Simulating the Formation of Semivolatile Primary and Secondary

Apr 9, 2009 - Hourly organic aerosol predictions are evaluated using data from the Pittsburgh Air Quality Study (PAQS), and daily averaged predictions...
0 downloads 0 Views 3MB Size
Download PDF
Environ. Sci. Technol. 2009, 43, 4722–4728

Simulating the Formation of Semivolatile Primary and Secondary Organic Aerosol in a Regional Chemical Transport Model BENJAMIN N. MURPHY† AND S P Y R O S N . P A N D I S * ,†,‡ Department of Chemical Engineering, Carnegie Mellon University, 5000 Forbes Ave, Pittsburgh, Pennsylvania 15213, and Department of Chemical Engineering, University of Patras, Patra, Greece

Received November 21, 2008. Revised manuscript received February 23, 2009. Accepted March 10, 2009.

Recent developments in our understanding of atmospheric organic particulate matter formation have been used to develop a state-of-the-art organic aerosol module for regional chemical transport models (CTMs). The module has been added to the regional CTM, PMCAMx, and has been evaluated against observations in the eastern U.S. The new module uses the volatility basis set framework to simulate primary organic aerosol (POA) partitioning between the gas and particulate phases and the gas-phase oxidation of the corresponding vapors. The formation and chemical aging of secondary organic aerosol (SOA) are modeled using the same volatility distribution approach. The module uses recent results from smog chamber studies for the formation of SOA from anthropogenic and biogenic hydrocarbons. Hourly organic aerosol predictions are evaluated using data from the Pittsburgh Air Quality Study (PAQS), and daily averaged predictions for July 2001 are compared to ambient measurements from the EPA Speciated Trends Network (STN) and the Interagency Monitoring of Protected Visual Environments (IMPROVE). The model reproduces both the absolute organic aerosol concentrations in urban and rural locations as well as the high degree of oxidation of these compounds. The chemical aging of anthropogenic SOA is consistent with the ambient organic aerosol concentration field and has a significant impact on the absolute ground-level concentrations of these compounds.

Introduction Organic particulate matter makes up more than 50% of the total mass concentration of ambient aerosols in locations throughout the world (1-3). Organic particles also affect human-health (4) and are possible drivers for global cooling (5). The effects and behavior of ambient aerosols are influenced by the intrinsic properties (hygroscopicity, volatility, reactivity, etc) of the organic compounds present. Despite its importance, organic aerosol is the least understood component of the atmospheric particulate matter system. Traditionally, organic aerosol has been split into two categories : (1) primary organic aerosol (POA) emitted in the * Corresponding author phone: 412-268-3531; fax: 412-268-7139; e-mail: [email protected]. † Carnegie Mellon University. ‡ University of Patras. 4722

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 43, NO. 13, 2009

particulate phase and (2) secondary organic aerosol (SOA) emitted as gas-phase volatile organic compounds (VOCs), oxidized in the atmosphere to form low-volatility products, and then condensed to the particle phase (1, 2, 6). Gaydos et al. (7) simulated the formation of organic aerosol (OA) in a regional-scale CTM, PMCAMx (Particulate Matter Comprehensive Air quality Model with Extensions), assuming that POA is nonvolatile. SOA formation was modeled assuming that the oxidation of gas-phase precursors produces two semivolatile surrogate products per precursor, consistent with previous models (8). Although this version of PMCAMx reproduced OA concentrations reasonably well throughout the Eastern U.S. during all seasons, it predicted that most of the organic aerosol compounds in urban centers were not oxidized, whereas ambient studies have reported that the nonoxidized organic material is a small fraction of the total (9). For example, Zhang et al. (10) analyzed ambient data using an aerosol mass spectrometer (AMS) at Pittsburgh and concluded that around 30% of organic aerosol mass was hydrocarbon-like OA (HOA), and 70% was oxygenated OA (OOA). PMCAMx predicted the opposite HOA-OOA split (70-30%), thereby calling our understanding of atmospheric OA into question. Recent laboratory studies have elucidated some of the causes of this discrepancy between CTM predictions and recent ambient observations. Lipsky and Robinson (11) showed that approximately half of the organic particle mass emitted by transportation sources and wood combustion evaporates when diluted to atmospheric concentrations. This vaporized mass can then be oxidized to lower volatility products that condense to the particulate phase (12). At the same time, smog chamber experiments have shown that early studies underestimated anthropogenic SOA formation under atmospherically relevant conditions as well as the effects of further oxidation (chemical aging) of these compounds (13-15). Representation of the above processes in existing CTMs is incompatible with the existing modeling frameworks. Departing from the Odum two-product oxidation model, Lane et al. (16) implemented the volatility basis set framework (16-18) in PMCAMx for just the SOA compounds. They used four surrogate compounds with saturation concentrations at 300 K equal to 1, 10, 100, and 1000 µg m-3. A chemical aging mechanism was also employed that shifted mass from the higher to the lower volatility bins by 1 order of magnitude (one volatility bin) (12). Simulating the further oxidation of both biogenic and anthropogenic organic compounds resulted in overpredictions of the measured OA concentrations throughout the domain. Lane et al. (16) assumed all POA to be completely nonvolatile and nonreactive. Shrivastava et al. (19) reassigned the originally assumed nonvolatile POA to volatility bins using the volatility basis set, thus relaxing the assumption of nonvolatile POA. Emissions were represented by nine surrogate species, spanning the volatility range (at 300 K) from 10-2 to 106 µg m-3. This represents most of the saturation concentrations likely to be found in combustion emissions (12). They also described the chemical aging of the gas-phase POA components by assuming that they react with the hydroxyl radical, OH, with a reaction rate constant equal to 40 × 10-12 cm3 molecule-1 s-1. This reaction was assumed to produce a corresponding oxygenated POA (OPOA) species with a saturation concentration that was a factor of 10 (one volatility bin) lower than the original POA surrogate species. The study reported that taking these processes into account in PMCAMx slightly decreased the model’s ability to reproduce observed OA concentrations throughout the domain compared to the 10.1021/es803168a CCC: $40.75

 2009 American Chemical Society

Published on Web 04/09/2009

FIGURE 1. (a) Timeseries of hourly averaged base-case model predictions and ambient measurements. Dashed lines represent midnight. (b) Diurnal profile of hourly averaged base-case model predictions and ambient measurements. Measurements were taken during the Pittsburgh Air Quality Study at Pittsburgh during July 2001.

results of Karydis et al. (9). With much of the POA mass evaporating, the model was prone to underestimating ambient measurements especially in the summertime. The Shrivastava et al. (19) study, though, used the traditional two-product model and mass yields for anthropogenic SOA formation now believed to underestimate atmospherically relevant SOA production (14, 19). There is no model that integrates the above recent developments for both POA and SOA and completely describes our current understanding of atmospheric OA formation. We describe here the revision of PMCAMx (now named PMCAMx-2008), which includes the treatment of both POA and SOA using the volatility basis-set framework and the chemical aging of POA and anthropogenic SOA. The results are evaluated against existing measurements and the sensitivity of the model to uncertain aging parameters is discussed.

Organic Aerosol Partitioning Theory Gas-phase oxidation of VOCs produces an array of semivolatile products that can then condense to the particle phase.

These reactions can be represented by oxidation

VOCj. 98 Rj,1Pj,1 + Rj,2Pj,2 + ... + Rj,nPj,n

(1)

where Pj,i is the ith product of the oxidation of VOCj and Rj,i is the corresponding stoichiometric yield. The existing models follow Odum et al. (8), describing oxidation using two surrogate species per VOC precursor (typically one each of high and low volatility). Problems have emerged with this approach when applied to CTMs. First, two species are needed for every VOC, so it is computationally expensive to incorporate a large number of SOA precursors. Second, the use of only two products limits the concentration range for which this approach gives accurate results. Finally, SOA vapors may undergo further gas-phase oxidation; representing this process would require introduction of even more species (13). The volatility basis set framework (18) simplifies and addresses these issues, describing the complete volatility range of OA compounds using logarithmically spaced bins, VOL. 43, NO. 13, 2009 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

4723

FIGURE 2. Scatter plot of model predictions vs ambient measurements of daily average PM2.5 OC in µgC m-3 from the IMPROVE and STN networks. Also shown are the 1:1, 2:1, and 1:2 lines.

characterized by an effective saturation concentration, C* (µg m-3). When used to model SOA formation, four condensable gas-phase products are typically chosen with values of C* equal to 1, 10, 100, and 1000 µg m-3 (20). These products represent all SOA species, regardless of the VOC precursor. The further oxidation of these species can be represented by shifting mass from a product of high volatility to one of lower volatility. This framework can be applied to the semivolatile reacting POA as well. POA emissions are split into the corresponding volatility bins based on the appropriate emission measurements using dilution samplers (19). The resulting POA surrogate species can then be treated as semivolatile and reactive, having a range of saturation concentrations and undergoing further gas-phase oxidation by the OH radical. The volatility basis set framework thus unifies treatments of POA and SOA into a single description of organic aerosol formation. The mass transfer of OA species between the gas and particulate phases can be approximated using vapor-liquid equilibrium theory (8, 21, 22). The gas-phase mass concentration of species i, Cigas, can be related to the species effective saturation concentration, Ci*, and the mass fraction of species i in the aerosol phase, xi, by assuming a pseudoideal solution (19): C igas ) xiC *i )

C iaer C* COA i

(2)

where concentrations are in µg m-3, Ciaer is the concentration of species i in the aerosol phase and COA is the total concentration of OA in the system. The use of mass fraction instead of mole fraction replaces the dependence Ci* on the average organic aerosol molecular weight, with a dependence on the average molecular weight of the surrogate species lumped with Ci*. The activity of every species, implicitly included in the calculation of Ci*, is assumed to be constant regardless of the composition of the condensed phase (18). Finally, Ci* is temperature-dependent, and this relationship is well-described by the Clausius-Clapeyron equation.

Model Description This study uses the revised PMCAMx-2008 to simulate July 12-28, 2001 for the eastern half of the United States. The 4724

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 43, NO. 13, 2009

FIGURE 3. Average predicted ground level concentrations (µg m-3) of (a) “fresh” primary organic aerosol (POA), (b) oxidized POA including the products of IVOCs, (c) OA transported from the boundaries, (d) traditional anthropogenic SOA, and (e) biogenic SOA for the simulation period. first two days of the simulation are used for initialization and are not included in the results presented here. The 3492 × 3240 km2 domain is made up of 36 × 36 km2 grid cells and includes 14 vertical layers extending approximately 6 km into the atmosphere. PMCAMx simulates advection, dispersion, gas-phase chemistry, emission, wet/dry deposition, aerosol dynamics, and aqueous-phase chemistry of atmospheric compounds (7, 9). Karydis et al. (9) evaluated the performance of the model for all seasons for other aerosol components and found it to vary from fair to excellent. The 2002 National Emissions Inventory (NEI) version 3 is used for emissions of VOCs and primary organic aerosol (23). This inventory has been updated with vehicular emissions from MOBILE6 (24), and biogenic emissions from the Biogenic Emissions Inventory System version 3.13 (25). Following Shrivastava et al. (19), a volatility distribution is applied to the emitted POA species with 10 simulated volatility bins, ranging from 0 to 106 µg m-3 saturation concentration, (Supporting Information Table S1). This simulation also includes emissions of intermediate volatility organic compounds (IVOCs), which are distributed among the 104, 105, and 106 µg m-3 saturation concentration bins with emissions rates equal to 0.2, 0.5, and 0.8 times the original nonvolatile POA emission rate, respectively (19). The gas-phase chemical mechanism in use, SAPRC-99 (Statewide Air Pollution Research Center), includes 77 gasphase species and 217 reactions (26). Additional details regarding OA gas-phase speciation are included in the Supporting Information. SOA is split between aerosol formed from anthropogenic (ASOA) and biogenic (BSOA) precursors. Each of these types is simulated with 4 volatility bins (1, 10, 100, 1000 µg m-3). Parameters for these species (molecular weights and enthalpies of vaporization) are taken from Lane et al. (16) and can be found in Supporting Information Table S1. POA is split into two types as well, fresh (unoxidized) and oxidized, each having 10 volatility bins (0 and 10-2-106 µg m-3). These species’ partitioning parameters are based on the work of Shrivastava et al. (19) and can be found in Supporting Information Table S1. The fresh POA, since it has not been oxidized, should correspond directly to HOA measured by the AMS, whereas the OOA concentration would be the sum of the traditional SOA, the oxidized POA, and any aged organic aerosol transported into the domain from the boundary conditions. This boundary OA is heavily processed, rather nonvolatile aerosol (2). Therefore, PMCAMx-2008 does not simulate the gas-particle partitioning of OA from the boundary conditions. Boundary OA (PM2.5) concentrations are set to 0.8 µg m-3 in every cell based on measured background concentrations in the eastern U.S. (9). Although this surrogate species is expected to include both SOA and

FIGURE 4. Predicted average diurnal profiles of predicted organic aerosol concentrations at (a) four urban sites and (b) three rural sites in the domain.

TABLE 1. Statistical Metrics Describing the Model’s Performance against Available Measurements network

predicted average OC (µgC m-3)

measured average OC (µgC m-3)

bias a (µgC m-3)

error b (µgC m-3)

fractional bias c

fractional error d

correlation coefficient

IMPROVE (daily averages) STN (daily averages) PAQS(hourly averages)

1.91 2.28 5.09e

1.62 3.16 5.86 e

0.29 -0.88 -0.77e

0.72 1.33 1.96e

0.17 -0.29 -0.13

0.39 0.52 0.34

0.62 0.52 0.50

Bias ) (1)/(N)∑iN) 1 (Ci,predicted - Ci,measured) b Error ) (1)/(N)∑iN) 1 |Ci,predicted - Ci,measured| c Fractional bias ) (2)/(N)∑iN) 1 (Ci,predicted - Ci,measured)/(Ci,predicted + Ci,measured) d Fractional error ) (2)/(N)∑iN) 1 (|Ci,predicted - Ci,measured|)/(Ci,predicted + Ci,measured) e The units for this comparison are in µg m-3, not µgC m-3. a

OPOA, there is little evidence for the magnitude or variation of this split. Thus, boundary OA will be represented separately in this study. The SOA yields used in PMCAMx-2008 (Supporting Information Table S2) are based on the NOx-dependent stoichiometric yields of Lane et al. (27). However, in this study, the anthropogenic yields for the low-NOx case correspond to the high-yield case investigated by Lane et al. (16). Recent laboratory studies (13, 14) have indicated that these higher yields are closer to the truth. An SOA density of 1.5 g cm-3 is assumed (28) so the yields in Supporting Information Table S2 have been calculated by multiplying the experimentally determined volume yields by 1.5. Further gas-phase oxidation of OA vapors (chemical aging) is modeled using a second-order reaction with hydroxyl radicals. To express the decrease of volatility with aging (assuming functional groups are added to the organic vapors during oxidation), products of this reaction are shifted down one volatility bin (factor of 10 decrease in effective saturation concentration) (16, 19). While oxidation of organic vapors will sometimes lead to fragmentation instead of functionalization, a uniform saturation concentration decrease is a good first step to a more realistic aging model. The base-case

simulation ages POA and SOA from anthropogenic sources only, using a rate constant k(298 K) ) 40 × 10-12 cm3 molec-1 s-1 and 10 × 10-12 cm3 molec-1 s-1 for POA and ASOA aging, respectively. The constant for POA aging is believed to be characteristic of reactions of long chain hydrocarbons with hydroxyl radicals while the ASOA aging rate constant is based on OH oxidation of the products of aromatic VOC oxidation (29, 30). Since the ASOA species are lumped, the model applies this aging rate regardless of the precursor VOC. No biogenic SOA aging is simulated in the base-case, an assumption based on both the available laboratory studies and the results of Lane et al. (13, 16, 31). The sensitivity of our results to these rather uncertain chemical aging parameter choices will be discussed in a subsequent section.

Results and Discussion Model Evaluation. PMCAMx-2008 predictions for total organic aerosol concentrations are evaluated using hourly measurements from the Pittsburgh Air Quality Study (PAQS) (32) as well as daily average measurements from the U.S. EPA’s Speciated Trends Network (STN) (33) and the IMPROVE VOL. 43, NO. 13, 2009 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

4725

monitoring network (34). Measurements from 100 sites (46 from the STN network and 54 from the IMPROVE network) are used. The timeseries of measured and predicted total organic aerosol concentrations in Pittsburgh, Figure 1a, indicates reasonable model performance considering the uncertainties in organic aerosol emissions. Uncertainty arises from the assumed ratio of organic carbon mass (measured) to organic matter (simulated by PMCAMx-2008) (OM/OC). Polidori et al. (35) reported an OM/OC ratio equal to 1.8 at Pittsburgh during July, 2001, so the results are reported in organic matter concentrations (all Pittsburgh OC measurements have been multiplied by 1.8). PMCAMx-2008 correctly estimates the magnitude of the observations throughout the simulation period; the average model predicted total organic aerosol concentration is 5.09 µg m-3 compared to the observed average, 5.86 µg m-3. This figure is somewhat similar to the results from Gaydos et al. (7), but those results showed a pronounced temporal behavior resembling primary emissions. Figure 1a indicates OA concentrations that vary less throughout the day. Moreover, the average predicted diurnal profile agrees with that measured (Figure 1b), especially in the afternoon when SOA formation is expected to be significant. There is some discrepancy evident at night, amounting to a difference of 1-2 µg m-3 between observed and predicted organic aerosol during nighttime. This discrepancy might be the result of emission errors, nighttime boundary layer effects, or unaccounted for oxidation of organic gases. Figure 2 shows a scatter plot comparing daily average ambient measurements taken by the STN and IMPROVE networks with model predictions. The STN and IMPROVE sites are representative of urban/suburban and rural locations, respectively. Like the ambient measurements made during the PAQS study, the organic aerosol mass concentrations for these studies are measured in µgC m-3, so an OM/ OC ratio must be used. The fresh POA species, being mostly hydrocarbon-like, are assumed to have an OM/OC ratio of 1.4, whereas all oxidized species (SOA, OPOA, and OA from the boundaries) are assumed to have an OM/OC ratio of 2 (36). The 2001 STN data had not been blank corrected originally. Based on subsequent measurements in the network a constant 1 µg m-3 was subtracted from the STN data to account for handling blanks (37-40). PMCAMx agreement with IMPROVE measurements is encouraging (Table 1), with a slight overprediction, (positive bias of 0.29 µgC m-3). This indicates good performance for the oxidized OA species that dominate in rural areas. However, there is some underprediction when comparing to the STN measurements (bias -0.88 µgC m-3). This disagreement could indicate problems with the semivolatile POA volatility distribution, or the absolute urban emissions rates. A significant fraction of the scatter (error 1.33 µgC m-3) is probably introduced by the emissions given that the same inventory is used for each weekday during the simulation. The use of the new modeling framework, updated yield parameters, and blank corrected measurements results in a statistically improved performance of the model for OA. The correlation coefficients between predicted and observed values increase from 0.37-0.48 for PMCAMx-2004 by Gaydos et al. (7) to 0.50-0.62 here for the different measurement networks. Agreement in the urban areas is improved where the bias is reduced (in absolute terms) from -1.8 to -0.88 µgC m-3 and the error from 2.01 to 1.33 µgC m-3. However, the most significant improvement is in the OOA/OA ratio. PMCAMx-2008 predicts the extent of oxygenation of organic aerosol compound well, as shown by its estimation of the oxidized organic aerosol (OOA) to total OA fraction. Zhang et al. (3) reported the average OOA fraction in a number of cities throughout the world. Supporting Information Table 4726

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 43, NO. 13, 2009

S3 compares their findings for urban and rural sites with this model’s predictions. PMCAMx-2008 predicts the OOA fraction quite well at all sites, especially when compared to the results of Gaydos et al. (7); most OOA predictions are within a few percent of their measured value. The small differences are partially due to the differing measurement and prediction period. For example, PMCAMx-2008 may overpredict the OOA fraction at Pittsburgh because the AMS measurements by Zhang et al. (10) were performed in September and the simulation results are for July. The reduction in photochemical activity in September explains, at least partially, this discrepancy. However, PMCAMx-2008 can reproduce the observed dominance of oxygenated OA over both urban and rural continental areas somewhat consistent with several AMS studies analyzed by Zhang et al. (2). Predicted Spatial, Diurnal, and Vertical Variations. The averaged ground-level concentrations of “fresh” primary OA (POA), oxidized primary OA (OPOA), boundary condition OA, and secondary OA predicted by PMCAMx-2008 are shown in Figure 3. Fresh POA concentrations are highest in the urban/suburban areas and negligible in the rural areas. Traditional SOA precursors are emitted from urban and rural areas and the products of their oxidation contribute to the total organic aerosol throughout the domain. Most of the mass from anthropogenic VOCs is emitted in or near cities, whereas most biogenic VOCs are emitted in rural areas, especially in the southeast U.S. Figure 3d-e show the concentrations of SOA from these two sources. Both concentration fields exhibit a regional profile and peaks where their precursor gases are emitted. For anthropogenic SOA, urban centers see a 1-2 µg m-3 increase beyond rural concentrations, whereas for biogenic SOA, an increase of 4-6 µg m-3 is predicted in the heavily forested areas in the southeast. OPOA, aerosol from the gas-phase oxidation of evaporated fresh POA and of IVOCs, shows characteristics similar to both fresh POA and traditional SOA. Fresh POA is emitted largely within cities; upon dilution, POA may evaporate and oxidize (and thereby be reassigned to OPOA) before it is transported as POA to the rural areas. PMCAMx-2008 indicates that OA concentrations from primary emissions, once thought to affect mainly urban centers, significantly influence the rural regions of the domain. Moreover, organic gases from POA evaporation that are transported from the cities to the rural areas are expected to be heavily oxidized. Thus, their molecular properties are expected to be different from those of the emitted particles (12, 18). These changes may cause different impacts on human health and climate than expected from the primary species. PMCAMx-2008 predicts almost uniform contributions of OA from areas outside the modeling domain for this summer period (Figure 3c); the ground-level concentration is approximately equal to the value specified at the boundary (∼0.8 µg m-3). The sensitivity of model results to boundary specifications and other input parameters is discussed in the Supporting Information. The modeling results have been used to explore the predicted diurnal profile of total OA concentrations at four major cities and three rural locations (Figure 4). The variety of trends indicates that observations made about the OA diurnal profile in Pittsburgh cannot be extrapolated to other sites in the domain as they are strongly dependent on the nature of the sources in the vicinity. This is most notable in Atlanta, GA, which is influenced by biogenic SOA formation. Despite the emissions rates’ diurnal variation, the photochemical production of both SOA and OPOA, and the changes in OA partitioning with temperature, the average OA concentration profile predicted by PMCAMx-2008 in a lot of areas is relatively flat on average. Figure 5 shows the source contributions to OA in Atlanta and Pittsburgh. All primary sources and anthropogenic SOA are about the same order

FIGURE 5. Diurnal profile of organic aerosol speciated by source at (a) Atlanta and (b) Pittsburgh. The sources are identified here as nonvolatile OA from the boundaries (BC OA), unoxidized POA (Fresh POA), POA formed by the aging mechanism (OPOA), biogenic SOA (BSOA), and anthropogenic SOA (ASOA). These last two species are produced from the oxidation of traditional VOC precursors corresponding to the SOA mass yields in Table 1.

FIGURE 6. Absolute concentrations of anthropogenic secondary organic aerosol for the (a) base-case (includes aging of ASOA) and (b) no aging case. of magnitude at both locations. Biogenic SOA, on the other hand, is dominant at Atlanta, and less so at Pittsburgh. These results indicate that the average diurnal profile of the OA concentrations can be used, if there are corresponding available measurements, as a tool for the evaluation of CTM performance given the uncertainties in day to day OA and VOC emissions. The vertical profiles of OA concentrations at five of these sites (Supporting Information Figure S1) show further distinction of OA concentration predictions based on location in the domain. The urban locations (Pittsburgh, New York, and Atlanta) are predicted to experience higher OA concentrations at ground level than the rural sites. The model predicts that Duke Forest, NC, a rural location, will have consistently smaller concentrations throughout the atmosphere than those in urban areas most likely due to low emission rates in the area. The New England Coast is characterized as a rural site downwind of an urban area (in this case, the Northeast US). PMCAMx-2008 predicts that OA concentrations will be elevated through a substantial

height of the atmosphere (about 400 m) above sites where anthropogenic OA sources contribute large fractions to the total. Anthropogenic SOA Aging. Anthropogenic SOA (ASOA) concentrations are quite sensitive to the assumed aging mechanism, as shown in Figure 6. As discussed above, compounds undergoing multiple atmospheric oxidations decrease in volatility thereby increasing total condensed mass. The implementation of aging in this model increases the predicted ASOA concentrations by 1-2 µg m-3 and in some cases 3-4 µg m-3. Understanding this aging mechanism could thus have significant impacts on air quality modeling applications and source attribution of organic aerosol. The three cases explored in this study (1) aging just ASOA (base-case), (2) not aging any SOA compounds, and (3) aging ASOA and biogenic SOA (BSOA) with the same rate constant (not shown) indicate that although aging is significant for the absolute concentration of the species being aged, it is not a significant driver for the OOA fraction. The three sensitivity studies all yielded average OOA fractions above VOL. 43, NO. 13, 2009 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

4727

0.9 for the modeling domain. The major factor contributing to the increased OOA fraction is the evaporation of the nonvolatile POA. It will be important to characterize the volatility distribution of significant primary organic emissions sources carefully in order to make a prediction of organic aerosolcompositionthatagreeswithambientAMSobservations.

Supporting Information Available Additional information including three tables and one figure. This material is available free of charge via the Internet at http://pubs.acs.org.

Acknowledgments This research was supported by the EPA STAR program through the National Center for Environmental Research (NCER).

Literature Cited (1) Kanakidou, M.; Seinfeld, J. H.; Pandis, S. N.; et al. Organic aerosol and global climate modeling: a review. Atmos. Chem. Phys. 2005, 5, 1053–1123. (2) Seinfeld, J. H.; Pandis S. N. Atmospheric Chemistry and Physics: From Air Pollution to Climate Change, 2nd ed.; John Wiley and Sons, Hoboken, NJ, 2006. (3) Zhang, Q.; Jimenez, J. L.; Canagaratna, M. R. Ubiquity and dominance of oxygenated species in organic aerosols in anthropogenically-influenced Northern Hemisphere midlatitudes. Geophys. Res. Lett. 2007, 34, L13801. (4) Pope, C. A.; Dockery, D. W. Health effects of fine particulate air pollution: lines that connect. J. Air Waste Manage. 2006, 56, 709–742. (5) IPCC Forth Assessment Report: Climate Change, 2007. (6) Chung, S. H.; Seinfeld, J. H. Global distribution and climate forcing of carbonaceous aerosols. J. Geophys. Res. 2002, 107, 4407. (7) Gaydos, T. M.; Pinder, R.; Koo, B.; Fahey, K. M.; Yarwood, G.; Pandis, S. N. Development and application of a threedimensional aerosol chemical transport model, PMCAMx. Atmos. Environ. 2007, 41, 2594–2611. (8) Odum, J. R.; Hoffman, T.; Bowman, F.; Collins, D.; Flagan, R. C.; Seinfeld, J. H. Gas/particle partitioning and secondary organic aerosol yields. Environ. Sci. Technol. 1996, 30, 2580–2585. (9) Karydis, V. A.; Tsimpidi, A. P.; Pandis, S. N. Evaluation of a three-dimensional chemical transport model (PMCAMx) in the eastern United States for all four seasons. J. Geophys. Res. 2007, 112, D14211. (10) Zhang, Q.; Worsnop, D. R.; Canagaratna, M. R.; Jimenez, J. L. Hydrocarbon-like and oxygenated organic aerosols in Pittsburgh: insights into sources and processes of organic aerosols. Atmos. Chem. Phys. 2005, 5, 3289–3311. (11) Lipsky, E. M.; Robinson, A. L. Effects of dilution on fine particle mass and partitioning of semivolatile organics in diesel exhaust and wood smoke. Environ. Sci. Technol. 2006, 40, 155–162. (12) 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. Rethinking organic aerosol: semivolatile emissions and photochemical aging. Science. 2007, 315, 1259–1262. (13) Ng, N. L.; Kroll, J. H.; Keywood, M. D.; Bahreini, R.; Varutbangkul, V.; Flagan, R. C.; Seinfeld, J. H. Contribution of first- versus second-generation products to secondary organic aerosols formed in the oxidation of biogenic hydrocarbons. Environ. Sci. Technol. 2006, 40, 2283–2297. (14) Hildebrandt, L.; Donahue, N. M.; Pandis, S. N. High formation of secondary organic aerosol from the photo-oxidation of toluene Atmos. Chem. Phys. Discuss. 2009, 9, 693-733. (15) Baltensperger, U.; Kalberer, M.; Dommen, J.; et al. Secondary organic aerosols from anthropogenic and biogenic precursors. Faraday Discuss. 2005, 130, 265–278. (16) Lane, T. E.; Donahue, N. M.; Pandis, S. N. Simulating secondary organic aerosol formation using the volatility basis-set approach in a chemical transport model. Atmos. Environ. 2008a, 42, 7439– 7451. (17) Stanier, C. O.; Donahue, N. M.; Pandis, S. N. Parameterization of secondary organic aerosol mass fraction from smog chamber data. Atmos. Environ. 2008, 42, 2276–2299. (18) Donahue, N. M.; Robinson, A. L.; Stanier, C. O.; Pandis, S. N. Coupled partitioning, dilution, and chemical aging of semivolatile organics. Environ. Sci. Technol. 2006, 40, 2635–2643.

4728

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 43, NO. 13, 2009

(19) Shrivastava, M. K.; Lane, T. E.; Donahue, N. M.; Pandis, S. N.; Robinson, A. L. Effects of gas-particle partitioning and aging of primary emissions on urban and regional organic aerosol concentrations. J. Geophys. Res. 2008, 113, D18301. (20) Pathak, R. K.; Presto, A. A.; Lane, T. E.; Stanier, C. O.; Donahue, N. M.; Pandis, S. N. Ozonolysis of R-pinene: parameterization of secondary organic aerosol mass fraction. Atmos. Chem. Phys. 2007, 7, 3811–3821. (21) Pankow, J. F. An absorption model of the gas/aerosol partitioning involved in the formation of secondary organic aerosol. Atmos. Environ. 1994, 28, 189–193. (22) Strader, R.; Lurmann, F.; Pandis, S. N. Evaluation of secondary organic aerosol formation in winter. Atmos. Environ. 1999, 33, 4849–4863. (23) U.S. EPA. : 1999 National Emission Inventory Documentation and Data, Office of Air Quality Planning and Standards, NEIv3; U.S. Environmental Protection Agency Research Triangle: Park, NC, 2002; http://www.epa.gov/ttn/chief/net/1999inventory.html. (24) MOBILE6: U.S. EPA, User’s Guide to MOBILE6.1 and MOBILE6.2: Mobile Source Emission Factor Model, October 2002, Report No. EPA420-R-02-028; Office of Transportation and Air Quality: Ann Arbor, MI, 2002; http://www.epa.gov/otaq/m6.htm. (25) BEISv3.13: Schwede, D., Pouliot G., and Pierce T. Changes to the Biogenic Emissions Inventory System version 3 (BEIS3)2005: 4th Annual CMAS Models-3 Users’ Conference, Friday Center, UNC-Chapel Hill, NC, September 26-28, 2005, Available at http://www.cmascenter.org/html/2005_conference/abstracts/ 2_7.pdf. (26) Carter, W. P. Documentation of the SAPRC-99 Chemical Mechanism for VOC Reactivity Assessment; University of California.: Riverside, CA 2000. (27) Lane, T. E.; Donahue, N. M.; Pandis, S. N. Effect of NOx on secondary organic aerosol concentrations. Environ. Sci. Technol. 2008, 42, 6022–6027. (28) Kostenidou, E.; Pathak, R.; Pandis, S. N. An algorithm for the calculation of secondary organic aerosol density combining AMS and SMPS data. Aerosol Sci. Technol. 2007, 41, 1002–1010. (29) Atkinson, R. Atmospheric chemistry of VOCs and NOx. Atmos. Environ. 2000, 34, 2063–2101. (30) Atkinson, R.; Arey, J. Atmospheric degradation of volatile organic compounds. Chem. Rev. 2003, 103, 4605–4638. (31) Presto, A. A.; Donahue, N. M. Investigation of R-pinene + ozone secondary organic aerosol formation at low total aerosol mass. Environ. Sci. Technol. 2006, 40, 3536–3543. (32) Wittig, A. E.; Anderson, N.; Khlystov, A. Y.; Pandis, S. N.; Davidson, C.; Robinson, A. L. Pittsburgh air quality study overview. Atmos. Environ. 2004, 38, 3107–3125. (33) U.S. EPA. User Guide: Air Quality System. Report prepared by the U.S. EPA. April 2002, http://www.epa.gov/ttn/airs/aqs/ softw/AQSUserGuide_v1.pdf. (34) IMPROVE. IMPROVE Data Guide, August, 1995; University of California Davis: Davis, CA, 1995; http://vista.cira.colostate.edu/ improve/Publications/OtherDocs/IMPROVEDataGuide/IMPROVEDataGuide.htm. (35) Polidori, A.; Turpin, B. J.; Lim, H.; Davidson, C.; Rodenburg, L. A.; Maimone, F. Organic PM2.5: fractionation by polarity, FTIR spectroscopy, and OM/OC ratio for the Pittsburgh aerosol. Aerosol Sci. Technol. 2008, 42, 233–246. (36) Aiken, A. C.; DeCarlo, P. F.; Kroll, J. H.; et al. O/C and OM/OC ratios of primary, secondary, and ambient organic aerosols with high-resolution time-of-flight aerosol mass spectrometry. Environ. Sci. Technol. 2008, 42, 4478–4485. (37) Chow, J. C.; Watson, J.G. ; Crow, D.; Lowenthal, D. H.; Merrifield, T. Comparison of IMPROVE and NIOSH carbon measurements. Aerosol Sci. Technol. 2001, 34, 23–34. (38) Graham, J. IMPROVE/STN Comparison & Implications for Visibility and PM2.5. Presented at the MANE-VU/ MARAMA 2004 Science Meeting, 27-29 January, Baltimore, MD, 2004. (39) Lane, T. E.; Pinder, R. W.; Shrivastava, M. K.; Robinson, A. L.; Pandis, S. N. Source contributions to primary organic aerosol: comparison of the results of a source-resolved model and the chemical mass balance approach. Atmos. Environ. 2007, 41, 3758–3766. (40) Shrivastava, M. K.; Subramanian, R.; Rogge, W. F.; Robinson, A. L. Sources of organic aerosol: positive matrix factorization of molecular marker data and comparison of results from different source apportionment models. Atmos. Environ. 2007, 41, 9353–9369.

ES803168A