Secondary aerosol formation from the oxidation of toluene by chlorine atoms

Atmospheric Environment 42 (2008) 7348–7359

Contents lists available at ScienceDirect

Atmospheric Environment journal homepage: www.elsevier.com/locate/atmosenv

Secondary aerosol formation from the oxidation of toluene by chlorine atoms Xuyi Cai a, Luke D. Ziemba a, Robert J. Griffin a, b, * a b

Climate Change Research Center, University of New Hampshire, Durham, NH 03824, USA Department of Earth Sciences, University of New Hampshire, Durham, NH 03824, USA

a r t i c l e i n f o

a b s t r a c t

Article history: Received 1 May 2008 Received in revised form 8 July 2008 Accepted 14 July 2008

Oxidation of toluene by chlorine atoms (Cl) was studied in a chamber. Secondary organic aerosol (SOA) yields ranged from 3.0 to 7.9% for aerosol concentrations up to 12.0 mg m3. SOA yields from toluene/Cl reactions and model parameters related to aerosol growth depend on the initial ratio of molecular chlorine to toluene. Data from an Aerodyne quadrupole Aerosol Mass Spectrometer (Q-AMS) indicate that a small fraction of the generated aerosol mass is inorganic chloride (approximately 4%), that inorganic chloride aerosol growth ceases contemporaneously with that of SOA, and that particles are mixed internally. Analysis of Q-AMS spectra indicates predominance of species that traditionally are thought to be representative of SOA but with increased importance of aromatic/ringretaining products. Mechanistic modeling of the toluene system indicates that final SOA products likely result from the oxidation of quinone type compounds derived via oxidation of the first-generation product benzaldehyde. Zero-dimensional calculations indicate Clinitiated oxidation could be as important as hydroxyl-radical-initiated oxidation in SOA formation from toluene in early morning in certain coastal or industrialized areas. Ó 2008 Elsevier Ltd. All rights reserved.

Keywords: Chamber studies Chlorine atom Modeling studies Secondary organic aerosol Toluene

1. Introduction Atmospheric particulate matter (PM) influences human health, climate, and visibility (Twomey et al., 1978; Charlson et al., 1992; Mazurek et al., 1997; Pope et al., 2002). Organic aerosol (OA), much of which is believed to be secondary (Zhang et al., 2007), is a ubiquitous component of PM (Murphy et al., 1998). Secondary OA (SOA) is formed from partitioning of secondary non- and semi-volatile organic compounds (SVOCs) into organic and/or aqueous aerosol phases (Pankow, 1994; Saxena and Hildemann, 1996). Oxidation of volatile organic compounds (VOCs) or the gas-phase portion of primary SVOCs (Robinson et al., 2007) results in products that form SOA. Due to the significant mass of OA in the troposphere, many studies * Corresponding author. Climate Change Research Center, University of New Hampshire, Durham, NH 03824, USA. Tel.: þ1 603 862 2021; fax: þ1 603 862 2124. E-mail address: rob.griffi[email protected] (R.J. Griffin). 1352-2310/$ – see front matter Ó 2008 Elsevier Ltd. All rights reserved. doi:10.1016/j.atmosenv.2008.07.014

have been attempted to understand the chemical and thermodynamic processes that lead to its formation (Seinfeld and Pankow, 2003). Aromatic hydrocarbons are important in the formation of urban photochemical smog because of their large emission rates, reactivities, and ozone (O3) and SOA-forming capabilities (Odum et al., 1996, 1997; Lewis et al., 2000; Hurley et al., 2001; Karlsson et al., 2001; Bahreini et al., 2005; Song et al., 2005; Alfarra et al., 2006; Ng et al., 2007). Toluene is a major component of aromatic emissions (Olivier et al., 1999), and SOA formed upon its oxidation accounts for a significant fraction of simulated SOA from mobile sources in urban areas (Dechapanya et al., 2004). Toluene in the atmosphere is oxidized by the hydroxyl radical (OH) which adds to the ring or abstracts a hydrogen atom (H) from the methyl substituent. In contrast, oxidation by chlorine atom (Cl) is expected to proceed only via the abstraction pathway (Wang et al., 2005). Efficient oxidation of VOCs by Cl affects net O3 formation and likely leads to SOA formation in some areas of the

X. Cai et al. / Atmospheric Environment 42 (2008) 7348–7359

troposphere (Ganske et al., 1992; Keene et al., 1996; Canosa-Mas et al., 1999; Karlsson et al., 2001; Knipping and Dabdub, 2003; Cai and Griffin, 2006; Pszenny et al., 2007). This study quantifies Cl-initiated SOA formation from toluene through calculation of the dimensionless SOA yield, Y (Pandis et al., 1992) ¼ DM0/DVOC where DM0 (mg m3) is the SOA mass concentration formed after the consumption of DVOC (mg m3) of the parent VOC by oxidation. This yield is an overall measure of the SOA-forming potential of a VOC. Based on absorptive partitioning (Pankow, 1994), Y is also expressed as follows (Odum et al., 1996)

Y ¼ D M0

X i

ai Kom;i 1 þ DM0 Kom;i



(1)

where Kom,i (m3 mg1) is the distribution coefficient of product i between the gas-phase and aerosol organic material (om) and ai is its mass-based stoichiometric factor. Yield data obtained from chamber experiments have been fit to Eq. (1) assuming two products (Odum et al., 1996, 1997; Song et al., 2005; Cai and Griffin, 2006; Ng et al., 2007). The four fitted parameters for this model are not associated with absolute physical meanings, but differences in yield curves between conditions and laboratories may be represented through such fits.

2. Experiments The system for this study (Cai and Griffin, 2006) includes a 6-m3 hemi-cylindrical chamber made of FEP TeflonÒ film and mounted in a frame 0.30 m above the floor. Pure air provided by a TEI (Franklin, MA) 111 zero air generator is dehumidified and stripped to remove nitrogen oxides (NOx) and VOCs. The number concentration measured in the zero air is less than 0.1 particles cm3. Blank experiments are conducted to verify that concentrations of VOCs, NOx, and O3 are below detection limits (1.0 ppb). A Shimadzu (Columbia, MD) GC-17A gas chromatograph (GC) with a flame ionization detector (FID) and an Agilent (Palo Alto, CA) DB-5 column is used to monitor VOC mixing ratios. Aerosol size distributions are collected with a TSI (St. Paul, MN) 3012 85Kr neutralizer, a TSI 3010 condensation particle counter (CPC), and a TSI 3080 nanodifferential mobility analyzer (DMA). In certain experiments, an Aerodyne (Billerica, MA) quadrupole Aerosol Mass Spectrometer (Q-AMS) (Jayne et al., 2000) characterizes the size and chemical composition of SOA. Twenty 4-ft, 40-W Sylvania 350BL lights are used to generate 365nm ultraviolet (UV) light to photo-dissociate molecular chlorine (Cl2) to form Cl. Between experiments, the chamber is conditioned for 48 h with the UV lamps on and flushed for 36 h in the dark with pure air. The experimental protocol is similar to that published previously (Cai and Griffin, 2006). Following a GC-FID calibration, the studied VOC is injected using a microliter syringe into a small glass tube and dispersed into the chamber using the zero air flow. After mixing, the initial VOC mixing ratio is measured by taking three replicates (generally within 2%) and averaging. Chlorine gas is injected from a certified cylinder of 1000 ppm in nitrogen.

7349

Once all of the gases are mixed homogeneously, the DMA/CPC, GC-FID, and/or Q-AMS programs are initiated and the UV lights are lit. When the aerosol mass reaches a plateau (based on correction for wall loss) and the VOC in the chamber is consumed completely (indicating that DVOC is generally equal to the initial mixing ratio), the experiment is terminated. Aerosol mass is calculated from the aerosol size distributions measured by the DMA/CPC assuming spherical, unit-density aerosol and is corrected for particle depositional loss (Odum et al., 1996; Cai and Griffin, 2006). Yields for SOA are not corrected for contribution from aerosol-phase inorganic chloride, as in Cai and Griffin (2006). In addition, loss of gases to chamber walls (Presto et al., 2005) is not considered because wall loss of gas-phase parent VOCs is less than 3% over a 2-h blank experiment (Cai and Griffin, 2006). While the loss rate of gas-phase SVOCs to chamber walls may be significantly larger than this, no adjustment to the data presented here has been made for two reasons. First, gas-phase SVOC loss data are unavailable. Second, previous data published using this system (Cai and Griffin, 2006) were not corrected in this manner; following identical protocols in the current study allows for data comparability. The ratio of the initial Cl2 to VOC mixing ratios ((Cl2/ VOC)0) is varied because previous studies (Karlsson et al., 2001; Cai and Griffin, 2006) found that SOA yields depend positively on this value. Cai and Griffin (2006) discuss in detail the relationship between this ratio and SOA yield. In the THR (toluene high ratio) set of experiments, (Cl2/VOC)0 is varied between 5.2 and 10.0; in the TLR (toluene low ratio) set of experiments, it is varied between 2.6 and 4.0. Experimental data are given in Table 1. Larger initial mixing ratios (though similar values of (Cl2/VOC)0) are used in the Q-AMS experiments. 3. Results and discussion 3.1. Yields SOA yields for toluene are described in Tables 1 and 2, and Fig. 1. Fig. 1 shows two yield curves for toluene depending on (Cl2/VOC)0. For THR experiments, the SOA yields range from 0.050 to 0.079 for generated aerosol mass ranging from 4.0 to 12.0 mg m3. In TLR experiments, the SOA yields range from 0.030 to 0.064, corresponding to a generated aerosol mass range of 3.0–11.0 mg m3. The yields associated with the two curves are similar, and the experimental error bars overlap. However, the observed yields and the predicted yield curve for THR experiments are systematically larger than those for TLR experiments. The two toluene SOA yield curves found in this study confirm the dependence of toluene SOA formation on (Cl2/ VOC)0 as observed by Karlsson et al. (2001), who found that the number and volume concentrations of SOA exhibited an increase as the initial Cl2 level increased. It should be noted that the split of yield behavior into two curves is somewhat arbitrary. Rather, it is expected that a continuum of yield curves exists as the (Cl2/VOC)0 value increases and approaches a critical value above which further oxidation is unlikely to occur, as discussed in Cai and Griffin (2006). It is also possible that the two yield curves result not from

7350

X. Cai et al. / Atmospheric Environment 42 (2008) 7348–7359

Table 1 Initial conditions and data for Cl-initiated oxidation yield experiments Experiment

(Cl2/VOC)0 (ppb ppb1)

Temperaturea (K)

DVOCb (ppb)

DVOC (mg m3)

DM0c (mg m3)

Y

t1/2d (min)

bd (min1/2)

THR-1 THR-2 THR-3 THR-4 THR-5 THR-6 THR-7 TLR-1 TLR-2 TLR-3 TLR-4 TLR-5 TLR-6 TLR-7

9.0 5.2 7.0 9.5 10.0 6.7 10.0 3.5 2.6 3.1 3.9 3.2 3.5 4.0

297.0 300.0 298.0 298.0 298.5 299.5 298.5 299.5 300.0 299.0 298.0 299.0 299.5 297.0

22.3 19.8 18.3 18.7 25.3 38.2 40.4 24.3 28.5 21.8 29.1 32.1 36.1 45.5

84.0 74.1 68.9 70.5 94.9 142.9 151.9 90.9 106.7 81.6 109.6 120.4 135.1 171.9

4.2 4.0 5.1 5.5 7.5 11.0 12.0 3.0 3.2 4.0 5.7 6.5 7.7 11.0

0.050 0.054 0.074 0.078 0.079 0.077 0.079 0.033 0.030 0.049 0.052 0.054 0.057 0.064

26.7 61.0 36.2 33.0 20.2 27.7 22.0 66.0 64.0 68.2 58.5 41.5 57.0 23.0

0.235 0.192 0.175 0.213 0.245 0.269 0.288 0.159 0.187 0.171 0.164 0.191 0.214 0.224

a The temperature is the average during the course of the experiment. Laboratory temperature is maintained at 23  C–27  C. The influence of this small range on the experimental results is neglected. b DVOC equals the initial VOC concentration, as discussed in the text. c DM0 represents the average measured mass concentration of SOA from the end of the growth period until the termination of the experiment. d Parameters used in Eq. (2).

changes in chemistry but from increased deposition of gasphase SVOCs to chamber walls in the cases where (Cl2/ VOC)0 is reduced because the resulting smaller and fewer particles provide a diminished condensational sink (Pirjola et al., 1999). While data to disprove this hypothesis definitively are not available, it is believed to be false because single yield curves were observed for pinene isomers despite varying (Cl2/VOC)0 (Cai and Griffin, 2006). Fig. 1 also shows a comparison of toluene SOA yields from different oxidants. The experimental results suggest yields from Cl oxidation that are larger than those from OH, O3, and NO3 oxidation (high-NOx (122–1,609 ppb)) (Odum et al., 1997). In the Odum et al. (1997) study, initial oxidation was predominantly by OH, but secondary and latter generation products could also have been oxidized by O3 and NO3. More recent OH-initiated oxidation data at lowNOx (<1.0 ppb) indicate significantly larger SOA yields for toluene (0.30–0.37) (Ng et al., 2007) compared to both this work and that of Odum et al. (1997). It is believed that the yields generated in the current study are greater than those of Odum et al. (1997) due to lower temperature and significantly smaller NOx concentrations (Song et al., 2005). The decrease in the yields here compared to those of Ng et al. (2007) likely results from Cl (unlike OH) only abstracting H from the methyl group (Wang et al., 2005). Therefore, less ring fragmentation chemistry (and, therefore, less SOA formation) is likely to occur in the Cl system, a hypothesis that would require product identification to be confirmed. This hypothesis, however, is based on the

Table 2 Aerosol yield parameters fit to Eq. (1) for oxidation of toluene Oxidant

a1

a2

Kom,1 (m3 mg1)

Kom,2 (m3 mg1)

Cl (THR) Cl (TLR) OHa OH þ O3 þ NO3b

0.114 0.095 0.360 0.071

0.001 0.017 N/A 0.138

0.324 0.148 N/A 0.053

0.203 0.262 N/A 0.002

N/A,not applicable. a Ng et al. (2007). b Odum et al. (1997).

apparent importance of ring opening products (and oligomerization thereof) in SOA from aromatic systems in which OH is the primary oxidant (Kalberer et al., 2004; Kleindienst et al., 2004). 3.2. Growth dynamics The mass of new particles from nucleation is small, and the bulk of the SOA mass in these experiments transfers to the particle phase via condensation, which inherently includes diffusion and convection. For this system, the observed SOA growth curves can be fit to:

1 2



t

t t

DMðtÞ ¼ DM0 erfc b 1=2pffiffi



(2)

where DM(t) is the formed aerosol mass (mg m3) at time t (min), t1/2 is the time (min) for the SOA concentration to reach one half of DM0, and b is the only fitting parameter (min1/2). This equation is derived by analogy to that for particle coagulation driven by diffusion (Seinfeld and Pandis, 2006). The parameter b is determined by a leastsquares minimization technique. Parameters t1/2 and b are included in Table 1. A characteristic time for aerosol mass growth is represented by t1/2, where smaller values represent very rapid growth (a steep slope of SOA versus time during the growth part of an experiment) and larger values represent less rapid growth. Values of t1/2 correlate strongly with ([VOC]0[Cl2]0)1/2, where the brackets represent initial mixing ratios. Values of t1/2 are controlled by concentrations at small initial reactant mixing ratios (Fig. 2a) because SOA formation is limited by availability of material. For larger initial mixing ratios, reactions proceed very quickly, and t1/2 values are limited only by mass transfer, which occurs rapidly. Fig. 2a shows that t1/2 values for THR experiments are smaller than those for TLR experiments, indicating that both the rate and amount of SOA formation are enhanced by higher oxidant levels (Ng et al., 2007). A characteristic time for diffusion is related to b, with larger values for rapid diffusion. The fitted b values are

X. Cai et al. / Atmospheric Environment 42 (2008) 7348–7359

7351

that diffusion is controlled by the rapid VOC oxidation, not by the oxidant itself. The larger VOC mixing ratios produce a larger concentration gradient between the gas and the particle surface, leading to more rapid diffusion. Fig. 2b demonstrates that the values for THR experiments are significantly larger than those for TLR experiments, indicating faster diffusion in THR experiments, consistent with smaller t1/2 values. 3.3. Q-AMS

Fig. 1. Summary of toluene (TLR and THR) SOA yields including error bars for the current study. Filled symbols show experimental data from this study, and fitted curves shown for comparison are based on the two-product model of Odum et al. (1996) with parameters listed in Table 2. The Odum curve is for O3 þ nitrate radical (NO3) þ OH (Odum et al., 1997). Essentially constant yield data from the study of Ng et al. (2007) are included on the right-hand y-axis.

correlated linearly to the initial VOC mixing ratios (Fig. 2b) because larger initial concentrations generate larger mixing ratios of condensable products; no such relationship was found with initial chlorine concentration, indicating

Fig. 2. (a) Correlation between t1/2 and ([Cl2]0[VOC]0)1/2. (b) Correlation between [VOC]0 and b.

The Q-AMS provides size-resolved particle chemical composition (Jayne et al., 2000). All data were analyzed using the tools of Allan et al. (2003) with corrections to ensure that signal was attributed only to inorganic chloride or organic mass. Time series of mass concentrations of inorganic chloride and organic aerosol for a typical experiment are shown in Fig. 3. In this experiment, the initial toluene mixing ratio is 59 ppb, and (Cl2/VOC)0 is 4.2. Fig. 3 shows that the maximum concentration of SOA greatly exceeds that of inorganic chloride (61.8 mg m3 versus 2.6 mg m3), confirming that inorganic chloride contributes a very small amount to the total aerosol mass concentration. Inorganic chloride growth also ceases when SOA growth has ceased. Because of the very similar temporal profiles in inorganic chloride and organic aerosol after the inflection point (when deposition to the walls begins to have a larger effect on concentrations than condensational growth), gas-phase chemistry may no longer be producing the hydrochloric acid (HCl) that likely leads to the observed inorganic chloride aerosol. It is also possible that uptake of chloride is limited by the lack of growth in surface area. Fig. 3 indicates a lag in measureable aerosol inorganic chloride compared to SOA because of either the higher vapor pressure of HCl compared to SOA species or the need for increased SOA surface area to allow for partitioning of HCl, as mentioned previously. Although not shown in Fig. 3, Q-AMS size distributions confirm that SOA and inorganic aerosol chloride likely are mixed internally given similar particle size distributions that peak between 100 and 200 nm in vacuum aerodynamic diameter (180 nm for the experiment shown in Fig. 3). These phenomena were observed in additional toluene experiments with the Q-AMS.

Fig. 3. Aerosol mass concentration time series as measured by the Q-AMS in a toluene–Cl oxidation experiment.

7352

X. Cai et al. / Atmospheric Environment 42 (2008) 7348–7359

An ion series analysis (McLafferty and Turecek, 1993; Drewnick et al., 2004; Bahreini et al., 2005) in which a delta value (D ¼ m/z  14n þ 1) is assigned to each mass-tocharge (m/z) ratio based on the nominal number of carbons (n) in the ion fragment provides the relative intensity for each D(7  D  6) from the spectrum. The relative intensity of D values from the spectrum provides insight into the likely size, nature, and volatility of the molecules analyzed. Larger values reflect oxidized material, and smaller numbers reflect cyclic and unsaturated (including aromatic) fragments. Delta analysis can also be segregated based on estimated n. The results of a delta analysis for the same toluene experiment discussed previously are shown in Fig. 4. Values shown in Fig. 4 were averaged over the entire experiment. The average delta pattern for toluene–Cl SOA exhibits strong contributions from D ¼ 0, 2 and 3. The large intensities of positive delta values in the toluene system result from strong contributions from m/z 27, 29, 43, and 44, as shown in the average spectrum also shown in Fig. 4, where m/z 44 traditionally is associated with more oxidized material (Zhang et al., 2007). When segmented by likely fragment size, the negative values for larger n indicate a degree of aromaticity and/or ring retention in the SOA from toluene. It should be noted that the delta pattern does not change significantly with time during the experiment, despite a shift in the dominant m/z value (29 to 44) and simulation results that show a shift in composition from aromatic species to those derived from oxidation of

quinone-like compounds. These phenomena are discussed subsequently. Bahreini et al. (2005) performed a Q-AMS analysis of the SOA generated by m-xylene via OH-initiated oxidation that focused on only the latter parts of the experiments. The previously published spectra are dominated by signals at m/z 43 and 44, with a delta analysis showing all positive values when segmented by n (confirming a larger degree of oxidation and less aromaticity, supporting the supposition that less ring fragmentation occurs in the Cl system). When a regression between the spectra published by Bahreini et al. (2005) for m-xylene (y-axis) and that derived here for toluene (x-axis) is performed, the slope is not greatly different than unity (1.1). The regression coefficient is significant (R2 ¼ 0.83), with deviation from unity driven almost exclusively by m/z 43, 44 (larger in the m-xylene experiments), and 29 (larger in the toluene experiments). This strong relationship indicates that SOA from different aromatic precursors under different oxidation scenarios is similar spectrally. 3.4. Mechanistic modeling of the toluene system In an effort to understand potential SOA formation mechanisms for the toluene–Cl system, a gas-phase mechanism to describe the product distribution when oxidation of toluene is initiated by Cl has been developed. This detailed mechanism was derived under the protocol of the Caltech Atmospheric Chemistry Mechanism (CACM,

Fig. 4. Results of a spectral analysis averaged over the entire toluene experiment described in Fig. 3.

X. Cai et al. / Atmospheric Environment 42 (2008) 7348–7359

Griffin et al., 2002) in which simplifications are made to balance the number of reactions and species in the system, the computational cost, and the accuracy of the proposed mechanisms. The number of predicted products is limited by disregarding those pathways with a probability of less than 5% (Jenkin et al., 1997). Fourth-generation products are considered non-reactive (Griffin et al., 2002), and oxidation products with similar molecular structures are lumped using a surrogate approach (Griffin et al., 2002). The self-reactions and permutation reactions of a given organic peroxy radical (RO2) are represented by a single parameterized reaction, in which an operator term, RO2T (the sum of all organic peroxy radicals), is used (Jenkin et al., 1997; Griffin et al., 2002). Finally, photolysis of hydroperoxides in the chamber simulations is not considered because of their small photolysis rates at the wavelength of the associated lamps (Jet Propulsion Laboratory, 1997). The development of the toluene–Cl mechanism is based on previous experiments and simple mechanisms of the reaction of toluene with Cl (Fantechi et al., 1998; Karlsson et al., 2001; Wang et al., 2005). The ring-retaining OHinitiated oxidation mechanisms of toluene (Calvert et al., 2002) also are used partially as an analogy for those of Clinitiated oxidation of toluene. The proposed oxidation mechanism in this study includes 52 reactions with eight organic radicals and 15 organic non-radical products that are described in Tables 3 and 4 and Fig. 5. The reaction list shown here includes those involving only inorganic species but does not include those reactions already included in CACM. In the chamber experiments, Cl2 is the photolytic source of Cl. In this modeling study, the rate of photolysis for Cl2 is calculated using the approach of Tanaka et al. (2003a) in which the photolysis rate constant (j) for Cl2 is scaled directly to that for nitrogen dioxide (NO2): jCl2 ¼ 0:033jNO2 , where jNO2 is the photolysis rate constant for NO2 at noon under clear-sky conditions in Los Angeles. The factor of 0.033 is used to account for the measured strength of lights in the chamber system, the differences in the spectrum of the lights compared to the actual solar spectrum, and differences between absorption spectra for Cl2 and NO2. The Cl-initiated oxidation of toluene proceeds via Hatom abstraction from the methyl group attached to the aromatic ring (Wang et al., 2005), leading to the formation of benzyl peroxy radical, which is converted to the corresponding alkoxy radical via its reactions with Cl. This alkoxy radical reacts with molecular oxygen to create benzaldehyde and HO2. Benzyl peroxy radical may also undergo self-reaction, yielding either benzyl alkoxy radical, benzaldehyde, and benzyl alcohol, or a benzyl dimer. The branching ratios for these three pathways are assumed to be 40%, 40%, and 20%, respectively (Karlsson et al., 2001). Benzyl hydroperoxide is generated from the reaction of benzyl peroxy radical with HO2. The reaction of benzyl peroxy radical with Cl also leads to the formation of a benzyl Criegee biradical via H-atom abstraction; the Criegee biradical is assumed to react with water to form benzoic acid. Benzaldehyde, benzyl alcohol, and benzylhydroperoxide are found to be three major gas-phase products in oxidation of toluene by Cl (Fantechi et al., 1998;

7353

Karlsson et al., 2001; Wang et al., 2005). The abstraction of the aldehydic H of benzaldehyde leads to the formation of a benzyl acyl radical, which leads to 1,4-benzoquinone and other quinone-like compounds. The two endocyclic double bonds of quinone-like compounds may then be further oxidized by Cl. Due to the existence of multiple reactive sites on the molecules of quinone-like compounds, generic products are used to represent the final products for their reaction with Cl in order to avoid excessive speculation about mechanistic pathways. Degradation products of benzoquinones are believed to be condensable (Dubtsov et al., 2006). Most rate constants used in the present study are based on the recommendations of Jenkin et al. (1997). Due to the lack of kinetic data for the reactions between Cl and specific peroxy radicals, the rate constants for the formation of the corresponding alkoxy radicals and Criegee biradicals (when possible to form mechanistically) are assumed to be 1.50E11 and 1.35E10 cm3 molecules1 s1, respectively (Maricq et al., 1994). The assumed rate constant for the reaction between Cl and an aldehyde is 2.50E10 cm3 molecules1 s1 based on the average measured values for aldehydes containing seven to ten carbon atoms (Thevenet et al., 2000). For reactions of Cl with alcohols, acids, and quinone-like compounds, rate constants available for compounds with similar molecular structures are used by analogy (Nelson et al., 1990; Olariu et al., 2000; Aranda et al., 2003). Secondary organic aerosol is assumed to form from the gas-to-particle partitioning of the SVOCs formed from Cl oxidation of toluene (Pankow, 1994). The coefficient Kom,i (m3 mg1) for condensable component i is determined by:

Kom;i ¼

Ai RT ¼ G i M0 MWom 106 gi p0L;i

(3)

where Ai and Gi are the aerosol and gas-phase concentrations (mg m3), respectively, of specie i, M0 (mg m3) is the total organic aerosol absorbent mass concentration (including both secondary and primary organic aerosol if present), R is the ideal gas constant (8.2  105 m3 atm mol1 K1), T is temperature (K), MWom is the average molecular weight (g mol1) of the absorbing organics (including both primary organic compounds and secondary products), and pL,0i is the pure component sub-cooled liquid vapor pressure (atm) of specie i at temperature T. The activity coefficient of specie i in the aerosol phase is represented by gi. As described by Colville and Griffin (2004), a mass balance constrains the total concentration of specie i (Ci, predicted by the gas-phase chemical mechanism) to be Ci ¼ Ai þ Gi. When it is recognized for chamber experiments that M0 is equal to DM0 and to the sum of the N individual Ai values, an implicit expression that can be used to estimate DM0 iteratively results: N X

Kom;i Ci 1 þ Kom;i DM0 i¼1

! 1 ¼ 0

(4)

The iteration procedures are to: (1) estimate MWom; (2) calculate Kom,i from Eq. (3) assuming unit activity coefficient (Seinfeld et al., 2001); (3) evaluate DM0 by solving Eq. (4) using Ci values from the gas-phase mechanism; (4)

7354

X. Cai et al. / Atmospheric Environment 42 (2008) 7348–7359

Table 3 Chemical species in the mechanism of the Cl-initiated oxidation of toluene Notation

Structure

Notation

Structure

Notation

Structure

Radical

Species

Radical

Species

Radical

Species

TRO21

CB-TRO21

TRO22

TRO23

TRO24

CB-TRO24

ZTRO21

TR-1

Non-radical

Products

Non-radical

Products

Non-radical

Products

TP1

TP2

TP3

TP4

TP5

TP6

TP7

TP8

TP9

TP10

TP11

TP12

No specific structure

TP15

No specific structure

TP13

No specific structure

TP14

evaluate Ai and Gi by solving simultaneously Eq. (3) and the mass balance; and (5) calculate MWom (based on the calculated Ai values) to see if it agrees with the initial value or to get a new estimate for MWom. These procedures are repeated until each equation has a tolerable error. The key parameter in Eq. (3) is the sub-cooled liquid vapor pressure, which is not available experimentally for

No specific structure

many species in the SOA system. If unavailable, the group contribution method of Joback is employed to estimate vapor–liquid critical temperatures, pressures, and boiling points, and the Riedel corresponding-states method is adopted to calculate sub-cooled liquid vapor pressures (Poling et al., 2000). However, the estimated vapor pressures are used as reference parameters and adjusted

X. Cai et al. / Atmospheric Environment 42 (2008) 7348–7359

7355

Table 4 Cl-initiated oxidation mechanism for toluene

TR1 TR2 TR3 TR4 TR5 TR6 TR7 TR8 TR9 TR10 TR11 TR12 TR13 TR14 TR15 TR16 TR17 TR18 TR19 TR20 TR21 TR22 TR23 TR24 TR25 TR26 TR27 TR28 TR29 TR30 TR31 TR32 TR33 TR34 TR35 TR36 TR37 TR38 TR39 TR40 TR41 TR42 TR43 TR44 TR45 TR46 TR47 TR48 TR49 TR50 TR51 TR52

Reactants

Products

Rate constant (cm3 molecule1 s1 or s1)

Toluene þ Cl TR-1 þ O2 TR-1 þ TRO21 ZTRO21 þ ZTRO21 TRO21 þ Cl TRO21 þ Cl TRO21 þ RO2T TRO21 þ HO2 CB-TRO21 þ H2O ZTRO21 þ O2 ZTRO21 TP1 þ Cl TP1 þ hv TRO22 þ Cl TRO22 þ RO2T TRO22 þ HO2 TRO23 þ RO2T TRO23 þ Cl TRO23 þ HO2 TRO24 þ Cl TRO24 þ Cl CB-TRO24 þ H2O TRO24 þ HO2 TRO24 þ RO2T TP2 þ Cl TP4 þ Cl TP4 þ hv TP5 þ Cl TP6 þ Cl TP7 þ Cl TP7 þ hv TP8 þ Cl TP8 þ hv TP9 þ Cl TP10 þ Cl TP10 þ hv TP11 þ Cl TP9 þ OH TP10 þ OH TP11 þ OH Toluene þ OH HCHO þ hv HCHO þ hv HCHO þ Cl Cl þ O3 HO2 þ Cl HO2 þ Cl HO2 þ ClO HOCl þ hv OH þ HCl OH þ HOCl Cl2 þ hv

HCl þ TR-1 TRO21 þ RO2T 2 ZTRO21 F4 TP1 þ F4 TP2 þ (1F4) TP3, F4 ¼ 0.67 ZTRO21 þ ClO CB-TRO21 þ HCl 0.6 ZTRO21 þ 0.2 TP1 þ 0.2 TP2 þ RO2T TP4 þ O2 TP5 þ H2O TP1 þ HO2 HCHO þ TRO22 þ RO2T HCl þ TRO23 þ RO2T HO2 þ CO þ TRO22 þ RO2T ClO þ TRO24 þ RO2T 0.7 TRO24 þ 0.3 TP6 þ 1.7 RO2T TP7 þ O2 0.3 TP5 þ 0.7 CO2 þ 0.7 TRO22 þ 1.7 RO2T ClO þ CO2 þ TRO22 þ RO2T 0.71 TP8 þ 0.71 O2 þ 0.29 TP5 þ 0.29 O3 ClO þ HO2 þ TP9 CB-TRO24 þ HCl H2O2 þ TP9 TP10 þ O2 0.8 TP9 þ 0.6 HO2 þ 0.2 TP11 þ RO2T TP1 þ HCl þ HO2 TRO21 þ HCl þ RO2T OH þ HO2 þ TP1 HCl þ CO2 þ TRO22 þ RO2T TRO24 þ HCl þ RO2T HCl þ TRO22 þ RO2T OH þ TRO24 þ RO2T TRO23 þ HCl þ RO2T OH þ CO2 þ TRO22 þ RO2T TP12 TP13 OH þ HO2 þ TP9 TP14 TP15 TP15 TP15 TP15 2 HO2 þ CO H2 þ CO HCl þ CO þ HO2 ClO þ O2 HCl þ O2 OH þ ClO HOCl þ O2 OH þ Cl H2O þ Cl H2O þ ClO 2Cl

6.20E11a 1.4E12a 5.0E11a 3.0E11a by analogy to the self-reaction of toluic peroxy radical 1.50E11a 1.35E10a 1.3E12b 1.0E11a 1.0E17b (2.5E14)  EXP(300/T) c (5.0E þ 13)  EXP(9108/T)c 2.5E10d 3.1E05e corrected for UV light strength and spectrum K5 (6.7E15)  EXP(416/T)b (3.0E13)  EXP(1250/T)  [1  exp(0.34  6)]b 5.0E12b K5 (4.3E13)  EXP(1040/T)b K5 K6 1.0E18b (3.0E13)  EXP(1250/T)  [1  EXP(0.34  6)]b 2.5E13b 4.79E11f (2.9E11)  EXP(190/T)b 1.42E07b corrected for UV light strength and spectrum (1.3E13)  EXP(1600/T)g K25 K26 K27 K26 K27 4.60E11h K34 K27 K34 4.60E12 h K38 K38 5.96E12i K13 K13 K12 (2.9E11)  EXP(260/T)i (1.8E11)  EXP(170/T)i (4.1E11)  EXP(450/T)i (4.8E13)  EXP(700/T)i 6.25E06i corrected for UV light strength and spectrum (2.6E12)  EXP(350/T)i (3.0E12)  EXP(500/T)i 3.28E03j corrected for UV light strength and spectrum

Reaction rate coefficients described as Ki are equivalent to those for reaction TRi. a Karlsson et al. (2001). b Jenkin et al. (1997). c Atkinson (2007). d Thevenet et al. (2000). e Griffin et al. (2002). f Nelson et al. (1990). g Aranda et al. (2003). h Olariu et al. (2000). i Jet Propulsion Laboratory (1997). j Tanaka et al. (2003b).

downward by two to three orders of magnitude in modeling SOA formation data. In modeling a single experiment (TLR-5), a single universal scaling factor of 0.002 is first applied to the vapor pressure of all products so that the

threshold for SOA formation is reached. Then, the vapor pressures for those compounds that have been identified in previous studies (Karlsson et al., 2001), those that are similar structurally (for example in Larsen et al. (2001)), or

7356

X. Cai et al. / Atmospheric Environment 42 (2008) 7348–7359

Fig. 5. Initial steps in the mechanism for the Cl-initiated oxidation of toluene as implemented into CACM.

that are generic quinone oxidation products (described above) are adjusted by compound specific values to fit the observed data, similarly to Jenkin (2004). This heuristic optimization approach is constrained by the observed time-dependent decay of precursor hydrocarbon concentrations, SOA mass growth data, and previous speciation data on SOA composition from toluene oxidation. A similar corrective approach was adopted in previous studies (Colville and Griffin, 2004; Jenkin, 2004; Chen and Griffin, 2005). Simulation results for experiment TLR-5 are shown in Fig. 6a,b. The toluene mixing ratio evolution is well

reproduced, indicating that Cl mixing ratios are also simulated appropriately (Fig. 6a). Simulated SOA formation is underpredicted during the initial 30 min of the experiment (Fig. 6b). From 40 to 80 min after the initiation of the experiment, the proposed mechanism significantly overestimates SOA mass. Model discrepancies reach as large as 50% of the observed value. After 80 min, simulated and observed concentrations agree within 5%, which would be expected due to modifications to product vapor pressures to match final data. The underprediction in the first 30 min is likely related to inaccuracies in applying partitioning theory to new particle formation. The overprediction

X. Cai et al. / Atmospheric Environment 42 (2008) 7348–7359

7357

Fig. 6. Simulation results for the Cl–toluene system. (a) Gas-phase measurements and simulation results for experiment TLR-5; (b) SOA measurements and simulation results for experiment TLR-5; (c) Measured and simulated SOA at 120 min in the THR series of experiments; (d) Measured and simulated SOA at 120 (TLR 4–7) or 150 min (TLR 1–3) in the TLR series of experiments.

between 40 and 80 min could be related to ignorance of particle size effects. Comparable results were found in a similar mechanistic modeling approach for SOA formation from oxidation of monoterpenes (Chen and Griffin, 2005). Experiment TLR-5 is used to calibrate the model, and the remaining toluene experiments are simulated with this calibrated version (with no further adjustments to model vapor pressures). Modeling results (using the calibrated version) for instantaneous SOA mass concentrations for all THR and TLR experiments are shown in Fig. 6c,d, respectively. (Again, the (Cl2/VOC)0 values in THR experiments are larger than those in TLR.) The overall relative errors for the simulation of all of THR and TLR experiments are 25% and 50%, respectively. Errors are larger for TLR, likely because simulated chemistry will be more sensitive to uncertainties in chlorine chemistry below the critical (Cl2/VOC)0 value. A mix of over- and underprediction is shown in Fig. 6c,d, indicating that there is no systematic bias for the model as a function of aerosol mass concentration. These results are comparable to those achieved previously using a similar mechanistic modeling approach for SOA formation from monoterpenes (Chen and Griffin, 2005). Because the mechanism and partitioning module were verified based on hydrocarbon decay, SOA as a function of time, and total SOA formation data over a range of experiments, the combined model was used to investigate the

potential chemical composition of the SOA formed in the toluene–Cl system. For experiment TLR-5, the composition of the simulated SOA is shown as a function of time in Fig. 7. The optimized calibration indicates that SOA consists predominantly of TP1 (benzaldehyde, average of 16%, maximum of 40%), TP2 (benzyl alcohol, average of 43%, maximum of 61%), and TP12 (generic quinone product, average of 33.4%, maximum of 81% at the end of the experiment), with varying contributions as a function of

Fig. 7. Simulated SOA composition as a function of time during experiment TLR-5.

7358

X. Cai et al. / Atmospheric Environment 42 (2008) 7348–7359

time. The TP1 and TP2 contributions, on a fractional basis, are largest earliest in the experiment when SOA concentrations are small. The first species to appear in the SOA phase is TP9, benzoquinone. Benzoquinone and its products are formed from the reaction of TRO24 (which comes from oxidation of benzaldehyde) with Cl and/or RO2T. Over the course of the experiment, it is clear that the nature of the SOA changes. At first, the SOA is dominated by aromatic ring-retaining products. As the mass of these compounds reacts away (to quinones and their products), the material in the SOA is replaced by the generic quinone products. However, non-zero concentrations of aromatic compounds in the SOA throughout the experiment qualitatively agree with the aromaticity of the SOA exhibited by the delta analysis. While no modeling of Q-AMS experiments was performed due to the lack of SMPS data, these results agree qualitatively with the data for the experiment described in Figs. 3 and 4. In this experiment, the earliest parts of the experiment show a spectra dominated by m/ z ¼ 29; later in experiments, the contribution of m/z 29 decreases and that of m/z 44 increases. It should be stressed, however, that the earliest parts of the experiment (less than an hour after experiments began) are associated with the largest uncertainties in Q-AMS data due to the reduced transmission efficiency associated with small particle sizes and the greatest uncertainty at smallest mass concentrations. The lack of a change in delta pattern over the course of the experiment, despite changes in m/z and simulated composition, indicate that aromatic compounds and the generic quinone oxidation products likely fragment similarly in the Q-AMS. The modeling results presented here are in contrast to the modeling of Karlsson et al. (2001) that emphasized the importance of Criegee biradical products in SOA production. No such influence is seen here. 3.5. Cl reaction importance in SOA formation Calculations of SOA formation rates from Cl- and OHinitiated oxidation of toluene in a hypothesized coastal urban area are made to evaluate their relative importance. If first-generation products are assumed to condense instantaneously to form SOA, SOA formation rates (Rj) for oxidant j are then the product of the SOA yield for the appropriate system (Yj) and the gas-phase reaction rate determined by reaction rate coefficients (kj) and the mixing ratios of the VOC and the oxidants: Rj ¼ kj[VOC][oxidant]jYj. The rate constant is 6.2e11 cm3 molecules1 sec1 for toluene with Cl (Wang et al., 2005); a corresponding value for OH oxidation is 5.63e12 (Calvert et al., 2002). Estimated peak early morning concentrations of Cl in marine areas are 104 molecules cm3 (Keene et al., 1996). A coincident concentration of OH of 5e5 molecules cm3 is assumed (Brauers et al., 1996). Based on data from urban areas (Calvert et al., 2002), a toluene mixing ratio of 30 ppbC (w16 mg m3) is assumed. The preexisting OA concentration in this hypothetical area is assumed to be 5.0 mg m3. SOA yields are calculated from the two-product model based on the parameters listed in Table 2 (highNOx$OH-initiated oxidation, low-NOx$OH-initiated oxidation, and TLR). For high-NOx$OH-initiated oxidation, the

yield used is 0.006; the corresponding value for low-NOx is 0.360. The yield used for the Cl system is 0.050. Values of Rj are 1.8, 1.0, and 58.8 ng m3 hr1, for Cl (TLR), high-NOx$OH, and low-NOx$OH oxidation, respectively. The large value for the low-NOx$OH case is dominated clearly by the very large SOA yield associated with this system. The high-NOx$OH case and the Cl case are comparable, however. For these two scenarios, the difference in gas-phase reaction rates (OH > Cl) is offset by the difference in SOA yields (Cl > OH-initiated). Thus, Cl oxidation of aromatics may contribute to OA mass growth under appropriate conditions in the early morning. Acknowledgments This work was funded by CAREER award grant ATM0327643 from the National Science Foundation. We thank J. Allan, T. Onasch, and M. Canagaratna for continued availability of Q-AMS software. Support of LZ and provision of the Q-AMS by the UNH/NOAA AIRMAP Cooperative Institute is acknowledged gratefully. References Alfarra, M.R., Paulsen, D., Gysel, M., Garforth, A.A., Dommen, J., Prevot, A.S. H., Worsnop, D.R., Baltensperger, U., Coe, H., 2006. A mass spectrometric study of secondary organic aerosols formed from the photooxidation of anthropogenic and biogenic precursors in a reaction chamber. Atmospheric Chemistry and Physics 6, 5279–5293. Allan, J.D., Jimenez, J.L., Coe, H., Bower, K.N., Williams, P.I., Worsnop, D.R., 2003. Quantitative sampling using an Aerodyne aerosol mass spectrometer: part 1: techniques of data interpretation and error analysis. Journal of Geophysical Research 108, D34090, doi:10.1029/ 2002JD002358. Aranda, A., Martinez, E., Diaz de Mera, Y., Rodriguez, A., Rodriguez, D., Cuartero, J., 2003. Low-pressure study of the reactions of Cl atoms with acrylic acid and allyl alcohol. Atmospheric Environment 37, 4361–4369. Atkinson, R., 2007. Rate constants for the atmospheric reactions of alkoxy radicals: an updated estimation method. Atmospheric Environment 41, 8468–8485. Bahreini, R., Keywood, M.D., Ng, N.L., Varutbangkul, V., Gao, S., Flagan, R.C., Seinfeld, J.H., Worsnop, D.R., Jimenez, J.L., 2005. Measurements of secondary organic aerosol from oxidation of cycloalkenes, terpenes, and m-xylene using an Aerodyne aerosol mass spectrometer. Environmental Science and Technology 39, 5674–5688. Brauers, T., et al., 1996. Intercomparison of tropospheric OH radical measurements by multiple-folded long-path laser absorption and laser induced fluorescence. Geophysical Research Letters 23, 2545–2548. Cai, X., Griffin, R.J., 2006. Secondary aerosol formation from the oxidation of biogenic hydrocarbons by chlorine atoms. Journal of Geophysical Research 111, D14206, doi:10.1029/2005JD006857. Calvert, J.G., Atkinson, R., Becker, K.H., Kamens, R.M., Seinfeld, J.H., Wallington, T.J., Yarwood, G., 2002. The Mechanisms of Atmospheric Oxidation of Aromatic Hydrocarbons. Oxford University Press, New York. Canosa-Mas, C.E., Hilary, R., Hutton-Squire, M.D., King, D.J.S., Katherine, C. T., Wayne, R.P., 1999. Laboratory kinetic studies of the reaction of chlorine atoms with species of biogenic origin: d-carene, isoprene, methacrolein, and methyl vinyl ketone. Journal of Atmospheric Chemistry 34, 163–170. Charlson, R.J., Schwartz, S.E., Hales, J.M., Cess, R.D., Coakley, J.A., Hansen, J. E., Hofmann, D.J., 1992. Climate forcing by anthropogenic aerosols. Science 255, 423–430. Chen, J., Griffin, R.J., 2005. Modeling secondary organic aerosol formation from oxidation of a-pinene, b-pinene, and d-limonene. Atmospheric Environment 39, 7731–7744. Colville, C.J., Griffin, R.J., 2004. The roles of individual oxidants in secondary organic aerosol formation from D3-carene: 2. SOA formation and oxidant contribution. Atmospheric Environment 38, 4013–4023.

X. Cai et al. / Atmospheric Environment 42 (2008) 7348–7359 Dechapanya, W., Russell, M., Allen, D.T., 2004. Estimates of anthropogenic secondary organic aerosol formation in Houston, Texas. Aerosol Science and Technology 38, 156–166. Drewnick, F., Schwab, J.J., Jayne, J.T., Canagaratna, M., Worsnop, D.R., Demerjian, K.L., 2004. Measurements of ambient aerosol composition during PTMACS-NY 2001 using an aerosol mass spectrometer. Part I: mass concentrations. Aerosol Science and Technology 38, 92–103. Dubtsov, S.N., Dultseva, G.G., Dultsev, E.N., Skubnevskaya, G.I., 2006. Investigation of aerosol formation during benzaldehyde photolysis. Journal of Physical Chemistry B 110, 645–649. Fantechi, G., Jensen, N.R., Saastad, O., Hjorth, J., Peeters, J., 1998. Reactions of Cl atoms with selected VOCs: kinetics, products and mechanisms. Journal of Atmospheric Chemistry 31, 247–267. Ganske, J.A., Berko, H.N., Finlayson-Pitts, B.J., 1992. Absorption crosssections for gaseous ClNO2 and Cl2 at 219 K d potential organic oxidant source in the marine troposphere. Journal of Geophysical Research 97, 7651–7656. Griffin, R.J., Dabdub, D., Seinfeld, J.H., 2002. Secondary organic aerosol, I. Atmospheric chemical mechanism for production of molecular constituents. Journal of Geophysical Research 107, 4332, doi:10.1029/ 2001JD000541. Hurley, M.D., Sokolov, O., Wallington, T.J., 2001. Organic aerosol formation during the atmospheric degradation of toluene. Environmental Science and Technology 35, 1358–1366. Jayne, J.T., Leard, D.C., Zhang, X., Davidovits, P., Smith, K.A., Kolb, C.E., Worsnop, D.R., 2000. Development of an aerosol mass spectrometer for size and composition analysis of submicron particles. Aerosol Science and Technology 33, 49–70. Jenkin, M.E., 2004. Modeling the formation and composition of secondary organic aerosol from alpha- and beta-pinene ozonolysis using MCM v3. Atmospheric Chemistry and Physics 4, 1741–1757. Jenkin, M.E., Saunders, S.M., Pilling, M.J., 1997. The tropospheric degradation of volatile organic compounds: a protocol for mechanism development. Atmospheric Environment 31, 81–104. Jet Propulsion Laboratory, 1997. Chemical Kinetics and Photochemical Data for Use in Stratospheric Modeling. JPL Publication. 97–4. Kalberer, M., et al., 2004. Identification of polymers as major components of atmospheric organic aerosols. Science 303, 1659–1662. Karlsson, R.S., Szente, J.J., Ball, J.C., Maricq, M.M., 2001. Homogeneous aerosol formation by the chlorine atom initiated oxidation of toluene. Journal of Physical Chemistry A 105, 82–96. Keene, W.C., Jacob, D.J., Fan, S.M., 1996. Reactive chlorine: a potential sink for dimethyl sulphide and hydrocarbons in the marine boundary layer. Atmospheric Environment 30, R1–R3. Kleindienst, T.E., Conver, T.S., McIver, C.D., Edney, E.O., 2004. Determination of secondary organic aerosol products from the photooxidation of toluene and their implications in ambient PM2.5. Journal of Atmospheric Chemistry 47, 79–100. Knipping, E.M., Dabdub, D., 2003. Impact of chlorine emissions from sea– salt aerosol on coastal urban ozone. Environmental Science and Technology 37, 275–284. Larsen, B.R., Bella, D.D., Glasius, M., Winterhalter, R., Jensen, N.R., Hjorth, J., 2001. Gas-phase OH oxidation of monoterpenes: gaseous and particulate products. Journal of Atmospheric Chemistry 38, 231–276. Lewis, A.C., Carslaw, N., Marriott, P.J., Kinghorn, R.M., Morrison, P., Lee, A.L., Bartle, K.D., Pilling, M.J., 2000. A large pool of ozone-forming carbon compounds in urban atmospheres. Nature 405, 778–781. Maricq, M.M., Szente, J.J., Kaiser, E.W., 1994. Reaction of chlorine atoms with methylperoxy and ethylperoxy radicals. Journal of Physical Chemistry 98, 2083–2089. Mazurek, M., Mason-Jones, M.C., Mason-Jones, H.D., Salmon, L.G., Cass, G. R., Hallock, K.A., Leach, M., 1997. Visibility-reducing organic aerosols in the vicinity of Grand Canyon National Park: properties observed by high resolution gas chromatography. Journal of Geophysical Research 102, 3779–3793. McLafferty, F.W., Turecek, F., 1993. Interpretation of Mass Spectra. University Science Books, Sausolito, CA. Murphy, D.M., Thomson, D.S., Mahoney, T.M.J., 1998. In situ measurements of organics, meteoritic material, mercury, and other elements in aerosols at 5 to 19 kilometers. Science 282, 1664–1669. Nelson, L., Rattigan, O., Neavyn, R., Sidebottom, H., Treacy, J., Nielsen, O.J., 1990. Absolute and relative rate constants for the reactions of hydroxyl radicals and chlorine atoms with a series of aliphatic alcohol and ethers at 298K. International Journal of Chemical Kinetics 22, 1111–1126. Ng, N.L., Kroll, J.H., Chan, A.W.H., Chhabra, P.S., Flagan, R.C., Seinfeld, J.H., 2007. Secondary organic aerosol formation from m-xylene, toluene, and benzene. Atmospheric Chemistry and Physics 7, 3909–3922.

7359

Odum, J.R., Hoffmann, T., Bowman, F., Collins, D., Flagan, R.C., Seinfeld, J.H., 1996. Gas/particle partitioning and secondary organic aerosol yields. Environmental Science and Technology 30, 2580–2585. Odum, J.R., Jungkamp, T.P.W., Griffin, R.J., Forstner, H.J.L., Flagan, R.C., Seinfeld, J.H., 1997. Aromatics, reformulated gasoline, and atmospheric organic aerosol formation. Environmental Science and Technology 31, 1890–1897. Olariu, R.I., Becker, K.H., Klotz, B., 2000. Rate coefficients for the gas phase reaction of OH radicals with selected dihydroxybenzenes and benzoquinones. International Journal of Chemical Kinetics 32, 696–702. Olivier, J.G.J., Bloos, J.P.J., Berdowski, J.J.M., Visschedijk, A.J.H., Bouwman, A.F., 1999. A 1990 global emission inventory of anthropogenic sources of carbon monoxide on 1 1 developed in the framework of EDGAR/GEIA. Chemosphere 1, 1–17. Pandis, S.N., Harley, R.A., Cass, G.R., Seinfeld, J.H., 1992. Secondary organic aerosol formation and transport. Atmospheric Environment 13, 2269–2282. Pankow, J.F., 1994. An absorption model of the gas/particle partitioning involved in the formation of secondary organic aerosol. Atmospheric Environment 28, 189–193. Pirjola, L., Kulmala, M., Wilck, M., Bischoff, A., Stratmann, F., Otto, E., 1999. Formation of sulphuric acid aerosols and cloud condensation nuclei: an expression for significant nucleation and model comparison. Journal of Aerosol Science 30, 1079–1094. Poling, B.E., Prausnitz, J.M., O’Connell, J.P., 2000. The Properties of Gases and Liquids. McGraw-Hill, New York. Pope, C.A., Burnett, R.T., Thun, M.J., Calle, E.E., Krewski, D., Ito, K., Thurston, G.D., 2002. Lung cancer, cardio-pulmonary mortality, and long-term exposure to fine particulate air pollution. Journal of the American Medical Association 287, 1132–1141. Presto, A.A., Huff Hartz, K.E., Donahue, N.M., 2005. Secondary organic aerosol production from terpene ozonolysis. 2. Effect of NOx concentration. Environmental Science and Technology 39, 7046–7054. Pszenny, A.A.P., Fischer, E.V., Russo, R.S., Sive, B.C., Varner, R.K., 2007. Estimates of Cl atom concentrations and hydrocarbon kinetic reactivity in surface air at Appledore Island, Maine (USA), during International Consortium for Atmospheric Research on Transport and Transformation/Chemistry of Halogens at the Isles of Shoals. Journal of Geophysical Research 112, D10S13, doi:10.1029/2006JD007725. Robinson, A.L., Donahue, N.M., Shrivastava, M.K., Weitkamp, E.A., Sage, A. M., Grieshop, A.P., Lane, T.E., Pierce, J.R., Pandis, S.N., 2007. Rethinking organic aerosols: semivolatile emissions and photochemical aging. Science 315, 1259–1262. Saxena, P., Hildemann, L.M., 1996. Water-soluble organics in atmospheric particles: a critical review of the literature and application of thermodynamics to identify candidate compounds. Journal of Atmospheric Chemistry 24, 57–109. Seinfeld, J.H., Erdakos, G.B., Asher, W.E., Pankow, J.F., 2001. Modeling the formation of secondary organic aerosol (SOA). 2. The predicted effects of relative humidity on aerosol formation in the a-pinene, b-pinene, sabinene, D3-carene, and cyclohexene/ozone systems. Environmental Science and Technology 35, 1806–1817. Seinfeld, J.H., Pandis, S.N., 2006. Atmospheric Chemistry and Physics: From Air Pollution to Climate Change. John Wiley, New York. Seinfeld, J.H., Pankow, J.F., 2003. Organic atmospheric particulate material. Annual Reviews of Physical Chemistry 54, 121–140. Song, C., Na, K., Cocker III, D.R., 2005. Impact of the hydrocarbon to NOx ratio on secondary organic aerosol formation. Environmental Science and Technology 39, 3143–3149. Tanaka, P.L., Allen, D.T., McDonald-Buller, E.C., Chang, S., Kimura, Y., Mullins, C.B., Yarwood, G., Neece, J.D., 2003a. Development of a chlorine mechanism for use in the carbon bond IV chemistry model. Journal of Geophysical Research 108, D44145, doi:10.1029/2002JD002432. Tanaka, P.L., et al., 2003b. Direct evidence for chlorine-enhanced urban ozone formation in Houston, Texas. Atmospheric Environment 37, 1393–1400. Thevenet, R., Mellouki, A., Bras, G.L., 2000. Kinetics of OH and Cl reactions with a series of aldehydes. International Journal of Chemical Kinetics 32, 676–685. Twomey, S., Davidson, A.K., Seton, K.J., 1978. Results of five years of observations of cloud nucleus concentration at Robertson, New South Wales. Journal of Atmospheric Sciences 35, 650–656. Wang, L., Arey, J., Atkinson, R., 2005. Reactions of chlorine atoms with a series of aromatic hydrocarbons. Environmental Science and Technology 39, 5302–5310. Zhang, Q., et al., 2007. Ubiquity and dominance of oxygenated species in organic aerosols in anthropogenically influenced Northern Hemisphere mid-latitudes. Geophysical Research Letters 34, L13801, doi: 10.1029/2007GL029979.