at SciVerse ScienceDirect

Atmospheric Environment 46 (2012) 271e278

Contents lists available

Atmospheric Environment

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

Secondary organic aerosol formation from photooxidation of a mixtureof dimethyl sulfide and isoprene

Tianyi Chen, Myoseon Jang*

Department of Environmental Engineering Sciences, P.O. Box 116450, University of Florida, Gainesville, FL 32611, USA

a r t i c l e i n f o

Article history:Received 10 August 2010Received in revised form5 September 2011Accepted 27 September 2011

Keywords:Secondary organic aerosolHeterogeneous reactionIsopreneDimethyl sulfideMethanesulfonic acid

* Corresponding author. Tel.: þ1 352 846 1744; faxE-mail address: mjang@ufl.edu (M. Jang).

1352-2310/$ e see front matter � 2011 Elsevier Ltd.doi:10.1016/j.atmosenv.2011.09.082

a b s t r a c t

Secondary organic aerosol (SOA) created from the photooxidation of a mixture of isoprene and dimethylsulfide (DMS) was studied at different NOx concentrations (40e220 ppb) and humidities (12%, 42% and80%) using a Teflon film indoor chamber. To study the effect of isoprene on DMS products, the major DMSphotooxidation products, such as sulfuric acid, methanesulfonic acid (MSA) and methanesulfinic acid(MSIA), were quantified in both the presence and the absence of isoprene using a Particle-Into-Liquid-Sampler coupled with Ion Chromatography (PILS-IC). The resulting PILS-IC data showed that the DMSaerosol yield significantly decreased due to the photooxidation of isoprene. A 35.2% DMS aerosol yieldreduction was observed due to 800 ppb isoprene in 185 ppb NOx and 140 ppb DMS. Among the aerosol-phase DMS oxidation products, MSA was the most sensitive to the presence of isoprene (e.g., 46%reduction). The DMS aerosol product analysis indicates that isoprene oxidation affects the pathways ofMSA formation on the aerosol surface. Using a new approach that implements an Organic Carbon (OC)analyzer, the isoprene SOA yield (Yiso) in the DMS/isoprene/NOx system was also estimated. The OC datashowed that Yiso increased significantly with DMS compared to the Yiso without DMS. For example, Yisowith 80 ppb NOx and 840 ppb isoprene was increased by 124.6% due to 100 ppb DMS at RH ¼ 42%. Ourstudy suggests that the heterogeneous reactions of isoprene oxidation products with the highly acidicproducts (e.g., MSA and sulfuric acid) from DMS photooxidation can considerably contribute to the Yisoincrease.

� 2011 Elsevier Ltd. All rights reserved.

1. Introduction

Dimethyl sulfide (DMS) is an important marine-borne reducedsulfur compound with an estimated flux of 15.4e28.0 Tg S yr�1

(Aumont et al., 2002; Bopp et al., 2003; Kloster et al., 2006). A largeamount of ambient monitoring data suggests that DMS photooxi-dation produces fine particles which consist primarily of sulfuricacid andmethanesulfonic acid (MSA) (Bardouki et al., 2003; Baroneet al., 1995; Gaston et al., 2010; Lukacs et al., 2009). These fineparticles have the potential effect on the Earth’s radiation budgetthrough the production of cloud condensation nuclei (CCN) overthe oceans, leading to increases in planetary albedo (Charlson et al.,1987). Thus, studies of DMS oxidation have been investigated bymany researchers to understand DMS products and their impact oncloud chemistry.

Despite many studies through controlled chamber experimentsand field observations, many details about DMS photooxidation

: þ1 352 392 3076.

All rights reserved.

mechanisms, the aerosol formation, and the kinetic model remaincontroversial. For example, most chamber studies of DMS oxidationproducts have been operated solely for the reactions of DMS withOH radicals under both NOx free and NOx containing conditions(Arsene et al., 1999; Barnes et al., 1996; Librando et al., 2004; Yinet al., 1990a, 1990b), but no impact of coexisting volatile organiccompounds (VOCs) has been considered. The atmospheric VOCscoexisting with DMS can modify gas phase chemistry of DMS, suchas RO2 radical chemistry, the production of OH radicals, and theozone formation associated with the production of O(1D). Thesemodified gas phase reactions due to VOCs can also influencechemical distributions of DMS oxidation products that are directlyrelated to DMS aerosol formation.

In addition to the complexity of DMS oxidation due to thecoexisting VOCs, the aerosol phase heterogeneous reactions of DMSproducts with atmospheric organic compounds have not beenelucidated yet. For example, Jang et al. (2002) have proposed thatatmospheric organics, which are partitioned to aerosols, can befurther transformed via heterogeneous reactions, particularly in thepresence of submicron sulfuric acid aerosols. The direct outcomeof heterogeneous acid catalyzed reactions is the formation of

T. Chen, M. Jang / Atmospheric Environment 46 (2012) 271e278272

oligomeric matter in aerosol and an increase in SOA mass. Besidessulfuric acid, DMS produces MSA which, being a strong acid(pKa ¼ �2), may also enhance the heterogeneous reactions ofatmospheric organic compounds increasing SOA production.

Up to date, the simulation of DMS photooxidation against fielddata has been conducted using an explicit gas-phase mechanismcoupledwith a few simple aerosol phase reactions (Karl et al., 2007;Lucas and Prinn, 2002, 2005) or the semiempirical gas and particlekinetic models based on the fitting parameters to field data (Chenet al., 2000; Davis et al., 1999; Mari et al., 1999). However, thesesimplified DMS oxidation mechanisms are not satisfactory inprediction of DMS oxidation products such as dimethyl sulfoxide(DMSO) (Chen et al., 2000), H2SO4 and MSA (Lucas and Prinn,2002). This deviation is caused by the uncertainty of the gasphase reaction rate constants for both DMS and DMS products andthe lack of MSA formation mechanisms through heterogeneousreactions of DMS oxidation products on the surface of aerosol. Inaddition, none of previous models included the aerosol phaseheterogeneous reactions of DMS products with atmospheric sem-ivolatile oxygenated compounds, which are mainly originated fromsecondary organic aerosol (SOA) due to the photooxidation of VOCs.

In this study, in an attempt to understand the interactionbetween the oxidations of DMS and SOA precursors, the photoox-idation of a mixture of isoprene and DMS was explored using anindoor Teflon film chamber under various NOx concentrations andrelative humidities (RH). Isoprenewas chosen as a representative ofbiogenic SOA precursors not only because it has a large emission(440e660 TgC yr�1, about same as the global methane flux)(Guenther et al., 2006), but also because its SOA yields are highlysensitive to aerosol acidity. For example, Czoschke et al. (2003)reported that the SOA yield produced from the ozonolysis ofisoprene increases by almost 200% when there is preexistingsulfuric acid aerosol. Edney et al. (2005) showed a 1000% increasein the isoprene SOA yield in the presence of SO2.

To investigate the impact of isoprene on the DMS aerosol, theDMS aerosol products (with and without isoprene) were quantifiedusing a Particle-Into-Liquid-Sampler coupled with Ion Chroma-tography (PILS-IC). To evaluate the DMS impact on the isopreneSOA formation, the isoprene SOA yields in the DMS/isoprene/NOx

system were also calculated using a new approach implementingan Organic Carbon (OC) analyzer, and compared with those in theabsence of DMS.

Table 1DMS aerosol phase products measured using a PILS-IC at two different RH levels (10% an

Exp.a NOxb

ppbIsopreneb

ppbDMSb ppb(mg m�3)

DDMS ppb(mg m�3)

OMmg m�3

MSIAd

mass(molar)

MSAmas(mo

RH ¼ 10%M-1 185 800 140 (354.0) 70 (177.0) 103.8 4.4 (3.4) 25.8D-1 200 0 170 (431.0) 71 (177.0) 108.3 4.8 (3.7) 48.0RH ¼ 45%M-2 181 847 139 (352.0) 56 (142.0) 126.7 6.8 (5.3) 33.4

a M: the mixture of isoprene and DMS experiment; D: the photooxidation of DMS onb Initial concentration.c All the aerosol yield data were corrected for the wall loss using the first-order decay ra

(�2%), PILS-IC (�6%), SOA density (�5%) and DROGiso (�6%).d Aerosol phase products only.e TRS: Total reduced sulfur including all the gas phase sulfur other than DMS. The TRSf Total sulfur yield¼MSIA aerosol yieldþMSA aerosol yieldþ H2SO4 aerosol yieldþ SO

loss.g xDMS is defined by Eq. (3). xDMS (IC) is determined using PILS-IC assuming that MSA

mass þMSA mass þ H2SO4 mass)/total aerosol mass (OM). The assumption here is valid bh OCDMS/OMDMS (IC) ¼ (0.15 �MSIA mass þ 0.125 �MSAmass)/(MSA mass þMSIA ma

are 0.15 and 0.125, respectively.

2. Experimental section

2.1. Teflon film indoor chamber experiments

All SOA experiments were conducted in a 2 m3 indoor Teflonfilm chamber equipped with UVeVisible lamps (Solarc SystemsInc., FS40T12/UVB) covering all wavelengths ranging between 280and 900 nm. Prior to each experiment, the chamber was flushedusing air from clean air generators (Aadco Model 737, Rockville,MD; Whatman Model 75-52, Haverhill, MA). Precursor organicswere added to the chamber by passing clean air through a T unionwhere the chemicals were injected using a syringe, while NOx gaswas introduced using a syringe through a certificate NO tank (99.5%nitric oxide, Linde Gas). SOA experimental conditions and resultingSOA data at different relative humidity (RH¼ 12%, 42% and 80%) aresummarized in Tables 1e3.

2.2. Chemicals and instruments

All the organic chemicals (purity levels > 98%) were purchasedfrom Aldrich (Milwaukee, WI). The ozone concentration wasmeasured by a photometric ozone analyzer (Teledynemodel 400E),while the NOx concentration was measured using a chem-iluminescence NO/NOx analyzer (Teledyne Model 200E). A fluo-rescence analyzer (Teledyne Model 102E) was used to measure SO2

and Total Reduced Sulfur (TRS, gas phase) concentration. Theparticle size distribution was monitored with a Scanning MobilityParticle Sizer (SMPS, TSI, Model 3080, MN) combined witha condensation nuclei counter (CNC, TSI, Model 3025A). The SMPSdata was corrected for the aerosol loss due to the chamber wallwith a first-order decay (Mcmurry and Grosjean, 1985). The gas-phase concentration of the precursors was measured using an HP5890 Gas Chromatography-Flame Ionization Detector (GC-FID).A PILS-IC (Metrohm, 761 Compact) was used to measure the majoraerosol products produced from DMS photooxidation. The PILS-ICsample was collected at the end of each chamber experiment(Exp. M-1, M-2 and D-1). The detection limit of PILS-IC is 0.2 mgm�3

and the associated error is�6%. The aerosol sampling flow rate was13 L min�1 and the liquid flow rate in the anion column was0.7 mL min�1. A basic denuder, coated with 1% glycerol and 2%K2CO3 in ethanol-water (1:1), was placed upstream of the PILS toremove gaseous acids. For measuring OC data, the aerosol was

d 45%).

Yield � 100%c YDMS �100%massd

Total Syield �100%molarf

xDMS �100%(IC)g

OCDMS/OMDMS

(IC)hd

slar)

H2SO4d

mass(molar)

SO2

molarTRSmolare

(16.7) 9.4 (7.3) 29.7 19.2 39.6 76.3 68 0.10(31.0) 8.4 (4.6) 37.3 12.3 61.1 88.9 100 0.11

(21.6) 9.8 (6.2) 32.8 14.0 50.0 79.8 56 0.10

ly.

te. The estimated uncertainty (�10%) is calculated from errors associated with SMPS

yield was not corrected for the wall loss.2 yieldþ TRS yield. MSIA, MSA and H2SO4 aerosol yield have been corrected for wall

, MSIA and H2SO4 are the only aerosol phase products of DMS. xDMS (IC) ¼ (MSIAecause the calculated xDMS of D-1 (no isoprene) reaches the theoretical value, 100%.ss þ H2SO4 mass). The MSIA and MSA carbon fractions of the total molecular weight

Table 2Isoprene SOA experiments with and without DMS in the presence of NOx at RH ¼ 12%, 42% and 80%.a

Exp.b NOx ppb DMS ppb Isoprene ppb DROG mg m�3 DDMS mg m�3 DOM mg m�3 OC/V c xiso � 100%d,h Yiso � 100%e,h YDMS � 100%f,h % DYisog,h

RH ¼ (12 � 2)%M-3 48 119 850 241.0 26.6 6.8 0.38 56.7 (53.2) 1.60 (1.50) 11.1 (11.90) 50.9% (41.6%)I-3 40 0 870 478.6 0 5.1 0.56 100 1.06 n.a.M-4 86 103 872 784.0 33.9 19.0 0.37 54.3 (50.7) 1.31 (1.23) 25.6 (27.50) 35.1% (26.7%)I-4 87 0 860 651.0 0 6.3 0.56 100 0.97 n.a.M-5 220 129 880 2355.1 203.5 116.3 0.31 37.8 (33.2) 1.87 (1.64) 35.5 (37.90) 98.9% (74.4%)I-5 210 0 920 2437.7 0 22.9 0.59 100 0.94 n.a.M-6 42 30 210 475.1 27.3 10.5 0.40 57.4 (54.2) 1.26 (1.20) 16.4 (17.50) 32.6% (26.1%)I-6 55 0 210 541.2 0 5.2 0.59 100 0.95 n.a.M-7 44 15 100 257.5 n.a. 6.3 0.42 61.3 (58.5) 1.5 (1.43) n.a. 68.5% (60.8%)I-7 57 0 100 257.0 0 2.3 0.59 100 0.89 n.a.RH ¼ (42 � 2)%M-8 40 100 820 446.4 n.a. 9.2 0.36 55.9 (52.1) 1.16 (1.07) n.a. 78.5% (65.2%)I-8 40 0 810 445.0 0 2.9 0.53 100 0.65 n.a.M-9 81 100 840 681.8 n.a. 18.6 0.31 46.1 (41.2) 1.28 (1.12) n.a. 124.6% (97.2%)I-9 95 0 880 1051.6 0 5.96 0.50 100 0.57 n.a.RH ¼ (80 � 2)%M-10 30 120 851 865.1 n.a. 23.4 0.31 41.8 (37.0) 1.13 (1.00) n.a. 98.2% (75.4%)I-10 40 0 1030 1635.6 0 9.4 0.54 100 0.57 n.a.M-11 85 130 1050 1335.6 n.a. 30.6 0.33 47.2 (42.7) 1.08 (0.98) n.a. 146.7% (124.4%)I-11 81 0 1040 1380.9 0 6.3 0.54 100 0.45 n.a.

a Temperature was 21e24 �C and the aerosol concentrations in the background air were 0.1e0.2 mg m�3. Refer to the superscripts in Table 1.b I: isoprene experiment (no DMS).c The ratio of the OC mass concentration to total volume concentration of aerosols. OC/V values (g cm�3) were determined using an OC/EC analyzer and SMPS.d xiso is defined by Eq. (2). Its uncertainty (�11.4%) was estimated from errors associated with SMPS, OC/EC (�10%) and SOA density.e The maximal isoprene SOA yield with and without DMS (see Eq. (4)). The uncertainty was estimated from errors associated with SMPS, OC/EC, SOA density and DROG. The

uncertainty for the single precursor SOA yield was �8% and that for SOA yields in the mixture was �13.8%.f The maximum DMS aerosol yield (YDMS) in the presence of isoprene. YDMS ¼ OM � (1 � xiso)/DDMS. The uncertainty associated with YDMS is �13.8%.g Percentage of Yiso increase due to DMS. %DYiso ¼ (Y 0

iso e Yiso)/Yiso, where Y 0iso is the isoprene SOA yield in the presence of DMS and Yiso is the yield in the absence of DMS.

h Data in brackets are corrected based on water loss due to the maximum esterification reactions. n.a.: not applicable.

T. Chen, M. Jang / Atmospheric Environment 46 (2012) 271e278 273

analyzed at the end of each experiment with a semi-continuousOC/EC carbon aerosol analyzer (Sunset Laboratory Model 4) usingthe NIOSH 5040 method. The detection limit of OC analysis is0.3 mgm�3 (for 120min sampling) and the associated error is�10%.The sampling flow rate for OC analyses was 8 L min�1. The samplingtime varied between 40 and 120 min, depending on the massconcentration of aerosols in the chamber.

3. Results and discussion

3.1. PILS-IC data to study the impact of isoprene on DMS aerosolformation

The products of DMS/NOx photooxidation with isoprene werecharacterized using a PILS-IC and compared to those withoutisoprene. Fig. 1 shows a typical ion chromatogram for DMSphotooxidation products, which mainly includes methanesulfinicacid (MSIA), MSA and sulfuric acid. For the analysis of MSIA andMSA, no interference due to gaseous carboxylic acids (e.g., formicacid and acetic acid) appears when a denuder coated with

Table 3SOA formation from DMS photooxidation using an indoor chamber at RH ¼ 12%.a

Exp. NOx

ppbDMSppb

DDMSmg m�3

OMmg m�3

OCDMS/VDMS

OCDMS/OMDMS

bYDMS �100%c

D-2 280 136 255.1 192 0.13 0.11 75.3D-3 310 127 209.6 156.8 n.a. n.a. 74.8D-4 210 170 113.9 73 0.14 0.12 64.1D-5 43 116 121.5 65.1 0.14 0.12 53.6

a Refer to the superscripts in Table 2.b OCDMS/OMDMS ¼ OCDMS/VDMS/1.2. The density of the DMS aerosol is 1.2 g cm�3.

OCDMS/VDMS values (g cm�3) were determined using an OC/EC analyzer and SMPS.c Aerosol yield for DMS. YDMS ¼ OM/DDMS. The uncertainty for the single

precursor SOA yield was �8%.

a mixture of 1% glycerol and 2% K2CO3 in ethanol-water upstreamthe PILS removes these carboxylic acids, which are produced fromphotooxidation of isoprene. Table 1 summarizes aerosol-phaseDMS product distribution in both the presence (M-1 and M-2)and the absence (D-1) of isoprene.

MSA accounts for the highest aerosol mass fraction (70% onaverage, see Fig. 2) among the three major DMS aerosol products.As shown in Table 1, the total aerosol mass yield of DMS (YDMS),estimated from the sum of MSA, MSIA and H2SO4 mass yield, was39.6% with isoprene (M-1) and 61.1% without isoprene (D-1),resulting in a 35.2% decrease in YDMS due to isoprene. The aerosol

Fig. 1. Anion chromatogram of the aerosol sample of D-1 showing MSA, MSIA andsulfate.

Fig. 2. Major acid (MSIA, MSA and sulfuric acid) mass fraction of the total acidicaerosol mass in M-1, M-2 and D-1 (assuming MSIA, MSA and sulfuric acid are the onlyacidic aerosol products).

T. Chen, M. Jang / Atmospheric Environment 46 (2012) 271e278274

phase MSA yield with isoprene (M-1) was 46% lower than thatwithout isoprene (D-1), indicating that isoprene significantlydecreases the DMS aerosol yield mainly through the reduction ofthe aerosol phase MSA.

It is currently unclear why the DMS aerosol yield becomes lowerin the presence of isoprene. There are two possible reasons. First,isoprene and its gas-phase products compete for atmosphericoxidants with the DMS and DMS gas-phase products. Due to thepresence of isoprene, the reaction system produces differentamounts of atmospheric oxidants and radicals, such as RO2, NO3,O3P and OH radicals. This modification changes DMS oxidationpathways that lead to different yields of MSA. Second, the aqueousfacilitated heterogeneous reaction of MSIA that produces MSA(Bardouki et al., 2002; Hopkins et al., 2008) is unfavorable on thesurface of the relatively hydrophobic aerosol containing isopreneoxidation products.

Unlike the DMS/NOx system mainly comprising MSA, MSIA andsulfuric acid, the aerosol produced from the DMS/isoprene/NOx

system may also contain irreversible organosulfates (nonelectro-lytes) that are formed through the reaction of isoprene productsand the acids originated from DMS photooxidation. The PILS-ICmethod may underestimate the sulfur compounds in the DMSaerosol in the presence of isoprene. However, the DMS yields usingthe PILS-IC data agree well with those obtained from the OC data(see “Section 3.3. Comparison of PILS-IC and OC data”). Sucha tendency suggests that the organosulfates produced, if any, arereversible during the PILS-IC analysis where hot water steam isused for the aerosol extraction.

The total sulfur molar yields in M-1 and M-2 were calculated(Table 1) without the wall loss correction for the total reducedsulfur (TRS). The total identified sulfur molar yield withoutisoprene (D-1) reaches almost 89% and that with isoprene (M-1 andM-2), on average 78% (Table 1). The wall loss of polar products (e.g.,MSA and MSIA) and the heterogeneous reaction of DMS productswith isoprene products on the chamber wall would contribute tothe unidentified sulfur molar yields (11% for D-1 and 22% for M-1and M-2).

3.2. OC analysis to study the impact of DMS on isoprene SOA yields

The typical SOA yield for a single precursor is described as(Odum et al., 1996)

Y ¼ OMDROG

(1)

where OM is the mass concentration (mg m�3) of organic matterformed and DROG is the mass concentration (mg m�3) of theconsumed reactive organic gas.

To estimate the SOA yield for each precursor in the mixture ofDMS and isoprene, the total aerosol mass, OMDMS�iso, is separatedinto the OM associated with DMS ðOM�

DMSÞ and the OM associatedwith isoprene ðOM�

isoÞ. We define the mass fraction of isoprene SOAin OMDMS�iso as xiso and the mass fraction of DMS aerosol inOMDMS�iso as xDMS.

xiso ¼ OM�iso

OMDMS�iso(2)

xDMS ¼ 1� xiso ¼ OM�DMS

OMDMS�iso(3)

Both xiso and xDMS range between 0 and 1. Then, the isoprene SOAyield (Yiso) in the presence of DMS can be expressed as:

Yiso ¼ OM�iso

DROGiso¼ xiso

OMDMS�isoDROGiso

(4)

where DROGiso is the consumed isoprene mass concentration.In this study, an OC analysis was used to estimate xiso and xDMS.

The organic carbon (OCDMS�iso) in the aerosol produced fromphotooxidation of the isoprene/DMS/NOx system is the sum of OCcontributed by isoprene ðOC�isoÞ and DMS ðOC�DMSÞ.

OCDMS�iso ¼ OC�DMS þ OC�

iso (5)

Each side in Eq. (5) is divided by OMDMS�iso.

OCDMS�isoOMDMS�iso

¼ OC�isoOMDMS�iso

þ OC�DMSOMDMS�iso

(6)

By adding Eqs. (2) and (3) into Eq. (6), the equation can be trans-formed into

OCDMS�isoOMDMS�iso

¼ OC�isoOM�

isoxiso þ

OC�DMSOM�

DMSð1� xisoÞ (7)

We assume that OC�iso=OM�iso for the DMS/isoprene/NOx system

is not significantly different from OCiso=OMiso for the isoprene/NOx

system, where OCiso is the organic carbon mass formed from theisoprene/NOx system (no DMS) at a given experimental conditionand OMiso is the total aerosol mass in the same system. Similarly,we assume OC�DMS=OM

�DMSzOCDMS=OMDMS, where OCDMS and

OMDMS are for the aerosol from the DMS/NOx system (no isoprene),respectively. Table 1 shows OCDMS/OMDMS values for M-1, M-2 andD-1, which are estimated using the chemical composition of DMSaerosol products (MSA, MSIA and sulfuric acid) (Fig. 2) and thecarbon fractions of each acid (0.125 for MSA, 0 for sulfuric acid and0.15 for MSIA). The resulting OCDMS/OMDMS values of the aerosolfrom the DMS/NOx system were 0.10 and close to OC�DMS=OM

�DMS

(0.11) of the DMS/isoprene/NOx aerosol, supporting the assumptionOC�DMS=OM

�DMSzOCDMS=OMDMS (within the �8.5% statistical error,

see superscript h in Table 1). Then, Eq. (7) can be rewritten as

OCDMS�isoOMDMS�iso

¼ OCisoOMiso

xiso þOCDMS

OMDMSð1� xisoÞ (8)

The organic mass (OMiso) in the isoprene/NOx system can be esti-mated from the isoprene SOA density (riso) and isoprene SOAvolume (Viso) using SMPS data.

OMiso ¼ Visoriso (9)

The organic mass (OMDMS) in the DMS/NOx system and the organicmass (OMDMS�iso) in the isoprene/DMS/NOx system can bedescribed in a similar way,

T. Chen, M. Jang / Atmospheric Environment 46 (2012) 271e278 275

OMDMS ¼ VDMSrDMS (10)

Fig. 3. xiso and YDMS values associated with M-1 (IC data) compared to those with M-5(OC/EC data). Experimental conditions of M-1 and M-5 are similar. xiso and YDMS

associated with OC/EC data are found in Table 2 and YDMS associated with IC data is inTable 1. xiso associated with IC data ¼ 1 � xDMS (Table 1).

OMDMS�iso ¼ VDMS�isorDMS�iso (11)

where VDMS is the DMS aerosol volume and rDMS is the DMS aerosoldensity. VDMS�iso and rDMS�iso correspond to the volume and thedensity of the aerosol of the isoprene and DMS mixture system,respectively. rDMS�iso is estimated approximately by the followingequation.

rDMS�iso ¼ risoxiso þ rDMSð1� xisoÞ (12)

Using Eqs. (9)e(12), Eq. (8) can be rewritten as

OCDMS�isoVDMS�iso½risoxiso þ rDMSð1� xisoÞ�

¼ OCisoVisoriso

xiso

þ OCDMS

VDMSrDMSð1� xisoÞ ð13Þ

Eq. 13 is rearranged to solve for xiso.

xiso¼1

2ðriso�rDMSÞ�

OCisoVisoriso

� OCDMS

VDMSrDMS

����OCisoVisoriso

rDMS

þ OCDMS

VDMSrDMSðriso�2rDMSÞ

�þ��

OCisoVisoriso

rDMS

þ OCDMS

VDMSrDMSðriso�2rDMSÞ

�2�4ðriso�rDMSÞ

�OCisoVisoriso

� OCDMS

VDMSrDMS

��OCDMS

VDMS�OCDMS�iso

VDMS�iso

��0:5�ð14Þ

where riso (1.1 g cm�3) and rDMS (1.2 g cm�3) are experimentallydetermined using SMPS data and the aerosol mass collected on thefilter. OCiso, OCDMS and OCDMS�iso are determined by the OC/ECanalyzer. Viso, VDMS and VDMS�iso are analyzed by the SMPS. TheOCiso/Viso values of the isoprene/NOx system range from 0.56 to 0.59at RH ¼ 12% and those of the isoprene/DMS/NOx system, 0.31e0.42(Table 2). The OCDMS/VDMS value in this study is fixed at 0.13 for NOx

reactions (OCDMS/VDMS in Table 3). The resulting xiso values arelisted in Table 2. The Yiso values reported in Table 2 are estimatedusing Eq. (2) with OMiso, OMiso�DMS, xiso and DROGiso.

When we take into consideration the esterification of DMS majorproducts (MSA and sulfuric acid) with isoprene products,OC�DMS=OM

�DMS can be slightly larger than OCDMS=OMDMS because

OM�DMS decreases with water evaporation e a concurrent process

associated with esterification. Considering esterification is thereforeneeded to correct the OC�

DMS=OM�DMS calculation. When we assume

that all the produced MSA and sulfuric acid are converted to esters(maximum esterification), the mass loss fraction of MSA due toesterification is 18/96 where 18 and 96 are the molecular weight ofwater and MSA, respectively; the loss fraction of sulfuric acid is 36/98where 36 is the molecular weight of 2 mol of water and 98 is that of1 mol of sulfuric acid. Using the chemical composition information(Fig. 2), the estimated mass loss fraction of OM�

DMS due to esterifica-tion is 0.20 suggesting that OC�DMS=OM

�DMS ¼ 1:25� OC�

DMS=OM�DMS

(no reaction). xiso (maximum esterification) is only 7.2% smaller thanxiso (no reaction), which is within the�11.4% error range (see “xiso” inTable 2). Thus, we will only report the data without water losscorrection in the following discussion.

3.3. Comparison of PILS-IC and OC data

The OC/EC data are compared to the PILS-IC data to evaluate thevalidity of the twomethods. For example, the PILS-IC measurement

in M-1 (Table 1) and the OC/EC analysis in M-5 (Table 2) wereconducted in similar experimental conditions. Fig. 3 summarizesxiso and YDMS fromM-1 andM-5. Bothmethods show similar resultsin either xiso or YDMS: e.g., xiso ¼ 32.0% based on IC data andxiso ¼ 37.8% from OC data; YDMS ¼ 39.6% based on IC data andYDMS ¼ 35.5% from OC data. Consistency between the two methodsis also observed for the OCDMS/OMDMS data: 0.11 for D-1 using IC(Table 1) and 0.12 for D-4 using OC/EC (Table 3). Thus, we concludethat the OC/EC approach is an appropriate method to estimate thedecoupled aerosol yields (Yiso and YDMS) in the binary mixture andthat the PILS-IC analysis reasonably quantifies the major DMSaerosol-phase products in the presence of isoprene.

3.4. The impact of DMS on the isoprene SOA yield: NOx effects

The impact of DMS on Yiso was analyzed by comparing Yiso withDMS to that without DMS at given NOx and isoprene concentra-tions. Fig. 4A presents Yiso at a given isoprene concentration(850 ppb) and humidity (RH ¼ 12%) with varying NOx concentra-tions (40e220 ppb) with and without DMS. Fig. 4B shows Yiso ata given NOx concentration (40 ppb) and humidity (RH ¼ 12%) withvarying isoprene concentrations (100e850 ppb) with and withoutDMS. The percentage increase of Yiso due to DMS (%DYiso) issummarized in Table 2. The resulting %DYiso values (between 32.6%and 98.9%) suggest that Yiso is significantly elevated by DMS ina wide range of NOx concentrations.

3.5. The impact of DMS on isoprene SOA formation: RH effects

In Fig. 5, the Yiso both with DMS and without DMS at RH ¼ 12%were compared with those at higher RHs (42% and 80%), with twoinitial NOx concentrations (40 ppb and 80 ppb) at a given isopreneconcentration (850 ppb). Overall, %DYiso (Fig. 5) was much higher athigh RH compared to that at low RH.

Humidity can influence both gas-phase chemistry and aerosol-phase reactions inside the chamber. In the presence of NOx,higher OH radicals can be formed though the reaction of O(1D) witha water molecule. Under our experimental condition, isoprene ispresent over the entire chamber experiment. Thus, the additionalOH radicals due to higher humidity are more likely consumed byisoprene instead of producing highly oxidized products. On theother hand, the higher humidity can reduce HO2 radicals in the gasphase (Kanno et al., 2005), influencing the formation of ROOH. It is

Fig. 4. (A) Isoprene SOA yields at different NOx concentration with and without DMS (isoprene concentration ¼ 850 ppb and RH ¼ 12%). (B) Isoprene SOA yields at different isopreneconcentration with and without DMS (NOx concentration ¼ 40 ppb and RH ¼ 12%). The isoprene SOA yield was calculated using Eq. (4).

T. Chen, M. Jang / Atmospheric Environment 46 (2012) 271e278276

known that ROOH can further react with an aldehyde and producehigh molecular weight hemiacetal in the aerosol (Johnson et al.,2004). Hence, isoprene SOA yields decrease with increasinghumidity.

The greater %DYiso at the high RH condition (Fig. 5) is mainly dueto the higher amount of acid formed fromDMS photooxidation. Theformation of MSA on the aerosol surface is enhanced by higherRH due to the aqueous processing from MSIA. The higher MSAaccelerates aerosol phase heterogeneous reactions of organiccompounds increasing SOA formation. This explanation is sup-ported by theMSAmass yield and DMS aerosol mass fraction (xDMS)of the total aerosolmass listed inTable 1. For example, theMSAmassyield is 25.8% at RH ¼ 10% (M-1 in Table 1) compared with 33.4% atRH ¼ 45% (M-2 in Table 1). In addition, the DMS aerosol fraction inthe total aerosol (xDMS) is 68.0% (M-1) at RH ¼ 10% and 56.0% (M-2)at RH ¼ 45%, indicating stronger heterogeneous reactions at higherRH that drive more volatile isoprene products onto the aerosol.

3.6. The impact of DMS on isoprene SOA formation: aerosol growthpattern

The impact of DMS on Yiso was also investigated for the aerosolgrowth pattern between the isoprene-only system and the mixturesystem. The time profiles of the SOA mass and isoprene concen-trations with DMS (M-4) and without DMS (I-4) are shown inFig. 6A. The time profile for DMS aerosol mass and DMS concen-tration (D-5) is shown in Fig. 6B. In Fig. 6A, the photoirradiation ofthe isoprene-only system shows a 40-min induction period(no significant SOA growth), the time to reach the saturation of the

Fig. 5. (A) Isoprene SOA yields at different RH with and without DMS (initial isoprene Concewithout DMS (initial isoprene concentration ¼ 850 ppb and NOx ¼ 80 ppb). The isoprene SOyield (%DYiso) due to DMS was marked in the figure.

gas-phase isoprene oxidation products. In contrast, rapid aerosolgrowth occurred from the beginning of SOA experiment in the DMSphotoirradiation system owing to the self-nucleation of H2SO4. Thedynamics of aerosol growth patterns shown in Fig. 6 indicate thatthe DMS aerosol forms before the condensation of isopreneoxidation products when the mixture of isoprene and DMS arephotoirradiated together. The pre-formed acidic products fromDMS photooxidation can serve not only as catalysts that facilitatethe formation of methyl-tetrols and nonvolatile oligomers fromreactive organic compounds (e.g., glyoxal and methyl glyoxal), butalso as reactants that produce organosulfates (Iinuma et al., 2007;Liggio and Li, 2008; Minerath et al., 2008, 2009; Paulot et al.,2009; Surratt et al., 2010).

4. Uncertainties

The uncertainties associated with the use of the indoor chamberinclude the temperature increase (w6 K) due to UV photo-irradiation lamps. The temperature change can affect the gas-particle partitioning of semivolatile products onto aerosols(Carlton et al., 2009). The artificial light sources producing wave-lengths in the range of 280 and 900 nm, different from the actualnatural sunlight, can lead to different product compositions(Warren et al., 2008). In addition, the progression of heterogeneouschemistry on the chamber wall between organic and inorganicspecies could potentially reduce the effects of heterogeneousreactions on the surface of the SOA. To minimize the uncertainties,a large outdoor chamber facility at the University of Florida will beused for future studies. In addition, the detailed clarification of the

ntration ¼ 850 ppb and NOx ¼ 40 ppb). (B) Isoprene SOA yields at different RH with andA yield (Yiso) was calculated using Eq. (4) and the percentage increase of isoprene SOA

Fig. 6. (A) Time profiles of aerosol growth and isoprene decay for the isoprene/NOx

photooxidation with and without DMS (Exp. I-4 and M-4). (B) Time profiles of aerosolgrowth and the DMS decay for the DMS/NOx photooxidation (Exp. D-5).

T. Chen, M. Jang / Atmospheric Environment 46 (2012) 271e278 277

mechanisms of the impact of DMS on isoprene SOA formation isneeded based on the analytical studies of both the gas and aerosolphase products.

5. Atmospheric implications

Our study is beneficial for evaluating the potential impact ofDMS as a reduced sulfur compound on SOA formation in theambient air especially near seaside areas. We have observedsignificant Yiso increases due to DMS at an atmospheric relevant RH(42%) conditions using the given mixing ratio (Table 2). Theisoprene concentration in the coastal area is 200e800 ppt, varyingwith season and location (Bottenheim and Shepherd, 1995;Holzinger et al., 2002; Yokouchi, 1994), and the coastal concen-tration of DMS is usually 50e200 ppt (Ramanathan et al., 2001).Thus, possible mixing ratios of isoprene to DMS can range from 1 to16 in coastal areas. Therefore, the mixing ratio of 7.5 used in ourstudy falls into the actual ambient mixing ratio range.

Based on the tendency in Yiso as a function of NOx (Fig. 4A), Yisovalue of this study was not strongly dependent of NOx concentra-tions but was significantly affected by the presence of DMS.Although concentrations of isoprene, DMS, and NOx are higher inour study than those in ambient, the similar impact of DMS on Yisowould occur in the ambient air. For example, Froyd et al. (2010)have recently discovered a significant loss of gaseous epoxide(characteristic product of isoprene photooxidation in ambient) tothe acidic aerosol during the maritime convective lofting of DMS.

Their observation suggests that heterogeneous reactions ofisoprene SOA products on the surface of the acidic aerosol, which isproduced from DMS oxidation, can increase isoprene SOA yields inthe ambient air. Further field studies are required in the future tofindmore evidence for the effect of DMS on isoprene SOA formationin the ambient atmosphere.

Acknowledgments

This work was supported by grants from the National ScienceFoundation (ATM-0852747). We thank David Smith in the SunsetLaboratory for correcting OC/EC data.

References

Arsene, C., Barnes, I., Becker, K.H., 1999. FT-IR product study of the photo-oxidationof dimethyl sulfide: temperature and O2 partial pressure dependence. PhysicalChemistry Chemical Physics 1, 5463e5470.

Aumont, O., Belviso, S., Monfray, P., 2002. Dimethylsulfoniopropionate (DMSP) anddimethylsulfide (DMS) sea surface distributions simulated from a global three-dimensional ocean carbon cycle model. Journal of Geophysical Research-Oceans107, 3029e3041.

Bardouki, H., Berresheim, H., Vrekoussis, M., Sciare, J., Kouvarakis, G.,Oikonomou, K., Schneider, J., Mihalopoulos, N., 2003. Gaseous (DMS, MSA, SO2,H2SO4 and DMSO) and particulate (sulfate and methanesulfonate) sulfurspecies over the northeastern coast of Crete. Atmospheric Chemistry andPhysics 3, 1871e1886.

Bardouki, H., da Rosa, M.B., Mihalopoulos, N., Palm, W.U., Zetzsch, C., 2002. Kineticsand mechanism of the oxidation of dimethylsulfoxide (DMSO) and meth-anesulfinate (MSIe) by OH radicals in aqueous medium. Atmospheric Envi-ronment 36, 4627e4634.

Barnes, I., Becker, K.H., Patroescu, I., 1996. FTIR product study of the OH initiatedoxidation of dimethyl sulphide: observation of carbonyl sulphide and dimethylsulphoxide. Atmospheric Environment 30, 1805e1814.

Barone, S.B., Turnipseed, A.A., Ravishankara, A.R., 1995. Role of adducts in theatmospheric oxidation of dimethyl sulfide. Faraday Discussions, 39e54.

Bopp, L., Aumont, O., Belviso, S., Monfray, P., 2003. Potential impact of climatechange on marine dimethyl sulfide emissions. Tellus Series B-Chemical andPhysical Meteorology 55, 11e22.

Bottenheim, J.W., Shepherd, M.F., 1995. C2eC6 hydrocarbon measurements at 4rural locations across Canada. Atmospheric Environment 29, 647e664.

Carlton, A.G., Wiedinmyer, C., Kroll, J.H., 2009. A review of secondary organicaerosol (SOA) formation from isoprene. Atmospheric Chemistry and Physics 9,4987e5005.

Charlson, R.J., Lovelock, J.E., Andreae, M.O., Warren, S.G., 1987. Oceanic phyto-plankton, atmospheric sulfur, cloud albedo and climate. Nature 326, 655e661.

Chen, G., Davis, D.D., Kasibhatla, P., Bandy, A.R., Thornton, D.C., Huebert, B.J.,Clarke, A.D., Blomquist, B.W., 2000. A study of DMS oxidation in the tropics:comparison of christmas island field observations of DMS, SO2, and DMSO withmodel simulations. Journal of Atmospheric Chemistry 37, 137e160.

Czoschke, N.M., Jang, M., Kamens, R.M., 2003. Effect of acidic seed on biogenicsecondary organic aerosol growth. Atmospheric Environment 37, 4287e4299.

Davis, D., Chen, G., Bandy, A., Thornton, D., Eisele, F., Mauldin, L., Tanner, D.,Lenschow, D., Fuelberg, H., Huebert, B., Heath, J., Clarke, A., Blake, D., 1999.Dimethyl sulfide oxidation in the equatorial Pacific: comparison of modelsimulations with field observations for DMS, SO2, H2SO4(g), MSA(g), MS, andNSS. Journal of Geophysical Research-Atmospheres 104, 5765e5784.

Edney, E.O., Kleindienst, T.E., Jaoui, M., Lewandowski, M., Offenberg, J.H., Wang, W.,Claeys, M., 2005. Formation of 2-methyl tetrols and 2-methylglyceric acid insecondary organic aerosol from laboratory irradiated isoprene/NOx/SO2/airmixtures and their detection in ambient PM2.5 samples collected in the easternUnited States. Atmospheric Environment 39, 5281e5289.

Froyd, K.D., Murphy, S.M., Murphy, D.M., de Gouw, J.A., Eddingsaas, N.C.,Wennberg, P.O., 2010. Contribution of isoprene-derived organosulfates to freetropospheric aerosol mass. Proceedings of the National Academy of Sciences ofthe United States of America 107, 21360e21365.

Gaston, C.J., Pratt, K.A., Qin, X.Y., Prather, K.A., 2010. Real-time detection and mixingstate of methanesulfonate in single particles at an inland urban location duringa phytoplankton bloom. Environmental Science & Technology 44, 1566e1572.

Guenther, A., Karl, T., Harley, P., Wiedinmyer, C., Palmer, P.I., Geron, C., 2006. Esti-mates of global terrestrial isoprene emissions using MEGAN (model of emis-sions of gases and aerosols from nature). Atmospheric Chemistry and Physics 6,3181e3210.

Holzinger, R., Sanhueza, E., von Kuhlmann, R., Kleiss, B., Donoso, L., Crutzen, P.J.,2002. Diurnal cycles and seasonal variation of isoprene and its oxidationproducts in the tropical savanna atmosphere. Global Biogeochemical Cycles 16,1074e1086.

Hopkins, R.J., Desyaterik, Y., Tivanski, A.V., Zaveri, R.A., Berkowitz, C.M.,Tyliszczak, T., Gilles, M.K., Laskin, A., 2008. Chemical speciation of sulfur inmarine cloud droplets and particles: analysis of individual particles from the

T. Chen, M. Jang / Atmospheric Environment 46 (2012) 271e278278

marine boundary layer over the California current. Journal of GeophysicalResearch-Atmospheres 113, D04209.

Iinuma, Y., Muller, C., Boge, O., Gnauk, T., Herrmann, H., 2007. The formation oforganic sulfate esters in the limonene ozonolysis secondary organic aerosol(SOA) under acidic conditions. Atmospheric Environment 41, 5571e5583.

Jang, M., Czoschke, N., Lee, S., Kamens, R., 2002. Heterogeneous atmospheric aerosolproduction by acid-catalyzed particle-phase reactions. Science 298, 814e817.

Johnson, D., Jenkin, M.E., Wirtz, K., Martin-Reviejo, M., 2004. Simulating theformation of secondary organic aerosol from the photooxidation of toluene.Environmental Chemistry 1, 150e165.

Kanno, N., Tonokura, K., Tezaki, A., Koshi, M., 2005. Water dependence of the HO2self reaction: kinetics of the HO2eH2O complex. Journal of Physical Chemistry A109, 3153e3158.

Karl, M., Gross, A., Leck, C., Pirjola, L., 2007. Intercomparison of dimethylsulfideoxidation mechanisms for the marine boundary layer: gaseous and particulatesulfur constituents. Journal of Geophysical Research-Atmospheres 112, D15304.

Kloster, S., Feichter, J., Reimer, E.M., Six, K.D., Stier, P., Wetzel, P., 2006. DMS cycle in themarine ocean-atmosphere systeme a global model study. Biogeosciences 3, 29e51.

Librando, V., Tringali, G., Hjorth, J., Coluccia, S., 2004. OH-initiated oxidation ofDMS/DMSO: reaction products at high NOx levels. Environmental Pollution 127,403e410.

Liggio, J., Li, S.M., 2008. Reversible and irreversible processing of biogenic olefins onacidic aerosols. Atmospheric Chemistry and Physics 8, 2039e2055.

Lucas, D.D., Prinn, R.G., 2002. Mechanistic studies of dimethylsulfide oxidationproducts using an observationally constrained model. Journal of GeophysicalResearch-Atmospheres 107, 4201e4226.

Lucas, D.D., Prinn, R.G., 2005. Parametric sensitivity and uncertainty analysis ofdimethylsulfide oxidation in the clear-sky remote marine boundary layer.Atmospheric Chemistry and Physics 5, 1505e1525.

Lukacs, H., Gelencser, A., Hoffer, A., Kiss, G., Horvath, K., Hartyani, Z., 2009. Quan-titative assessment of organosulfates in size-segregated rural fine aerosol.Atmospheric Chemistry and Physics 9, 231e238.

Mari, C., Suhre, K., Rosset, R., Bates, T.S., Huebert, B.J., Bandy, A.R., Thornton, D.C.,Businger, S., 1999. One-dimensional modeling of sulfur species during the first

aerosol characterization experiment (ACE 1) lagrangian B. Journal of Geo-physical Research-Atmospheres 104, 21733e21749.

Mcmurry, P.H., Grosjean, D., 1985. Gas and aerosol wall losses in teflon film smogchambers. Environmental Science & Technology 19, 1176e1182.

Minerath, E.C., Casale, M.T., Elrod, M.J., 2008. Kinetics feasibility study of alcoholsulfate esterification reactions in tropospheric aerosols. Environmental Science& Technology 42, 4410e4415.

Minerath, E.C., Schultz, M.P., Elrod, M.J., 2009. Kinetics of the reactions of isoprene-derived epoxides in model tropospheric aerosol solutions. EnvironmentalScience & Technology 43, 8133e8139.

Odum, J.R., Hoffmann, T., Bowman, F., Collins, D., Flagan, R.C., Seinfeld, J.H., 1996.Gas/particle partitioning and secondary organic aerosol yields. EnvironmentalScience & Technology 30, 2580e2585.

Paulot, F., Crounse, J.D., Kjaergaard, H.G., Kurten, A., St Clair, J.M., Seinfeld, J.H.,Wennberg, P.O., 2009. Unexpected epoxide formation in the gas-phase photo-oxidation of isoprene. Science 325, 730e733.

Ramanathan, V., Crutzen, P.J., Kiehl, J.T., Rosenfeld, D., 2001. Atmosphere e aerosols,climate, and the hydrological cycle. Science 294, 2119e2124.

Surratt, J.D., Chan, A.W.H., Eddingsaas, N.C., Chan, M.N., Loza, C.L., Kwan, A.J.,Hersey, S.P., Flagan, R.C., Wennberg, P.O., Seinfeld, J.H., 2010. Reactive inter-mediates revealed in secondary organic aerosol formation from isoprene.Proceedings of the National Academy of Sciences of the United States ofAmerica 107, 6640e6645.

Warren, B., Song, C., Cocker, D.R., 2008. Light intensity and light source influence onsecondary organic aerosol formation for the m-xylene/NOx photooxidationsystem. Environmental Science & Technology 42, 5461e5466.

Yin, F.D., Grosjean, D., Flagan, R.C., Seinfeld, J.H., 1990a. Photooxidation of dimethylsulfide and dimethyl disulfide .2. Mechanism evaluation. Journal of Atmo-spheric Chemistry 11, 365e399.

Yin, F.D., Grosjean, D., Seinfeld, J.H., 1990b. Photooxidation of dimethyl sulfide anddimethyl disulfide .1. Mechanism development. Journal of AtmosphericChemistry 11, 309e364.

Yokouchi, Y., 1994. Seasonal and diurnal-variation of isoprene and its reaction-products in a semirural area. Atmospheric Environment 28, 2651e2658.