gas/particle partitioning of water-soluble organic aerosol...

Atmos. Chem. Phys., 9, 3613–3628, 2009www.atmos-chem-phys.net/9/3613/2009/© Author(s) 2009. This work is distributed underthe Creative Commons Attribution 3.0 License.

AtmosphericChemistry

and Physics

Gas/particle partitioning of water-soluble organic aerosol in Atlanta

C. J. Hennigan1,*, M. H. Bergin1,2, A. G. Russell1, A. Nenes2,3, and R. J. Weber2

1School of Civil and Environmental Engineering, Georgia Institute of Technology, Georgia, USA2School of Earth and Atmospheric Sciences, Georgia Institute of Technology, Georgia, USA3School of Chemical and Biomolecular Engineering, Georgia Institute of Technology, Georgia, USA∗

now at: The Center for Atmospheric Particle Studies, Carnegie Mellon University, Pittsburgh, Pennsylvania, USA

Received: 29 October 2008 – Published in Atmos. Chem. Phys. Discuss.: 8 January 2009Revised: 6 April 2009 – Accepted: 18 May 2009 – Published: 4 June 2009

Abstract. Gas and particle-phase organic carbon compoundssoluble in water (e.g., WSOC) were measured simultane-ously in Atlanta throughout the summer of 2007 to inves-tigate gas/particle partitioning of ambient secondary organicaerosol (SOA). Previous studies have established that, in theabsence of biomass burning, particulate WSOC (WSOCp) ismainly from secondary organic aerosol (SOA) production.Comparisons between WSOCp, organic carbon (OC) andelemental carbon (EC) indicate that WSOCp was a nearlycomprehensive measure of SOA in the Atlanta summer-time. WSOCp and gas-phase WSOC (WSOCg) concentra-tions both exhibited afternoon maxima, indicating that pho-tochemistry was a major route for SOA formation. An ad-ditional nighttime maximum in the WSOCg concentrationindicated a dark source for oxidized organic gases, but thiswas not accompanied by detectable increases in WSOCp. Tostudy SOA formation mechanisms, WSOC gas/particle par-titioning was investigated as a function of temperature, RH,NOx, O3, and organic aerosol mass concentration. No clearrelationship was observed between temperature and parti-tioning, possibly due to a simultaneous effect from othertemperature-dependent processes. For example, positivetemperature effects on emissions of biogenic SOA precur-sors and photochemical SOA formation may have accountedfor the observed similar proportional increases of WSOCp

and WSOCg with temperature. Relative humidity data indi-cated a linear dependence between partitioning and predictedfine particle liquid water. Lower NOx concentrations wereassociated with greater partitioning to particles, but WSOCpartitioning had no visible relation to O3 or fine particle OCmass concentration. There was, however, a relationship be-tween WSOC partitioning and the WSOCp concentration,

Correspondence to:R. Weber([email protected])

suggesting a compositional dependence between partitioningsemi-volatile gases and the absorbing organic aerosol. Com-bined, the overall results suggest two dominant SOA forma-tion processes in urban Atlanta during summer. One wasthe photochemical production of SOA from presumably bio-genic precursors that increased with the onset of sunrise andpeaked in the afternoon. The other, which showed no appar-ent diurnal pattern, involved the partitioning of semi-volatilegases to liquid water, followed by heterogeneous reactions.The co-emission of water vapor and biogenic VOCs fromvegetation may link these processes.

1 Introduction

The gas/particle partitioning of oxidized semi-volatile or-ganic compounds (SVOCs) is an integral process in the for-mation of secondary organic aerosol (SOA), making it oneof the most important routes for producing fine atmosphericparticles. At present, there is little data that characterizesthe total gas/particle partitioning of ambient SOA, largelydue to the plethora of species involved in partitioning andthe highly complex nature of SOA. Ambient studies have,however, examined the partitioning of individual secondaryorganic compounds. For example, Matsunaga et al. (2005)simultaneously measured ten carbonyls in the gas and parti-cle phases at a site near Tokyo, while Fisseha et al. (2006)made simultaneous measurements of four carboxylic acidsin the gas and particle phases in Zurich. One limitation tothese studies, and others like them, is that they characterize avery minor fraction of the total SOA (and hence, organic car-bon aerosol (OC)) mass, so their behavior may not be repre-sentative of the SOA. Additionally, heterogeneous reactionsmay alter the condensed phase component to forms difficultto detect, which would then not be considered in single com-ponent analysis partitioning studies.

Published by Copernicus Publications on behalf of the European Geosciences Union.

http://creativecommons.org/licenses/by/3.0/

3614 C. J. Hennigan et al.: Gas/particle partitioning of ambient WSOC

Smog chamber experiments have described thegas/particle partitioning behavior of total SOA gener-ated from a given volatile organic compound (VOC) or VOCmixture. Evidence from chamber experiments suggeststhat the partitioning of SVOCs between gas and particlephases is highly dependent upon the organic aerosol mass(Odum et al., 1996; Odum et al., 1997; Hoffmann et al.,1997), consistent with SOA gas/particle partitioning models(e.g., Pankow, 1994). The strong temperature dependenceof SOA formation in smog chamber experiments is alsoconsistent with partitioning theory (Takekawa et al., 2003).The translation of these studies to the ambient atmosphereremains a challenge, as current models employing thesemechanisms systematically under-predict ambient SOAconcentrations (de Gouw et al., 2005). The cause(s) of thisdiscrepancy is still not clear, and is not necessarily the resultof uncertainties in the partitioning model: other factorscould be responsible, including additional SOA precursorsnot currently considered (Goldstein and Galbally, 2007;Volkamer et al., 2009). The lack of ambient data that mayprovide clear and comprehensive mechanistic insights intothe SOA formation process inhibits an understanding of theexact causes for the modeling discrepancies.

In this study, we analyze the gas/particle partitioningof ambient SOA through extensive time-resolved measure-ments of water-soluble organic carbon compounds in the gas(WSOCg) and particle (WSOCp) phases in summertime At-lanta. In making a bulk measurement like WSOC, chemicalspecificity related to individual compounds is lost, but in-stead a major fraction of the SOA, and hence, a large fractionof total OC mass, is characterized.

2 Methods

Ambient gas and particle measurements were conductedfrom 11 May–20 September 2007 on the Georgia Instituteof Technology Campus, located near the center of Atlanta.The measurement platform was∼30–40 m above ground,and the instruments were located in the rooftop laboratoryof the Ford Environmental Science & Technology Building.

WSOCp measurements were conducted via a particle-into-liquid sampler (PILS) coupled to a Total Organic Car-bon analyzer, a method which has been detailed by Sulli-van et al. (2004). Briefly, particles with aerodynamic diam-eter less than 2.5µm, selected using a non-rotating micro-orifice impactor (Marple et al., 1991), were collected intothe liquid phase by a PILS (Orsini et al., 2003) operated at15 L min−1. The aqueous sample was then transferred to aTotal Organic Carbon (TOC) analyzer (GE Analytical) forthe semi-continuous determination of the water-soluble frac-tion of the carbonaceous aerosol. Two studies have shownthat this approach agrees to within∼10% with WSOCp ex-tracted from integrated filters (Sullivan and Weber, 2006;Miyazaki et al., 2006). The TOC analyzer was operated with

a six-minute integration time, the maximum sampling rateof the analyzer employed. Twice daily, the sample airstreamwas diverted through a filter to determine the organic car-bon content of the collection water and any interference fromVOCs potentially collected by the PILS. This dynamic blankwas subtracted from sample organic carbon values to quan-tify the ambient concentration of WSOCp. The limit of de-tection was approximately 0.1µg C m−3 and the estimateduncertainty of the method is∼10% (Sullivan et al., 2004).

WSOCg concentrations were measured with a methodsimilar to that of Anderson et al. (2008), but modified andconfigured for the present study. Sample air was pulled at21±1 L min−1 through a Teflon filter for particle removal,and immediately entered a glass mist chamber (MC) (Coferand Edahl, 1986) initially filled with 10 m L of ultra-pure(>18.0 M�, low carbon) water. The MC, which was oper-ated with a sample residence time of approximately 0.45 s,efficiently collects gases with a Henry’s Law constant greaterthan 103 M atm−1 (Spaulding et al., 2002). The collec-tion efficiency was tested using nitric acid, and found to be95.1%±2.1% (mean± 1σ ). One five-minute batch samplewas collected every ten minutes using an automated valve toopen and close flow from the MC to the vacuum pump. Oncecollected, the sample was analyzed by a Total Organic Car-bon analyzer (GE Analytical). In between sample analyses,the carbon content of the 18.0 M� ultra-pure water was de-termined and the MC was rinsed with the ultra-pure water aswell. The carbon content of the pure water was subtractedfrom the organic carbon content of the subsequent sample toquantify the organic carbon of the captured ambient organicgases.

Factory calibrations were performed on the two TOC ana-lyzers used to measure WSOCp and WSOCg, and sucrosestandards were also periodically run over the full range(50–2000 ppb C) of expected liquid concentrations to verifyproper analyzer accuracy throughout the measurement pe-riod.

Accurate determination of the concentrations measuredwith the MC depends on an accurate measure of the vol-ume of water used to collect the sample. Consequently, wa-ter evaporation effects during sample collection need to beaccounted for. Calibrations of water evaporation were per-formed using gravimetric analysis on three separate days (n

total = 56). It was found that evaporation was approximatelyconstant, regardless of ambient temperature and relative hu-midity (RH) conditions. During the calibrations, ambienttemperature exhibited a range of 18.5–32.4◦C while ambi-ent RH exhibited a range of 23–66%. With an initial vol-ume of 10 mL, the ending volume after five minutes of sam-pling was 8.75 mL± 0.12 mL (1.4%) at the 95% confidenceinterval. Because the ambient temperature and RH condi-tions encountered during calibrations covered much of therange observed during sampling, the use of a constant end-ing water volume is assumed to be valid throughout the studyperiod. At an air flow rate of 21 L min−1 and a five-minute

Atmos. Chem. Phys., 9, 3613–3628, 2009 www.atmos-chem-phys.net/9/3613/2009/

C. J. Hennigan et al.: Gas/particle partitioning of ambient WSOC 3615

sample integration time, the LOD for the WSOCg measure-ment was 0.9µg C m−3. Based on uncertainties in the air-flow (±5%), ending water volume (±1.4%), TOC analyzeruncertainty (±2%), and uncertainty in the TOC backgroundconcentration (±2%), the overall estimated uncertainty forthe WSOCg measurement is 7%.

Measurements of aerosol organic carbon (OC) and ele-mental carbon (EC) were made with a Sunset Labs OCECField Analyzer (Sunset Laboratory Inc., Tigard, OR) in ac-cordance with NIOSH method 5040 (NIOSH, 1996). Themeasurements consisted of a 45-min collection period fol-lowed by 15 min of analysis time. The data are not blank-corrected and may be systematically overestimated by about0.5µg C m3 (Peltier et al., 2007). At a sampling timeof 45 min, the method uncertainty is estimated at 20%(Peltier et al., 2007). NOx concentrations measured viachemiluminescence were taken from the Georgia Depart-ment of Natural Resources (DNR) Ambient Monitoring Net-work (http://www.georgiaepd.org/air/amp/) site co-locatedwith the aerosol measurements discussed above, while O3data were taken from a Georgia DNR site approximately5 km from the Georgia Institute of Technology campus.

Measurements were carried out for approximately 4.5months, with intermittent periods for instrument mainte-nance/repair and calibrations, such that the total number ofobservations was equivalent to approximately 3 months ofcontinuous sampling. As detailed above, the measurementswere carried out at a high time resolution; six minutes forthe WSOCp measurement, a five-minute WSOCg sample ev-ery 10 min, and hourly for the OC and EC measurements.The high time resolution and extended length of the mea-surements resulted in the collection of extensive data setsthat provided high statistical significance for the analysesperformed (n=19 049 for WSOCp, n=12 298 for WSOCg,n=1478 for OC and EC, the sample size for an averageddataset corresponds to the instrument with lower samplingrate). This may provide insights into overall trends thatwould not have been detected with a shorter data set due tothe many short-term factors that cause day-to-day variabilityin ambient SVOC and particle concentrations.

3 Results and discussion

3.1 WSOC in Atlanta

In the following analysis, infrequent periods of biomassburning influence were removed from the data set.For the entire summer, WSOCp was in the range of0.2–12.1µg C m−3 and had an average concentration of3.3±1.8µg C m−3 (mean±1σ , for n=19 049). WSOCp washighly correlated with OC (R2=0.73) and the slope (0.70µgC/µg C) indicates that a high fraction of the OC was solubleand hence likely to be secondary. (Note, the results are basedon a Deming linear regression that minimizes the distance

Fig. 1. Scatter plot of WSOCp versus WSOCg for the entire sum-mer data set and Deming linear regression results (note, using stan-dard linear regression analysis, the slope is 0.23 if the fitted line isforced through 0).

between the observed data and fitted line in both the x and ydirections and is more appropriate than regular linear regres-sion for this type of data (Cornbleet and Gochman, 1979)).The range of WSOCg concentrations observed for the en-tire summer was 1.1–73.1µg C m−3, and the mean WSOCgconcentration was 13.7µg C m−3. Overall, WSOCp andWSOCg were correlated during the summer, though therewas a large amount of scatter in the data (Fig. 1). A Dem-ing linear regression gives a slope of 0.24 and an intercept of0.27µg C m−3, as shown in the WSOCp vs. WSOCg plot ofFig. 1 (if forced through 0, standard linear regression gives aslope of 0.23).

Since WSOCp and WSOCg have oxygenated functionalgroups, it is expected that both are mostly secondary. In At-lanta, particle and gas-phase WSOC originate mainly frombiogenic VOC oxidation, since they dominate the VOC emis-sions budget (Environmental Protection Agency, NationalEmissions Inventory; Guenther et al., 1994). The correla-tion between WSOCp and WSOCg (Fig. 1) supports a com-mon source (biogenic VOCs). The large amount of scatter inthe data, however, may indicate differences that include theirformation mechanisms and atmospheric lifetimes.

The average diurnal trends of WSOCp and WSOCg areshown in Fig. 2. WSOCp reached an average daily max-imum of 3.8µg C m−3 at 14:00 Local Time (Eastern Day-light Time). On average, WSOCp steadily increased in con-centration from a morning minimum of 3.1µg C m−3 at06:00 Local Time to a daily maximum of 3.8µg C m−3 at14:00. Following the afternoon concentration maximum, theaverage WSOCp concentration decreased until it reached alevel of 3.2µg C m−3 at 20:00. During the evening and

www.atmos-chem-phys.net/9/3613/2009/ Atmos. Chem. Phys., 9, 3613–3628, 2009

http://www.georgiaepd.org/air/amp/

3616 C. J. Hennigan et al.: Gas/particle partitioning of ambient WSOCFigure 2

4.0

3.8

3.6

3.4

3.2

3.0

WS

OC

p (

µg

C m

-3)

2015105

Hour (Local time)

16

15

14

13

12

WS

OC

g (µg

C m

-3)

70

60

50

40

30

20

10

O3 (

pp

b)

WSOCp

WSOCg

O3

Fig. 2. Average diurnal profiles for O3, WSOCp, and WSOCg . The vertical bars on each data point represent the standard error (standarddeviation/

√N).

night (between 20:00 and 06:00 the next morning) the meanWSOCp concentration was relatively constant, ranging be-tween 3.1–3.3µg C m−3. Between the hours of 06:00 and20:00, diurnal profiles of WSOCp and WSOCg were highlysimilar. The average WSOCg peak concentration also oc-curred at 14:00 (14.5µg C m−3) in between minima at 06:00(12.4µg C m−3) and 19:00 (12.3µg C m−3). The mid-dayincreases in WSOCp and WSOCg concentrations suggestthat photochemical formation was one of the primary sourcesfor secondary organic gases and particles in Atlanta, butthe relatively modest enhancement (∼20%) over pre-sunriseconcentrations indicates a substantial regional WSOC back-ground and a relatively long lifetime of both classes of ma-terial. This is consistent with the findings of Weber etal. (2007), who showed the highly regional nature of WSOCp

in and around Atlanta.

Figure 2 also indicates that substantial WSOCg formationoccurred through nighttime processes that did not lead to sig-nificant enhancements in the WSOCp concentration. UnlikeWSOCp, the WSOCg concentration increased at night andhad another maximum (15.4µg C m−3) at 23:00. This night-time WSOCg increase was approximately equal to the day-time (i.e., photochemical) concentration increase, indicatingthe importance of this nighttime WSOCg source. This dif-ference led to a lower WSOCp/WSOCg slope at night (0.22)than during the day (0.26).

Ambient studies suggest that nighttime sources for bio-genic VOCs may be relatively common. Significant darkemissions of monoterpens from pine species have been ob-served (e.g., Simon et al., 1994; Hakola et al., 2000; Janson etal., 2001) and, though isoprene emissions are driven by lightand temperature (Tingey et al., 1979), Warneke et al. (2004)observed a buildup of isoprene concentrations in the late-afternoon/early evening as the OH concentration decreased.Dark biogenic VOC oxidation is known to occur through re-

actions with both O3 and the nitrate radical (NO3) (Griffinet al., 1999), though our observations imply that NO3 wasthe dominant oxidant initiating nighttime WSOCg formationfor this study. From 19:00 to midnight, when the WSOCg

nighttime concentration peaked, the average O3 concentra-tion decreased sharply from 46 ppb to 22 ppb (Fig. 2). Also,the average nighttime trend of WSOCg closely resembledthe average diurnal profile of NO3 observed in other studies(Stark et al., 2007). It is noteworthy that a reaction mecha-nism, presumably between biogenic VOCs and NO3, wouldproduce significant WSOCg and yet little WSOCp relative todaytime production. Yields ofα–pinene or isoprene reactionwith NO3 are consistent with our observations of minimalincreases in nighttime WSOCp. Hallquist et al. (1999) ob-served SOA yields from the reaction of NO3 with α – pineneof less than 1%. If the 3.1µg C m−3 increase in WSOCg(between 19:00–midnight) was due largely to anα – pinene−NO3 reaction that had an aerosol yield of 1% or less, thena correlated nighttime increase in WSOCp would not be ex-pected since the signal would not be detected above randomvariability. Since isoprene emissions are high in the South-east (Guenther et al., 1994), and since it may build up in theevening due to afternoon emissions combined with dimin-ishing OH (Warneke et al., 2004), the nighttime WSOCg in-crease may also have been due to reactions of isoprene withNO3. Ng et al. (2008) found SOA yields from the isoprene-NO3 reaction that ranged from 4.3–23.8% in terms of organicmass, or roughly 2 to 12% on a carbon mass basis (Brown etal., 2009). The isoprene SOA yields were a strong functionof the amount of isoprene reacted and total organic loading,indicating that yields in the ambient atmosphere are likely tobe on the low end of the range reported by Ng et al. (2008)since the chamber concentrations were, in general, higherthan ambient. Thus, likeα–pinene−NO3, yields of only afew percent for isoprene−NO3 reactions would not produce


C. J. Hennigan et al.: Gas/particle partitioning of ambient WSOC 3617Figure 3

3.0

2.5

2.0

1.5

1.0

0.5

Co

nce

ntr

atio

n (

µg

C m

-3)

20151050

Hour (Local time)

EC

WIOCp

Fig. 3. Average diurnal profiles of EC and Water-Insoluble Organic Carbon (WIOCp). WIOCp was not measured directly, but was calculatedas OC–WSOCp. The vertical bars on each data point represent the standard error (standard deviation/

√N).

an appreciable amount of WSOCp at the observed nighttimeWSOCg concentrations. Recent field observations suggestthat background WSOCp concentrations may limit our abil-ity to detect SOA enhancements from nighttime isoprene ox-idation (Brown et al., 2009). For relatively high WSOCp

background conditions, nighttime isoprene oxidation did notappear to contribute significantly to WSOCp concentrations.This was opposed to observations with a relatively low back-ground WSOCp concentration where nighttime isoprene ox-idation did produce a detectable WSOCp concentration in-crease. In each case, the amount of isoprene oxidized wascomparable. Although the relatively high and persistent con-centrations of WSOCp in Atlanta may be a factor, the datasuggest that if nighttime isoprene oxidation occurred, it mayhave been a significant source for nighttime WSOCg, but itwas not a significant source of WSOCp compared to daytimeproduction. Finally, smog chamber studies indicate that SOAyields from NO3 reaction with biogenic VOCs other than iso-prene andα- pinene (i.e.,β-pinene,13-carene, or sabinene,Griffin et al., 1999) are likely to be higher. For these com-pounds, the production of a significant mass of oxidized gas-phase species (e.g., WSOCg) would be accompanied by asignificant increase in SOA (e.g., WSOCp) as well. Becausethis was not observed in our ambient data, the yields are ei-ther lower, or the concentrations of these VOCs too low tosignificantly contribute to WSOCg. Overall, the reaction ofNO3 with some combination ofα-pinene and isoprene waslikely responsible for the nighttime increase in WSOCg con-centration.

3.2 WSOC partitioning

The fraction of total WSOC in the particle phase,Fp, is usedto investigate partitioning, by,

Fp=WSOCp

WSOCp+ WSOCg

(1)

Fp is related to the WSOCp/WSOCg concentration ratio:

fp=WSOCp

WSOCg

=Fp

1−Fp

(2)

There is evidence that, in the absence of significant biomassburning influence, the compounds that make up WSOCp

are largely secondary (Sullivan et al., 2006; Weber et al.,2007), although WSOCp may not be a completely compre-hensive measurement of SOA in certain locations. For ex-ample, SOA estimated by the EC tracer method in Tokyowas highly correlated with WSOCp (R2=0.70–0.79) and ob-served WSOCp/SOA slopes were 0.67–0.75 (Miyazaki et al.,2006). The same study also found 90% or more of primaryorganic aerosol to be insoluble in water. A study by Favez etal. (2008) observed prominent formation of water-insolubleSOA in Cairo, a highly arid urban center. In Atlanta how-ever, WSOCp is accounting for most, if not all, of the SOA.This can be seen from Fig. 3, where strong similarities be-tween the mean diurnal profiles of water-insoluble organiccarbon (WIOCp=OC–WSOCp) and elemental carbon (EC)are evident, indicating that it is mostly primary. Clear dif-ferences between the diurnal trends of WIOCp and WSOCpindicate very different sources between the two. Many stud-ies have established that organic compounds constitute oneof the largest fractions of PM2.5 in Atlanta (e.g., Weber et al.,2003; Chu et al., 2004), thus at∼70% (gC/gC) of the sum-mertime OC, processes contributing to WSOCp (i.e., SOA)have a large impact on the region’s air quality. Addition-ally, because of the high contribution of WSOCp to the totalSOA, these results describe the partitioning behavior of a sig-nificant fraction of SOA in an ambient atmosphere. Variousfactors that may influenceFp are investigated in the follow-ing sections.



3.2.1 TheFp-temperature relationship

According to equilibrium gas/particle partitioning theory,temperature (T ) imparts a strong effect on the gas/particlepartitioning of organic compounds (Pankow and Bidleman,1991). From Pankow (1994), an expression for the partition-ing constant,Kp, is

Kp=Cp/Mo

Cg

=760RTfom

106MWomζp◦

i

(3)

where Kp has units of m3 µg−1, Cp is the compound’sparticle-phase concentration,Mo is the mass concentration ofthe absorbing organic phase (including water), andCg is thecompound’s gas-phase concentration.Kp can also be pre-dicted from the properties of the partitioning species, whereR is the ideal gas constant,T is temperature,fom is the ab-sorbing fraction of total particulate mass,MWom is the av-erage molecular weight of the absorbing organic (and aque-ous) phase,ζ is the particle-phase activity coefficient, andp◦

i

is the saturation vapor pressure. The vapor pressures of or-ganic compounds approximately double for every 10 K tem-perature increase (Seinfeld et al., 2001). Thus, the tempera-ture effect on partitioning in the ambient atmosphere will bedominated by the temperature effect on individual compoundsaturation vapor pressures, with higher temperature favoringthe gas-phase and lower temperature favoring the particlephase. A strong temperature effect on SOA formation wasobserved in chamber studies using anthropogenic (toluene,m-xylene, 1,2,4-trimethylbenzene) and biogenic (α-pinene)VOCs (Takekawa et al., 2003).

For our ambient measurements, the box plot of Fig. 4shows that there was no well-defined relationship betweenFp and temperature (for medianFp vs. T , R2=0.02), (alsono relationship betweenfp and temperature). The ambienttemperature range was large enough (greater than 20◦C) thatany effect of temperature on WSOC partitioning should bepresent. It is likely, however, that a discernable temperatureeffect on partitioning was obscured by other temperature-dependent processes affecting WSOC and its precursor emis-sions.

Overall, concentrations of both WSOCp (R2=0.22) andWSOCg (R2=0.24) were weakly correlated with tempera-ture, however median concentrations of WSOCp (R2=0.91)and WSOCg (R2=0.97) were highly correlated with temper-ature (Fig. 5a, b). This indicates that multiple factors influ-enced WSOC concentrations, but the central tendency overthe entire summer was for higher WSOC concentrations tobe associated with higher temperatures. The high correla-tion with median concentrations accompanied with a largeamount of scatter in the overall data (evident from the spreadin 10th, 25th, 75th, and 90th percentile values in Fig. 5) im-plicates temperature as one of multiple factors that affectedWSOC concentrations. The emissions of biogenic precursorsto WSOC (Tingey et al., 1980) and the photochemical for-mation of WSOC (Tsigaridis et al., 2005) are both positively

Figure 4

0.30

0.25

0.20

0.15

0.10

Fp

3530252015

Temperature (°C)

Bin width

Fig. 4. Box plot of particulate WSOC fraction,Fp, versus Tem-perature. The box plot was generated by segregating the data intoten equally-spaced temperature bins (averageN per bin = 1040),and median values (thick horizontal bar), 25th and 75th percentiles(lower and upper box bounds, respectively), and 10th and 90th per-centiles (lower and upper whiskers, respectively) for each bin areplotted.

related to temperature, and likely explain these trends. Thissuggests that the positive effects of temperature on biogenicemissions and reactivity may dominate over the negative ef-fect of volatility on WSOCp, demonstrated by the increasein both WSOCg and WSOCp with temperature.

It is possible that a temperature effect on WSOCgas/particle partitioning did exist, but was not visible dueto the relationships WSOCp and WSOCg exhibited withtemperature (Fig. 5a, b). The slopes of the medianWSOCp-T and the median WSOCg-T correlation lines were0.199µg C/m3/◦C and 0.821µg C/m3/◦C, respectively. Ifbackground concentrations of WSOCp and WSOCg are bothassumed zero, then, using the median relationships, a nomi-nal temperature change would produce a constantFp value of0.195, which compares closely with the meanFp for the en-tire study (0.203). This supports the view that a temperature-partitioning effect was obscured and highlights the dynamicnature of the ambient system in which the measurementswere made. Specifically, an increase in temperature led toadditional inputs of WSOC (particle and gas) into the sys-tem that made the determination of a specific temperatureeffect on partitioning ambiguous. This difficulty with ambi-ent sampling underscores the need to pair ambient measure-ments with controlled laboratory experiments that can isolateexperimental variables.

3.2.2 TheFp-RH and liquid water relationship

Hennigan et al. (2008a) observed, at RH levels>70%, astrong increase inFp with RH that was likely due to liquidwater uptake by fine (PM2.5) particles. This was evident from



the similarities between theFp-RH curve and the two otherindependent predictions of liquid water uptake as a functionof RH (Fig. 6). The RH effect on gas/particle partitioningled to increases in SOA mass (0.3–0.9µg C m−3) that weresignificant in the context of ambient concentrations. Figure6 indicates a positive linear relationship betweenFp and pre-dicted liquid water concentration (linear regressionR2 be-tween medianFp and liquid water for RH>45% is 0.997,n=5). The relationship betweenFp and RH had no appar-ent diurnal variability, as it was nearly constant across day-time and nighttime conditions (Hennigan et al., 2008a). Thisexplains why, though this may be a significant source ofWSOCp in Atlanta, the only diurnal signature evident forWSOCp was the obvious photochemical source. Althoughthese results point to a Henry’s Law-type partitioning pro-cess (by Henry’s Law the ratio of particle to gas concentra-tion is directly proportional to liquid water concentration),we have noted that Henry’s Law partitioning alone cannotaccount for the significant increase inFp observed at ele-vated RH levels (Hennigan et al., 2008a). Additionally, ap-plication of Henry’s Law assumes dissolution into an idealsolution, while ambient particles are non-ideal. The liquidwater increase, even in a polluted atmosphere under highRH (>90%), is not enough to simply dissolve soluble semi-volatile gases (e.g., acetic acid or formic acid, etc.) and ac-count for the SOA increase (Hennigan et al., 2008a). Using10µg m−3 liquid water concentration and anFp value of 0.3(approximately equal to the observedFp above 90% RH),we calculate an effective Henry’s Law constant for the to-tal WSOC of∼2×109 M atm−1. This value agrees well withthe experimentally-derived value of Volkamer et al. (2009),and suggests that condensed phase reactions are playing animportant role in WSOCp formation in Atlanta.

Seinfeld et al. (2001) predict a substantial influence of liq-uid water on SOA formation due to the effect of the uptakeof water on two terms in Eq. (3),MWom andζ . Though theeffects ofMWom andζ due to water uptake may be in compe-tition, depending on the hydrophilic nature of the partition-ing SVOCs, Seinfeld et al. (2001) predicted overall higherKp values (more partitioning to the particle phase) at higherliquid water levels. Pun and Seigneur (2007) also predict anenhancing effect of aerosol water on SOA formation due tothe greater absorptive capacity of added water (the role inMo, Eq. 3), its decreasing effect on theMWom term (thus in-creasingKp), and acid-catalyzed oligomer formation in theaqueous phase from glyoxal. Chang and Pankow (2008) pro-pose an SOA model that incorporates the effects of water onMWom andζ , as shown by Seinfeld et al. (2001), while alsoincluding phase separation considerations.

At present, understanding the effect of humidity on SOAformation is incomplete. It is known that water vapor af-fects the oxidation pathway of terpene ozonolysis, and thus,SOA formation by this mechanism (e.g., Bonn and Moort-gat, 2002; Jonsson et al., 2006; Stenby et al., 2007). Bonn etal. (2002) showed that, strictly through an effect on gas-phase

Figure 5

7

6

5

4

3

2

1

WS

OC

p (

µg

C m

-3)

35302520

Temperature (°C)

R2 = 0.91 (median line)

Bin width

30

25

20

15

10

5

WS

OC

g (

µg

C m

-3)

35302520

Temperature (°C)

R2 = 0.97 (median line)

Bin width

A

B

Fig. 5. Box plots of WSOCp (a) and WSOCg (b) versus Tem-perature. The plots show median values (thick horizontal bar),25th and 75th percentiles (lower and upper box bounds, respec-tively), and 10th and 90th percentiles (lower and upper whiskers,respectively) for each bin. For the WSOCp-T plot (a), the averageN per bin = 1803, and for the WSOCg-T plot, the averageN perbin = 1165.

chemistry, water vapor is likely to influence SOA formationonly through oxidation processes initiated by O3, and not OHor NO3. Experimentally, the effect of liquid water on SOAformation appears to be variable. Smog chamber studies byCocker et al. (2001a, b) found no enhancement in SOA for-mation fromα–pinene, m-xylene, or 1,3,5-trimethylbenzenedue to liquid water. Edney et al. (2000) also found nowater effect on SOA formation in smog chamber studieswith toluene as the precursor VOC. In contrast, Volkameret al. (2009) observed a positive linear dependence betweenacetylene (C2H2) SOA yields and the liquid water content ofparticles in chamber experiments. This was attributed to the



gas phase hydroxyl radical (OH) oxidation of C2H2 to pro-duce glyoxal and its subsequent uptake and reaction in theaqueous phase. The variable effects of liquid water observedbetween smog chamber studies may be due to differing ex-perimental conditions, or to differences in the water solubili-ties of the oxidation products from different precursor VOCs.Overall, our ambient results showed a strong effect of RH onFp, and so are in better agreement with the smog chamberresults of Volkamer et al. (2009) and less-so with the tradi-tional partitioning theory involvingKp.

3.2.3 TheFp-NOx relationship

In addition to relative humidity, the data demonstrate thatNOx levels impactedFp as well. A box plot ofFp versusRH was constructed after sorting the data according to NOxmixing ratios, and mean values are plotted for the high NOxand low NOx data (Fig. 7a). The mean NOx mixing ratio forthe high NOx data was 33±23 ppb (mean±1σ ) while thatfor the low NOx data was 6±2 ppb (mean±1σ ). Below80% RH, low NOx conditions were associated with highervalues of Fp (mean=0.208), while high NOx conditionswere associated with lower values ofFp (mean = 0.180).Overall, the meanFp difference (0.028) below 80% RHwas statistically significant at the 99.9% confidence inter-val (t=7.793 from student’s t-test,df =1046). Multiplyingthis 1Fp by the mean total WSOC concentration below80% RH (18.58µg C m−3) yields a WSOCp enhancementfrom this difference of 0.52µg C m−3. The difference inFp

values between high NOx and low NOx regimes (0.028) mayappear small, despite its statistical significance. However,the difference is non-negligible given that the mean WSOCp

concentration for the entire summer was 3.3µg C m−3.

The difference in WSOC partitioning with NOx levelscould explain theFp differences observed between the night-time and the daytime, where generally more WSOC was par-titioned in the particle phase during the day than at night(Hennigan et al., 2008a). Though it is possible that theday/night differences inFp were not driven by changes inNOx levels, the evidence for a NOx effect is compelling.Median Fp values had a strong negative linear correlation(R=−0.97) to the NOx mixing ratio (Fig. 7b). (Note, corre-lation remains strong (R=−0.84) even with the highest NOxpoint removed). The data was also subdivided into day/nightperiods to investigate the role of NOx during these times.When only daytime data was considered, the meanFp valuecorresponding to the top 15% of NOx mixing ratios was sta-tistically lower (at the 99.5% confidence level) than the meanFp value for the bottom 15% of NOx mixing ratios (t=3.111from student’s t-test,df =264). This was also the case fornighttime only data (t=4.676 from student’s t-test,df =335)and it supports the existence of a NOx effect onFp versusan indirect impact from another effect, such as oxidant (OH,O3) concentration.

Figure 6

4

3

2

1

0

f(RH

)

9080706050403020

Relative Humidity (%)

0.40

0.35

0.30

0.25

0.20

0.15

Fp

50

40

30

20

10

0

-10

ISO

RR

OP

IA liq

uid

wa

ter (µ

g m

-3) f(RH), from Malm and Day [2001]

ISORROPIA-predicted liquid water Bin width

Fig. 6. Relationship between partitioning and liquid water uptake,adapted from Hennigan et al. (2008a). The box and whiskers repre-sent ambientFp data, with median values (thick horizontal bar),25th and 75th percentiles (lower and upper box bounds, respec-tively), and 10th and 90th percentiles (lower and upper whiskers,respectively) shown for each bin. ISORROPIA-predicted (Neneset al., 1998; Fountoukis and Nenes, 2007) liquid water as a func-tion of RH is based on an average inorganic aerosol composition(see Hennigan et al. (2008a) for discussion), and the f (RH) traceis a measure of the scattering enhancement due to particulate wateruptake reported by Malm and Day (2001).

In general, for a given precursor VOC concentration, SOAyields from the reaction of hydrocarbons with ten or fewercarbon atoms are significantly higher at low NOx levels com-pared to high NOx levels (Kroll and Seinfeld, 2008). Thesesmall hydrocarbons include isoprene (Kroll et al., 2006; Pan-dis et al., 1991) and monoterpenes (Presto et al., 2005; Ng etal., 2007a), biogenic species with high emission rates in theSoutheast (Guenther et al., 1994). Conversely, SOA yieldsfrom the reaction of larger hydrocarbons are, in general,highest at high NOx levels (Kroll and Seinfeld, 2008). TheNOx effect on SOA yields arises from its substantial influ-ence on the oxidant initiating SOA formation (i.e., hydroxylradical, nitrate radical, or ozone) and on the fate of the RO2and RO radicals (Kroll and Seinfeld, 2008). In general, reac-tion of smaller VOCs in the presence of higher NOx concen-trations leads to the formation of higher volatility products(e.g., fragmentation and formation of short aldehydes) com-pared to smaller VOC reactions in low NOx conditions thatproduce lower volatility products (e.g., hydroperoxides andacids from reaction of RO2+HO2). As isoprene and monoter-penes are expected to be the major SOA precursors in thesoutheastern US, the observed relationship betweenFp andNOx is qualitatively consistent with the smog chamber stud-ies which observed a similar phenomenon. It is noted thatthe absolute NOx concentration may not be a critical param-eter in VOC oxidation; rather, it is likely the VOC: NOx ratioin the air mass during the period of SOA formation which



impacts the oxidation product distribution (Kroll and Sein-feld, 2008). For this study, parent VOCs were not measured,and thus an analysis ofFp as a function of the VOC: NOxratio is not possible. This may partly explain the large spreadobserved in Fig. 7b, as the parent VOC concentrations werelikely highly variable for a given NOx concentration.

A limitation with the non-chemically specific approachused here is thatFp can depend on both the oxidation processleading to SVOCs and partitioning, similar to the process de-scribed by the Yield (Seinfeld and Pandis, 1998). Thus, thisdata does not provide evidence whether the observed NOxeffect is due to the types of SVOCs produced under differ-ent NOx concentrations, or if NOx plays a role in partition-ing. Given the smog chamber results, the NOx effect on theSVOCs product distribution seems the most likely cause forthe observed effect onFp.

In contrast to NOx, ozone (O3) concentrations did not ap-pear to impact WSOC partitioning at all. The meanFp

at O3 concentrations above 70 ppb (0.211) was not statisti-cally different from the meanFp at O3 concentrations be-low 20 ppb (0.214). This result is in contrast to high aerosolyields from O3 reactions with biogenic VOCs that have beenobserved in smog chamber studies (e.g., Griffin et al., 1999).It is possible, though, that O3 was prominently involved inVOC oxidation but that O3 was abundant, and VOCs lim-iting, in O3-VOC reactions and thus little dependence wasseen betweenFp and O3.

3.2.4 TheFp-organic aerosol mass relationship

Equation (3), along with extensive chamber experimental re-sults (Odum et al., 1996; Odum et al., 1997; Hoffmann et al.,1997), suggests that the mass concentration of the absorbingorganic phase,Mo, will also impact equilibrium gas/particlepartitioning. The relationship suggests that, at a given tem-perature, a higherMo will lead to a higher fraction of a par-titioning compound in the particle phase due to a greater ca-pacity (i.e., volume) of absorbing medium for the partition-ing of SVOCs. Figure 8a shows a box plot ofFp as a functionof the organic carbon aerosol (OC) mass concentration andsuggests that anFp−Mo (i.e., Fp−OC) relationship did notexist across the entire summer. When the data were binnedinto smaller temperature ranges (<20◦C, 20–25◦C, 25–30◦C,>30◦C), the same trend was observed as that for the overallanalysis; generally no correlation observed betweenFp andOC. The total OC concentration was used because of (1) thestrong dependence of SOA yields and partitioning on the to-tal concentration (by mass) of organic aerosol in smog cham-ber studies (Odum et al., 1996), and (2) the substantial pre-dictive capability of this method when used for subsequentsmog chamber experiments of SOA formation from complexVOC mixtures (Odum et al., 1997). For these reasons, thisformulation has been widely applied to air quality models,but these data suggest this may not be appropriate in all en-vironments.

Figure 7

0.35

0.30

0.25

0.20

0.15

Fp

90807060504030


High NOx (mean Fp)

Low NOx (mean Fp)

0.30

0.25

0.20

0.15

0.10

Fp

806040200NOx (ppb)

r = -0.97(median line)

A

B

Fig. 7. Relationship between NOx andFp. MeanFp versus RHfor data sorted by NOx mixing ratio(a) (Data represent the highestand lowest 35% of NOx mixing ratios). Particulate WSOC fraction,Fp, as a function of NOx mixing ratio (b). The data were binnedaccording to NOx mixing ratios and median (thick horizontal line),25th and 75th percentiles (lower and upper box), and 10th and 90thpercentiles (lower and upper whiskers) are shown for each bin (av-erageN per bin = 197).

Following the traditional line of reasoning on SOA for-mation, becauseFp describes the partitioning of only water-soluble carbon compounds, WSOCp may be a more appro-priate partitioning medium than OC due to the higher chemi-cal similarities between WSOCp and WSOCg compared tosimilarities between OC and WSOCg. This greater simi-larity arises because WSOCg is oxygenated and thus is pre-dominantly secondary like WSOCp, while OC is comprisedof significant fractions of both secondary and primary com-ponents. Alternatively, heterogeneous reactions involvingWSOCp would also be expected to make investigatingFp



relative to WSOCp a more appropriate parameter than OC.Using WSOCp concentrations as the mass of the absorb-

ing organic phase, Fig. 8b does suggest a relationship be-tween gas/particle partitioning and existing organic aerosolmass. Higher WSOCp concentrations were associated witha larger fraction of total WSOC partitioned in the particlephase, however, a dependence was only seen for WSOCp

concentrations below∼4µg C m−3. That the dependencebetweenFp and WSOCp appeared to diminish and end atWSOCp above∼4µg C m−3 may suggest that the “average”WSOCp sampled in the Atlanta summertime was relativelynon-volatile (Seinfeld and Pandis, 1998; Ng et al., 2007b).Lanz et al. (2007) found that approximately 2/3 of the oxy-genated aerosol (interpreted as SOA) was non-volatile inSwitzerland, where, like Atlanta, most of the SOA is frombiogenic precursors in summer.

Note thatFp and WSOCp are related through Eq. (1) andit cannot be ruled out that the behavior of Fig. 8b was an ar-tifact of this Fp-WSOCp relationship. However, since themagnitude of WSOCg was significantly larger than WSOCp,weakening anyFp-WSOCp direct dependence, there is rea-son to believe it is real. Additional analyses support thisview, and suggest that WSOC partitioning was impacted bythe WSOCp concentration. (A number of tests were under-taken, including applying the Deming regression results ofFig. 1 to predictFp, and testing randomly generated log-normal distributions of WSOCp and WSOCg based on ob-served medians and geometric standard deviations. Neithertest resulted in anFp-WSOCp dependence similar to the ob-servations of Fig. 8b. Furthermore, plottingFp vs. WSOCpis directly analogous to the plots of SOA Yield vs.Mo (e.g.,Odum et al., 1996) used to formulate current partitioning the-ory).

An Fp dependence on WSOCp was also evident whenthe WSOCp concentration was incorporated into theFp-RHanalysis. When the data were sorted according to WSOCp

concentrations (Fig. 9a), there was a substantial difference inmeanFp values plotted against RH for high WSOCp con-centrations (top 35% of data, mean WSOCp=5.3µg C m−3)compared with low WSOCp concentrations (bottom 35% ofdata, mean WSOCp=1.7µg C m−3). All of the eight pointsare statistically different at the 99.9% confidence interval.Note that this high versus low WSOCp range spans the tran-sition at∼4µg C m−3 in Fig. 8b (i.e., the transition from aFp–WSOCp dependence to no dependence).

The differences between theFp–OC andFp–WSOCp re-lationships may be due to composition differences betweenthe two systems and may suggest that the chemical natureof the absorbing organic phase is a meaningful parameterto the partitioning process. Specifically, it appears as ifthe total organic aerosol may not be acting as a partitioningmedium for the condensation of water-soluble organic com-pounds. Rather, only condensed-phase compounds that arewater-soluble (and water, itself) participate in the uptake ofWSOCg. This is in general agreement with the results of

Figure 8

0.35

0.30

0.25

0.20

0.15

0.10

Fp

12108642

OC (µg C m-3

)

Bin width

0.35

0.30

0.25

0.20

0.15

0.10

0.05

0.00

Fp

1086420WSOCp (µg C m

-3)

Bin width

A

B

Fig. 8. WSOC partitioning (Fp) as a function of OC concentra-tion (a) and WSOCp concentration(b). For the box plot of par-ticulate WSOC fraction,Fp, versus OC (a), which shows medianvalues (thick horizontal bar), 25th and 75th percentiles (lower andupper box bounds, respectively), and 10th and 90th percentiles(lower and upper whiskers, respectively) for each bin, averageN

per bin = 139. For the box plot ofFp versus WSOCp (b), averageN per bin = 1220. The curve is forced through zero based on thedefinition ofFp (Eq. 1).

Song et al. (2007), who observed no relationship betweenthe mass of primary organic aerosol (POA) and the yieldof SOA from α-pinene, implying that the POA did not ac-tively participate in the SOA formation. Additionally, Volka-mer et al. (2009) observed a significant seed composition ef-fect on SOA formation from C2H2: yields using ammoniumsulfate plus fulvic acid seed particles were a factor of 3–4higher than yields using only ammonium sulfate seed. Theabsorbing organic plus aqueous phase has also demonstrateda compositional effect on the partitioning of semi-volatile



organics in model predictions. Seinfeld et al. (2001) foundthat the partitioning of hydrophilic compounds to the parti-cle phase was enhanced by particulate water uptake whilethat for hydrophobic compounds was diminished by wateruptake. Bowman and Karamalegos (2002) found a compo-sitional effect on SOA mass concentrations in model simu-lations based on the effect of the composition-dependent ac-tivity coefficient, ζ , on Kp. It is noteworthy that the dif-ferences inFp versus WSOCp and OC were present de-spite an observed high correlation between WSOCp and OC(R2=0.73, see above), and despite WSOCp accounting for,on average, 70% of OC. This implies that approximately30% of OC was insoluble in water, and similarities betweenthis water-insoluble organic carbon fraction and the EC diur-nal profiles suggest that these insoluble organic compounds(calculated as OC-WSOCp) were largely primary (Fig. 3).These primary compounds are chemically much differentfrom WSOCp, including significant differences in oxida-tion, functional groups, and polarity (Saxena and Hildemann,1996). The primary component of OC, and its differentchemical character from that of WSOC, may in part explainthe lack of anFp dependence on OC. Though the generaltrends in data and fit in Fig. 8b are somewhat similar to thesmog chamber SOA Yield-Mo relationship found by Odumet al., (1996), where in that caseMo was OC mass formedin the chamber, it is also possible that theFp-WSOCp re-lationship resulted from a completely different process thanabsorptive partitioning. Additionally, the results in Fig. 9aand b suggest that RH and WSOCp impact WSOC parti-tioning through different means. The high RH analysis inFig. 9b only included data with RH levels above 65% (meanRH=78.2%) while the low RH analysis only included datawith RH levels below 50% (mean RH=38.5%). This con-trasts how WSOCp affected partitioning for very wet parti-cles and those with significantly less water.

The results in Fig. 9a and b may indicate that the pro-cess is similar on particles containing an abundance of liquidwater and particles with a modest liquid water contribution.This has two implications. First, the role that the particleorganic component (e.g., WSOCp) plays in partitioning ispresent even when particle H2O concentrations likely far ex-ceeded WSOCp concentrations. This behavior would not beexpected if partitioning were described solely by the parti-tioning coefficient,Kp. As the liquid water content of par-ticles increases, the importance of water in the partitioningprocess should increase and the importance of the WSOCp

concentration should diminish. This is due to the significantimpact of water on the average molecular weight of the ab-sorbing organic plus aqueous phase,MWom, and on the in-crease it would bring to the total mass of the absorbing or-ganic plus aqueous phase,Mo (Eq. 3). For example, at RHlevels above 90%, the liquid water concentration may be anorder of magnitude higher than the WSOCp concentration,and so would dictate any relationship withMo. Our ambi-ent results suggest that this did not occur with WSOC par-

Figure 9

0.40

0.35

0.30

0.25

0.20

0.15

Fp

9080706050403020


High WSOCp (mean Fp)

Low WSOCp (mean Fp)

0.35

0.30

0.25

0.20

0.15

0.10

0.05

0.00

Fp

86420WSOCp (µg C m

-3)

High RH (mean Fp)

Low RH (mean Fp)

High RH = RH > 69% (mean = 78.2%)

Low RH = RH < 50% (mean = 38.5%)

A

B

Fig. 9. WSOC partitioning as a function of RH andWSOCp concentration. MeanFp values versus RH forthe highest (mean WSOCp = 5.3µg C m−3) and lowest (meanWSOCp = 1.7µg C m−3) 35% of WSOCp concentrations(a).Mean Fp values as a function WSOCp, segregated for high andlow RH levels(b). The two data points at (0, 0) are not measured,but are based on the definition ofFp (Eq. 1). Note that the curvesonly represent the central tendency and that there is considerablescatter about the mean (e.g., see Fig. 8b)

titioning: at the highest RH levels, the WSOCp concentra-tion still appears to be important (Fig. 9a, b). It indicates thatWSOCp and liquid water both somewhat independently con-tribute to WSOC gas/particle partitioning. As an example,the liquid water may affect absorption of the water-solublegases, and the WSOCp in the particle may influence theheterogeneous production of lower vapor pressure products.This qualitatively agrees with the conceptual models pro-posed by Kroll et al. (2007) and Chan et al. (2007), who sug-gest that condensed phase reactions may significantly affect



the partitioning of SVOCs. Our results indicate that particle-bound water is critical to enhanced SVOC partitioning andheterogeneous reaction. The types of heterogeneous reac-tions that take place in the liquid particle may account forthe observedFp dependence on WSOCp. Secondly, becausethe curve for the much dryer particles (Fig. 9b) is essentiallyidentical in shape to the wet particles, the overall partitioningmechanism appears to be similar throughout the RH rangemeasured: the role of water is not just confined to RH>80%when particles rapidly absorb water and are very wet. In anambient study, Khlystov et al. (2005) found that fine parti-cles contained water at RH levels as low as 20–30% in thesummer, compared to winter conditions in which fine parti-cles were essentially dry below∼60% RH. This water asso-ciated with fine particles, even at low abundances, may playan important role in heterogeneous chemistry leading to SOAformation.

There is substantial evidence from both laboratory andambient studies suggesting that heterogeneous chemical re-actions are important. Limbeck et al. (2003) observed en-hanced isoprene SOA yields in the presence of acidic seedparticles compared to yields using neutral seed particles.Kroll et al. (2007) found systematically higher SOA yieldsfrom aromatic VOCs in the presence of ammonium sulfateseed particles compared to experiments in which no seedparticles were present. In experiments designed to simu-late the chemistry occurring within cloud droplets, Altieri etal. (2008) observed the formation of oligomers from methyl-glyoxal, while Volkamer et al. (2009) observed SOA for-mation from acetylene in smog chamber experiments andattributed this to the formation of glyoxal and its subse-quent uptake and reaction in particles. The results of Volka-mer et al. (2009) included a linear correlation between SOAyields and aerosol liquid water concentrations, and an ap-parent effect of the existing organic aerosol composition onSOA yields as well, very similar to our findings. Perhapsthe strongest proof of heterogeneous reactions is the pres-ence of macromolecular organic compounds, or oligomers,as components of organic aerosol (e.g., Kalberer et al.,2006; Denkenberger et al., 2007). It is highly unlikely thatoligomer formation occurs in the gas-phase, thus their pres-ence in ambient particles confirms the occurrence of particle-phase reactions as well. In relation to heterogeneous chem-istry and the gas/particle partitioning of SVOCs, model sim-ulations by Chan et al. (2007) demonstrate that the rates ofSOA formation and the amount of SOA formed may be im-pacted by extensive heterogeneous reactions.

3.3 General discussions/implications

The above findings are based on measurements made overan extended period in the ambient atmosphere, hencethey are relevant for regulatory and modeling applications.First, the observed NOx effect onFp could be importantfor control strategies of O3 and PM2.5, criteria pollutants

which annually exceed EPA attainment limits in Atlanta(http://www.epa.gov/oar/oaqps/greenbk/). At high NOx lev-els, Fp values were lower compared to those at low NOxlevels, indicating that a higher fraction of the total WSOCwas in the particle phase when NOx levels were low. Atlantais a NOx-limited environment, which suggests NOx reduc-tions may be the most effective means of controlling O3 con-centrations. The present results indicate that mitigating O3concentrations through NOx reductions could increase PM2.5mass concentrations. The importance of a NOx feedback onPM2.5 levels to NOx control is noteworthy and should be ex-amined in more detail. (Note that decreasing NOx may alsodecrease SOA yields of certain compounds like sesquiterpensand large (C>12) alkanes (Kroll and Seinfeld, 2008), thoughthe overall contribution of these compounds to the SOA bud-get is small in the Southeast. Thus, the feedback potential islikely greater with VOCs which exhibit inverse SOA yieldswith NOx, though this question should be examined as well).There is evidence, however, that some other anthropogeniccomponent may also influence WSOCp production, signify-ing that there may be additional unidentified processes lead-ing to SOA formation that could still restrict accurate predic-tion (Weber et al., 2007; Hennigan et al., 2008b).

The results presented in this study suggest that traditionalSOA theory based solely on the partitioning of SVOCs asdescribed byKp formulation (Eq. 3) may not be a suitablepredictor of SOA in Atlanta. SOA partitioning theory pre-dicts a strong relationship betweenKp (i.e., Fp) and OC(e.g., Odum et al., 1996) that was not seen in our ambientdata. Though we note apparent differences between the ap-plication of theory based on the smog chamber results ofOdum et al. (1996, 1997) and the present study, the smogchamber results, themselves, are actually in agreement withour observations. The absorbing organic medium,Mo, inthose experiments is the SOA which was formed in the ab-sence of any primary organic aerosol. It is thus likely thatMo, even the low-volatility components, is chemically sim-ilar to the SVOCs whose partitioning is enhanced by addedorganic aerosol since they arise from oxidation of the sameVOCs. The real discrepancy arises when this theory is ap-plied to the ambient atmosphere and the total organic aerosolis assumed to contribute wholly and equally to the absorptivepartitioning of different SVOCs. Our results suggest that thechemical composition ofMo is a critical factor in the par-titioning process. This was apparent through the relation-ship observed betweenFp and WSOCp, though it differedfrom the relationships observed in chamber studies.Kp the-ory predicts a strong effect of water on partitioning, and astrongFp-RH effect was seen in the ambient results. How-ever,Kp theory also predicts that as the mass of liquid waterexceeds WSOCp (or OC), Fp should become increasinglyinsensitive to WSOCp. This behavior was not observed inthe ambient data, where the WSOCp concentration remainedan important factor for WSOC gas/particle partitioning evenat the highest RH levels. A mechanism that incorporates both


http://www.epa.gov/oar/oaqps/greenbk/


a partitioning of SVOCs to fine particle water and the reac-tion of organic compounds in the aqueous phase may bet-ter represent the partitioning that occurs in the Atlanta sum-mer. This enhancement in particle phase partitioning, alongwith a photochemical mechanism, represent the major SOAformation pathways identified in the present study. A pre-vious study has shown that these processes may be related(Hennigan et al., 2008b). In that study, periodic episodes ofhigh correlation between WSOCp and water vapor were ob-served throughout the summer. The co-emission of biogenicVOCs and water vapor from vegetation may have been re-sponsible for the correlation. Subsequent oxidation of thebiogenic VOCs and uptake of the water vapor to the aerosolphase upon ambient temperature changes may have increasedSOA formation further, according to the mechanism sug-gested above.

Further ambient studies are required to investigate if thefindings presented here apply to other locations as well, andidentify the SVOCs and possible heterogeneous chemicalmechanisms. Finally, an important issue not investigated isthe reversibility of the observed partitioning, and how thisvaries with RH. There is evidence from an earlier study inMexico City that a significant fraction (20 to 50%) of theWSOCp formed during a period of high RH (∼90%) sub-sequently evaporated when RH dropped a few hours later(Hennigan et al., 2008c). A translation of results from Mex-ico City to Atlanta is uncertain, as Mexico City VOCs arepredominantly anthropogenic (Volkamer et al., 2006) whileAtlanta’s are predominantly biogenic. Additionally, the pro-cesses affecting aerosol formation (i.e., emission profiles andphotochemistry) may be highly different in the two cities aswell. Nevertheless, the potential reversibility of the liquidwater/heterogeneous reaction mechanism in Atlanta is a sig-nificant question and should be investigated.

4 Conclusions

We provide a detailed characterization of SOA partitioningin summertime Atlanta, a large urban center with substan-tial biogenic VOC emissions. WSOC in the particle andgas phases both had strong photochemical sources, whileWSOCg also appeared to have a unique nighttime source,possibly involving NO3 chemistry. Partitioning, analyzedthrough the fraction of total WSOC in the particle phase,Fp, was found to have no net dependence on temperature,though this was likely due to the emissions of precursorVOCs and the formation of WSOC (in the gas and particlephase), which both could have temperature dependences.Fp

was related to NOx concentrations, with higher NOx levelscorresponding to lower values ofFp. Although not likelyrelated to the partitioning process, this behavior could be at-tributed to the effect of NOx on the product distribution inVOC oxidation. In previous work,Fp was shown to havea very strong dependence on relative humidity at RH levels

above 70% due to the uptake of liquid water by fine parti-cles. The current work also shows a strong partitioning de-pendence on the existing WSOCp concentration, but not onthe OC concentration, possibly indicating the importance ofthe chemical similarity between partitioning SVOCs and theabsorbing aerosol phase, or the role of heterogeneous chem-istry in SOA formation, or both. The relationship betweenFp

and WSOCp was present even at the highest RH levels, whenthe liquid water concentration was likely far greater thanthe WSOCp concentration. Collectively, the results pointto liquid-phase heterogeneous chemical reactions as anothermajor SOA formation process and are remarkably similar tothe recent chamber experiments of Volkamer et al. (2009).

Finally, the summertime data indicate that WSOCp in-corporated most, if not all, of the SOA in Atlanta. To ourknowledge, this represents the first detailed characterizationof SOA formation processes based on ambient data. Pre-vious studies have described the gas/particle partitioning ofthe total SOA within a smog chamber, and previous ambi-ent studies have reported on the partitioning of individualcompounds. The results provide important insight into SOAformation and the gas/particle partitioning process beyondthe current scientific understanding. Specifically, that fine-particle water and heterogeneous chemical reactions may bea significant mechanism for SOA formation in summertimeAtlanta. This mechanism may also play an important role inother locations (e.g., Hennigan et al., 2008c), and may ex-plain the systematic under-prediction of SOA by most mod-els.

Acknowledgements.R. Weber thanks the National Science Foun-dation for their financial support through grant ATM-0802237. Thegenerous financial gift from F. Cullen and L. Peck also made thiswork possible.

Edited by: A. S. H. Prevot

References

Altieri, K. E., Seitzinger, S. P., Carlton, A. G., Turpin, B. J., Klein,G. C., and Marshall, A. G.: Oligomers formed through in-cloudmethylglyoxal reactions: Chemical composition, properties, andmechanisms investigated by ultra-high resolution FT-ICR massspectrometry, Atmos. Environ., 42(7), 1476–1490, 2008

Anderson, C. H., Dibb, J. E., Griffin, R. J., Hagler, G. S. W.,and Bergin, M. H.: Atmospheric water-soluble organic carbonmeasurements at Summit, Greenland, Atmos. Environ., 42(22),doi:10.1016/j.atmosenv.2008.03.06, 2008.

Bonn, B. and Moorgat, G. K.: New particle formation during a-and b-pinene oxidation by O3, OH and NO3, and the influenceof water vapour: particle size distribution studies, Atmos. Chem.Phys., 2, 183–196, 2002,http://www.atmos-chem-phys.net/2/183/2002/.

Bonn, B., Schuster, G., and Moortgat, G. K.: Influence of Water Va-por on the Process of New Particle Formation during Monoter-pene Ozonolysis, J. Phys. Chem. A, 106(12), 2869–2881, 2002.


http://www.atmos-chem-phys.net/2/183/2002/


Bowman, F. M. and Karamalegos, A. M.: Estimated effects of com-position on secondary organic aerosol mass concentrations, Env-iron. Sci. Technol., 36(12), 2701–2707, 2002

Brown, S. S., deGouw, J. A., Warneke, C., Ryerson, T. B., Dube, W.P., Atlas, E., Weber, R. J., Peltier, R. E., Neuman, J. A., Roberts,J. M., Swanson, A., Flocke, F., McKeen, S. A., Brioude, J., Som-mariva, R., Trainer, M., Fehsenfeld, F. C., and Ravishankara,A. R.: Nocturnal isoprene oxidation over the Northeast UnitedStates in summer and its impact on reactive nitrogen partitioningand secondary organic aerosol, Atmos. Chem. Phys., 9, 3027–3042, 2009,http://www.atmos-chem-phys.net/9/3027/2009/.

Chan, A. W. H., Kroll, J. H., Ng, N. L., and Seinfeld, J. H.: Ki-netic modeling of secondary organic aerosol formation: effectsof particle- and gas-phase reactions of semivolatile products, At-mos. Chem. Phys., 7, 4135–4147, 2007,http://www.atmos-chem-phys.net/7/4135/2007/.

Chang, E. I. and Pankow, J. F.: Organic particulate matter forma-tion at varying relative humidity using surrogate secondary andprimary organic compounds with activity corrections in the con-densed phase obtained using a method based on the Wilson equa-tion, Atmos. Chem. Phys. Discuss., 8, 995-1039, 2008,http://www.atmos-chem-phys-discuss.net/8/995/2008/.

Chu, S.-H., Paisie, J. W., and Jang, B. W.-L.: PM data analysis –a comparison of two urban areas: Fresno and Atlanta, Atmos.Environ., 38, 3155–3164, 2004.

Cocker, D. R., Clegg, S. L., Flagan, R. C., and Seinfeld, J. H.:The effect of water on gas-particle partitioning of secondary or-ganic aerosol, Part I: alpha-pinene/ozone system, Atmos. Envi-ron., 35(35), 6049–6072, 2001a.

Cocker, D. R., Mader, B. T., Kalberer, M., Flagan, R. C., and Sein-feld, J. H.: The effect of water on gas-particle partitioning of sec-ondary organic aerosol: II. m-xylene and 1,3,5-trimethylbenzenephotooxidation systems, Atmos. Environ., 35(35), 6073–6085,2001b.

Cofer, W. R. and Edahl, R. A.: A new technique for collection, con-centration and determination of gaseous tropospheric formalde-hyde, Atmos. Environ., 20(5), 979–984, 1986.

Cornbleet, P. J. and Gochman, N.: Incorrect least-squares regres-sion coefficients in method-comparison analysis, Clin. Chem.,25, 432–438, 1979.

de Gouw, J. A., Middlebrook, A. M., Warneke, C., Goldan, P.D., Kuster, W. C., Roberts, J. M., Fehsenfeld, F. C., Worsnop,D. R., Canagaratna, M. R., Pszenny, A. A. P., Keene, W. C.,Marchewka, M., Bertman, S. B., and Bates, T. S.: Budget oforganic carbon in a polluted atmosphere: Results from the NewEngland Air Quality Study in 2002, J. Geophys. Res.-Atmos.,110(D16), doi:10.1029/2004JD005623, 2005.

Denkenberger, K. A., Moffet, R. C., Holecek, J. C., Rebotier, T.P., Prather, K. A.: Real-time, single-particle measurements ofoligomers in aged ambient aerosol particles, Environ. Sci. Tech-nol., 41(15), 5439–5446, 2007.

Edney, E. O., Driscoll, D. J., Speer, R. E., Weathers, W. S., Klein-dienst, T. E., Li, W., and Smith, D. F.: Impact of aerosolliquid water on secondary organic aerosol yields of irradiatedtoluene/propylene/NOx/(NH4)(2) SO4/air mixtures, Atmos. En-viron., 34(23), 3907–3919, 2000.

Environmental Protection Agency, National Emissions Inventory,www.epa.gov/ttn/chief/net/2002inventory.html, 2002.

Favez, O., Sciare, J., Cachier, H., Alfaro, S. C., and Abdelwahab,M. M.: Significant formation of water-insoluble secondary or-ganic aerosols in semi-arid urban environment, Geophys. Res.Lett., 35(15), 2008.

Fisseha, R., Dommen, J., Gaeggeler, K., Weingartner, E., Sam-burova, V., Kalberer, M., and Baltensperger, U.: Online gas andaerosol measurement of water soluble carboxylic acids in Zurich,J. Geophys. Res.-Atmos., 111(D12), 2006.

Fountoukis, C. and Nenes, A.: ISORROPIA II: a computa-tionally efficient thermodynamic equilibrium model for K+-Ca2+-Mg2+-NH+

4 -Na+-SO2−

4 -NO−

3 -Cl−-H2O aerosols, At-mos. Chem. Phys., 7, 4639–4659, 2007,http://www.atmos-chem-phys.net/7/4639/2007/.

Georgia Department of Natural Resources Ambient MonitoringNetwork,http://www.georgiaepd.org/air/amp/, 2008.

Goldstein, A. H. and Galbally, I. E.: Known and unexplored organicconstituents in the earth’s atmosphere, Environ. Sci. Technol.,41(5), 1514–1521, 2007.

Griffin, R. J., Cocker, D. R., Flagan, R. C., and Seinfeld, J. H.:Organic aerosol formation from the oxidation of biogenic hydro-carbons, J. Geophys. Res.-Atmos., 104(D3), 3555–3567, 1999.

Guenther, A., Zimmerman, P., and Wildermuth, M.: Natural volatileorganic compound emission rate estimates for United Stateswoodland landscapes, Atmos. Environ., 28(6), 1197–1210, 1994.

Hakola, H., Laurila, T., Rinne, J., and Puhto, K.: The ambient con-centrations of biogenic hydrocarbons at a northern European, bo-real site, Atmos. Environ, 34(29–30), 4971–4982, 2000.

Hallquist, M., Wangberg, I., Ljungstrom, E., Barnes, I., and Becker,K. H.: Aerosol and product yields from NO3 radical-initiated ox-idation of selected monoterpenes, Environ. Sci. Technol., 33(4),553–559, 1999.

Hennigan, C. J., Bergin , M. H., Dibb, J. E. , and Weber, R.J.: Enhanced secondary organic aerosol formation due to wa-ter uptake by fine particles, Geophys. Res. Lett., 35, L18801,doi:10.1029/2008GL035046, 2008a.

Hennigan, C. J., Bergin, M. H., and Weber, R. J.: Correlations be-tween water-soluble organic aerosol and water vapor: A syner-gistic effect from biogenic emissions?, Environ. Sci. Technol.,42(24), 9079–9085, 2008b.

Hennigan, C. J., Sullivan, A. P., Fountoukis, C. I., Nenes, A., Heco-bian, A., Vargas, O., Peltier, R. E., Case Hanks, A. T., Huey, L.G., Lefer, B. L., Russell, A. G., and Weber, R. J.: On the volatil-ity and production mechanisms of newly formed nitrate and wa-ter soluble organic aerosol in Mexico City, Atmos. Chem. Phys.,8, 3761–3768, 2008c,http://www.atmos-chem-phys.net/8/3761/2008/.

Hoffmann, T., Odum, J. R., Bowman, F., Collins, D., Klockow, D.,Flagan, R. C., Seinfeld, J. H.: Xb, Formation of organic aerosolsfrom the oxidation of biogenic hydrocarbons, J. Atmos. Chem.,26(2), 189–222, 1997.

Janson, R., Rosman, K., Karlsson, A., and Hansson, H. C.: Bio-genic emissions and gaseous precursors to forest aerosols, Tellus.B., 53(4), 423–440, 2001.

Jonsson, A. M.; Hallquist, M.; Ljungstrom, E., Impact of humidityon the ozone initiated oxidation of limonene, Delta(3)-carene,and alpha-pinene, Environ. Sci. Technol., 40(1), 188–194, 2006.

Kalberer, M. Sax, M., and Samburova, V.: Molecular size evolutionof oligomers in organic aerosols collected in urban atmospheresand generated in a smog chamber, Environ. Sci. Technol., 40(19),




http://www.atmos-chem-phys-discuss.net/8/995/2008/

www.epa.gov/ttn/chief/net/2002inventory.html


http://www.georgiaepd.org/air/amp/



5917–5922, 2006.Khlystov, A., Stanier, C. O., Takahama, S., and Pandis, S.

N.: Water content of ambient aerosol during the Pitts-burgh Air Quality Study, J. Geophys. Res., 110, D07S10,doi:10.1029/2004JD004651, 2005.

Kroll, J. H.; Seinfeld, J. H., Chemistry of secondary organic aerosol:Formation and evolution of low-volatility organics in the atmo-sphere, Atmos. Environ., 42(16), 3593–3624, 2008.

Kroll, J. H., Chan, A. W. H., Ng, N. L., Flagan, R. C., and Sein-feld, J. H.: Reactions of semivolatile organics and their effectson secondary organic aerosol formation, Environ. Sci. Technol.,41(10), 3545–3550, 2007.

Kroll, J. H., Ng, N. L., Murphy, S. M., Flagan, R. C., and Sein-feld, J. H.: Secondary organic aerosol formation from isoprenephotooxidation, Environ. Sci. Technol., 40(6), 1869–1877, 2006.

Lanz, V. A., Alfarra, M. R., Baltensperger, U., Buchmann, B.,Hueglin, C., and Prvt, A. S. H.: Source apportionment of submi-cron organic aerosols at an urban site by factor analytical mod-elling of aerosol mass spectra, Atmos. Chem. Phys., 7, 1503–1522, 2007,http://www.atmos-chem-phys.net/7/1503/2007/.

Limbeck, A.; Kulmala, M.; Puxbaum, H., Secondary organicaerosol formation in the atmosphere via heterogeneous reactionof gaseous isoprene on acidic particles. Geophys. Res. Lett.,30(19), doi:10.1029/2003GL017738, 2003.

Malm, W. C. and Day, D. E.: Estimates of aerosol species scat-tering characteristics as a function of relative humidity, Atmos.Environ., 35(16), 2845–2860, 2001.

Marple, V. A., Rubow, K. L., and Behm, S. M.: A microorificeuniform deposit impactor (MOUDI) – description, calibration,and use, Aerosol. Sci Tech., 14(4), 434–446, 1991.

Matsunaga, S. N., Kato, S., Yoshino, A., Greenberg, J. P., Kajii, Y.,and Guenther, A. B.: Gas-aerosol partitioning of semi volatilecarbonyls in polluted atmosphere in Hachioji, Tokyo, Geophys.Res. Lett., 32(11), 2005.

Miyazaki, Y., Kondo, Y., Takegawa, N., Komazaki, Y., Fukuda,M., Kawamura, K., Mochida, M., Okuzawa, K., and We-ber, R. J.: Time-resolved measurements of water-soluble or-ganic carbon in Tokyo, J. Geophys. Res.-Atmos., 111(D23),doi:10.1029/2006JD007125, 2006.

Nenes, A., Pandis, S. N., and Pilinis, C.: ISORROPIA: A new ther-modynamic equilibrium model for multiphase multicomponentinorganic aerosols, Aquat. Geochem., 4(1), 123–152, 1998.

Ng, N. L., Chhabra, P. S., Chan, A. W. H., Surratt, J. D., Kroll, J.H., Kwan, A. J., McCabe, D. C., Wennberg, P. O., Sorooshian,A., Murphy, S. M., Dalleska, N. F., Flagan, R. C., and Seinfeld,J. H.: Effect of NOx level on secondary organic aerosol (SOA)formation from the photooxidation of terpenes, Atmos. Chem.Phys., 7, 5159–5174, 2007a,http://www.atmos-chem-phys.net/7/5159/2007/.

Ng, N. L., Kroll, J. H., Chan, A. W. H., Chhabra, P. S., Flagan,R. C., and Seinfeld, J. H.: Secondary organic aerosol formationfrom m-xylene, toluene, and benzene, Atmos. Chem. Phys., 7,3909–3922, 2007b,http://www.atmos-chem-phys.net/7/3909/2007/.

Ng, N. L., Kwan, A. J., Surratt, J. D., Chan, A. W. H., Chhabra, P.S., Sorooshian, A., Pye, H. O. T., Crounse, J. D., Wennberg, P.O., Flagan, R. C., and Seinfeld, J. H.: Secondary organic aerosol(SOA) formation from reaction of isoprene with nitrate radicals

(NO3), Atmos. Chem. Phys., 8, 4117–4140, 2008,http://www.atmos-chem-phys.net/8/4117/2008/.

NIOSH: Elemental Carbon (Diesel Particulate): Method 5040, in:NIOSH Manual of Analytical Methods, edited by: Eller, P. M.and Cassinelli, M. E., National Institute for Occupational Safetyand Health, Cincinnati, USA, 1996.

Odum, J. R., Jungkamp, T. P. W., Griffin, R. J., Flagan, R. C., andSeinfeld, J. H.: The atmospheric aerosol-forming potential ofwhole gasoline vapor, Science, 276(5309), 96–99, 1997.

Odum, J. R., Hoffmann, T., Bowman, F., Collins, D., Flagan, R. C.,and Seinfeld, J. H.: Va, Gas/particle partitioning and secondaryorganic aerosol yields, Environ. Sci. Technol., 30(8), 2580–2585,1996.

Orsini, D. A., Ma, Y. L., Sullivan, A., Sierau, B., Baumann, K., We-ber, R. J.: Refinements to the particle-into-liquid sampler (PILS)for ground and airborne measurements of water soluble aerosolcomposition, Atmos. Environ., 37(9–10), 1243–1259, 2003.

Pandis, S. N., Paulson, S. E., Seinfeld, J. H., and Flagan, R. C.:Aerosol formation in the photooxidation of isoprene and beta-pinene, Atmos. Environ. A-Gen., 25(5–6), 997–1008.1991

Pankow, J. F.: An absorption model of the gas aerosol partitioninginvolved in the formation of secondary organic aerosol, Atmos.Environ., 28(2), 189–193, 1994.

Pankow, J. F. and Bidleman, T. F.: Effects of temperature, TSP andpercent nonexchangeable material in determining the gas parti-cle partitioning of organic compounds, Atmos. Environ. A-Gen.,25(10), 2241–2249, 1991.

Peltier, R. E., Weber, R. J., Sullivan, A. P.: Investigating a liquid-based method for online organic carbon detection in atmosphericparticles, Aerosol. Sci. Tech., 41(12), 1117–1127, 2007.

Presto, A. A., Hartz, K. E. H., and Donahue, N. M.: Secondaryorganic aerosol production from terpene ozonolysis, 2. Effect ofNOx concentration, Environ. Sci. Technol., 39(18), 7046–7054,2005.

Pun, B. K. and Seigneur, C.: Investigative modeling of new path-ways for secondary organic aerosol formation, Atmos. Chem.Phys., 7, 2199–2216, 2007,http://www.atmos-chem-phys.net/7/2199/2007/.

Saxena, P. and Hildemann, L. M.: Water-soluble organics in atmo-spheric particles: A critical review of the literature and appli-cation of thermodynamics to identify candidate compounds, J.Atmos. Chem., 24(1), 57–109, 1996.

Seinfeld, J. H., Erdakos, G. B., Asher, W. E., and Pankow, J. F.:Modeling the formation of secondary organic aerosol (SOA). 2.The predicted effects of relative humidity on aerosol formationin the alpha-pinene-, beta-pinene-, sabinene-, Delta(3)-Carene-,and cyclohexene-ozone systems, Environ. Sci. Technol., 35(9),1806–1817, 2001.

Seinfeld, J. H. and Pandis, S. N.: Atmospheric Chemistry andPhysics, From Air Pollution to Climate Change, John Wiley &Sons, New York, NY, USA, 1998.

Simon, V., Clement, B., Riba, M. L., and Torres, L.: The LANDESexperiment – Monoterpenes emitted from the maritime pine, J.Geophys. Res.-Atmos., 99(D8), 16501–16510, 1994.

Song, C., Zaveri, R. A., Alexander, M. L., Thornton, J. A.,Madronich, S., Ortega, J. V., Zelenyuk, A., Yu, X.-Y., Laskin,A., and Maughan, D. A.: Effect of hydrophobic primary or-ganic aerosols on secondary organic aerosol formation fromozonolysis of alpha-pinene, Geophys. Res. Lett., 34, L20803,








doi:10.1029/2007GL030720, 2007.Spaulding, R. S., Talbot, R. W., and Charles, M. J.: Optimization

of a mist chamber (cofer scrubber) for sampling water-solubleorganics in air, Environ. Sci. Technol., 36(8), 1798–1808, 2002.

Stark, H., Brown, S. S., Goldan, P. D., Aldener, M., Kuster, W. C.,Jakoubek, R., Fehsenfeld, F. C., Meagher, J., Bates, T. S., andRavishankara, A. R.: Influence of nitrate radical on the oxidationof dimethyl sulfide in a polluted marine environment, J. Geophys.Res., 112, D10S10, doi:10.1029/2006JD007669, 2007.

Stenby, C., Poschl, U., von Hessberg, P., Bilde, M., Nielsen, O. J.,and Moortgat, G. K.: Temperature and humidity dependence ofsecondary organic aerosol yield from the ozonolysis ofβ-pinene,Atmos. Chem. Phys. Discuss., 7, 2091–2132, 2007,http://www.atmos-chem-phys-discuss.net/7/2091/2007/.

Sullivan, A. P., Peltier, R. E., Brock, C. A., de Gouw, J. A., Hol-loway, J. S., Warneke, C., Wollny, A. G., Weber, R. J.: Airbornemeasurements of carbonaceous aerosol soluble in water overnortheastern United States: Method development and an inves-tigation into water-soluble organic carbon sources, J. Geophys.Res.-Atmos., 111(D23), D05314, doi:10.1029/2005JD006485,2006.

Sullivan, A. and Weber, R. J.: Chemical characterization ofthe ambient organic aerosol soluble in water Part 2: Isola-tion of acid, neutral, and basic fractions by modified size ex-clusion chromatography, J. Geophys. Res., 111(D5), D05315,10.1029/2005JD006486, 2006.

Sullivan, A. P., Weber, R. J., Clements, A. L., Turner, J. R., Bae,M. S., and Schauer, J. J.: A method for on-line measurement ofwater-soluble organic carbon in ambient aerosol particles: Re-sults from an urban site, Geophys. Res. Lett., 31(13), L13105,doi:10.1029/2004GL019681, 2004

Takekawa, H., Minoura, H., and Yamazaki, S.: Temperature depen-dence of secondary organic aerosol formation by photo-oxidationof hydrocarbons, Atmos. Environ., 37(24), 3413–3424, 2003.

Tingey, D. T., Manning, M., Grothaus, L. C., and Burns, W. F.: In-fluence of light and temperature on monoterpene emission ratesfrom slash pine, Plant. Physiol., 65(5), 797–801, 1980.

Tingey, D. T., Manning, M., Grothaus, L. C., and Burns, W. F.: In-fluence of light and temperature on isoprene emission rates fromlive oak, Physiol. Plantarum., 47(2), 112–118, 1979.

Tsigaridis, K., Lathire, J., Kanakidou, M., and Hauglustaine, D.A.: Naturally driven variability in the global secondary organicaerosol over a decade, Atmos. Chem. Phys., 5, 1891–1904, 2005,http://www.atmos-chem-phys.net/5/1891/2005/.

Volkamer, R., Ziemann, P. J., and Molina, M. J.: Secondary Or-ganic Aerosol Formation from Acetylene (C2H2): seed effect onSOA yields due to organic photochemistry in the aerosol aque-ous phase, Atmos. Chem. Phys., 9, 1907–1928, 2009,http://www.atmos-chem-phys.net/9/1907/2009/.

Volkamer, R., Jimenez, J. L., San Martini, F., Dzepina, K., Zhang,Q., Salcedo, D., Molina, L. T., Worsnop, D. R., and Molina, M.J.: Secondary organic aerosol formation from anthropogenic airpollution: Rapid and higher than expected, Geophys. Res. Lett.,33(17), L17811, doi:10.1029/2006GL026899, 2006.

Warneke, C., de Gouw, J. A., Goldan, P. D., Kuster, W. C.,Williams, E. J., Lerner, B. M., Jakoubek, R., Brown, S. S.,Stark, H., Aldener, M., Ravishankara, A. R., Roberts, J. M.,Marchewka, M., Bertman, S., Sueper, D. T., McKeen, S. A.,Meagher, J. F., and Fehsenfeld, F. C.: Comparison of daytimeand nighttime oxidation of biogenic and anthropogenic VOCsalong the New England coast in summer during New Eng-land Air Quality Study 2002, J. Geophys. Res., 109, D10309,doi:10.1029/2003JD004424, 2004.

Weber, R. J., Sullivan, A. P., Peltier, R. E., Russell, A., Yan, B.,Zheng, M., de Gouw, J., Warneke, C., Brock, C., Holloway, J.S., Atlas, E. L., and Edgerton, E.: A study of secondary organicaerosol formation in the anthropogenic-influenced southeasternUnited States, J. Geophys. Res.-Atmos., 112(D13), D13302,doi:10.1029/2007JD008408, 2007.

Weber, R., Orsini, D., Bergin, M., Kiang, C. S., Chang, M., John, J.S., Carrico, C. M., Lee, Y. N., Dasgupta, P., Slanina, J., Turpin,B., Edgerton, E., Hering, S., Allen, G., Solomon, P., and Chamei-des, W.: Short-Term Temporal Variation, in: PM2.5 Mass AndChemical Composition During The Atlanta Supersite Experi-ment, 1999, J. Air. Waste. Manage., 53, 84–91, 2003.


http://www.atmos-chem-phys-discuss.net/7/2091/2007/



gas/particle partitioning of water-soluble organic aerosol...

Documents