Water-soluble SOA from Alkene ozonolysis

13 Pages·2010·1.05 MB·English

Atmos. Chem. Phys.,10,1585–1597,2010
Atmospheric
www.atmos-chem-phys.net/10/1585/2010/
Chemistry
©Author(s)2010. Thisworkisdistributedunder
theCreativeCommonsAttribution3.0License. and Physics
Water-soluble SOA from Alkene ozonolysis: composition and
droplet activation kinetics inferences from analysis of CCN activity
A.Asa-Awuku1,*,A.Nenes1,2,S.Gao3,**,R.C.Flagan3,4,andJ.H.Seinfeld3,4
1SchoolofChemicalandBiomolecularEngineering,GeorgiaInstituteofTechnology,Atlanta,GA,USA
2SchoolofEarthandAtmosphericSciences,GeorgiaInstituteofTechnology,Atlanta,GA,USA
3DepartmentofEnvironmentalScienceandEngineering,CaliforniaInstituteofTechnology,Pasadena,CA,USA
4DepartmentofChemicalEngineering,CaliforniaInstituteofTechnology,Pasadena,CA,USA
*nowat: DepartmentofChemicalandEnvironmentalEngineering,UniversityofCalifornia-Riverside,CA,USA
**nowat: DivisionofMath,Science,andTechnology,NovaSoutheasternUniversity,FortLauderdale,FL,USA
Received: 7June2007–PublishedinAtmos. Chem. Phys.Discuss.: 26June2007
Revised: 11January2010–Accepted: 27January2010–Published: 15February2010
Abstract. Cloud formation characteristics of the water- 1 Introduction
soluble organic fraction (WSOC) of secondary organic
aerosol(SOA)formedfromtheozonolysisofalkenehydro- Aerosols, by acting as cloud condensation nuclei (CCN),
carbons (terpinolene, 1-methlycycloheptene and cyclohep- have a profound impact on the hydrological cycle and cli-
tene) are studied. Based on size-resolved measurements of mate.Carbonaceousmaterial(organiccarbon,OC)cancom-
CCNactivity(ofthepureandsaltedWSOCsamples)wees- priseupto90%ofaerosolmass(AndreaeandCrutzen,1997;
timatetheaveragemolarvolumeandsurfacetensiondepres- Cachier et al., 1995; Yamasoe et al., 2000), 10–70% of
sionassociatedwiththeWSOCusingKo¨hlerTheoryAnaly- which may be water-soluble (WSOC). Studies have shown
sis(KTA).Consistentwithknownspeciation,theresultssug- that WSOC can influence aerosol hygroscopicity and sur-
gestthattheWSOCarecomposedoflowmolecularweight face tension (Decesari et al., 2003; Saxena and Hildemann,
species, with an effective molar mass below 200gmol−1. 1996; Shulman et al., 1996) and must be characterized to
The water-soluble carbon is also surface-active, depressing quantify the impact of these aerosols on cloud droplet for-
surface tension 10–15% from that of pure water (at CCN- mation. WSOC can be present in primary organic carbon
relevant concentrations). The inherent hygroscopicity pa- butalsoformedduringtheoxidationofvolatileorganiccar-
rameter, κ, of the WSOC ranges between 0.17 and 0.25; if bon(VOC)tosecondaryorganicaerosol(SOA)(Kanakidou
surfacetensiondepressionandmolarvolumeeffectsarecon- et al., 2005; Robinson et al., 2007; Saxena and Hildemann,
sideredinκ,aremarkablyconstant“apparent”hygroscopic- 1996).
ity ∼0.3 emerges for all samples considered. This implies Natural VOC emissions (e.g., monoterpernes, sesquiter-
that the volume fraction of soluble material in the parent penes),estimatedtobe1150Tgyr−1(Guentheretal.,1995),
aerosol is the key composition parameter required for pre- are a major source of SOA. Alkene ozonolysis is well es-
dictionoftheSOAhygroscopicity,asshiftsinmolarvolume tablishedasasourceofSOA(e.g., Alfarraetal., 2006; As-
across samples are compensated by changes in surface ten- chmann et al., 2002; Baltensperger et al., 2005; Claeys et
sion. Finally,using“thresholddropletgrowthanalysis”,the al., 2004; Dommen et al., 2006; Forstner et al., 1997; Gao
water-soluble organics in all samples considered do not af- etal.,2004a;Hamiltonetal.,2006;Huff-Hartzetal.,2005;
fectCCNactivationkinetics. Kalbereretal.,2006;Kanakidouetal.,2005;Keywoodetal.,
2004;Krolletal.,2006;Limbecketal.,2003;Shillingetal.,
2007;Varutbangkuletal.,2006). AttemptstospeciateSOA
(Alfarraetal.,2006;Aschmannetal.,2002;Dommenetal.,
Correspondenceto: A.Nenes 2006; Gao et al., 2004a; Kalberer et al., 2006) have been
([email protected]) met with limited success, as 80 to 90% of the aerosol mass
PublishedbyCopernicusPublicationsonbehalfoftheEuropeanGeosciencesUnion.
1586 A.Asa-Awukuetal.: Water-solubleSOAfromAlkeneozonolysis
can remain uncharacterized (Kalberer et al., 2006; Rogge impedeactivationkinetics; infact,Asa-Awukuetal.(2009)
et al., 1993; Seinfeld and Pandis, 1998). The potential for found evidence where the insoluble fraction was correlated
forming oligomeric or polymeric structures (Baltensperger withdelayedactivationkinetics.
etal.,2005;Gaoetal.,2004a;Gaoetal.,2004b;Kalbereret In this study we report the experimental investigation of
al.,2004)hasbeensuggestedtoexplaintheuncharacterized the CCN activity of the water soluble fraction of SOA gen-
SOAfraction. Oligomershavethepotentialtoexhibitchar- erated in laboratory chamber ozonolysis of alkenes; these
acteristics similar to humic-like substances (HULIS) (Bal- measurements are then used to constrain the hygroscopic-
tenspergeretal., 2005), whichstronglydepresssurfaceten- ityofthewater-solubleorganicfraction,usingκ-Ko¨hlerthe-
sion(Asa-Awukuetal.,2008;Dinaretal.,2007;Graberand ory(PettersandKreidenweis,2007). Measurementsarealso
Rudich,2006; Kissetal.,2005; Salmaetal.,2006)andpo- usedtoinferaveragethermodynamicproperties(molarmass
tentially, droplet growth kinetics. All of these can have im- andsurfacetensiondepression),usingKo¨hlerTheoryAnaly-
portant impacts on CCN activity. Adding electrolytes (par- sis(KTA)(Asa-Awukuetal.,2008;Padro´etal.,2007;Engel-
ticularly the bivalent SO2− ion) to surfactant-rich solutions hartetal.,2008;Mooreetal.,2008;Asa-Awukuetal.,2009).
4
canenhancepartitioningtothesurfacelayer(Asa-Awukuet Thefirstmethod,κ-Ko¨hlertheory,isusedtoparameterizethe
al.,2008;Kissetal.,2005),andfurtherfacilitatedropletfor- hygroscopicityoftheaerosolforuseinatmosphericmodels,
mation. Limited organic mass within the CCN may reduce andthelatter,KTA,isemployedtounderstandthesourcesof
theextentofsurfacetensiondepression(Lietal.,1998;Sor- hygrosocopicity (solute versus surface tension effects) and
jamaaandLaaksonen,2006; Sorjamaaetal.,2004),butde- theirinfluenceonthehygroscopicityparameter. Finally,we
pressionisstillrequiredtoreconcilethedifferenthygroscop- characterizethedropletactivationkineticsbycomparingac-
icitesseeninsubsaturatedandsupersaturatedconditionsfor tivateddropletsizesfromWSOCCCNagainstthoseformed
SOA aerosol (e.g., Jura´nyi et al., 2009; King et al., 2009; from(NH ) SO calibrationaerosol.
4 2 4
Wexetal.,2009).
Numerous studies quantifying and parameterizing the
cloud droplet formation potential of SOA has increasingly 2 Experimentalmethodsandtheoreticalanalysis
appeared in the literature (e.g., but not limited to Prenni et
al.,2007;Duplissyetal.,2008;Jura´nyietal.,2009;Kinget 2.1 Filterextractionandchemicalcomposition
al., 2009; Wex et al., 2009). The properties of the water-
soluble fraction of the aerosol, which is by definition its Secondary organic aerosol is generated from the seedless
dominant source of hygroscopicity, has received less atten- dark ozonolysis of three parent alkenes (cycloheptene, 1-
tion. Engelhart et al. (2008), Asa-Awuku et al. (2009) ex- methylcycloheptene and terpinolene) and collected upon
tractedthewater-solublefractionfromSOAgeneratedfrom Teflon filters. The ozonolysis experiments were performed
darkozonolysisofbiogenicVOC,andfoundthatwhencon- in the Caltech dual 28m3 teflon chambers under dry con-
sidering surface tension depression, the water-soluble frac- ditions (>5% relative humidity), a detailed description of
tionhadfairlyconstantmolarmassthatwasconsistentwith which can be found in Keywood et al. (2004). The ozone
knownspeciation,andwithinthelimitswheremolecularsize mixing ratio was three times that of the reactant concen-
strongly correlates with CCN activity (Petters et al., 2009). tration (Table 1) to ensure adequate oxidation (Gao et al.,
Howeverdiscrepancieshavebeenobservedforlargermolec- 2004a). SOA chemicalspeciation informationmeasured by
ular weight compounds inferred from growth factor mea- liquidchromatography/mass-spectrometry(LC-MS)andion
surements (Brooks et al., 2004; Gysel et al., 2004; Bal- trapmassspectrometryareavailableforthecyclohepteneand
tensperger et al., 2005). Asa-Awuku et al. (2008) and Car- 1-methylcycloheptene precursors from Gao et al. (2004a)
ricoetal.(2008)performedsimilaranalysistowater-soluble (Table 1). The LC-MS analysis from methanol extracts are
organics extracted from biomass burning aerosol and found comparedtoatomizedwaterextractspresentedinthisstudy;
that most of its hygroscopicity can be attributed to low although the exact speciation may differ between extracts,
molecular weight compounds, with an inferred molar mass the species cited by Gao et al. (2004a) are hydrophilic and
of∼250gmol−1. Mooreetal.(2008)alsostudieddissolved areexpectedtobepresentintheaqueousphase. Nochemi-
marineorganicmatterisolatedfromthesurfaceocean;asex- calspeciationdataareavailableforSOAgeneratedfromter-
pected, the matter exhibited very little hygroscopicity, but pinolene. Thepresentedanalysisisthefirststudytocharac-
still could affect CCN activity by reducing surface tension terizetheCCN-relevantpropertiesofWSOCfromcyclohep-
by∼10%. Finally,Padro´ etal.(2010)studiedwater-soluble teneand1-methylcyclohepteneozonolysis. Table1presents
organicsextractedfromMexicoCityaerosol,andfoundthat the estimated average (mole fraction weighted) molar mass
the material exhibited remarkably constant hygroscopicity, andcarbontoorganiccarbonmassratioforthespeciatedor-
regardless of location or time of day; this invariance was ganics.
attributed to shifts in molar mass being compensated for Followingtheprotocolsspecifiedforfilterextractionout-
changesinsurfacetensiondepression. Allthesestudiesalso lined in Sullivan and Weber (2006), the WSOC in the fil-
foundthatthewater-solubleorganicmatterstudieddoesnot tersampleswasextractedinpurewater(18MOhms)during
Atmos. Chem. Phys.,10,1585–1597,2010 www.atmos-chem-phys.net/10/1585/2010/
A.Asa-Awukuetal.: Water-solubleSOAfromAlkeneozonolysis 1587
Table1.Characteristicsofparenthydrocarbonsandwater-solublefractionofSOA.InformationobtainedfromGaoetal.(2004a).

Hydrocarbonprecursor Table 1: Characteristics of paCrenty hycdrlocoarbhones apndt weatner-esolublTe afbralec t1io: nC ohfa rSaOctAer i stics of parent hydrocarbons and wate1r-s-olmublTee afbratlceht i1o:yn C ohlfa SrcaOcAtye r icstilcso ofh paerepnt htyedrnocearbons and water-soluble fraction of SOA Terpinolene
Hydrocarbon precursor Cycloheptene 1-methylH cyydclroohceaprbteonne p recursToerr pinolene Cycloheptene 1-methyl Hcyycdlroohceaprtbeonne precurTseorrp inolene Cycloheptene 1-methyl cycloheptene Terpinolene

Structure
Structure Structure Structure
Reactantconcentration(ppb) R(pepabc)t ant concentration 22000 0 171 R(pepabc)t ant concentrat1io8n8 200 11717 1R(pepabc)t ant concentra1ti8o8n 200 171 11888 8
M(MCMo(><rlgaaaa0ajjj0sooo.ns.1rrr1eicsgsTsgol/ycsi/1lpg1opu0eh0meb0ct0llopiegyagfotoHsnCeHroed2gol2nOuamOtnbs)p)iliceoducenondmtisfipIeoddneNms%smmMc(C Moi(AcfcMCOa d ><rppaooIreloonbvuaeoen greaeaa mmaL00mmsabtenmlfjccajjsaensoooo..sroppotiisn11tp i bcaiuaerrrreswoonndfi ou g metsrcipsiggTssnn eeteee urelt elof//dMreedfiaacos ydmi 11nriAclmgpnn rto vpo uf00diico oeehttmovarweembao00s srenlocttnd e gesllm epeoit mIeggyr scatasiddooffa do g p utn ro re HHsgn betGohoaliornaeedtac22mtgtnallatdnir OO uiaeiooco St f nwbnnDas)) fi Ooeee ilrtetceMsdbtfdhAi. o g efaA nhr lo .t m(2 01G[(dd2[P[A[dL P[ A[ 13dcdh 2[G[0LC 0-(1111o54.iy imi1111<<4oold5Haa6473mniii0 udiaw2mcico0e0662tyld trp)Heaaa5iiat 3764ma o pdddihlggggM0uircixw2osn ryi c ccica oc,ymmmmgui2606oAl ln ycAxc tp knlAdooooe5mcgyiiieayydacillllicr cc a----dddlodsidagggigbPM1111ulli dc0 cle]]]]odorilc- i iisma1h dnssdr)tirecsey mmmmse gx Worc,,llo s-Aie c anea xAcieAlnoooomgAl deocchay kiztlllldc oci r−−−−ocnydiobPu PA[ AM 6[[[ dcdh [LC(G401ill11111<do,4.4iyiylio1111oldd75aa35664mdo2−nua7dsolwmiciic-0 28006n5ittarppdeiisaaaa]]]] o mo0pddii l iggggMgirnccdixom osn 1i cga ruc,ymmmmmexuioAA l en Aec oNms%smmMc(C Moi(AcfcMCOa myknlAdooooto)d cc><regepp aaooIhrhsydtclooWibvulllllaeiioenoc gr creaeaa ommaea----y lddsLia00mmbsalbtenm1111-ullfi jccajjdls pct 1ala cee]]]]-dsoooo..sirooppotliisn 1] ai- 11tpt ii bcaisEaiuaeaerrr)r d ensawsoonnddfi t ou g amretsrcp sinssggceyTsscnn ee ese see urelt taWeloro,fr//ldMreedatacos ey dmi 11s-nrriAcli mgpnn rtoo r ivpo uucf00sdiiceo oaaeehttmovarweemcA b ao00is sgrnnlocAttgnd et geslnlm epeodiut mIeggyrh sccatasiddooff ar oh g pt uitn ro are HHs gn bdecttGohoalilornaaeedtaac 22mtgtnaat llastdnir OO i uiaeiooico St f nmnwbnnDas)) fdii NNNNNNNOCoeee ilrteitceMsndbtfdooooooolhAdi. a o g tttttttefarA nhor AAAAAAAiilot .t y nvvvvvvvm( maaaaaaa2 iiiiiii01glllllllh P[ A[ 13dcdh 2[G[0LC aaaaaaa0-(1111obbbbbbb54.iy imi<4oold5Haap6473mnylllllll0 udaw2mcic0eeeeeeee0662ty trpa)e5ii at a o pdddiholggggMd0ircixosn ryic c a oc,ymmmmguioAl lun cAxrc knlAdioooomcgyeayydciollllicr ccd an----lodsiabP1111uli dc cle]]]]odoil- i xisma1sh ddnd)tresey se ,x Wyr,lls s-ie c nea ieAng doch iztd o no PA[ AM 6[[[ dcdh [LC(G401[6[G[P[AML(dddA[l11111<o,4.4iyiyiooldd75aa11111<35664mo2nua7dswmc,iiiiic-0 28006in5itotarppdeiisalddaa 7 o aaao0pddii65364m l iggggMgoirccdixom uosn i wcg2a uc,ymmmmmxiccciuioAA- l e08206n Anec omknNms%smmMc(C Moi(AcfcMCOa lAtppdooootoccgd e da><repp5hhaooyIdriiitcillllllooiiobvucr aaecoeno geareaeaa ----y lddmmasiaobL1111ddd-00ummli sabtenmdggggiiglf pcjcc ajj1lslMa ce]]]]-ida0ioesooool.. 1sr] oppotai riisnttisEcc11ap ai) b cadi inauaesodrrrrmeswoonn tdssfiar o u g mnssetseyircp ssi cggsTssnn eemmmmmetaWee rurelo,tl gteloe fc,a//dMr xeedaacos-s yrdmi o11i nAAriAcrl umgpnn rtoc evpo aauef00diico oec eh ttmiAovarwacneemoAbao00s sgrooooomltnlocttAnd edu gesltlm hepeoict mIcceeggyr slacratasiddthoioffah oa lllll g p udtn roc rke HHsgn beliitGochoaalir−−−−o orcnaeedt asce22ymtgtnddallai−tidnir OOy b uiuameiooco iSlt f dNNNNNNNnwpbnnDas)) fi1111 laOiodeee − illrtoooooooolelt1ceMsdabtfdhaAtttttttit. ]]]]r a oE g eaAAAAAAAfianA a]tnhr 1lyvvvvvvvor .t aaaaaaamsny ()eiiiiiii2lllllll tWaaaaaaa01aol P[ A[ 13dcdh 2[G[0LC bbbbbbbs0e-(1111o-lllllll54.iy imi<4oold5Heeeeeeeaa6473mtin0 udaw2rm ccic0e0662tye trpec)e5iiat a o apdddihlggggM0iorcixosn riyri c a oc,ymmmmguioAl Agln csAxnc knlAdoooomcgyeayydcillllicr cch a----lodstiabP1111uli cdc cle]]]]odoail- i istma1h dndi)treseyi sea dx Wr,lCln s-ie cn neaa ieiAnog ddnoch iztmd og no PA[ AM 6[[[ dcdh [LC(G401pl11111<o,4.4iyiyiaoohldd75aa35664mo2nua7dswmciic-0 28006on5ittarppcdeiisaaay o o0pddii l iggggMgirccdixom osn uii cga uc,ymmmmmxduioAA l edn Aec omknlAdooootonccge ahhydtcrillllliiocr cosea----y lddsiab1111-uli dd poc 1la ce]]]]-diol 1], ai ttisEaa) d nasd tarsx nssey s staWro,lte s-ri yr uceaac inAgtduhc rtia dla simNNNNNNNNNNNiooooooolatttttttr ooooAAAAAAAityvvvvvvvtttt aaaaaaa iiiiiiilllllllAAAAaaaaaaabbbbbbbllllllleeeeeeevvvv aaaaiiiillllaaaabbbblllleeee
%LowMolecularweightspeciesfrom 34 44a NotAvailable
total SOA mass derived from DMA 26 26
measurements 26
NumberAveragedmolecularweightof 150 147a NotAvailable
thespeciatedcomponents
Average mass ratio of carbon to or- 0.50 0.50a NotAvailable
ganic carbon from speciated compo-
nents
aInformationobtainedfrom1-methylcyclohexeneozonolysisduetoitsstructuralsimilarity.
a 1.25h sonication process with heat (water bath temper-
ature ∼60◦C). WSOC concentration was then measured
with a Total Organic Carbon (TOC) Turbo Siever ana-
lyzer (Sullivan and Weber, 2006). Anion concentrations Table2. SummaryofWSOCandionconcentrations,αandβ pa-
(SO2−, Cl− and NO−) of the extracted sample were mea- rametersoftheSzyszkowski-LangmuirisothermforallSOAcon-
4 3
sured with the Dionex DX-500 ion chromatograph with sidered. Measured Cl−, SO2− and NO− concentrations were all
4 3
Na2CO3/NaHCO3 eluent and Metrosep A Supp 5-100 an- below2.55×10−5mgL−1.
alytical column (Metrohm, Switzerland). Table 2 provides
a summary of the offline WSOC chemical composition
measurements and nominal anion concentrations (less than Parent WSOC αa β
2.55×10−5mgL−1) in the extracted samples. It is possible Hydrocarbon (mgCL−1) (mNm−1K−1) (Lmg−1)
thatsomecarbonlossandchemicalalterationsofthesample
Cycloheptene 13 2.59 1×10−6
mayhaveoccurredduringaerosolcollection,dissolution,at-
1-methyl- 6 4.82×10−5 4.63×10−19
omization,andsubsequentdrying. Applicationofthistech-
cycloheptene
nique however to aerosol collected from chamber SOA ex- Terpinolene 10 92.5 1×10−13
periments (Engelhart et al., 2008; Asa-Awuku et al., 2009)
provided properties for the WSOC that is consistent with
known speciation. This suggests that water extraction may aMeasurements are taken at room temperature between 296 and
299K.
notsubstantiallyalterthepropertiesoftheWSOC.
www.atmos-chem-phys.net/10/1585/2010/ Atmos. Chem. Phys.,10,1585–1597,2010
1588 A.Asa-Awukuetal.: Water-solubleSOAfromAlkeneozonolysis
2.2 CCNactivityofSOA cut-off diameter, d, for the (NH ) SO aerosol for which
4 2 4
half of the classified aerosol activates (i.e., CCN/CN=0.5)
The instruments and experimental set-up used to measure have a critical supersaturation, s , that characterizes the in-
c
WSOCCCNactivityareidenticaltothosedescribedinAsa- strumentsupersaturation. Ko¨hlertheoryisthenusedtocom-
Awukuetal.(2008,2009),Engelhartetal.(2008),Mooreet pute s from d, assuming that the density of (NH ) SO is
c 4 2 4
al.(2008)andPadro´etal.(2007).3–5mlofextractedsample equal to 1760kgm−3, surface tension of water and a molar
isatomizedinacollisiontypeatomizer(UniversityofMin- massof0.132kgmol−1. FollowingthesuggestionsofRose
nesota), dried with two diffusional driers and subsequently etal.(2008),theeffectivevan’tHofffactor,v ,wasassumed
s
classifiedwithascanningmobilityparticlesizer(TSISMPS tobe2.5(additionalsubsequentanalysiswithcomprehensive
3080). A0.71cmimpactorisplacedontheaerosolinletand thermodynamictheory,gavenegligibledifferences;Bougia-
aerosol are charged with a Kr-85 neutralizer (TSI 3077A). tioti et al., 2009) The supersaturation uncertainty was ob-
Thesheathtoflowratiowithinthedifferentialmobilityana- tained from the standard deviation in observed d and never
lyzer(TSIDMA3081)iskeptataconstant10Lmin−1 to1 exceeded3%(relativeuncertainty).
Lmin−1. ScanningMobilityCCNAnalysis(SMCA;Nenes
etal.,2010),isusedtoobtainfastsize-resolvedCCNactivity 2.3 Additionofinorganicsalts
(which is particularly important, given the small amount of
The impact of adding electrolytes to the CCN activity of
aerosolmassavailable). Aswith“stepping”modemeasure-
the WSOC is explored by mixing a pre-calculated amount
ments, theclassifiedaerosolissplittobecountedbyacon-
of(NH ) SO tothedissolvedSOAsample(sothatthesalt
densationparticlecounter(TSI3010)andalsoactivatedinto 4 2 4
mass fraction in the atomized aerosol is known). The mass
droplets using a DMT Continuous Flow Stream-wise Ther-
oforganiccarbon,m ,intheextractedsampleisdetermined
malGradientChamber(RobertsandNenes, 2005; Lanceet o
bymultiplyingthemeasuredWSOCcarbonconcentrationby
al., 2006; Rose et al., 2008). The CCN instrument was op-
anorganiccarbon-to-carbonratioof2,estimatedfromspeci-
erated at a 10:1 sheath-to-aerosol ratio, with a sample flow
of0.5Lmin−1 andatop-bottomcolumndifference,1T,be- ationinformationprovidedinGaoetal.(2004a)anditssub-
sequentsupplementalmaterial(Table1). Foreachidentified
tween4and15K.Thetotalconcentration(CN)ofsizedpar-
compound,thecarbontoorganiccarbonratioisdetermined
ticles measured by the CPC is used to determine the ratio
fromitsmolecularformulaandweightedbyitsabundancein
ofCCNtoCN.Theprocessisrepeatedfordifferentparticle
the speciated compounds. The inorganic mass to be added,
sizes. For each supersaturation, s, the cut-off diameter, d,
m ,toobtaintheresultinginorganicmassfraction,α,iscom-
(definedasthepointatwhichthenormalizeddistributionof i
putedas,
CCN/CN=0.5, i.e., the particle diameter with a critical su-
persaturationisequaltotheinstrumentsupersaturation)pro- α
m = m (1)
videsaquantitativecharacterizationoftheSOACCNactiv- i (1−α) o
ity(i.e.,thes−drelationship).Theexperimentsarerepeated
where m =2[WSOC]V , [WSOC] is the WSOC con-
o sample
a minimum of four times for each supersaturation and any centration (mgCL−1), and V is the sample volume
sample
influence of doubly charged particles is neglected, as dis-
(ml).Aspreviouslymentioned,filteredultrapure(18MOhm)
cussed by Moore et al. (2008), Engelhart et al. (2008) and
waterwasusedtoextracttheWSOCduringsonication,con-
Asa-Awukuetal.(2009).
firmedbythenegligibleionspresentinthefilterextracts(Ta-
Calibration of the instrument supersaturation was deter-
ble2).
mined by the minimum diameter of classified (NH ) SO
4 2 4
aerosol that activated at given chamber flow rate and 1T. 2.4 MeasuringandinferringsurfacetensionoftheCCN
(NH ) SO aerosol was generated by atomizing an aque-
4 2 4
oussolutionandsubsequentlydryingthedropletstreamwith ACAM200pendantdropmethodgoniometerisusedtodi-
silica-geldiffusionaldriers(operatingat∼5%RH).There- rectlymeasuresurfacetension. Adescriptionofthemethod
sulting polydisperse dry aerosol was then charged using a and procedure can be found in Asa-Awuku et al. (2008).
Kr-85 neutralizer (TSI 3077A), and classified using a TSI Since the surface tension depression strongly depends on
3080SMPSwithaTSI3081DMAoperatingunderasheath- WSOC (Decesari et al., 2003; Henning et al., 2005; Kiss
to-aerosol ratio of 10:1. The classified aerosol flow was et al., 2005), surface tension, σ, is measured at numer-
split and introduced into the CPC and the CCN instrument; ous concentrations. The measurements are then fit to the
the concentration of total particles (or condensation nuclei, Szyskowski-Langmuir isotherm (Asa-Awuku et al., 2008;
CN) and CCN were measured, and the activation fraction Langmuir,1917),
(CCN/CN) was computed. This process was repeated over
σ=σ −αTln(1+βc) (2)
w
many classified particle sizes (between 10 and 200nm), so
that an “activation curve” (i.e., CCN/CN as a function of whereσ isthesurfacetensionofpurewaterattemperature,
w
mobility diameter) was obtained. The activation curve ex- T, (obtained by infinitely diluting our sample with deion-
hibits a characteristic sigmoidal shape. The characteristic ized ultra-filtered water), and α, β are empirical constants
Atmos. Chem. Phys.,10,1585–1597,2010 www.atmos-chem-phys.net/10/1585/2010/
A.Asa-Awukuetal.: Water-solubleSOAfromAlkeneozonolysis 1589
obtainedfromthefit. Unfortunately, directmeasurementof If s , σ , ε and d are known from the CCN activity and
c w i
σ ofWSOCsolutionsatconcentrationsrelevantforCCNac- chemicalcompositionmeasurements,Eqs.(6)and(7)canbe
tivation (103ppm and above) requires significant amount of combinedtogiveσ:
mass(103µgandabove)whichwasnotavailableinthefilter
(cid:18) s (cid:19)2/3
sample. If α, β are based on usingdilute samples, extrapo- σ=σ c (8)
w s ∗
lation of Eq. (2) to higher concentrations is often subject to c
substantialuncertainty. Equation(8)representstheextensionofKo¨hlerTheoryAnal-
Ifthesaltmassfractionexceeds50%,themajorityofdis- ysis(Sect.2.5)toinfersurfacetensionfromactivationexper-
solved solute, n , in the aerosol is from the inorganic salt, iments. Thevalueofσ obtainedfromEq.(8)correspondsto
s
andanysurfacetensiondepressionatthedropletlayercould the concentration of dissolved carbon at the critical wet di-
be attributed to the presence of organics. Hence, one could ameteroftheCCN(i.e.atthepointofactivation),whichcan
then infer the droplet surface tension, σ, at the point of ac- also be computed from Ko¨hler theory (Padro´ et al., 2007;
tivationusingacombinationofCCNactivationexperiments Mooreetal.,2008). IftheorganiccontributiontotheRaoult
andKo¨hlertheory,asfollows.Forparticlescomposedofsol- term(Eq.4)isnotnegligible, thenitmustbeaccountedfor
uble and insoluble fractions, the critical supersaturation, s , inEqs.(6–8),usingtheprocedureofMooreetal.(2008).
c
is(Ko¨hler,1936;SeinfeldandPandis,1998), When σ is inferred over a range of supersaturations (i.e.
range of concentration of dissolved carbon), one can in-
4A3!1/2 fer the relationship between surface tension and concentra-
sc= (3) tionofdissolvedorganiccarbonrelevantforCCNactivation.
27B
This approach has shown to successfully reproduce surface
tension measured directly with a pendant-drop goniometer
(cid:16) (cid:17) (cid:16) (cid:17)
where A= 4Mwσ , B= 6nsMw , R is the universal gas (Mooreetal.,2008)forprimarymarineorganicsamples.
RTρw πρw
constant, T is droplet temperature, n are the moles of dis-
s
2.5 Ko¨hler theory analysis (KTA) and molar volume
solvedsolute. M andρ arethemolecularweightandden-
w w
uncertainty
sityofwater,respectively,andσ isthesurfacetensionofthe
dropletatthepointofactivation. Theassumptionthatthein-
Ko¨hlerTheoryAnalysis(KTA;Padro´ etal.,2007)isusedin
organicsaltcontributesthedominantsoluteimpliesitisthe
thisworktoinferaveragemolarvolume(molecularweight,
onlycomponentthatcontributestoB,
M, over density ρ) of the water-soluble organic fraction of
the SOA. KTA (method b ) employs measurements of dry
M ρ 1
B= Mw ρi d3εiυi (4) diameterversuscriticalsupersaturation,sc,whicharethenfit
i w totheexpression,s =ωd−3/2.FromtheFittedCCNactivity
c
where M , ν is the average molecular weight and effective (FCA) parameter, ω, estimates of σ and measurements of
i i
van’t Hoff factor of the inorganic constituent “i”, respec- ionicandWSOCconcentrations, Mo isobtainedas,
ρo
tively. ε is volume fraction of the inorganic in the aerosol,
i M ε υ
calculatedfromthemassfraction,m,anddensity,ρ,ofeach o = o o (9)
ρ (cid:16) (cid:17)2(cid:16) (cid:17)3
componentas o 256 Mw 1 σ3ω−2− ρi ε υ
27 ρw RT Mi i i
(cid:14) KTA has been shown to reproduce molecular weights (to
m ρ
i i
εi=m (cid:14)ρ +m (cid:14)ρ (5) within20%)forCCNcomposedofknownmixturesofinor-
i i o o ganicandorganiccompounds(Padro´ etal.,2007). KTAhas
also been applied to complex biomass burning WSOC with
where “i” and “o” subscripts refer to inorganic and organic
anestimated40%uncertainty(Asa-Awukuetal.,2008). The
components, respectively. Iftheorganicisnotastrongsur-
factant,thenthecriticalsupersaturations∗ forthedrydiam- methodhasalsoprovidedmolarvolumeestimatesofwater-
c solubleextractsfrombiogenicSOA(Engelhartetal., 2008;
eterd andvolumefractionε ,shouldbegivenby
i
Asa-Awukuetal.,2009)thatisconsistentwithknownspeci-
2(cid:18)4M σ (cid:19)3/2(cid:18) M ρ (cid:19)−1/2 ation.
s∗= w w 3 w i d3ε υ (6) The measured variables employed in the KTA analysis
c 3 RTρ M ρ i i
w i w are summarized in Table 3. In applying KTA, we assume
thattheeffectiveorganicvan’tHofffactor, v =1. Molecu-
where σ corresponds to the surface tension of pure water. o
w
lar weights are presented assuming an average organic den-
However, iftheorganicdepressessurfacetensiontoσ (less
sity of 1.4gcm−3 (Turpin and Lim, 2001). The uncer-
thanσw),thenthecriticalsupersaturationisgivenby (cid:16) (cid:17)
tainty in inferred molar volume is computed as 1 Mo =
ρo
sc=2(cid:18)4Mwσ(cid:19)3/2(cid:18)3Mw ρi d3εiνi(cid:19)−1/2 (7) r P (8x1x)2,where1x istheuncertaintyinofeachof
3 RTρw Mi ρw forallx
www.atmos-chem-phys.net/10/1585/2010/ Atmos. Chem. Phys.,10,1585–1597,2010
1590 A.Asa-Awukuetal.: Water-solubleSOAfromAlkeneozonolysis
Table3.ResultsfromtheanalysisofCCNactivityofallWSOCsamples.
Property(units) Cycloheptene 1-methylcycloheptene Terpinolene
ω(m1.5) 7.14×10−14 5.67×10−14 6.53×10−14
σ (Nm−1)a 5.99×10−2 6.52×10−2 6.11×10−2
(cid:16) (cid:17)
Mo (m3mol−1) 1.44×10−4 5.69×10−5 7.54×10−5
ρo
Mo(gmol−1)b 207(126)c 101(80)c 162(106)c
κ(σ) 0.25±0.04, 0.17±0.02, 0.16±0.03,
(κ(σwater)c) (0.33±0.06)c (0.30±0.04)c (0.26±0.04)c
ainferredfromactivationexperiments(Table5).
bAssumingthedensityofthesoluteisassumedtobe1400kgm−3(TurpinandLim,2001).
cResultsbasedonσ=σwater(72mNm−1)
Table4.FormulaefortheSensitivityofMolarVolumetothedependantparametersσ,ω,andνo.
(cid:16) (cid:17)
Property Sensitivity,8x=∂∂x Mρoo
σ 8σ=(cid:18)3×22756(cid:16)Mρww(cid:17)2(cid:16)R1T(cid:17)3σε2oωυ−o2(cid:19)(cid:16)Mρoo(cid:17)2
ω 8ω=(cid:18)2×22756(cid:16)Mρww(cid:17)2(cid:16)R1T(cid:17)3σε3oωυ−o3(cid:19)(cid:16)Mρoo(cid:17)2
νo 8υo=22576(cid:16)Mρww(cid:17)2(cid:16)R1T(cid:17)3σ3ωε−o2νo−2(cid:16)Mρoo(cid:17)2+ iP6=j Mρiiεεoiνi!νo−2(cid:16)Mρoo(cid:17)2
(cid:16) (cid:17)
the measured parameters x,(i.e., any of σ, ω, and υ) and is where Mo istheaveragemolarvolumeofthesolubleor-
(cid:16) (cid:17) ρo
thesensitivityofmolarvolumetox,8x=∂∂x Mρoo ,derived ganic fraction and νo is the effective van’t Hoff factor of
fromEq.(9). Table4providesalistof8 . the soluble organic. If organics depress surface tension, to
x
a value σ at the point of activation, application of Eq. (10)
(cid:16) (cid:17)−3
2.6 Hygroscopicityparameter,κ overestimatesκ byafactorof 1−σw−σ .
σw
TheobservedCCNactivityoftheWSOCcanbeparameter- 2.7 Dropletactivationkinetics
izedintermsofasinglehygroscopicityparameter,κ (Petters
andKreidenweis,2007), When exposed to the same s profile (that exceeds their s )
c
twoCCNwillactivateandgrowtoclouddropletsofsimilar
4A wetdiameter,D ,providedthattheircriticalsupersaturation
κ= (10) p
27dln2s and mass transfer coefficient of water vapor to the growing
dropletsisthesame. TheCCNinstrumentmeasuresdroplet
where A=(cid:16)4Mwσw(cid:17)3, and σ being the surface tension of sizesbyanopticalparticlecounterandthereforecanbeused
RTρw w to explore the impact of organics on the droplet growth ki-
waterattheaverageinstrumenttemperature. κ andωarere-
netics. By comparing the droplet sizes of activated SOA
h i1/2
latedbyω= 4A3 .Forinsolublenon-wettablematerials, particlesagainst(NH ) SO particlesatidenticals (andfor
27κ 4 2 4 c
κ=0; hygroscopicSOAexhibitsκ≈0.1,whilesolubleinor- identicalconditionsofinstrumentoperation),wedirectlyas-
ganicsaltsexhibitmuchhighervalues(e.g.,(NH ) SO has sesstheimpactoforganicsonCCNgrowthkinetics. Inthis
4 2 4
κ≈0.6). IfhygroscopicSOAareassumedtobecomposedof study, we select the wet diameter, Dp, that corresponds to
a soluble fraction with volume fraction εo, and the surface particles with sc equal to the instrument supersaturation, s,
tension at their point of activation is equal to σ , one can (i.e., CCN with a dry diameter equal to the cut-off diam-
w
show, eter, d). This method, termed “Threshold droplet Growth
Analysis”(TDGA),hasbeensuccessfullyappliedinnumer-
(cid:18)M (cid:19)(cid:18)ρ (cid:19)ε ousstudies(Asa-Awukuetal.,2008,2009; Engelhartetal.,
κ= w o o (11)
ρ M ν 2008; Moore et al., 2008; Bougiatioti et al., 2009; Lance et
w o o
Atmos. Chem. Phys.,10,1585–1597,2010 www.atmos-chem-phys.net/10/1585/2010/
A.Asa-Awukuetal.: Water-solubleSOAfromAlkeneozonolysis 1591
Table5.σ valuesinferredatthepointofactivation/
Sample σ (mNm−1) ±1σ (mNm−1)
CyclohepteneSOAwith90%(NH4)2SO4 73.6 4.6
CyclohepteneSOAwith33%(NH4)2SO4 59.9 1.9
1-methylcyclohepteneSOAwith33%(NH4)2SO4 65.2 2.5
TerpinoleneSOAwith98%(NH4)2SO4 74.4 4.9
TerpinoleneSOAwith90%(NH4)2SO4 70.5 4.0
TerpinoleneSOAwith33%(NH4)2SO4 61.1 7.9

FigFurieg 1.: 1C.CN CacCtiNvitya ocft iWviStOyCo gfenWeraSteOd Cfromg eonzeonraoltyesdis ofrf ocmycloohzepotnenoel. yRseissulotsf are FFigiugre. 22: .CCCN CacNtivitayc otfi vWitSyOCo fgeWnerSateOd Cfromg eonzeonroaltyesdis offr otemrpinoolzenoen. oRleysuslitss aoref
shocwync floorh peupret eWneSO.C Raneds umlitxstuarerse ofs h(NoHw4)n2SOfo4.r Tpheu rceut-WoffS dOiamCetaern, dd, mis ithxet uproeinst at stheorwpni nfoor lpeunree W. SORCe asnud lmtsixtaurrees osf h(NowH4n)2SOfo4. r Tpheu rcuet-oWff dSiaOmCetera, dn,d is tmhei xpotiunrt eats
whiocfh C(NCNH/C4N)2=S0.O5)4 i.s pTlohtteed cvuerts-uosf fsupdeirasamtueratetior,n. dL,ineiss atrhe efitp oofi netxpaertimwenhtaicl hdata wohfich( NCHCN4/C)2NS=O0.54) .is pTlohtteed cvuerts-uos fsfupderisaamtureattieorn,. Ldin,esi saret hfiet opf oeixnpetrimatenwtalh diactah
poiCntsC. NIns/Cet Ngra=ph0 .p5r)esiesntps loextatemdplvese rosf uasctsivuaptieonrs cautruvreas ti(io.en.,. CLCinNe/CsNa rves.fi dtryo fmeoxb-ility pCoiCntsN. /CN=0.5) is plotted versus supersaturation. Lines are fit of
diampeetreirm) weintht aclordreastpaonpdoinign stisg.mIonidsaelt figtsr faopr hpuprer esasmepnltes aeerxoasoml. plesofactivation experimentaldatapoints.
curves(i.e.,CCN/CNvs.drymobilitydiameter)withcorrespond-

ingsigmoidalfitsforpuresampleaerosol.

Forallthreeparenthydrocarbons,theoriginalSOAsamples
activate at diameters larger than that of (NH ) SO ; this is
4 2 4
al.,2009;Sorooshianetal.,2008;Murphyetal.,2009;Padro´ expectedasorganicsaregenerallylesshygroscopicthansol-
etal.,2010)toassesstheimpactofcompositiononactivation ubleelectrolytes. Theactivationcurvesarewellrepresented
kinetics. with a power law consistent with a d−3/2dependence; this
implies that the water-soluble SOA does not exhibit limited
solubility over the range of critical supersaturations consid-
3 Resultsanddiscussion ered(Padro´ etal.,2007). 33
32
3.1 CCNactivity 3.2 Surfacetension
Thecut-offdiameter,d,asafunctionofsupersaturationand Figure 4 shows the direct measurements of surface tension
(NH ) SO massfractionareshownforallSOAsamplesin for all SOA samples and the Szyskowski-Langmuir fits to
4 2 4
Figs.1–3. Asthemassfractionof(NH ) SO increases,the the data (α and β parameters of the fits are given in Ta-
4 2 4
aerosol smoothly transitions to pure (NH ) SO behavior. ble2). Noneofthesamplesdemonstratesignificantsurface
4 2 4
ThissuggeststhattheSOAaresolubleandhygroscopic(low tension depression at measured concentrations, even when
molecularweight)compoundsthatarenotstrongsurfactants. extrapolated to concentrations relevant for CCN activation
www.atmos-chem-phys.net/10/1585/2010/ Atmos. Chem. Phys.,10,1585–1597,2010
1592 A.Asa-Awukuetal.: Water-solubleSOAfromAlkeneozonolysis

FFigigur.e 44:. DiDrecitr eσ cmteaσsurmemeeanstsu orfe SmOeAn atss a ofufncStiOonA of awsatear-sfoulunbclet icoanrboonf cowncaetnetrra-tion

(scololsuebd lseymcbaorlbs)o anndc oinnfecrreend tvraaltuieosn fr(ocml o(Tsaebdle s5y) m(obpeonl sS)OaAn sdyminboflesr) raesd a vfuanlcutieosn of
FFigiugr.e 33.: CCCCN Nactaivcittyi voift yWoSOfCW gSenOeraCtedg efrnomer aotzeondolfyrsios mof o1z-moenthoyllycyscilsohoepfte1ne-. wfraotemr s(oTluabblel eca5rb)on(o cpoenncenStrOatiAon sayt macbtiovalsti)ona. sCaurvfuesn cretpiroensenot fSwzyaskteowrsskoi-lLuabnglemuir
Rmeseutlhtsy alrec yshcolwonh feopr pteurnee W.SROCe saundlt msixatrueress ohfo (wNHn4)2fSoOr4.p Tuhree cuWt-oSffO diCamaetnerd, dm, isi xth-e icsaotrhbeormn fictso onfc eexnpetrriamteinotnal daattaa. cHtUivLaISti doanta. frComu rAvsea-sAwreupkure est eanl. t(2S00z8y) sisk porwovsidkeid- for
ptuoirnet sat owfhi(cNh CHC4N)/2CSNO=04.5.) Tis hpleotcteud tv-eorsfufs dsuiapemrseattuerar,tiodn,. Lisinetsh earep foiti onft eaxtpewrimheincthal cLoamnpgarmisouni. r isotherm fits of experimental data. HULIS data from
dCatCa pNoi/nCtsN. =0.5) is plotted versus supersaturation. Lines are fit of Asa-Awukuetal.(2008)isprovidedforcomparison.
experimentaldatapoints.
Lim, 2001), KTA gives average organic molecular weights
(100mgCL−1 andabove)(Fig.4). Ifsurfactantsdoexistin of 162±28, 101±20, 207±54gmol−1 for terpinolene, 1-
theSOA,itislikelytheyarenotconcentratedenoughinthe methyl cycloheptene, and cycloheptene SOA, respectively
extracted samples to have a notable impact on surface ten- (Tables 3 and 6). This agrees (to within uncertainty) with
sion; even for strong surfactants extracted from a biomass the Gao et al. (2004a) speciation (Table 1), which in-
burningsample(Asa-Awukuetal.,2008),thedepressionfor cludes low molecular weight diacids, carbonyl-containing
concentrations up to 100mgCL−1 is within the measure- acids,diacidalkeylsters,andhydroxyldiacids(e.g.,pimelic
34
mentuncertainty(Fig.4). Thus,directsurfacetensionmea- acid, 160gmol−1; adipic acid, 146gmol−1; glutaric acid,
surements for dilute samples would not conclusively reveal 132gmol−1; succinic acid,35118gmol−1; pimelic acid
the presence of surfactants. Acquiring sufficient sample for monomethylester,174gmol−1;adipicacidmonomethyles-
σ measurement is challenging, so we infer surface tension ter, 160gmol−1; 2-hydroxypimelic acid, 176gmol−1; 2-
usingthemethoddescribedinSect.2.4. Forlargemassfrac- hydroxyglutaricacid,148gmol−1;6-oxohexanoicacid,130
tionsofsalt(>90%),theinferredsurfacetensionapproaches gmol−1 and6-oxo-7-hydroxyheptanoicacid,158gmol−1).
that of water used to extract the WSOC from the SOA fil- Themodestdepressionofinferredσ fromthevalueofpure
tersamples(∼72mNm−1)(Table5). However,forthe33% water(∼10%),andthevalueofinferredmolarvolumebeing
mixtureofsulfatewithcyclohepteneandterpinolene,thein- intherangewherehygroscopicitycorrelateswithmolarsize
ferred σ is ∼60mNm−1 (Table 5), ∼15% depression from (Petters et al., 2009) implies that the KTA-based estimates
purewater,suggestingthatsurfaceactivecomponentsindeed are representative for the WSOC. Indeed, if water surface
existintheWSOC.Theextentofsurfacetensiondepression tensionwasusedinKTAtoinferthemolecularweightofthe
suggeststhatthesurfactantsareappreciablystrong,whichis organics, large deviations arise from the Gao et al. (2004a)
expectedgiventheamphiphilicnatureoftheoxidationprod- speciation(Tables1and3).
ucts. Furthermore,ifmicellesareformed,theabilityofsur-
The molar volumes of water-soluble alkene SOA in-
face active organics to depress σ is limited and consistent ferred here (5.69×10−5–1.44×10−4 m3 mol−1) are within
withinferredvalues(Tabazadeh,2005). the range inferred for WSOC from α-pinene (8.18×10−5
m3 mol−1), monoterpene (8.36×10−5 m3mol−1) and β-
3.3 Molarvolumeestimatesanduncertainty caryophyllene SOA (1.02×10−4m3mol−1) dark ozonoly-
sis (Engelhart et al., 2008; Asa-Awuku et al., 2009). The
Using the inferred values of surface tension (Table 5) and WSOC alkene SOA is also consistent with HULIS sam-
assuming an aerosol density of 1.4gmol−1 (Turpin and ples (1.08×10−4–2.06×10−4m3mol−1; Wex et al., 2007).
Atmos. Chem. Phys.,10,1585–1597,2010 www.atmos-chem-phys.net/10/1585/2010/
A.Asa-Awukuetal.: Water-solubleSOAfromAlkeneozonolysis 1593
Table6.MolarVolumeSensitivityAnalysisforSOA.
SOAPrecursor Propertyx 1x 8x Molarvolume
Hydrocarbon (units) (m3mol−1x−1) uncertainty(%)
Terpinolene
σ 1.41×10−3 3.73×10−3 7.0
ω 2.21×10−15 2.69×109 7.9
νorganic 0.20a 8.78×10−5 23.3
TotalUncertainty 26.2
1-methylcycloheptene
σ 1.40×10−3 2.76×10−3 6.8
ω 1.90×10−15 2.26×109 7.5
νorganic 0.20a 6.41×10−5 22.5
TotalUncertainty 25.1
Cycloheptene
σ 1.21×10−3 8.34×10−3 7.0
ω 1.58×10−15 4.71×109 5.2
νorganic 0.20a 1.68×10−4 23.3
TotalUncertainty 26.9
aErrorbasedonobservationsof20%dissociationoforganicHULISintitrationexperimentsDinaretal.(2006).
Assuming an organic density of 1.5gcm−3, these molar (Table3). Suchlevelsofuncertaintyareconsistentwiththe
volumes correspond to a molar mass range of about 180– studies of Engelhart et al. (2009), Padro´ et al. (2010), and
250gmol−1,whichissimilartotheaveragemolarmassfrom Jura´nyietal.(2009).
thealkeneSOA. Theκ forthewater-solublefraction(regardlessofsurface
The uncertainty in inferred molar volume is affected by tensionassumption)isconsistentlyhigherthanthe“typical”
theuncertaintyinallparametersusedforKTA.First,theas- κ∼0.1 found for most hygroscopic SOA (Petters and Krei-
sumptionthatν =1,doesnotaccountforthepartialdissoci- denweis, 2007). This is because the samples analyzed here
o
ationoftheorganicspecies,andtendstounderestimatemolar reflect the water-soluble fraction of the parent SOA, and is
volume. The potential dissociation of organics (up to 20% theprimesourceofitshygroscopicity. TheparentSOA,be-
as measured in HULIS titration experiments; Dinar et al., ingamixtureof“insoluble”(non-hygroscopic)and“soluble”
2006),contributesroughly23%uncertaintytothemolarvol- components inevitably has a lower κ, because the volume
umeestimatesandisthegreatestestimatedsourceofuncer- fractionoftheWSOC,ε ,islessthanunity(Eq.11).
o
taintyinmolarvolume(Table6).AsinpreviousKTAstudies Astrikingfeatureofκ(assumingsurfacetensionofwater)
(e.g.,Asa-Awukuetal.,2008,2009; Padro´ etal.,2007; En- is its invariance (κ∼0.30) with respect to parent hydrocar-
gelhart et al., 2008, Moore et al., 2008, Padro´ et al., 2010), bonandspeciation. Evenmorestrikingisitsagreementwith
thecontributionsofσ andωvariabilitytotheinferredmolar the hygroscopicity of WSOC from monoterpene SOA (En-
volumeuncertaintyarearound10%each. Ifonewantstoes- gelhart et al., 2008), sesquiterpene SOA (Asa-Awuku et al.,
timatemolarmass,uncertaintyalsoarisesfromthevalueof 2009) and Mexico City aerosol (Padro´ et al., 2010). Given
density; varying from 1.4gcm−3 to 1.6gcm−3 (Turpin and thevariablecompositionoftheWSOCinallofthesesamples
Lim,2001)increasesmolarmassesby14%,whichisimpor- andthenatureofthecompoundsinvolved(i.e.,lowmolecu-
tantbutstillsmallerthantheuncertaintyfromνo. Thetotal larweightcompounds,withvaryingsurfactantcapability),a
estimateduncertaintyinmolarvolumeisapproximately25% constantκ impliesthatshiftsinmolarvolumebetweensam-
forallSOAsamples(Table6). plesareaccompaniedbycompensatingshiftsinsurfaceten-
siondepression.
3.4 Hygroscopicityparameter,κ
3.5 Dropletactivationkinetics
Table 3 reports κ values for water-soluble alkene SOA. If
Eq. (10) is applied assuming a surface tension of water, κ Figure 5 presents the droplet size measurements at the in-
∼0.30 for all samples. If the inferred σ is used, κ ∼0.16– strument OPC for all supersaturations and samples consid-
0.25. The sensitivity of κ to the value of surface tension ered. Forallpoints, theflowratewithintheinstrumentwas
can provide an uncertainty for the real value of the param- maintainedconstantat0.5Lmin−1 andthesheathtoaerosol
eter; whichonaverageis37%fortheSOAconsideredhere ratiois10:1; thisensuresthatalltheparticleswereexposed
www.atmos-chem-phys.net/10/1585/2010/ Atmos. Chem. Phys.,10,1585–1597,2010
1594 A.Asa-Awukuetal.: Water-solubleSOAfromAlkeneozonolysis

This finding is consistent with the studies of Engelhart et
al. (2008), Asa-Awuku et al. (2009), Moore et al. (2008)
andPadro´ etal.(2010). Second, KTAresultsareconsistent
withavailablecompositiondatawhenusinginferredsurface
tensionvalues;assumingsurfacetensionofwateryieldstoo
low molecular weights. Given this and that inferences are
intherangewherehygroscopicityisstronglycorrelatedwith
molar volume (Petters et al., 2009) suggests that appropri-
ate application of KTA (when applied to the water-soluble
fractionof the aerosol)is a powerful tool forunderstanding
thesourcesofhygroscopicity(i.e.,surfacetensionandsolute
contributions) of organic aerosol. Furthermore, the appar-
entκ valuesareremarkablyconstant,andinagreementwith
valuesobtainedinpreviousstudiesforwater-solublecarbon
extractedfromozonolysisSOA(e.g.,Engelhartetal.,2008;
Asa-Awukuetal., 2009)andMexicoCityaerosol(Padro´ et
al., 2010). When KTA is used to deconvolute the source of
thishygroscopicityintermsofsoluteandsurfacetensionde-
pressioncontribution,thestrongestsurfactantscoincidewith

Figure 5: Droplet growth measurements for WSOC SOA and (NH4)2SO4 CCN with critical thelargestmolarmass;thisimpliesthatlossofhygroscopic-
Fig. 5. Droplet growth measurements for WSOC SOA and
supersaturation equal to the instrument supersaturation. ityfromincreasesinmolarvolumemaybecompensatedby
(NH4)2SO4 CCNwithcriticalsupersaturationequaltotheinstru- shiftsinsurfacetensiondepression. Thesurfacetensionde-
mentsupersaturation.
pression(withrespecttopurewater)inferredforthesamples
consideredherearesimilartothosefromotherSOAsystems
(Engelhartetal.,2008;Asa-Awukuetal.,2009;Kingetal.,
tosimilarsupersaturationprofiles. FromFig.5weconclude
2007), urban and regional aerosol (Padro´ et al., 2010), and
thatthedropletgrowthkineticcurvesforallSOAsamplesare
biomassburningaerosol(Asa-Awukuetal.,2008).
virtually indistinguishable for all s values examined; com-
Based on the observations presented here, a simple pa-
pared to (NH ) SO , SOA particles grow to very similar
4 2 4
rameterizationforthe“apparent”κ (i.e.,thatderivedassum-
sizes. Thedropletgrowthofwatersolublecompoundsfrom
ing a surface tension of water) for the organic fraction, κ ,
alkeneSOAissimilartowatersolubleorganicspresentedin o
is κ ∼0.3ε , where ε is the volume fraction of the water-
Engelhartetal.(2008)andAsa-Awukuetal.(2009). Almost o o o
soluble organics in the aerosol. When applying this param-
allofthegrowthkineticsexperimentsliewithinthemeasure-
36 eterization to complex aerosol, special attention should be
ment uncertainty, so we conclude that the growth kinetics
giventothenonlineardependenceofsurfacetensiondepres-
(andcorrespondingwatervapormassuptakecoefficient)are
sion with organic fraction. Nevertheless, the invariance of
uniformandequaltothatof(NH ) SO .
4 2 4
theapparenthygroscopicityparameterfortheWSOCacross
manystudiessuggeststhatthissimpleparameterizationmay
4 Summaryandimplications be quite general in nature. This exciting prospect requires
support from a broader CCN activity dataset of CCN activ-
Inthisstudy,weexplorethesize-resolvedCCNactivityand ity,andwillbethefocusoffuturestudies.
droplet activation kinetic characteristics of SOA generated
fromtheozonolysisofbiogenicprecursors. Thedataisthen Acknowledgements. The work in this study was supported by
parameterized using κ-Ko¨hler theory. The data is also pro- a NSF CAREER Award and a NASA Earth and Space Science
Fellowship.Additionalfundingfortheozonolysisexperimentsand
cessed with Ko¨hler theory analysis (KTA) to infer surface
speciationwasprovidedbytheUSDepartmentofEnergyBiologi-
tensionandtheaveragemolarvolumeofthesolubleorganics
cal and Environmental Research Program DE-FG02-05ER63983,
extractedfromthefiltersamples. Theresultsoftheanalysis
Electric Power Research Institute, and US Environmental Protec-
are compared against known speciation (whenever known)
tion Agency RD-83107501-0. We would like to thank Rodney
to evaluate the method. Finally, we compare the activated
WeberandChrisHenniganoftheGeorgiaInstituteofTechnology
dropletsizesagainstareferencefromammoniumsulfatecal- for the use of their Total Organic Carbon (TOC) Turbo Siever
ibrationaerosol(usingTDGA)toevaluatetheimpactofthe analyzerandDionexDX-500ionchromatograph.
dissolvedorganicsonactivationkinetics.
A number of important findings emerge from this study. Editedby:N.Pirrone
First,usingTDGAwefindthatthewater-solubleorganicsdo
notaffectdropletgrowthkinetics,asCCNfromalltheSOA
samplesgrowtosimilardropletsizesas(NH ) SO aerosol.
4 2 4
Atmos. Chem. Phys.,10,1585–1597,2010 www.atmos-chem-phys.net/10/1585/2010/