used so that products sorbed onto the back filter could be subtracted from the front filter to remove a positive artifact (27, 47, 48). Controversy, however, still exists with respect to positive and negative artifacts (49). To confirm the presence of particle phase 3634

9


products, aerosols were also collected on an impactor surface (ungreased 25 mm square glass plate). Tables 2-6 also show the major and minor products which were found in each sampling component: front filter, backup filter, and denuder from a filter-filter-denuder sampling train as well as the impactor system. Some of the compounds found in the front filter were not seen in the impactor system or contributed differently to the tentatively determined product compositions (Tables 2-6). Because many of the identified compounds are not commercially available and difficult to synthesize, major components of photooxidation products were determined based on the normalized GC area intensity (base peak area of a target analyte/internal standard area). Most important, however, is that products observed in the impactor samples can be regarded with certainty as being in the aerosol phase. Table 7 lists the predicted major components observed in the impactor particle system along with the vapor pressure (ipoL), predicted log partitioning coefficient (log iKp), and experimental log iKp values. The predicted vapor pressure was calculated by (50, 51)

ln ipoL )

[ (

)

( )]

∆Svap(Tb) Tb Tb (1.8) 1 + (0.8) ln R T T

(atm) (1)

SCHEME 4. Reaction Pathways for Ring Opening Products (I)

where ∆Svap is the entropy of vaporization, R is a gas constant (8.314 J/K/mol), Tb is a boiling point (K), and T is an ambient temperature (K) for a given organic compound. ∆Svap of an organic compound was calculated using a modified Trouton’s method developed by Yalkowsky and co-workers, considering parameters related to molecular geometry and association (52, 53). The Tb values of organic compounds were calculated by a group contribution method originally developed by Joback and Reid (54) with a modified equation and group contribution parameters (55).

Experimental iKp values were calculated by [(iCp/TSP)/ iC and iC have units of nanograms per cubic p g meter, TSP has units of micrograms per cubic meter. iKp has units of cubic meters per microgram. The iCp and iCg values are indirectly described using the normalized GC area of analytes and an internal standard (decafluorobibenzyl). We assumed that the normalized GC intensity was linear withthe analyte concentration and that a zero concentration was equal to a GC intensity of zero.

iC

g] (56). When


9

3635

SCHEME 5. Reaction Pathways for Ring Opening Products (II)

To illustrate predicted gas/particle partitioning behaviors of SOA products, the absorptive partitioning coefficients (iKp,) of SOA products for major components were calculated and compared with experimentally observed iKp. The predicted absorptive equilibrium iKp of a semivolatile organic compound was originally described by Pankow (56) and presented below using units for the gas constant (R) of 8.314 JK-1mol-1 (37), i

Kp )

7.501RTfom 10 MWomiγ∞om ip0L 9

(2)

MWom is the average molecular weight of the given organic matter (om), and iγ∞om is the activity coefficient of an organic compound (i) at infinite dilution in a given liquidlike particle 3636

9


medium. In our case, we have estimated that MWom is 120 (g/mol), fom is 0.88, iγ∞om is 1, and T is 303 K for calculating the predictive iKp of major SOA products. As shown in Table 7, predicted vapor pressures for major SOA products in the impactor sample range from 10-2 to 10-5 mmHg at 303 K. This suggests that products from the gas-phase photooxidation of toluene in the presence of NOx/ 1-propene should not partition to a great extent to atmospheric particulates, because of their relatively “high” vapor pressures. The predicted iKp values of ketones (TOL-R7 and TOL-R8) and phenols (TOL-AR10 and TOL-AR14) tended to be closer to the experimental iKp values. However, Table 7 shows that the experimental log iKp values of the aldehyde compounds were much higher and deviated more from predicted iKp values.

TABLE 7. Estimated Vapor Pressures and Partitioning of Major Products Found in the Impactor Sampling System at 303 K

TABLE 8. Predicted log Kp Values for Hemiacetals from the Reaction of 2-Hydroxypropanedial with Low Molecular Weight Chamber Rich Aldehydes hemiacetal reactions

MW

HC(O)CH(OH)C(O)H HC(O)CH(OH)C(O)H + HCHO HC(O)CH(OH)C(O)H + HC(O)C(O)H HC(O)CH(OH)C(O)H + HC(O)C(O)CH3 HC(O)CH(OH)C(O)H + HC(O)CH(OH)C(O)H Note: HC(O)CH(OH)C(O)H *experimental

88 118 146 160 176 88

a

predicted log ipo (mmHg) L

predicted iK (A) p

predicted log iKp

iK a (A)/(B) p

-0.76 -2.28 -3.80 -3.80 -6.85

8.01E-7 (B) 2.61E-5 2.56E-4 1.54E-4 9.72E-1 *2.19E-3

-6.10 -4.58 -3.59 -3.81 -0.01 *-2.66

1 32.6 320 192 1.0E6 3650

relative

The relative Kp value is calculated by the ratio of the predicted Kp for demonstrated hemiacetal to the predicted one for HC(O)CH(OH)C(O)H.

One way to explain how relatively high vapor pressure compounds end up in the aerosol phase is further heterogeneous reaction of high volatility compounds in the particle phase. These subsequent reactions result in lower vapor pressure products. Furthermore, nitric or sulfuric acid formed atmospherically deposits on the particle phase and can catalyze heterogeneous reactions of aldehydes: e.g., hydration, hemiacetal/ketal reactions, polymerization (57). For example, Tobias and Ziemann (58, 59) analyzed SOA using a thermal desorption particle beam mass spectrometer and have also shown the possible formation of much larger peroxyhemiacetals from the reaction of an R-acyloxyalkyl

hydroperoxide with formaldehyde. It is also known that alkyl or acyl peroxides react with aldehydes to give alkyl/acyl peroxyhemiacetals (57, 58). The gas-phase polymerization of formaldehyde has been well-known in the polymer resin industry to produce polyoxymethylene (60). In upper tropospheric/lower stratospheric atmospheric chemistry, the loss of formaldehyde to sulfuric acid aerosol has also been reported by Iraci and Tolbert (61). Compared to aldehydes, ketones are much less reactive for heterogeneous reactions. Hemiacetals or acetals are viewed in polymer chemistry as an equilibrium reaction and called as “zipping/unzipping” process via forward and backward equilibrium reactions. In VOL. 35, NO. 18, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

3637

our workup/analytical system, these hemiacetals/acetals may be easily hydrolyzed and return to their original aldehydes and alcohols. There are also large numbers of reaction combinations between different aldehydes and alcohols to produce hemiacetals/acetals resulting in lower vapor pressure SOA products. For example, Table 8 illustrates predicted partitioning coefficients for hemiacetals from the reaction of 2-hydroxy1,3-propandial (TOL-RO-C3) with some of the gas-phase products. If product TOL-RO-C3 is dimerized and leads to a cyclohemiacetal (a substituted dioxane), its predicted ipoL is 6 orders of magnitude lower than one for monomeric hydroxy propanal at the experimental temperature of this study. What seems most important is that SOA yields can potentially increase via heterogeneous reactions of chamber rich aldehyde species such as glyoxal, methylglyoxal, hydroxyacetaldehyde, and some ring opening products shown in Table 7 (TOL-RO-B3, B4, C3, C5, C6, and C8). Table 8 illustrates that the hemiacetal from TOL-RO-C3 results in a huge reduction in ipoL (increase in iKp values) compared to unreacted TOL-RO-C3. It strongly suggests that it is possible to obtain the observed iKp value by the reaction of some TOL-RO-C3 with a variety of different product alcohols and aldehydes.

Acknowledgments This work was supported by a Grant from National Science Foundation (ATM 9708533) to UNC-CH. We also acknowledged the gift of a GC/MS system from the Hewlett-Packard Corp. and an GC-ITMS system from the Varian Corp. to the Department of Environmental Sciences & Engineering at UNC-CH. The authors would like to thank Bharadwaj Chandra, Mohammed Jaoui, Sangdon Lee, and Sirakarn Leungskakul for helping with the chamber experiments.

Supporting Information Available Appendixes of PFBHA, PFBBr, and PFBBH/BSTFA derivatives of aromatic and nonaromatic ring retaining products and figures of NO, NO2, NOx, and ozone time profiles and GC chromatograms. This material is available free of charge via the Internet at http://pubs.acs.org.

Literature Cited (1) Twomey, S. Atmos. Environ. 1991, 25A, 2435-2442. (2) Pilinis, C.; Pandis, S.; Seinfeld, J. H. J. Geophys. Res. 1995, 100, 18739-18754. (3) Eldering, A.; Larson, S. M.; Hall, J. R.; Hussey, K. J.; Cass, G. R. Environ. Sci. Technol. 1993, 27, 626-635. (4) Eldering, A.; Cass, G. R. J. Geophy. Res. 1996, 101, 19343-19369. (5) EPA. Air Quality Criteria for Particulate Matter; EPA/600/P95/001cF; Environmental Protection Agency: Washington, DC, 1996. (6) Altshuller, A. P. Atmos. Environ. 1983, 17, 2131-2165. (7) Noziere, B.; Barnes, I. J. Geophys. Res. 1998, 103, 25587-25597. (8) Noziere, B.; Barnes, I.; Becker, K. H. J. Geophys. Res. 1999, 104, 23645-23656. (9) Gery, M. W.; Fox, D. L.; Jeffries, H. E. Int. J. Chem. Kinet. 1985, 17, 931-955. (10) Forstner, H. J. L.; Flagan, R. C.; Seinfeld, J. H. Environ. Sci. Technol. 1997, 31, 1345-1358. (11) Odum, J. R.; Hoffmann, T.; Bowman, F.; Collins, D.; Flagan, R. C.; Seinfeld, J. H. Environ. Sci. Technol. 1996, 30, 2580-2585. (12) Odum, J. R.; Jungkamp, T. P. W.; Seinfeld, J. H. Science 1997, 276, 96-99. (13) Odum, J. R.; Jungkamp, T. P. W.; Seinfeld, J. H. Environ. Sci. Technol. 1997, 31, 1890. (14) Hoffmann, T.; Odum, J. R.; Bowman, F.; Collins, D.; Klockow, D.; Flagan, R. C.; Seinfeld, J. H. Atmos. Environ. 1997, 26, 189222. (15) Kamens, R. M.; Jang, M.; Chien, C. J.; Leach, K. Environ. Sci. Technol. 1999, 33, 1430-1438. (16) Kamens, R. M.; Jaoui, M. Environ. Sci. Technol. 2001, 35, 13941405. 3638

9


(17) Yu, J.; Cocker, D. R.; Griffin, R. J.; Flagan, R. C.; Seinfeld, J. H. J. Atmos. Chem. 1999, 34, 207-258. (18) Kurthenbach, R.; Brockmann, K. J.; Lorzer, J.; Niedojadlo, A.; Becker, K. H. VOC-Measurements in urban air of the city of Wuppertal, in TFS-LT3 auunal report 1998 (German); Becker, K. H., Ed.; University of Wuppertal: 1999. (19) Klotz, B.; Becker, K. H. Chem. Phys. 1998, 231, 289-301. (20) Klotz, B.; Barnes, I.; Becker, K. H.; Golding, B. T. J. Chem. Soc.: Faraday Trans. 1997, 93, 1507-1516. (21) Kwok, E. S. C.; Aschmann, S. M.; Atkinson, R.; Arey, J. J. Chem. Soc.: Faraday Trans. 1997, 93, 2847-2854. (22) Andino, J. M.; Smith, J. N.; Flagan, R. C.; Goddard, W. A.; Seinfeld, J. H. J. Phys. Chem. 1996, 100, 10967-10980. (23) Seuwen, R.; Warneck, P. Int. J. Chem. Kinet. 1996, 28, 315-332. (24) Yu, J.; Jeffries, H. E.; Sexton, K. G. Atmos. Environ. 1997, 31, 2261-2280. (25) Yu, J.; Jeffries, H. E. Atmos. Environ. 1997, 31, 2281-2287. (26) Smith, D. M.; Kleindienst, T. E.; McIver, C. D. J. Atmos. Chem. 1999, 34, 339-364. (27) Smith, D. M.; McIver, C. D.; Kleindienst, T. E. J. Atmos. Chem. 1998, 30, 209-228. (28) McDow, S. R.; Huntzicker, J. J. Atmos. Environ. 1990, 24, 25632571. (29) Kleindienst, T. E.; Smith, D. F.; Li, W.; Edney, E. O.; Driscoll, D. J.; Speer, R. E.; Weathers, W. S. Atmos. Environ. 1999, 33, 36693681. (30) Hurley, M. D.; Sokolov, O.; Wallinton, T. J.; Takekawa, H.; Karasawa, M.; Klotz, B.; Barnes, I.; Becker, K. H. Environ. Sci. Technol. 2001, 35, 1358-1366. (31) Barthelmie, R. R.; Pryor, S. C. J. Geophys. Res. 1999, 104, 2365723699. (32) Heidi, L. B.; Atkinson, R.; Arey, J. J. Phys. Chem. 2000, 104, 89228929. (33) Calvert, J. G.; Atkins, R.; Becker, K. H.; Kamens, R. M.; Seinfeld, J. H.; Wallington, T. J.; Yarwood, G. Reactions of aromatic compounds with OH radicals (Chapter 2). In Mechanisms of atmospheric oxidation of aromatic hydrocarbons; CRC: New York, 2000. (34) Le Lacheur, R. M.; Sonnenberg, L. B.; Singer, P. C.; Christman, R. F.; Charle, M. J. Environ. Sci. Technol. 1993, 27, 2745-2753. (35) Weintraub, S. T. Mass spectrometry of lipids (Chapter 8). In Mass spectrometry of biological materials; McEwen, C. N., Larsen, B. S., Eds.; Marcel Dekker Inc.: New York, 1990. (36) Jang, M.; Kamens, R. M. Atmos. Environ. 1999, 33, 459-474. (37) Jang, M.; Kamens, R. M.; Leach, K.; Strommen, M. R. Environ. Sci. Technol. 1997, 31, 2805-2811. (38) Chien, C.; Charles, M. J.; Sexton, K. G.; Jeffries, H. E. Environ. Sci. Technol. 1998, 32, 299-309. (39) Pitts, B. J. F.; Pitts, J. N. Chemistry of the upper and lower atmosphere: Theory, Experiments, and Applications; Academic Press: New York, 2000; pp 207-213. (40) Atkinson, R.; Aschmann, S. M. Int. J. Chem. Kinet. 1994, 26, 929-944. (41) Mebel, A. M.; Lin, M. C. J. Am. Chem. Soc. 1994, 116, 95779584. (42) Lay, T. H.; Bozzelli, J. W.; Seinfeld, J. H. J. Phys. Chem. 1996, 100, 6543-6554. (43) Klotz, B.; Barnes, I.; Golding, B. T.; Becker, K. H. Phys. Chem. Chem. Phys. 2000, 2, 227-235. (44) Platz, J.; Nielsen, O. J.; Wallington, T. J.; Ball, J. C.; Hurley, M. D.; Straccia, A. M.; Schneider, W. F.; Sehested, J. J. Phys. Chem. A 1998, 102, 7964-7974. (45) Shepson, P. D.; Edney, E. D.; Corse, E. W. J. Phys. Chem. 1984, 88, 4122-4126. (46) Bierbach, A.; Barnes, I.; Becker, K. H.; Wiesen, E. Environ. Sci. Technol. 1994, 28, 715-729. (47) Kenneth, M. H.; Pankow, J. F. Environ. Sci. Technol. 1994, 28, 655-661. (48) Turpin, B. J.; Hering, S. V. Atmos. Environ. 1994, 28, 3061-3071. (49) Turpin, B. J.; Saxena, P.; Andrews, E. Atmos. Environ. 2000, 34, 2983-3013. (50) Mackay, D.; Bobra, A.; Chan, D. W.; Shiu, W. Y. Environ. Sci. Technol. 1982, 16, 645-649. (51) Schwarzenbach, R. P.; Gschwend, P. M.; Imboden, D. M. Environmental Organic Chemistry; John Wiley & Sons: New York, 1993. (52) Zhao, L.; Li, P.; Yalkowsky, S. H. J. Chem. Inf. Comput. Sci. 1999, 39, 1112-1116. (53) Zhao, L.; Ni, N.; Yalkowsky, S. H. Ind. Eng. Chem. 1999, 39, 324-327. (54) Joback, K. G.; Reid, R. C. Chem. Eng. Commun. 1987, 57, 233.

(55) Stein, S. E.; Brown, R. L. J. Chem. Inf. Comput. Sci. 1994, 34, 581-587. (56) Pankow, J. F. Atmos. Environ. 1994, 28, 2275-2283. (57) Aldehyde and Ketones. In Comprehensive Organic Chemistry: The synthesis and reactions of organic compounds; Barton, D., Ollis, W. D., Eds.; Pergamond Press: New York, 1979; 9601013. (58) Tobias, H. J.; Ziemann, P. J. Anal. Chem. 1999, 71, 3428-3435. (59) Tobias, H. J.; Ziemann, P. J. Environ. Sci. Technol. 2000, 34, 2105-2115. (60) Walker, F. Formaldehyde, 3rd ed.; Reinhold Publishing: New York, 1964; pp 140-205. (61) Iraci, L. T.; Tolbert, M. A. J. Geophys. Res. 1997, 102, 1609916107.

(62) Atkinson, R.; Aschmann, S. M.; Carter, J.; Arey, W. Int. J. Chem. Kinet. 1989, 21, 801-827. (63) Tuazon, E. C.; MacLeod, H.; Atkinson, R.; Carter, W. Environ. Sci. Technol. 1986, 20, 383-387. (64) Dumdei, B.; Kenny, D.; Shepson, P.; Kleindienst, T.; Nero, C.; Cupitt, L.; Claxton, L. Environ. Sci. Technol. 1988, 22, 1493.

Received for review February 22, 2001. Revised manuscript received June 4, 2001. Accepted June 25, 2001. ES010676+


9

3639

Characterization of Secondary Aerosol from the Photooxidation of

Recommend Documents