Indoor Air. 2019;29:913–925.  | 913wileyonlinelibrary.com/journal/ina

Received:10April2019  |  Revised:1July2019  |  Accepted:11August2019DOI: 10.1111/ina.12594

O R I G I N A L A R T I C L E

Indoor ozone/human chemistry and ventilation strategies

Christian Mark Salvador1 | Gabriel Bekö2  | Charles J. Weschler2,3  | Glenn Morrison4 | Michael Le Breton1 | Mattias Hallquist1 | Lars Ekberg5,6 | Sarka Langer6,7

ThisisanopenaccessarticleunderthetermsoftheCreativeCommonsAttributionLicense,whichpermitsuse,distributionandreproductioninanymedium,providedtheoriginalworkisproperlycited.©2019TheAuthors.Indoor AirpublishedbyJohnWiley&SonsLtd

1DepartmentofChemistryandMolecularBiology,AtmosphericSciences,UniversityofGöteborg,Göteborg,Sweden2InternationalCentreforIndoorEnvironmentandEnergy,DepartmentofCivilEngineering,TechnicalUniversityofDenmark,Lyngby,Denmark3EnvironmentalandOccupationalHealthSciencesInstitute,RutgersUniversity,Piscataway,NJ,USA4DepartmentofEnvironmentalSciencesandEngineering,GillingsSchoolofGlobalPublicHealth,TheUniversityofNorthCarolinaatChapelHill,ChapelHill,NC,USA5CITEnergyManagementAB,Göteborg,Sweden6DivisionofBuildingServicesEngineering,DepartmentofArchitectureandCivilEngineering,ChalmersUniversityofTechnology,Göteborg,Sweden7IVLSwedishEnvironmentalResearchInstitute,Göteborg,Sweden

CorrespondenceSarkaLanger,IVLSwedishEnvironmentalResearchInstitute,P.O.Box53021,SE‐40014Göteborg,Sweden.Email:sarka.langer@ivl.se

Present addressMichaelLeBreton,VolvoGroupTrucksandTechnologyMethodandTechnicalDevelopment,Göteborg,Sweden

Funding informationSvenskaForskningsrådetFormas,Grant/AwardNumber:2015‐1509

AbstractThisstudyaimedtobetterunderstandandquantifytheinfluenceofventilationstrat‐egiesonoccupant‐relatedindoorairchemistry.Theoxidationofhumanskinoilcon‐stituentswasstudiedinacontinuouslyventilatedclimatechamberattwoairexchangerates(1h−1 and 3 h−1)andtwoinitialozonemixingratios(30and60ppb).Additionalmeasurementswereperformedtoinvestigatetheeffectofintermittentventilation(“off”followedby“on”).Soiledt‐shirtswereusedtosimulatethepresenceofoccu‐pants.Atime‐of‐flight‐chemicalionizationmassspectrometer(ToF‐CIMS)inpositivemodeusingprotonatedwaterclusterswasusedtomeasuretheoxygenatedreactionproductsgeranylacetone,6‐methyl‐5‐hepten‐2‐one(6‐MHO)and4‐oxopentanal(4‐OPA).Themeasurementdatawereusedinaseriesofmassbalancemodelsaccount‐ingforformationandremovalprocesses.Reactionsofozonewithsqualeneoccurringonthesurfaceofthet‐shirtsaremasstransportlimited;ventilationratehasonlyasmalleffectonthissurfacechemistry.Ozone‐squalenereactionsonthet‐shirtspro‐ducedgas‐phasegeranylacetone,whichwassubsequentlyremovedalmostequallybyventilationandfurtherreactionwithozone.About70%ofgas‐phase6‐MHOwasproducedinsurfacereactionsonthet‐shirts,theremainderinsecondarygas‐phasereactionsofozonewithgeranylacetone.6‐MHOwasprimarilyremovedbyventila‐tion,whilefurtherreactionwithozonewasresponsibleforaboutathirdof itsre‐moval.4‐OPAwasformedprimarilyonthesurfacesoftheshirts(~60%);gas‐phasereactionsofozonewithgeranylacetoneand6‐MHOaccountedfor~30%and~10%,respectively.4‐OPAwasremovedentirelybyventilation.Theresultsfromtheinter‐mittentventilationscenariosshoweddelayedformationofthereactionproductsandlowerproductconcentrationscomparedtocontinuousventilation.

K E Y W O R D S

airexchangerate,indoorenvironment,oxygenatedvolatileorganiccompounds,ozone,squalene,ToF‐CIMS

914  |     SALVADOR et AL.

1  | INTRODUC TION

The importanceof indoorairchemistryhasbeen increasingly rec‐ognizedoverthepastdecades.1‐4Buildingmaterials,furnishingsandcarpeting,cleaningproducts,personalcareproducts,humanactivi‐ties,aswellasoutdoorairhavebeenconsideredmajorsourcesofindoorairpollutants.Duringthepastdecade,therehasbeengrow‐ingrecognitionthathumansthemselvesconsiderablycontributetoindoorairpollutionandimpactindoorairchemistry.5,6

Squaleneconstitutesapproximately12%(byweight)ofhumanskinlipids;otherconstituentsincludefattyacids,glycerides,waxesters,ceramides,andcholesterolesters.7Havingsixcarbon‐car‐bon double bonds in its structure, squalene serves as a naturalantioxidant to protect our skin fromatmospheric oxidants suchasozone(O3),whichreactswiththeunsaturatedsitesofthemol‐ecule.Squaleneisresponsibleforabouthalfofallunsaturationsavailable inskin lipids,8andgiven theslightdifferences in reac‐tion probabilities between unsaturations in squalene and fattyacidsandtheiresters,itisresponsibleforapproximatelyhalftheozoneuptake.

Ozoneisthemostabundantoxidantintheindoorenvironment.Itsmajor source indoors is outdoor air, entering the buildings viaventilationand infiltration.The indoor‐to‐outdoor (I/O)concentra‐tionratioofozoneisusuallybetween0.1and0.7,dependingonthebuildingairexchangerate(AER)andthepropertiesoftheexposedsurfaces.9

Ozonolysis of surface‐bound squalene was first investigatedonMediterraneanvegetation.10Thecomplex reactionof squalenewithO3producesa largespectrumofoxygenatedvolatilespecies,includingacetone, geranyl acetone (GA),6‐methyl‐5‐hepten‐2‐one(6‐MHO)and4‐oxopentanal(4‐OPA).Thesewerereportedinasim‐ulatedaircraftcabincontainingsoiledt‐shirts in200511and inthesamesimulatedcabinwithhuman “passengers” in2007.12 Furtherdetails of ozone/human occupant chemistry were reported byWisthalerandWeschler13inacontrolledexperimentconductedinaclimatechamberrepresentingasimulatedoffice.

In subsequent studies,4‐OPAwas shown tobea sensory irri‐tant and capable of inducing inflammatory cytokine expression invitro.14‐16Thederivedhumanreferencevalueforsensoryirritationby6‐MHOis0.3ppm,andthereferencevaluesforairflowlimitationare0.03and0.5ppmfor4‐OPAand6‐MHO,respectively.17

Othervolatilesecondaryproductsofthisreactioninclude4‐methyl‐8‐oxo‐4‐nonenal (4‐MON), 4‐methyl‐4‐octene‐1,8‐dial(4‐MOD),1,4‐butanedialandcarboxylicacidssuchas4‐oxo‐bu‐tanoic acid and 4‐oxopentanoic (levulinic) acid. In addition tothevolatileproducts, theozonolysisof squalenealsogeneratesa number of long‐chained polyunsaturated aldehydes and car‐boxylicacids suchasC27‐pentaenal,C22‐tetraenal,C17‐trienal,C27‐pentaenoicacid,C22‐tetraenoicacid,C17‐trienoicacid,thatremain bound to the surface. Surface reactions remove ozonefromindoorenvironmentsmoreeffectivelythangas‐phasereac‐tions.18Thehumansurfacehasbeenshowntobeveryeffectiveatremovingozone.12,13,19

The oxygenated and highly oxygenated products of squaleneozonolysis are difficult to measure in indoor air by conventionaltechniquessuchassamplingonadsorbenttubesfollowedbyther‐maldesorptionintoagaschromatographwith,forexample,massse‐lectivedetector(TD/GC‐MS).Thesecompoundsareoftenthermallyunstable and undergo decomposition in the thermal desorptionstage.Even if someof these “stealthcompounds”maybesuitablefortheclassicalTD/GC‐MSanalysis,thetimeresolutionisinminutestohoursduetosamplingontheadsorbents.

Developmentofmonitoringinstrumentswithtimeresolutioninminutestoseconds,basedonmassspectrometry,hasenabledthedeterminationandconcentration‐timeprofilingofa rangeofvola‐tileorganiccompounds inair.Proton transfer reactionmassspec‐trometer(PTR‐MS)wasused,forexample,intheabove‐mentionedsimulatedaircraftcabin11,12andofficeexperiments.13Otherstudieslookedatozoneremovalandproductformationonclothes,20 human hair8,piecesofclothessoakedwithmethanolextractsofhumanskinlipids,21ontheheterogeneousmechanismofsqualene/ozonereac‐tion,22andonmodelingoftheprocesseswithrelevanceto indoorenvironments.18,19Building ventilation is used to control the ther‐mal environment and the concentrations of indoor air pollutants.Experimentalandmodelingstudieshavedemonstratedthe impactofventilationonindoorchemistryandproductformation.18,23,24

Theair exchange rate (AER)hasonlya small impactonchem‐ical reactions that occur on indoor surfaces. However, for indoorreactions in the gas‐phase,AERdetermines the time available forreactions to occur.While higher AER shortens the time availablefor reactions, it increases indoor ozone concentrations relative totheiroutdoorlevel.Thatis,lowAERtransportslessozonefromout‐doorstoindoors,butallowsmoretimeforchemistry.1Energysavingmeasureshaveledtotheimplementationofalternativeventilation

Practical Implications• Inoccupiedbuildings,ventilationstrategiesdonotoffera simple approach for controlling products that mayarise from ozone/human chemistry. This reflects thefactthatsuchproductsaregeneratedfrombothsurfacereactionsandgas‐phasereactions.

• Surface reactions are relatively insensitive to ventila‐tionrateswhilethegas‐phasereactionscompetewithventilation.

• Exposure to noxious products of ozone/human chem‐istry can be reduced by decreasing ventilation duringperiodswithhighoutdoorozonelevels.Whenbuildingsareunoccupied(eg,overnightoronweekends),decreas‐ingorstoppingventilationreducesindoorozonechem‐istryandsurfaceaccumulationofoxidationproducts.

• Results from this study can serve as inputs to predic‐tive models of ozone/human interaction in indoorenvironments.

     |  915SALVADOR et AL.

strategies, such as intermittent ventilation, where the ventilationsystemoperateswithminimalornoairflowovernight.Differentven‐tilationstrategiesareanticipatedtoaffectchemicaltransformationsinindoorairdifferently.

Theobjectiveofthisstudywastoinvestigatetheeffectofcon‐tinuousandintermittentventilationstrategiesonoccupant‐relatedozonechemistry.Dynamicandsteady‐statemixingratiosofoxygen‐atedproductsofskinoilozonolysisweremeasured.Togainabetterunderstanding of simultaneous formation and removal processes,productyieldswerederivedbyfittingaseriesofmassbalanceequa‐tionstothedata.

2  | METHODS

2.1 | Climate chamber

Theexperimentswereperformedina30m3stainlesssteelclimatechamber.Inordertoremovereactivecompoundsfromthesurfaces,the chamberwas treatedwith 375 ppb of ozone during a periodof72hourspriortothefirstexperiment.Duringtheexperiments,thetemperatureinthechamberwasmaintainedat23°C.Therela‐tivehumidity (RH)wasnot controlled,butwas relatively constantaround 35%. The air exchange rate (AER) was set to 1 or 3 h−1 (Table1).ItwasmeasuredregularlyduringtheexperimentalmonthwithanInnova1312PhotoacousticMulti‐gasMonitor(LumaSenseTechnologiesA/S,Ballerup,Denmark)usingFreon®134aastracergas.Thechamber'sHVACsystemcontainedaparticlefilterandanactivated carbon filter. The background ozonemixing ratio in thechamberwas<2ppb.

Ozonewasgenerated in theHVACsystemdownstreamof theactivated carbon filter by delivering air through a Jelight 600UVozonegenerator (JelightCompany Inc).Theozonegenerationratewas controlled by changing the fraction of theUV lamp thatwasexposed. The amount of air drawn through the ozone generator(0.5Lmin−1)wascontrolledwithaflowmeter.Ozonewascontinu‐ouslymonitoredwithanEnvironicsSeries300UVabsorptionozone

monitorwithatimeresolutionof1minuteandaccuracyof±2ppb(Environics Inc). Temperature and relative humiditywere continu‐ouslymonitoredwith1‐minute time resolutionusingWöhlerCDL210sensors(WöhlerTechnikGmbH).

2.2 | Experimental conditions

Humanoccupancywassimulatedusingrecentlywornt‐shirts.Plain,white, cotton t‐shirts (exposed surface area of one t‐shirt is ap‐proximately0.85m2)werewashedat60°C,tumbledried,andstoredin zip‐lockedplastic bagsbefore theexperiments.Before ameas‐urement, thesamefourpersonswereaskedtowear fresh t‐shirtsovernightandcometothelaboratoryinthemwithoutshowering.11 Thefourt‐shirtswerethenturnedinsideoutandstretchedoverthebacksofwiremeshchairsinthechamber.Twomixingfanswereop‐eratinginthechamber.

Twosetsofexperimentswereperformed (Table1).The firstset(Conditions1‐7)examinedtheeffectoftwoairexchangerates(1and3h−1)andthreelevelsofozone(<2,30,and60ppb)inthechamber on occupant‐related indoor air chemistry. The experi‐mental conditionswereestablished theeveningbefore, insuringsufficient time to reach steady‐state ozone mixing ratio in thechamber before the commencement of the experiment in themorning (“continuous ventilation”).At the endof eachmeasure‐mentday,thechamberwasflushedwithfreshair(AER=20h−1for20minute)beforesettingtheconditionsforthenextexperimentalday (except afterCondition4; seebelow). Theovernightozona‐tionoftheemptychamberalsoservedascleaning/passivationofthechamberwallsforthenextexperiment.Thesecondsetofex‐periments (Conditions8‐9)examinedthe impactof “intermittentventilation.”Theseexperimentsincludedtwoconsecutivedaysofmeasurements.AnintermittentventilationschemewassimulatedbystoppingtheozonegenerationandloweringtheAERtoapprox‐imately0.15h−1(simulatinginfiltration)attheendofthefirstdayofmeasurements. Conditions 4 and 7 (identical) served as basedaysinordertohaveidenticalstartingpointsforConditions8and

Condition*  AER (h−1) ozone (ppb) Note

1 1 0 Referencecondition—witht‐shirts

2 1 60 Referencecondition—withoutt‐shirts

3 1 30

4 1 60 BasedayforCondition8** 

5 3 30

6 3 60

7 1 60 BasedayforCondition9** 

8 1 60 VentilationandO3generationstartedatt=0h

9 1 60 VentilationandO3generationstartedatt=−2h

*Condition8followedCondition4,Condition9followedCondition7.Conditions4and7wereidentical.Duetoanalyticalissues,datafromCondition4wereunavailable.DatafromCondition7werenotconsistentwithdataobtainedintheotherexperiments—see‘LimitationsoftheStudy’.**O3generationstopped,t‐shirtsremovedandAERloweredto0.15h

−1attheendofexperiment.

TA B L E 1  Experimentalmatrix,forallConditionsexcept2,t‐shirtsplacedinchamberatt=0h

916  |     SALVADOR et AL.

9.The followingmorning,ozonegenerationwasstartedand theAERwasincreasedto1h−1eitheratthetimewhentheshirtswereplacedinthechamber(t=0hour,Condition8)or2hoursbefore(t=−2hour,Condition9).

2.3 | ToF‐CIMS measurement

TheAerodyneToF‐CIMSwasutilizedforthemeasurementofgas‐phasespeciesproducedinthechamber.TheworkdescribedinthispaperusedtheToF‐CIMS inpositivemodewithprotonatedwaterclustersasreagentions.Adetaileddescriptionoftheinstrumentcanbefoundelsewhere.25

Thesampleairwasbroughtintotheionmoleculeregion(IMR)attwostandardlitersperminute(SLM)througha10mmouterdiame‐terand2‐meter‐longinlet.Nitrogen(99.9%purity)flowedat2.2LPMthroughafrittedglassbubblerfilledwithdeionizedwatertocreateahighconcentrationofwatervapor.Itwassubsequentlyionizedbyapoloniumsourcetoproducethe(H2O)nH

+ ionsviatheadditionofahydrogenatomwhichthenenteredtheIMRorthogonaltothesampleflow.Theionizationthenproceededviathefollowingreaction:

TheIMRpressurewasmaintainedat150mbarandatempera‐tureof50°C.Thepressureof the collisionaldissociationchamber(CDC),thechamberaftertheIMRandcontainingthefirstquadru‐pole (small segmented quadrupole),was held at 2mbar. The fieldstrengthswithin theToF‐CIMSwere tuned tobeasweakaspos‐sible to reduce fragmentation of product ions.Within this setup,weobservedthe(H2O)H

+,(H2O)2H+andthe(H2O)3H

+clusterswiththedimerclustercontributingto65%ofthetotalionizationclustercounts.ThisdiffersfromthestandardoperationofaPTR‐MSwherethe(H2O)H

+ ionisthedominant ionizationsignal, likelyduetothenascenceofadrifttubeandthereforelowercontrolof ionenergywithin thesystem.Wethereforeoptimizedthe (H2O)H

+ counts toincreasesensitivityandnormalizethespectratothesumofallthreeclusterionizationions.ThesensitivityoftheCIMSalsodependsonthedistributionof reagent ions,which is indicatedby the ratioofthedimer((H2O)2H

+)toitsmonomer((H2O)H+).Thisratiowasmain‐

tainedthroughoutthecampaignwithlessthan10%variation.Datawereacquiredat1stimeresolutionandaveragedto1min‐

utefordataanalysisusingthesoftwareTofware3.1.1(AerodyneInc).Post‐campaignmasscalibrationofthespectrawasconductedusingfour constant peaks on the spectra; (H2O)H

+, (H2O)2H+, (H2O)3H

+,andC5H9O2

+,enablinganaccuracyof5ppmforpeakidentification.Theaveragespectralmassresolution(m/Δm)was3500.

Permeationsourcesheldatconstanttemperaturewereutilizedforcalibration.Compoundstandardsforacetone,6‐MHO,andgeranylac‐etone(GA)werepurchasedfromSigma‐Aldrichand5mLwasplacedinasmallvialwithanorifice in the lid.AnN2 flowof2.5SLMwaspassedthrougha largeglassvialholdingthepermeationsourceandthenpartialflowwassampledbytheTOF‐CIMSwiththeexcessbeingevacuatedtoanexhaust.Themixingratiointheflowwascalculated

using amass loss approach (see Supporting Information for amoredetaileddescription).Thesensitivitiesforacetone,6‐MHO,andgera‐nylacetonewerecalculatedtobe12.77,4069,and2785ioncountsperppb,respectively.Wequantified4‐OPAusingthesensitivityfactorof6‐MHO,ensuringthathumiditynormalizationandtotal ioncountsignalwereconstantforeachexperiment.

2.4 | Experimental procedure

Eachexperimentwasinitiatedbyplacingfourt‐shirtsinthecham‐ber.Thepressurizedchamberdoorwasopenforlessthan3sec‐onds,apersonenteredandthedoorwasclosed.Afterplacingthet‐shirts on the chairs, the chamber doorwas open again as thepersonwasleaving.Openingthechamberdoormayhaveaffectedtheconditioninthechamber.However,nosubstantialimpactontheozonemixing ratiowasobserved;ozonemixing ratiodecaysfollowedfromthebeginningthepatternexpectedafterintroduc‐ing the soiled t‐shirts (see Figure 2). The experiment continuedfor 3 hours withmeasurements of ozone and the organic com‐poundswith theToF‐CIMS. This time intervalwas judged to besufficientforthesystemtoreachat least95%steady‐statemix‐ingratiosofozoneandprimaryreactionproductsatbothtargetairexchangerates.Duringthisphase,concentration‐timeprofilesofozoneandtheprimaryandsecondarysqualene/ozonereactionproducts were monitored with 1‐minute time resolution. After90minutes,desorptionofparticlescollectedduring thepreced‐ing30minutesof gas‐phasemeasurementwasperformedusingtheFilterInletforGasesandAEROsols(FIGAERO).26Weplantopresent results from theparticlemeasurements in a future arti‐cle.AftertheFIGAEROdesorptionprocess,ittookapproximately10minutesforthemeasuredmixingratiosofgas‐phasespeciestoreachthosevaluesimmediatelypriortoparticledesorption.Aftertheexperiment,thet‐shirtswereremovedfromthechamberandthechamberwasflushedwithoutdoorair(20h−1for20minutes).

2.5 | Mass balance model

Aseriesofmassbalancemodelswereusedtodescribethesteady‐state conditionsofozoneand theoxygenatedproductsofozone‐squalene reaction inside the chamber during the continuousmeasurements(Conditions3,5‐7).ThemodelsaredescribedbelowandvisualizedinFigure1.

2.5.1 | Ozone

Equation(1)canberewrittenforsteady‐statecondition(d[O3]/dt=0)as:

(H2O)nH+

+M→n(H2O)+MH+

(1)

d[O3]

dt=Emission rate−�[O3]−ksur[O3]−k6MHO[O3][6MHO]−kGA[O3][GA]

(2)Emission Rate=�[O3]ss+ksur[O3]ss+k6MHO[O3]ss[6MHO]ss+kGA[O3]ss[GA]ss

     |  917SALVADOR et AL.

whereemissionrate(ppbh−1)istheozonegenerationrateinthecham‐ber,λ (h−1)istheairexchangerate,ksur(h

−1)istheoverall,first‐orderozoneremovalrateduetoreactionswiththestainlesssteelchambersurfacesandthesoiledt‐shirts,k6MHO and kGA(ppb

−1 h−1)arethegas‐phase, second‐order reaction rate coefficients for 6‐MHO and GAwithozone,respectively,[O3]ss,[6MHO]ss,and[GA]ssarethesteady‐statemixing ratios of ozone, 6‐MHO, and geranyl acetone, respec‐tively (ppb).ksurwasbasedonsteady‐stateozoneconditions,where[O3]ss=O3emissionrate/sumofsinks~O3emissionrate/(λ+ksur). The averagevalueofksurwasestimatedtobe(1.37±0.11)h

−1. The calcula‐tionisshowninSupportingInformation.

Theozoneremovalrateduetoitslossonthechamberwallswascalculatedfromthedecayofozoneintheemptychamberattheendof theexperiments forConditions4 and7,whenozonegenerationwasstoppedandtheAER loweredto0.15h−1.Themeasuredvalueofkchamberwas0.16h

−1 and 0.15 h−1 in these twoexperimentsandisconsistentwithozonelossratespreviouslyreported27‐30 inemptystainlesssteelchambers(between0.06and0.35h−1).InthefollowingEquations (3‐8), kshirt, the first‐order ozone removal rate coefficientbythet‐shirts,wascalculatedaskshirt=ksur−kchamber=(1.2±0.1)h

−1. Theerrorforksurwasestimatedtobe8%,andweusedthisvaluetocalculatetheuncertaintyofkshirt.

Allparametersintheequationsareknown,measuredintheseexper‐imentsorestimatedbasedonthemeasurements,withtheexceptionof

kGA.SimilartoLakeyetal,18wehavechosentouse0.068ppb−1 h−1for

kGA;thisistwicethevalueofk6MHO=0.034ppb−1 h−1fromFruekildeet

al,10consistentwiththetwodoublebondsinthegeranylacetonemol‐eculeandgiventhemolecularsimilarityofGAto6‐MHO.

2.5.2 | Geranyl acetone (GA)

Geranyl acetone is primarily produced through surface reactionofozonewithsqualeneonthet‐shirtswhenozoneaddstodoublebonds3or3′inthesqualenemolecule(Figure1).Additionalgeranylacetoneisproducedthroughsecondarysurfacereactionsbetweenozoneandthelong‐chainedpolyunsaturatedaldehydesandcarbox‐ylic acids (C27‐pentaenal, C22‐tetraenal, C17‐trienal, C27‐pentae‐noicacid,C22‐tetraenoicacid,andC17‐trienoicacid) remainingontheshirts.

Equation(3)canberewrittenforsteady‐statecondition(d[GA]/dt=0)as:

where YieldGA (unitless) istheyieldofgeranylacetonefromthepri‐maryandthesecondaryreactionsofozoneontheshirts.

(3)d[GA]

dt=kshirtYieldGA[O3]−�[GA]−kGA[O3][GA]

(4)kshirtYieldGA[O3]ss=�[GA]ss+kGA[O3]ss[GA]ss

F I G U R E 1  Visualizationoftheformationandlossofgeranylacetone,6‐MHO,and4‐OPA(f–branchingratio;1°,2°,3°–primary,secondary,tertiaryreaction;λ–airexchangerate)

918  |     SALVADOR et AL.

2.5.3 | 6‐methyl‐5‐hepten‐2‐one (6‐MHO)

6‐MHOisformedasaprimaryproductofozone/squalenereactionsonthesurfaceofthet‐shirtswhenozoneaddstothedoublebonds2or2′ in the squalenemolecule (Figure1). It canalsobe formedthrough secondary reactions both on the t‐shirts and in the gas‐phaseasozonereactswithgeranylacetone.

Equation (5) can be rewritten for steady‐state condition(d[6MHO]/dt=0)as:

where Yield6MHO (unitless) is the yield of 6‐MHO from primary(squalene) and secondary (C27‐pentaenal, C22‐tetraenal, C17‐trienal,C27‐pentaenoicacid,C22‐tetraenoicacid,C17‐trienoicacid,andgeranylacetone)reactionsofozoneonthet‐shirts,andf6MHO isthefractionofthegas‐phasereactionofozonewithgeranylac‐etonethatproduces6‐MHO(branchingratio).Thebranchingratiosfor6‐MHOand4‐OPAfromthegas‐phasereactionofozoneandGAareunknown;wehavesetthevalueoff6MHOto0.3inaccordancewiththefindingsdescribedbyGrosjeanandGrosjean(1997).31 The sumofformationyieldsofprimarycarbonylsisclosetoone,andthebranching ratio favors theprimarycarbonylwith themoresubsti‐tutedassociatedbiradical.

2.5.4 | 4‐oxopentanal (4‐OPA)

4‐OPAisformedasasecondaryproductfromozonereactingwithC27‐pentaenalandC22‐tetraenalonshirts,asasecondaryproductofozonereactingwithgeranylacetoneinthegas‐phase,andassec‐ondaryortertiaryproductfromozonereactingwith6‐MHOinthegas‐phase(Figure1).

Equation (7) can be rewritten for steady‐state condition(d[4OPA]/dt=0)as:

where Yield4OPA(unitless)istheyieldof4‐OPAfromozonereactionswithproductsof squaleneozonolysison the t‐shirts (ie,C27‐pen‐taenal,C22‐tetraenal, andgeranyl acetone). fOPA is the fractionofgas‐phasereactionofozonewithgeranylacetonethatproduces4‐OPA.Giventhat0.3wasusedforf6MHO,wesetf4OPA=0.7(assumingthatthesumofthebranchingratiosisunity).Thetertiaryformationof4‐OPAfrom6‐MHOisnottreatedinthemodel.

Yield6MHO, YieldGA, andYield4OPAwere adjusted in themodelcalculations forConditions3,5,6,and7toobtainsource‐to‐sinkratios near unity for 6‐MHO, geranyl acetone, and 4‐OPA (seeSection3.2).

3  | RESULTS AND DISCUSSION

3.1 | Ozone mixing ratio

Therateatwhichozonewasremovedbythestainlesssteelchambersurfaces(kchamber,0.15h

−1)wasaround10%oftheremovalbyallsur‐faceswhenfoursoiledt‐shirtswerepresent(ksur,1.37h

−1). Table 2 liststhemeasuredinitialandsteady‐statemixingratiosofozoneforallConditions.Nosignificantchangeinozonemixingratiowasob‐servedduringthemeasurementperiodforCondition2,inwhichnot‐shirtswerepresentinthechamber.Thesteady‐statemixingratiosofozonewerebetween12and37ppbwhent‐shirtswerepresent.ThehighestvaluewasobservedforCondition6withaninitialozonemixing ratioof 54ppb and an elevated ventilation rate (3h−1).Atidentical initial ozone mixing ratios, the ozone generation rate isthree timeshigheratAER3h−1compared toAER1h−1.Althoughlow, the measured mixing ratios of the ozone‐squalene reactionproductsforCondition1(soiledt‐shirts,noozone)indicatethepres‐enceofozone‐initiatedreactionsonthesurfacesofthet‐shirtspriortotheexperiments.

Figure 2 shows the time series of ozone mixing ratios duringcontinuousventilationexperiments(Conditions1,2,3,5,6and7).Steady‐statemixingratiosforozonewereachieved30‐80minutesaftertheplacementoft‐shirtsinthechamber.Itisapparentinthefigure that it took substantially longer to reach steady‐state at anAERof1h−1comparedtoanAERof3h−1.

Table 3 presents the parameters in the model that addressessteady‐stateozonemixingratios.Themeasuredsteady‐stateozonemixingratiosfitwellwiththeknownozonegenerationratesandtheidentifiedsinks(source‐to‐sinkratiosclosetounity;Table3).Ozoneremoval rates at steady‐stateweredominatedbyAER (λ) and theremovalbyreactionsonthesurfacesofthechamberwallsandthet‐shirts (ksur) (Equation (2); Table 3).Air exchangewas responsiblefor~40%oftheozoneremovalrateat1h−1andfor~70%at3h−1. Thesumofairexchangerateandozoneremovalratebythesurfacesaccountedformorethan96%ofthe lossofozone inthechamberunderallconditions.Gas‐phasereactionswithGAand6‐MHOhadanegligibleimpactonozonelosscomparedtoventilationandsurfaceremoval.Thetimetosteady‐stateforeachCondition isconsistentwiththeozonelossratebeingdominatedbyventilationandsurfaceremoval(λ+ksur).

Thewornt‐shirtsaccountedforapproximately90%oftheremovalofozonebyallsurfaces(kshirt/ksur=1.22h

−1/1.37 h−1=0.89).Therateofozoneremovaltoahighlyreactivesurfaceislimitedprimarilybytheresistancetomasstransportacrosstheboundarylayerofairadjacenttothesurface.32Thismaybethecaseforthet‐shirts(kshirt),wherethetransport‐limitedozone flux controls themaximum rateof reactiveuptakeofozonewithskinlipids.Thisissupportedbytheobservationthat, fromthe introductionof theshirts into thechamberuntil theendof themeasurements,ozoneconsumptionoccurredatanearlyconstantrate.Wealsoobservedthatkshirtvariedverylittlefromcon‐ditiontocondition.Giventhesurfaceareaofthefourt‐shirts(3.4m2),

(5)

d[6MHO]

dt=kshirtYield6MHO[O3]+kGAf6MHO[O3][GA]−�[6MHO]−k6MHO[O3][6MHO]

(6)kshirtYield6MHO[O3]ss+kGAf6MHO[O3]ss[GA]ss=�[6MHO]ss+k6MHO[O3]ss[6MHO]ss

(7)

d[4OPA]

dt=kshirtYield4OPA[O3]+k6MHO[O3][6MHO]+kGAf4OPA[O3][GA]−�[4OPA]

(8)kshirtYield4OPA[O3]ss+kMHO[O3]ss[6MHO]ss+kGAfOPA[O3]ss[GA]ss=�[4OPA]ss

     |  919SALVADOR et AL.

thevolumeofthechamber(30m3),andthevalueofkshirt(1.22h−1),the

calculateddepositionvelocityforozonetothet‐shirtsis10.8mh−1. Thisisslightlyhigherthanpreviouslyreportedozone/t‐shirtdeposi‐tionvelocities,perhapsreflectingthepresenceoftwomixingfansinthechamber.Tamasetal(2006)33measureddepositionvelocitiesof

6.8 and 9.7 m h−1undertwosetsofexperimentalconditions,whileRaietal(2014)34measureddepositionvelocitiesrangingfrom5.4to10.4 m h−1inaseriesofchamberexperiments.Arecentcomputationalfluiddynamicsmodelingstudy35calculatesozonedepositionveloci‐tiestohumansurfacesof8‐9mh−1atanAERof3h−1.

Theassumptionthatkshirtisconstantduringthecourseofanex‐periment(approx.3hoursintotal)issupportedbytheobservationthattheozonemixingratiodoesnotvary,withinthe±2ppbaccuracyofthemeasurement,duringthefinal100minutesofanexperiment(seeFigure2).However,itshouldbenotedthattheassumptionthattheozonemixingratioisatsteady‐state(ss)isnotnecessarilyvalidforprolongedperiodsoftime.Afterreachingitslowestmixingratiofollowingtheplacementofthet‐shirtsinthechamber,atsomepointtheozonemixingratiobeginstoslowlyincreaseassqualeneandtheunsaturatedreactionproductsonthesurfaceofthet‐shirtsbecomedepleted.Thus,kshirtisnotconstantoveralong‐term(tensofhours)measurement.Forexample,inacompanionmeasurementofozonemixing ratiooveraperiodof45hours theozone level rapidlyde‐creasedfrom63to26ppb(41%)afterplacing4soiledt‐shirtsinthechamberandthenslowlyincreasedto41ppb(64%)after45hours(seeSupportingInformation).Withasinglesoiledt‐shirtinthecham‐ber,ittook~90hourstoreachtheinitialozonelevelestablishedinthechamberbeforeintroducingthet‐shirt(datanotshown).

F I G U R E 2  Timeprofilesofozoneconcentrationduringcontinuousventilationexperiments(Conditions1,2,3,5,6,and7).T‐shirtswereplacedinthechamberatt=0

Time of experiment (minutes)

0 50 100 150

Ozo

ne

(pp

b)

0

20

40

60

Cond 3: AER 1 ozone 30Cond 7: AER 1 ozone 60Cond 5: AER 3 ozone 30Cond 6: AER 3 ozone 60Cond 1: t-shirts/no ozone Cond 2: ozone/no shirts

TA B L E 2  Mixingratiosofozoneandmajorreactionproducts

Condition AER (h−1)Target Ozone (ppb)

Target Emission Rate (ppb h−1)

Ozone concentration (ppb) VOC concentration (ppb)* 

Initial Steady‐state GA 6‐MHO 4‐OPA

1 1 0 0 <2.0 <2.0 0.16 0.04 0.06

2**  1 60 60 58 58 ND ND ND

3 1 30 30 31 12 1.1 0.65 1.9

4 1 60 60 62 28 NA NA NA

5 3 30 90 28 19 0.77 0.27 0.69

6 3 60 180 54 37 0.93 0.58 2.2

7***  1 60 60 63 28 0.29 0.32 2.0

8 1 60 60 0 28 0.083 0.097 0.37

9 1 60 60 53 26 0.027 0.055 0.32

Note: NA,Notavailable;ND,Notdetected.*Theaverageofthelast40minforeachexperimentwereusedasthesteady‐statemixingratiosofVOCs.Themixingratioswerecorrectedforchamberbackground.**Nosoiledt‐shirtswereinthechamberduringthisexperiment.***SeediscussionofCondition7in‘LimitationsoftheStudy’.

TA B L E 3  Modelparametersforozonemassbalance(Equation(2))

ConditionAER, ozone (h−1, ppb)

Source term Sink terms

Source/sink ratioTrue emission rate (ppb h−1)

λ[O3]ss (ppb h−1)

ksur[O3]ss (ppb h−1)

k6MHO[O3]ss[MHO]ss (ppb h−1)

kGA[O3]ss[GA]ss (ppb h−1)

1,30 31 12 17 0.27 0.90 1.0

1,60 62 28 38 0.30 0.55 0.93

3,30 83 57 26 0.17 0.99 0.99

3,60 161 110 50 0.73 2.33 0.98

Note: ThenumbersinthecolumnlabeledConditionrefertotheAER(h−1)andtheozonemixingratio(ppb).

920  |     SALVADOR et AL.

Figure3comparesthetimeseriesofozonemixingratiosfor theintermittentventilationconditionswiththecorrespondingcontinuousventilationcondition(Conditions4and7(identical),AER1h−1,60ppbozone). InConditions4and7 (continuousventilation, t=−16hours)and in Condition 9 (intermittent ventilation, t = −2 hours), steady‐stateozonemixingratiowasapproachedfroman“ozonerich”state.InCondition8(intermittentventilation,t=0hour),ozoneapproachedsteady‐statefroman“ozonepoor”state.Allthreeconditionsreachedsimilarsteady‐statemixingratiosabout60minutesafterplacing thet‐shirtsinthechamberattimet=0hour.Thisisconsistentwiththeozoneremovalratecoefficientforthesoiledt‐shirtsbeingdeterminedprimarilybymasstransportacrosstheboundarylayerofairadjacenttothesurfaceoftheshirtasopposedtoreactionsonthesurfacesoftheshirts.However,bythetimesteady‐statehadbeenachieved,sub‐stantiallylesssurfacechemistryhadoccurredduringCondition8thanduringConditions4,7,and9.Between0and75minutes,theozonefluxtotheshirtsforCondition8was46%ofthefluxforConditions4and7,whiletheozonefluxforCondition9was89%ofthatforConditions4 and 7.

3.2 | Oxygenated VOCs

Figure4showsthetimeseriesforgeranylacetone,6‐methyl‐5‐hepten‐2‐one, and 4‐oxopentanal, three major volatile oxygen‐atedproductsofsqualeneozonolysis,fromtheexperimentswithcontinuous ventilation. 1,4‐butanedial, 4‐MON, and 4‐MODwere also observed in these experiments. 1,4‐butanedial wasnotfurtheranalyzedduetoits lowandunreliablesignalsduringthemeasurements.Timeseriesplotsfor4‐MONand4‐MODarepresented in theSupporting Information (FigureS3).These twocompoundshavenotbeen included in themassbalancemodels

sincedoing sowould requireextensiveassumptions.Secondaryproductssuchas4‐OPA,4‐MON,and4‐MODshowcasethedif‐ferent influenceofairexchange ratesongas‐phaseandsurface

F I G U R E 3  Timeprofilesofozoneconcentrationunderconditionswithintermittentventilation(Conditions8(t=0h)and9(t=−2h)).Datafromthecorrespondingcontinuousventilationexperiment(t=−16,Conditions4and7,60ppbozone,1h−1 AER) areincludedforreference.T‐shirtswereplacedinthechamberatt=0

Time of experiment (minutes)

–100 –50 0 50 100 150

Ozo

ne

(pp

b)

0

20

40

60

Cond 8: t = 0 hCond 9: t = –2 hConds. 4 and 7: t = –16 h

F I G U R E 4  Timeseriesofgeranylacetone,6‐MHO,and4‐OPAfordifferenttargetozoneconcentrationsandAERsduringexperimentswithcontinuousventilation.Thegapinthetimeseriesisduetothermaldesorptionofaerosolsandthesubsequentmemoryeffect

     |  921SALVADOR et AL.

reactions for secondaryproducts.The importanceofgas‐phasechemistryfollowstheorder4‐OPA>4‐MON>>4‐MOD.There‐sultsshowninFigureS3areconsistentwiththisorder,which isexplainedbytheprecursorsforthesesecondaryproducts.4‐OPAisformedfrom6‐MHOandgeranylacetone(primarilyinthegas‐phase) and fromC27‐pentaenal andC22‐tetraenal (primarily onsurfaces); 4‐MON is formed from geranyl acetone (primarily inthe gas‐phase) and from C27‐pentaenal (primarily on surfaces);4‐MODisformedfromC22‐tetraenalandC‐17‐trienal(almostex‐clusivelyonsurfaces).

Table2 lists the average steady‐statemixing ratios for gera‐nylacetone,6‐MHO,and4‐OPA(seeSupportingInformationfortherationalebehindthesteady‐stateassumptions).Thevaluesareapproximatelysteady‐stateexceptforCondition3,where4‐OPAdid not reach steady‐state within the three‐hour measurement.However,itsrateofchangeinmixingratioduringthelast40min‐uteswassmallcomparedtothefirst75minutesoftheexperiment.DependingontheCondition,theaveragemixingratioforthelast40minuteswas between0.29 and1.1 ppb for geranyl acetone,between 0.27 and 0.65 ppb for 6‐MHO, and between 0.69 and2.2ppbfor4‐OPA.

Table4showssteady‐statemodelparameters forgeranylace‐tone.Theyield (YieldGA),which is that fractionofozone removedby theshirts thatproducesgeranylacetone,hasbeenadjusted toproduce source‐to‐sink ratios near unity. The estimated effectiveyieldforgeranylacetonewascloseto12%forConditions3,5,and6,whileitwasonly2.5%forCondition7.

The highest steady‐state mixing ratio of geranyl acetone(Table2)occurredunderCondition3 (lowAER, lowO3),whilethelowestoccurredunderCondition7(lowAER,highO3). The primary andsecondaryreactionsthatgenerategeranylacetonearetogethercapturedintheterm“kshirtyieldGA[O3]ss”(seeEquation(4)).

Air exchange rate and gas‐phase reactionswith ozone are thesinks for geranyl acetone in the chamber. For Condition 3, thesesinks are of similarmagnitude; this is also the case for Condition6.ForCondition7,removalviagas‐phasereactionoccursatabouttwicetherateofremovalviaventilation;forCondition5theoppo‐siteistrue.Thesereactionsformfurtherproducts,suchasacetone,6‐MHO,4‐OPA,4‐methyl‐8‐oxo‐4‐nonenal(4MON),and4‐oxopen‐tanoic(levulinic)acid.Inourmodeling,wefocusedontheozonoly‐sisofgeranylacetoneinthegas‐phasetoform6‐MHOand4‐OPA(Figure1).

Table5showssteady‐statemodelparametersfor6‐MHO.Theyield (Yield6MHO),which is that fraction of ozone removed by theshirtthatproduces6‐MHO,hasbeenadjustedtoproducesource‐to‐sinkratiosnearunity.Theestimatedyieldfor6‐MHOfromsur‐facereactionswasbetween3.0%and4.4%forConditions3,5,and6,whileitwasonly1.3%forCondition7.

The highest steady‐statemixing ratio of 6‐MHO (Table 2) oc‐curredunderCondition3 (lowAER, lowO3),while the lowestoc‐curred under Condition 4 (high AER, low O3). The production of6‐MHOontheshirts(thetermkshirtyield6MHO[O3]ssinEquation(6))constitutedabout~70%percentofthetotalproductionof6‐MHOinsidethechamber.Thegas‐phasereactionofgeranylacetonewith

TA B L E 4  Steady‐statemodelparametersforgeranylacetone

ConditionAER, ozone (h−1, ppb) YieldGA (‐)

Source term Sink terms

Source/sink ratiokshirtyieldGA[O3]ss (ppb h−1) λ[GA]ss (ppb h−1)

kGA[O3]ss[GA]ss (ppb h−1)

1,30 0.13 1.9 1.1 0.90 0.97

1,60 0.025*  0.84 0.29 0.55 0.99

3,30 0.14 3.2 2.3 0.99 0.97

3,60 0.11 4.9 2.8 2.3 0.96

Note: ThenumbersinthecolumnlabeledConditionrefertotheAER(h−1)andtheozonemixingratio(ppb).*SeediscussionofCondition7in‘LimitationsoftheStudy’.

TA B L E 5  Steady‐statemodelparametersfor6‐MHO

ConditionAER, ozone (h−1, ppb) Yield6MHO (−) f6MHO (−)

Source terms Sink terms

Source/sink ratio

kshirtyield6MHO[O3]ss (ppb h−1)

kGAf6MHO[O3]ss [GA]ss (ppb h−1)

λ[6MHO]ss (ppb h−1)

k6MHO[O3]ss[6MHO]ss (ppb h−1)

1,30 0.044 0.3 0.65 0.27 0.65 0.27 1.00

1,60 0.013*  0.3 0.44 0.17 0.32 0.30 0.99

3,30 0.030 0.3 0.69 0.30 0.81 0.17 1.00

3,60 0.040 0.3 1.8 0.70 1.8 0.73 1.01

Note: ThenumbersinthecolumnlabeledConditionrefertotheAER(h−1)andtheozonemixingratio(ppb).*SeediscussionofCondition7in‘LimitationsoftheStudy’

922  |     SALVADOR et AL.

ozone (the term kGAf6MHO[O3]ss[GA]ss) constituted the remaining~30%ofitsproduction.

Thesinksof6‐MHOareairexchangeandinteractionwithozonein thegas‐phase.Airexchangeaccounted for70%‐80%of the re‐moval and dominated the approach of 6‐MHO mixing ratios tosteady‐state for Conditions 3, 5, and 6. Reactionwith ozonewasresponsiblefortheremaining20%‐30%ofthesinks.Thetwosinkterms were of similar magnitude for Condition 7 (low AER, highO3). The reaction additionally forms secondary and tertiary prod‐ucts such as 4‐oxopentanal, 4‐oxopentanoic acids, acetone, andhydroxyacetone.

Table6 shows steady‐statemodel parameters for4‐OPA.Theyield (Yield4OPA), which is that fraction of ozone removed by theshirt thatproduces4‐OPA,hasbeenadjusted toproduce source‐to‐sink ratiosnearunity.The fittedyield for4‐OPAvariedamongtheconditionsandwashighestforCondition6(9.2%)andlowestforCondition7(3.8%).

The highest steady‐statemixing ratios of 4‐OPA (Table 2) oc‐curredunderConditions3(lowAER,lowO3),7(lowAER,highO3),and6(highAER,highO3),whilethelowestoccurredforCondition5(highAER,lowO3).Theproductiononthesurfacesofthet‐shirts(thetermkshirtyield4OPA[O3]ssinEquation(8))accountedfor50%‐65%ofthe total production rate. Thegas‐phase reactionof6‐MHOwithozoneaccountedfor10%‐15%(termk6MHO[O3]ss[6MHO]ss)andthegas‐phase reaction of geranyl acetone with ozone for 20%‐30%(termkGAf4OPA[O3]ss[GA]ss).ThereportedyieldforCondition3maybeslightlylowbecause4‐OPAmixingratiohasnotreachedsteady‐state.Unlike6‐MHOandgeranylacetone,4‐OPAisremovedonlythroughventilation; itdoesnothaveC‐Cdoublebonds to furtherreactwithozone.Thus,therateofremovalof4‐OPAisdeterminedbyairexchange.

Summing up, geranyl acetone was formed entirely throughozone‐initiatedreactionsonthet‐shirts (withsqualeneand itsun‐saturatedreactionproducts),while~70%of6‐MHOand50%‐65%of4‐OPAwereformedthroughsuchsurfacereactions.Theremovalof4‐OPAwasentirelythroughventilation,whiletheremovalof6‐MHOwas70%‐80%throughventilation.Therelativecontributionofventilationandgas‐phasereactionstotheremovalofgeranylac‐etonevariedwiththeCondition.

Figure 5 shows the mixing ratios of geranyl acetone, 6‐MHO,and 4‐OPA over time for Conditions 7, 8, and 9. The steady‐statemixingratiosarepresented inTable2.Byt=180minutes,geranylacetone appears to have reached a semi‐steady‐statemixing ratioin Conditions 7 (ozone generation began at t = −16 hours) and 9(t=−2hours),butnotinCondition8(t=0hours).Thesameappearstobetruefor6‐MHOand4‐OPA.However,theacceleratedmixingratiogrowthobservedbetween150and180minutesfor4‐OPAandtoasmallerextentforgeranylacetonewasunexpectedandmayre‐flectchemistrytakingplacethatisnotaccountedforinthecurrentmechanisticmodel.

InCondition8,theincreasein6‐MHO’smixingratiowasnotap‐parentuntilabout10minutesafterozoneandthet‐shirtsweresi‐multaneouslyintroducedtothechamber.ThisisconsistentwiththefindingsofWisthalerandWeschler,13whoobserveda30‐40min‐utes delay in the increase of the oxygenatedVOC after the startofozonegeneration inanoccupiedsimulatedoffice. It ishoweverunclearwhyasimilardelaywasnotobservedforgeranylacetone.InCondition8,anelevatedbackgroundmixingratiowasobservedfor4‐OPAbeforetheexperimentcommenced.Thismayreflectoff‐gassingof4‐OPAfromthechamberwallsduringthenightwhentheventilationratewasreduced.Whentheventilationwasincreasedto1 h−1atthestartoftheexperiment,the4‐OPAmixingratioinitiallydecreasedbeforeanincreasewasobserved.

Intermittentventilation(reducedAER,noozonegenerationthenight before experiment) resulted in substantially lower steady‐statemixingratiosofgeranylacetone,6‐MHO,and4‐OPA.WhileinCondition7(t=−16hours)thesoiledt‐shirtswereplaced inan“ozonerich”andchamber,inCondition8(t=0hour)theywereplacedinan“ozonepoor”chamber.Condition9(t=−2hours)was,however,relatively “ozone rich”,with [O3] =53ppb at the time the t‐shirtswereintroduced,andwewouldanticipatesteady‐statemixingratiosclosertothoseofCondition7thanactuallyobserved.Additionally,afterasufficientlylongperiod,allthreeexperimentswouldbeantic‐ipatedtoapproachasimilarsteady‐statemixingratiosunderother‐wiseidenticalfinalconditions(AER=1h−1,60ppbofozone).

Thesteady‐statemixingratiosunderintermittentconditionsaresomewhatsensitivetothemixofunsaturatedproductspresentonthesurfacesofthet‐shirtsandthechamberwalls.Thedifferences

TA B L E 6  Steady‐statemodelparametersfor4‐OPA

ConditionAER, ozone (h−1, ppb) Yield4OPA (−) f4OPA (−)

Source terms Sink term

Source/sink ratio

kshrtyield4OPA[O3]ss (ppb h−1)

k6MHO[O3]ss[6MHO]ss (ppb h−1)

kGAf4OPA[O3]ss[GA]ss (ppb h−1)

λ[4OPA]ss (ppb h−1)

1,30 0.066 0.7 0.98 0.27 0.63 1.9 1.00

1,60 0.038*  0.7 1.3 0.30 0.39 2.0 1.00

3,30 0.053 0.7 1.2 0.17 0.69 2.1 1.00

3,60 0.092 0.7 4.1 0.73 1.6 6.5 1.00

Note: ThenumbersinthecolumnlabeledConditionrefertotheAER(h−1)andtheozonemixingratio(ppb).*SeediscussionofCondition7in‘LimitationsoftheStudy’

     |  923SALVADOR et AL.

maybecaused, inadditiontothedifferentstartingozonemixingratiosatthetimethet‐shirtswereplacedinthechamber,bydif‐ferencesinsoilinganddifferencesintheextenttowhichskinlipidshavebeenoxidizedbeforethet‐shirtswereplacedinthechamber.

4  | LIMITATIONS OF THE STUDY

Acetone is known to be a major product of squalene ozonolysis.However,wewereunable toquantify it, since the instrumenthadpoorresponsetothismolecule.

Theyieldsofformationforgeranylacetone,6‐MHO,and4‐OPAinthemodelswereadjustedtoachievesource‐to‐sinkratiosclosetounity.TheseyieldsweresimilarforConditions3,5,and6,whiletheyweresubstantiallylowerforCondition7.Thislatterconditionoccurredearlyinthecampaignafterapoweroutage.However,thehumidity and temperaturewithin the chamberwas unaffected bytheoutage,andthemasscalibrationoftheToF‐CIMSdidnotchangesubstantially. The total ion count was approximately 10% lowerduringCondition7,and the ratioofwaterdimer tomonomerwas20%lower.Thesignalnormalizedbythemixingratioofthereagentionpartially takes this into account.Nonetheless, the accuracyofthemeasurementsdependsonmanyparameters.WedecidedtostillreportthedatafromCondition7sinceitprovidedvaluableinforma‐tiononthedynamicsoftheproductdistributionwithinthechamber.

Althoughthemeasuredvaluesofkchamber(0.16h−1 and 0.15 h−1)

were consistent with ozone loss rates (0.06‐0.35 h−1) previouslyreported27‐30 inempty stainless steel chambers, thesevaluesmayrepresentreactivityofthenormalchambersurfacesplusnewlyde‐positedreactants.Theyweredeterminedunderconditionsinwhichtherewere likely freshlygeneratedSVOCproductssorbedon thechambersurfaces.The impactofsuchfreshlygeneratedSVOCs isanticipatedtobesmall,sincemostoftheproductsthatpartitionedtothechambersurfacesdidnotcontainunsaturatedcarbonbondsandwouldnotreactwithozone.

Whiletheuseofasteady‐statemodelisjustifiedfor6‐MHOandgeranylacetone,thisisnotalwaysthecasefor4‐OPA.ForCondition3, it is apparent that themixing ratio of 4‐OPA is still slightly in‐creasingduringthefinal40minutesofmeasurement.Inhindsight,itwouldhavebeenvaluabletohavemademeasurementsoveralongertimeperiodforthiscondition.Giventhat4‐OPAwasnotatsteady‐state,theresultsfor4‐OPA,Condition3,shouldbeviewedasroughapproximations.

Weacknowledgethatthelackofreplicationsisamajorlimitationofourstudy.WescheduledCondition4asareplicateforCondition7(1AER/60ppbO3).Unfortunately,duetoanalyticalissues,nore‐liablemeasurements of oxygenated productswere obtained fromthisexperiment.

Thesuddenincreaseinthemixingratiosofgeranylacetoneand4‐OPAobservedtowardtheendofthemeasurementsinCondition8(Figure5)remainsunexplained.

F I G U R E 5  Timeseriesofgeranylacetone,6‐MHO,and4‐OPAduringexperimentswithintermittentventilation(Conditions8and9).Condition7(continuousventilation)isshownforcomparison.Time(t)isthetimewhenozonegenerationbeganrelativetowhenthet‐shirtswereplacedinthechamber.Notethattheconcentrationsarenotcorrectedforbackgroundmixingratiosinthechamber.BackgroundmeasurementswereunavailablepriortoConditions8and9duetothenatureoftheexperiments(intermittentventilation:reducedAERandnoozonegenerationovernight)

924  |     SALVADOR et AL.

5  | CONCLUSIONS

Occupant‐related indoor chemistry is complex. In the case of un‐saturated species such as geranyl acetone and 6‐MHO, ozone canbothproduceandconsumeamolecule.Steady‐statemixingratiosofsqualene‐ozonereactionproductsarearesultofaninterplaybetweenprimary, secondary,and tertiarygas‐phaseandsurface reactions,aswell as ventilation. In order to quantify the formation and removalprocesses,aseriesofmassbalanceequationswasapplied tomeas‐urementdataobtainedforanumberofexperimentalconditionsinaclimatechamber.

Whenpreviouslywornt‐shirtswereinthechamber,morethan96%of theozone removalwasattributable to thecombinationofsurfacereactionsandventilation(about40%atAER=1h−1and70%atAER=3h−1),whilegas‐phasereactionsaccountedfor lessthan4%oftheozoneremoval.Theventilationstrategieshaveonlyamar‐ginaleffectontheozoneremovalratebythet‐shirts,whichismasstransportlimited.

Geranylacetoneisproducedonlybyozonereactionsoccurringonthesurfaceofthet‐shirts.Ventilationwasresponsibleonaver‐ageforabout50%ofitsremovalfromthegas‐phase,whiletherestwas attributable to further reactions with ozone. Approximately70%of6‐MHOwasproducedonthesurfacesofthet‐shirts,whilesecondary gas‐phase reaction of ozonewith geranyl acetonewasresponsiblefortheremainingformation.Airexchangewasthedom‐inantremovalprocessfor6‐MHO(upto80%),asopposedtofurtherreactionswith ozone in the gas‐phase. 4‐OPA is a secondary andtertiaryproductofsqualeneozonolysisoccurringonthesurfacesofthet‐shirts.About60%of4‐OPAwasproducedthroughthispath‐way,20%‐30%throughgas‐phasereactionsofgeranylacetoneandtherestfromgas‐phasereactionof6‐MHO.4‐OPAisonlyremovedbyventilation.

ExcludingCondition7,thefittedformationyields(thefractionofozoneremovedbytheshirtsthatultimatelyproducedgeranylacetone, 6‐MHO, and4‐OPA)were approximately 12%, 3%‐4%,and 5%‐10% for geranyl acetone, 6‐MHO, and 4‐OPA, respec‐tively.TheformationyieldforGAisamixofyieldsfromsurfacechemistry and is relatively insensitive to ventilation rates. Theoverallyieldsof6‐MHOand4‐OPAareamixofyieldsfromsur‐facechemistryandyieldsfromgas‐phasechemistryandaremod‐eratelysensitivetoventilationrates.Theyieldsarevaluableinputstopredictivemodels.

Turningofftheventilationovernightoronweekendsmayleadtotheaccumulationofcertainpollutantswithindoorsources,butcouldalso limit the extent to which ozone‐derived products are formedandwill delay their generationwhen the ventilation is turned backon again. In applying such a practice, it should be recognized thatozoneconcentrationsareoftenattheirlowestdailyvaluesovernight.Preventingozone‐initiatedchemistryinindoorenvironmentsisbetterachievedby limiting indoorozoneconcentrationvia filtration ratherthanadjustingventilationrates.

ACKNOWLEDG EMENTS

ThisstudywasfinancedbytheSwedishResearchCouncilFORMAS(Dnr942‐2015‐1509).Wewould like to thankElliotGall (PortlandStateUniversity,USA)forsharingdataonozoneremovalrates.

ORCID

Gabriel Bekö https://orcid.org/0000‐0001‐6107‐8336

Charles J. Weschler https://orcid.org/0000‐0002‐9097‐5850

Sarka Langer https://orcid.org/0000‐0002‐6580‐8911

R E FE R E N C E S

1. WeschlerCJ,ShieldsHC.Potentialreactionsamongindoorpollut‐ants.Atmos Environ.1997;31:3487‐3495.

2. WeschlerCJ.Reactionsamongindoorpollutants:What’snew?Sci World J.2001;1:443‐457.

3. Weschler CJ. Chemistry in indoor environments: 20 years of re‐search.Indoor Air.2011;21:205‐218.

4. Weschler CJ, Carslaw N. Indoor chemistry. Environ Sci Technol. 2018;52:2419‐2428.

5. Tang X,Misztal P, NazaroffWW, Goldstein AH. Volatile organiccompound emissions from humans indoors. Environ Sci Technol. 2016;50:12686‐12694.

6. Weschler CJ. Roles of the human occupant in indoor chemistry.Indoor Air.2016;26:6‐24.

7. Nicolaides N. Skin lipids: their biochemical uniqueness. Science. 1974;186:19‐26.

8. Pandrangi LS, Morrison GC. Ozone interactions with humanhair: Ozone uptake rates and product formation. Atmos Environ. 2008;42(20):5079‐5089.

9. Weschler CJ. Ozone in indoor environments: concentration andchemistry.Indoor Air.2000;10:269‐288.

10. Fruekilde P, Hjort J, Jensen NR, et al. Ozonolysis at vegetationsurfaces: A source of acetone, 4‐oxopentanal, 6‐methyl‐5‐hep‐ten‐2‐oneandgeranylacetoneinthetroposphere.Atmos Environ. 1998;32:1893‐1902.

11. WisthalerA,TamasG,WyonDP,etal.Productsofozone‐initiatedchemistryinasimulatedaircraftenvironment.Environ Sci Technol. 2005;39(13):4823‐4832.

12. WeschlerCJ,WisthalerA,CowlinS,etal.Ozone‐initiatedchem‐istry in an occupied simulated aircraft cabin. Environ Sci Technol. 2007;41(17):6177‐6184.

13. Wisthaler A, Weschler CJ. Reactions of ozone with human skinlipids: Sourcesof carbonyls, dicarbonyls andhydroxycarbonyls inindoor air. PNAS.2010;107:6568‐6575.

14. AndersonSE,WellsJR,FedorowiczA, et al. Evaluationof the con‐tact and respiratory sensitization potential ofvolatile organic com‐pounds generated by simulated indoor air chemistry. Toxicol Sci. 2007;97:355‐363.

15. AndersonSE,JacksonLG,FrankoJ,WellsJR.Evaluationofdicar‐bonylsgeneratedinasimulatedindoorairenvironmentusinganinvitroexposuresystem.Toxicol Sci.2010;115(2):453‐461.

16. AndersonSE,FrankoJ,JacksonLG,etal.Irritancyandallergicre‐sponsesinducedbyexposuretotheindoorairchemical4‐oxopen‐tanal.Toxicol Sci.2012;127:371‐381.

17. WolkoffP,LarsenST,HammerM,etal.Humanreferencevaluesforacuteairwayeffectsoffivecommonozone‐initiatedterpenereac‐tionproductsinindoorair.Toxicol Lett.2013;216(1):54‐64.

     |  925SALVADOR et AL.

18. Lakey P, Wisthaler A, Berkemeier T, et al. Chemical kinetics ofmultiphase reactions between ozone and human skin lipids:Implications for indoor air quality and health effects. Indoor Air. 2017;27(4):816‐828.

19. KruzaM, Lewis AC,Morrison GC, CarslawN. Impact of surfaceozoneinteractionsonindoorairchemistry:amodelingstudy.Indoor Air.2017;27(5):1001‐1011.

20. ColemanBK,DestaillatsH,HodgsonAT,NazaroffWW.Ozonecon‐sumptionandvolatilebyproductformationfromsurfacereactionswith aircraft cabin materials and clothing fabrics.Atmos Environ. 2008;42(4):642‐654.

21. YangS,GaoK,YangX.Volatileorganiccompounds(VOC)formationdue to interactions betweenozone and skin‐oiled clothing:mea‐surements by extraction‐analysis‐reaction method. Build Environ. 2016;103:146‐154.

22. ZhouSM,ForbesMW,AbbattJ.Kineticsandproducts fromhet‐erogeneousoxidationofsqualenewithozone.Environ Sci Technol. 2016;50(21):11688‐11697.

23. WeschlerCJ,ShieldsHC.Theinfluenceofventilationonreactionsamongindoorpollutants:Modelingandexperimentalobservations.Indoor Air.2000;10(2):92‐100.

24. FischerA,LangerS,LjungströmE.Chemistryandindoorairqualityinamulti‐storeywoodenpassive(lowenergy)building:Formationofperoxyacetylnitrate.Indoor Build Environ.2014;23(3):485‐496.

25. LeeBH,Lopez‐HilfikerFD,MohrC,etal.An iodide‐adducthigh‐resolution time‐of‐flight chemical‐ionization mass spectrometer:application to atmospheric inorganic and organic compounds.Environ Sci Technol.2014;48(11):6309‐6317.

26. Lopez‐HilfikerFD,MohrC,EhnM,etal.Anovelmethodforonlineanalysisofgasandparticlecomposition:descriptionandevaluationofafilterinletforgasesandAEROsols(FIGAERO).Atmos Meas Tech. 2014;7:983‐1001.

27. KunkelDA,Gall ET, Siegel JA, et al. Passive reduction of humanexposuretoindoorozone.Build Environ.2010;45:445‐452.

28. WangJ,DoussinJF,PerrierS,etal.Designofanewmulti‐phaseex‐perimentalsimulationchamberforatmosphericphotosmog,aerosolandcloudchemistryresearch.Atmos Meas Tech.2011;4:2465‐2494.

29. Gall ET, Darling E, Siegel JA, et al. Evaluation of three commongreenbuildingmaterials forozoneremoval,andprimaryandsec‐ondaryemissionsofaldehydes.Atmos Environ.2013;77:910‐918.

30. MassabòD, Danelli SG, Brotto P, et al. ChAMBRe: a new atmo‐sphericsimulationchamberforaerosolmodellingandbio‐aerosolresearch.Atmos Meas Tech.2018;11:5885‐5900.

31. GrosjeanE,GrosjeanD.Gasphasereactionofalkeneswithozone:Formation yields of primary carbonyls and biradicals. Environ Sci Technol.1997;31(8):2421‐2427.

32. Cano‐RuizJA,KongD,BalasRB,NazaroffWW.Removalofreac‐tivegasesat indoorsurfaces:Combiningmass transportandsur‐facekinetics.Atmos Environ.1993;27(13):2039‐2050.

33. TamasG,WeschlerCJ,Bakó‐BiróZ,etal.Factorsaffectingozoneremoval rates in a simulated aircraft cabin environment. Atmos Environ.2006;40:6122‐6133.

34. RaiAC,GuoB,LinC‐H,etal.OzonereactionwithclothinganditsinitiatedVOCemissions in an environmental chamber. Indoor Air. 2014;24:49‐58.

35. RimD,GallET,AnanthS,WonY.Ozonereactionwithhumansur‐faces:Influencesofsurfacereactionprobabilityandindoorairflowcondition.Build Environ.2018;130:40‐48.

SUPPORTING INFORMATION

Additional supporting information may be found online in theSupportingInformationsectionattheendofthearticle.

How to cite this article:SalvadorCM,BeköG,WeschlerCJ,etal.Indoorozone/humanchemistryandventilationstrategies.Indoor Air. 2019;29:913–925. https://doi.org/10.1111/ina.12594