Ionic-strengthandpHdependentreactivitiesofascorbicacidtowardozonein

aqueousmicro-dropletsstudiedbyaerosolopticaltweezers

Yuan-PinChang,1,2,3*Shan-JungWu,1Min-SianLin,1Che-YuChiang,1GeninGary

Huang3

1.DepartmentofChemistry,NationalSunYat-senUniversity,Kaohsiung80424,

Taiwan

2.AerosolScienceResearchCenter,NationalSunYat-senUniversity,Sizihwan,

Kaohsiung80424,Taiwan

3.DepartmentofMedicinalandAppliedChemistry,KaohsiungMedical

University,Kaohsiung80708,Taiwan

*E-mail:ypchang@mail.nsysu.edu.tw

AbstractThe heterogeneous oxidation reaction of single aqueous ascorbic acid (AH2)

aerosol particleswith gas-phase ozonewas investigated in this study utilizing

aerosoloptical tweezerswithRamanspectroscopy.Themeasured liquid-phase

bimolecular rate coefficients of the AH2 + O3 reaction exhibit a significant pH

dependence,andthecorrespondingvaluesationicstrength0.2Mare(3.1±2.0)

×105M−1s−1and(1.2±0.6)×107M−1s−1 forpH≈2and6, respectively.These

resultsmeasuredinmicron-sizeddropletsagreewiththosefrompreviousbulk

measurements, indicating that the observed aerosol reaction kinetics can be

solelyexplainedby liquidphasediffusionandAH2+O3 reaction.Furthermore,

theresultsindicatethathighionicstrengthscouldenhancetheliquid-phaserate

coefficients of theAH2+O3 reaction.The results also exhibit a negative ozone

pressure dependence that can be rationalized in terms of a Langmuir–

Hinshelwood typemechanism for the heterogeneous oxidation of AH2 aerosol

particles by gas-phase ozone. The results of the present work imply that in

acidifiedairway-lining fluids theantioxidantabilityofAH2againstatmospheric

ozonewillbesignificantlysuppressed.

IntroductionAscorbicacid (AH2) isoneofmainantioxidants inepithelial lining fluids (ELF)

which work as thin fluid layer (typical ~0.1-0.2 μm thick) on the surfaces of

airwaysandalveolitoprotectagainstatmosphericoxidants,suchasozone.The

reaction pathways of the ozonolysis of aqueous AH2 (pKa = 4.1) significantly

dependonpH.ForthecaseofpH>pKa,suchaspH≈7, thedominant formof

AH2 is monoanion, AH−, and the ozonolysis of AH− produces dehydroascorbic

acid(DHA)andsingletO2inhighyields(>90%):1,2

AH! + O! → DHA+ O!(!Δ!)+ OH! (1)

ThesingletO2productassecondaryoxidantmaytransduceoxidationdamages

throughELF,3asitiswellknowntoreactactivelywithbiomoleculesinproteins

andDNA.4,5Ontheotherhand,theozonolysisofAH2inthefreeacidformfirstly

produces an unstable primary 1,2,3-trioxolane ozonide (POZ), which could

subsequently subject to unimolecular decomposition and form a secondary

ascorbateozonide(AOZ)orthreonicacid(THR):1

AH2+O3→POZ→AOZorTHR (2)

AOZmay also be qualified as harmful secondary oxidant, as the finding of its

shortlifetimeinbulkwaterimpliesitspotentialhighreactivity.1

Besides the reaction mechanisms, pH can also affect the reactivities of AH2

toward ozone. The reaction rate coefficients of the AH2 + ozone reaction at

different pHhave beendeterminedby various experimental studies,while the

measurement results seem to highly depend on the experimental designs or

conditions.ThemeasurementsinbulkwatercarriedoutbyGiamalvaetal.6show

thattheratecoefficientsatpH=2.0and4.8are6.9×105and5.6×107M−1s−1,

respectively.ThemeasurementsofKanofskyandSima7utilizingatypeofreactor

where the liquid contactedwith gaseousozoneyielded similar values: atpH=

2.0and7.0themeasuredratecoefficientswere5.6×105and4.8×107M−1s−1,

respectively.Twolaterstudies8,9utilizedthesimilartypeofreactorsbutcarried

out the reactions in ELF model solutions (pH = 7.4) instead. However, the

measuredratecoefficientsweresignificantlysmallerthanabovereportedvalues

ofpH>pKa:5.5×104M−1s−1fromKermanietal.9Recently,Enamietal.1studied

this reaction inmicro-droplets as amodel system of air-water interfaces. The

aqueousAH2dropletsinsizeoffewμmwereexposedtogaseousozoneforvery

shorttime,andthenthecompositionoftheinterfaciallayersofreactingdroplets

wasanalyzedbyonlineelectrospraymassspectrometry.Theyobservedthatthe

reaction rates at the interfacial layer (a few nanometers) were at least two

ordersofmagnitudelargerthanthosefrombulkmeasurementsofGiamalvaetal.

Ontheotherhand,theirobservedratioofreactionratecoefficientsatpH>pKa

and pH < pKa was about one order of magnitude smaller than that in bulk

water.1,6

Inthisstudy,weutilizedanaerosolopticaltweezers(AOT)apparatus,whichcan

trapsingleliquiddropletsviaopticalgradientforce,tostudythereactiveuptake

processes of single micro-droplets of aqueous AH2 to gaseous ozone. The

interactionofmicro-dropletswithgaseousenvironmentscanbeamodelsystem

fortheinteractionbetweengasandfluidfilms1andexhaledbioaerosols,asthe

reactiveuptakebehaviorofmicro-droplets involves thedynamicsprocessesof

both thebulkphaseand thegas-liquid interface.10Also the rate equationsand

kineticsmodels for interpreting the reactive uptakes of aerosol droplets have

beenwellestablished,providingastraightforwardstrategytoretrievingtherate

coefficientsof interest fromexperimentaldataofAOTdirectly.10,11Particularly,

oneofthemainadvantagesofAOTisthatthemicrophysicalpropertiesofeach

singletrappedmicro-droplet,suchasradius,refractiveindex,viscosity,diffusion

coefficient,surface tension, temperature,pH,vaporpressureandcompositions,

canbedeterminedviaspectroscopicmeansinhighaccuracyandinrealtime.12–25

Thus,AOTcanbeutilizedasawall-lessreactor to investigatingheterogeneous

reactionsindetails,asthereactionkineticsandmicrophysicalpropertiesofeach

particlecanbemonitoredsimultaneously.26

Therehavebeenseveralexperimental investigationsof thereactionkineticsof

heterogeneous oxidations in single droplets via AOT combined with Raman

spectroscopy.26–31 King et al.27 studied the ozonolysis of aqueous and organic

dropletscontainingunsaturatedorganiccompoundsandobtainedtheirreactive

uptakecoefficients,andthe liquid-phasebimolecularratecoefficientsretrieved

from the reactive uptake coefficients agreed with literature values.

Dennis-Smither et al.28 studied theheterogeneousoxidationof aqueousmaleic

aciddropletsbyozone,and theirmeasuredreactiveuptakecoefficientsagreed

with literature values. Recently, Hunt et al.26 investigated the heterogeneous

oxidation of nitrite anion in aqueous droplets by ozone, attempting to clarify

whetherthesurfaceexcessordepletionofnitriteanionscouldaffecttheaerosol

reactionkinetics.Theirmeasuredratecoefficientagreedwithliteraturevaluesof

bulk measurements, implying that it is not necessary to include the surface

effects in the reaction kinetics of aqueous nitrite droplets under atmospheric

conditions.

In thisworkweutilizedAOTcombinedwithRamanspectroscopy to study the

kineticsof thereactionofaqueousAH2micro-dropletswithgaseousozone.We

measured the bimolecular reaction rate coefficients at different values of pH,

ozonepressuresandionicstrengths,andtheireffectstotheratecoefficientsof

theAH2+ozonereactionwillbeaddressedinthisreport.Finally,wediscussthe

implications of these results for antioxidant kinetics in ELF and exhaled

bioaerosols.

ExperimentalTheaerosoloptical tweezersapparatusused in thiswork issimilar to thoseof

previous studies of AOT,32 consisting a trapping laser, an invertedmicroscope

andaRamanspectrometer.WeutilizedaCWNd:YVO4laseroperatingat532nm

(Coherent, Verdi V2) as the trapping laser. The typical laser power before

entering the microscope was about 30 to 130 mW. For achieving an optimal

trapping force, the laser beamwas expandedby twoplano-convex lenses. The

expanded laserbeamwasguided into theepi-illuminationportof the inverted

microscope (Nikon,EclipseTi2), and then itwas reflected to anoil-immersion

objective(Nikon,CFIPlanApochromatlambda,60X,NAof1.4,WDof0.13mm)

by a dichroicmirror (Chroma, ZT532dcrb). The focused light passed from the

objective through an index matching fluid and through a coverslip into an

aerosol trapping chamber that was mounted on the sample stage of the

microscope. TheRaman scattering light emitted from the trapped dropletwas

collectedbytheobjective,passingthroughaholographicnotchfilter(Semrock,

NF03-532E-25), and imaged onto the entrance slit of a 0.5 m spectrograph

(Andor, SR-500i, using 1200 l/mm grating in this work) coupled with a

TE-cooled spectroscopic CCD (Andor, DU401-BVF). The integrated time for

acquiring Raman spectra was 1 second. The brightfield image of the trapped

dropletwasacquiredbyutilizinga455nmLED(Thorlabs,M455L3)asthelight

source, a CCD camera (Basler, ICDA-ACA640-90UM), and a 451 nm bandpass

filter(Semrock,FF01-451/106-25).

AdenseflowofaqueousL-ascorbicacid(Sigma-Aldrich)aerosolwasgenerated

by a medical nebulizer (Sumo, V-15), and the sizes of aerosol droplets were

about3to8μmindiameter.Theaerosolflowwasintroducedintothetrapping

chamber,andafterwaitingaboutafewminutes,asingledropletwillbecaught

by the focused laser. The relative humidity (RH) in the trapping chamberwas

maintained around85%via injecting awet nitrogen gas flowwith a flow rate

about300sccm,whichwascontrolledbyamassflowcontroller(MKS,1179A),

and itwasmonitoredbyahumiditysensor (Honeywell,HIH-4602-C,accuracy:

3.5%)whichwascalibratedbyamoreaccurateone(Rotronic,HC2A-S,accuracy:

0.5%). To maintain this high RH inside the trapping chamber, all of its

componentswereimmersedinwaterpriorexperiments.Afterthechamberwas

assembled, it was further flashedwith the wet nitrogen flow at least over 30

minutes,beforestartingthetrappingexperiments.Duringatypicalexperimental

time, such as a couple of hours, ~85% RH in the chamber may slowly drift

around 5%, while supplying the constant wet nitrogen flow.We chose not to

change the flow rate to regulate RH during trapping experiments, as a stable

trapping is also affected by the gas flow. The temperature inside the chamber

wasmaintainedat297K.Anozonegenerator(Airphysics,C-L010-DSI)wasused

to produce gaseous O3 from a pure O2 gas flow, and the generated O2/O3 gas

mixturewasfurtherdilutedwithdryN2inastainlesssteelchamber.Duringthe

reactionkineticsmeasurements, thepremixedO2/O3/N2gas flowof fewsccms

wasmixedwiththewetN2gasflowpriorflowingintothetrappingchamber.The

concentrationofozonewason-linemonitoredviaitsUVabsorptionpeakat250

nmbyusinganabsorptioncell(lengthof75cm),adeuteriumlamp(Hamamatsu,

S2D2module),andaCCDspectrometer(OceanOptics,Flame-S-UV-VIS-ES).For

themeasurementsatpH≈2,thesolutionforgeneratingaerosolcontained0.3M

AH2and0.1MNaCl.ForthecaseofpH≈6,the0.3MAH2solutioncontained0.5

M sodium phosphate buffer and 0.1M NaCl, and pH of the final solutionwas

adjustedto6.0usingsolutionsofHClandNaOH.

Results Ramanspectraoftrappeddroplets

Figure 1 shows the representative Raman spectra of a trapped aqueous AH2

dropletinpH≈2,beforeandafterthereactionwithozone.ThemeasuredRaman

bandsofaqueousAH2,suchasthecoupledC=CandC=Ostretchingmodeat1690

cm−1, indicated that thedominate formofaqueousAH2 is freeacid.33After the

trapped AH2 droplets were exposed to ozone, the Raman peak at 1690 cm−1

gradually decreased, due to the cycloaddition of ozone toward theC=Cdouble

bondofAH2.1,34Therelativelybroadbandsofproductspeakedataround1790

cm-1 and 3000 cm−1 were also formed, as shown in Figure 1, and they were

tentatively assigned to C=O and C-H stretching modes of threonic acid,

respectively.Enamiatal.observedthattheotherreactionproductAOZdepicted

inreaction(2)onlysurvivedontheair-waterinterfaceduetoitsshortlifetimein

water.1TheseinterfacialAOZwereexpectedtohaverelativelylowconcentration

andthusnotobservedinthisstudybecauseofthelimitedsensitivityofRaman

spectroscopy. For the trappedAH2 dropletswith pH≈ 6, the coupledC=C and

C=OstretchingmodeofAH− iscenteredat1590cm−1,asshowninFigureS1.33

AftertheAH-dropletswereexposedtoozone,severalRamanbandsofproducts

were formed, such as those centered at 1400 cm−1, 1720 cm−1 and 3000 cm−1

(seeFigureS1andS2), and theywere tentativelyassigned todehydroascorbic

acid,whichisthemainproductofO3+AH−reactiondepictedinreaction(1).

Figure1.RepresentativeRamanspectraofanopticallytrappedaqueousAH2dropletinpH=1.8

before(topdata)andafter(bottomdata)thereactionwithozoneat85%RH.Twospectraare

offsetforclarity.NotethatrelativelysharppeaksinthespectraareCERSsignals(labelledwith

asterisk).Theinsetshowsthebrightfieldimageofthedropletbeforethereactionwithozone.See

TableS1(ESI)(expt.01)fordetailsoftheexperimentalconditions.

Determiningmicrophysicalpropertiesoftrappeddroplets

In theRamanspectraofaqueousAH2droplets therearealsoseveral relatively

sharp peaks superposed on the relatively broad molecular Raman band (see

Figures1andS1).Theyaresocalledcavity-enhancedRamanscattering(CERS),

which are whisper-gallery modes amplified by stimulated Raman scattering.

32,35,36 These CERS signals can be used to retrieve the values of radius and

refractiveindex(RI)ofthedropletwithahighaccuracy.WeutilizedaMietheory

simulation programmrfit developed by Preston et al.13 to analyse the spectral

positionsofCERSsignalsandobtainedfittedRIandradius.ThefittedRIvalues

were further used to retrieve the absolute concentration of AH2 before the

reaction,[AH2]0,andionicstrengthinsidethedroplet.Wealsoutilizedthetime

evolutionof fittedradius to trackthechangeofdroplet ionicstrengthduringa

reactionprogress.WhennoobviousCERSsignalsexisted,theradiusand[AH2]0

of the trapped droplet were determined by brightfield imaging and Raman

intensityratiosofAH2andwater,respectively.Asallsolutesindroplets,suchas

AH2,Na+,Cl−andphosphates,areinvolatile,thedropletionicstrengthcanalsobe

determinedviacomparingtheconcentrationsofAH2indropletsandinthebulk

solution. The values of [AH2]0 determined by CERS and spontaneous Raman

1600 2000 2400 2800 3200 3600

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

Ram

an In

tens

ity (a

rb. u

.)

Raman shift (cm-1)

Before reaction with ozone After reaction with ozone

*

*

*

*

*

*

CH stretch

OH stretch

C=C and C=O stretch

*

spectroscopyarealsoingoodagreement,justifyingbothmethodsutilizedinthis

study.

Tables S1 and S2 list the values of radius, [AH2]0 and droplet ionic strength

determined by above methods for each trapped droplet. The typical droplet

radius in the present work is about 3 μm. Both [AH2]0 and ionic strength of

trappeddropletswere found to be several times larger than those of the bulk

solutiononaverage,probablycausedbyavaporizingenrichmentbeforeflowing

intothemoisturizedtrappingchamber.

Figure2.Representativetimeprofilesofradius(crosses)andintegratedRamanintensitiesofa

trapped AH2(aq) droplet which was exposed to gaseous ozone after 0th second. The Raman

intensitiesofAH2(squares)andproducts(triangles)wereobtainedfromintegratingtheareasof

theRamanpeakscenteredat1690cm-1and1790cm-1,respectively.Mostintensityjumpsinthe

RamanintensitytimeprofilesareascribedtodriftingCERSpeaks.Hollowsymbolsrepresentthe

intensity interferences caused by frequency-drifting CERS peaks. For clarity, only the typical

errorbarsareprovidedat theendofdata.Theozonepressure is1.13ppm.SeeTableS1(ESI)

(expt.17)fordetailsoftheexperimentalconditions.

DeterminingpHoftrappeddroplets

ForthekineticsmeasurementsatpH≈6,weestimatedthevaluesofpHinside

thetrappeddropletsviacomparingmolecularRamansignalsof inorganicacids

andtheirconjugatebases, followingthemethodsdevelopedbypreviousworks

utilizing sodium bisulfate buffer. We utilized the Raman peak area ratios of

H!PO!! (874cm-1)and HPO!!! (987cm-1)(seeFigureS3)todeterminepHover

0 1000 2000 3000 4000 5000 6000 7000

0

0.2

0.4

0.6

0.8

1

1.2

Ram

an in

tens

ity /

arb.

u.

Time / s

2100

2200

2300

2400

2500

2600

2700

Rad

ius

/ nm

therangeof5.5to7.3,withthecalibrationcurveshowninFigureS4.Notethat

the Raman intensities of H!PO!! and HPO!!! at pH ≈ 6 are very similar,

facilitating the determination of pH around this range. The pH measurement

resultsshowthatthevaluesofpHinsidedropletsbeforeexposedtoozonewere

equal to about 6.2 on average. After the reactionswith ozone finished, the pH

values of droplets slightly increased to about 6.7 on average (see Table S2),

agreeingwith thepredictedmechanismofreaction(1).For thecaseofpH≈2,

the values of pH before reactionwere estimated from themeasured values of

[AH2]andknownpKa(seeTableS1).

It shouldbenoted that the above estimationsof pHperformedby thepresent

work did not consider the effect of high ionic strengths insidemicro-droplets.

Two recent studies utilizing AOT to determine aerosol pH have demonstrated

that thenon-ideal thermodynamicsdue tohigh ionicstrengthshas tobe taken

into account via thermodynamic model calculations, such as specific ion

interactiontheoryorE-AIMaerosolthermodynamicsmodel,inordertopredict

theproton activity correctly.23,25Otherwise, the estimated values of pH canbe

lessaccurate.For thecaseofaqueoussulfate/bisulfatesystems,suchdeviation

couldbeuptoabout2pHunits.23,25Weassumethatthismayberegardedasthe

maximum deviation of pH reported in this work. The main limitation of the

presentwork topredict theprotonactivitymoreappropriately isa lackof the

thermodynamic coefficients of AH2 required by the non-ideal thermodynamic

modelsdescribedabove,andlotsofexperimentalandtheoreticaleffortsarestill

neededforsuchpurpose.

KineticsoftheAH2+ozonereaction

Figures2andS5showthetimeevolutionsofradiusandRamanintensitiesofthe

aqueousAH2dropletsexposedtoozoneatpH≈2andpH≈6,respectively.After

exposed to ozone at 0 second, the Raman intensities of AH2 decreased

immediately, and the product Raman bands emerged at the same time. Their

Raman intensity timeprofiles furthershowthat the formationofproductswas

concomitantlyaccompaniedwiththedecayofAH2,indicatingthattheobserved

productwassolelycreatedfromtheAH2+ozonereaction.Thisobservationalso

indicatesthattheproductsareinerttowardozone,agreeingwiththefindingsof

Enamietal.1TheRaman intensity timeprofiles inFigure2alsoexhibitseveral

intensity fluctuations for short periods of time, and they are caused by the

progressofCERSpeaksthroughthecorrespondingmolecularRamanbands.28In

dataanalysis, suchCERS interferencewas identifiedby tracing theprogressof

CERSpeaks inRamanspectra timeseriesand labelling the timeswhenaCERS

peak superposeswith themolecular Raman band of interest, such as the data

pointslabelledwithhollowsymbolsinFigure2.Suchassignmentofprogressing

CERS peaks can be assisted and justified by Mie theory simulation, when

assuming continuous changes of droplet size and RI and nomode hopping of

CERSpeaks.Figure2alsoshowsthat theradiusonlydecreasedslightlyduring

the reaction progress, implying involatile reaction products and slow

evaporationofwater.Forsimplifyingthefollowingkineticsanalysis,theradiusis

assumed tobea fixedvalue in thedataanalysis,andweassume thechangeof

radiusduringthereactiontobetheerrorofradius.

Figure3.SquarerootofthenormalizedRamanintensitiesoftheAH2C=Cbandversusreaction

time(symbols)atpH=1.8andozonepressureP=1.4ppm(expt.19)andthefit(straightline)

basedonequation(3).Eachdatapointrepresentstheaverageover20consecutivespectra.Two

rugged lines represent an error band which corresponds to 1σ uncertainties of data. The

significantburstsofintensities(symbolsnotingreyareas)wereattributedtofrequency-drifting

CERSpeaks,andthustheywerenotincludedinthefit.

AccordingtoSmithetal.10andKingetal.,27thereactiveuptakeofgaseousozone

ontheaqueousAH2dropletstudiedinthepresentworkcanbeapproximatedas

thediffusion-limitedspecialcase,wherethedissolvedaqueousozonereactswith

the aqueous AH2 near the surface and thus the reaction rate depends on the

bimolecularratecoefficientoftheAH2+ozone reactioninaqueousphase(k2)

andthediffusioncoefficientofaqueousozone(DO).Theintegratedrateequation

canbeexpressedasfollowing:

[AH2](t)[AH2]0

=1− 3HP2r

DOk2[AH2]0

t (3)

where[AH2](t) is [AH2]as functionoft, [AH2]0represents[AH2]at0secondof

reactiontime,HistheHenry'sLawconstantofozoneintheaqueoussolution,P

is the pressure of ozone and r is the radius of the droplet. The value ofDO is

about1.8×10−9m2s−1.35ThevaluesofHatvarioussoluteconcentrationsused

here can be estimated from a Sechenov relation (see ESI).38 According to

equation(3),aplotof([AH2]/[AH2]0)1/2versust shouldyieldastraight line,as

shown in Figure 3, where [AH2] is represented by the C=C stretching mode

RamanintensityofAH2,andthefittedgradientcanbeusedtoderivethevalueof

k2. The values ofk2 determined from all data of this study are summarized in

TablesS1andS2,forpH≈2and≈6,respectively.Finally,notethatweonlyfit

the datawithin the reaction timewhere the values of ([AH2](t)/[AH2]0)1/2 are

between1and~0.3(seeFigure3),becauseofsignificantstandarddeviationsof

data at the longer reaction times. Note that the CERS interferences described

previously can cause significant deviations of theRaman intensity timeprofile

fromEquation(3),asshowninFigure3.Thus,thedatapointsusedtoderivek2

did not include those associatedwith CERS interferences, such as the symbols

notingreyareasinFigure3.Finally,forthecaseofpH≈2measurements,when

assuming that pH of the trapped droplet was solely regulated by [AH2], the

estimatedchangeofaerosolpH in this fittingrange isabout0.5,whichdefines

theuncertaintyofpHforthefittedvaluesofk2.

Figure 4. Bimolecular reaction rate coefficientk2 (symbols) of pH≈2plotted as a functionof

dropletionicstrengthandthefit(line)usingequation(4).Thepressureofozoneappliedtoeach

dataisbelow3.5ppm.Eachplottederrorbarincludesallexperimentalerrors(seeESI).

DiscussionEffectofionicstrengthonbimolecularratecoefficients

Toclarifytheeffectsofionicstrength,wemeasuredthereactionratecoefficient

k2atdifferentionicstrengths.Figure4plotsallthemeasuredratecoefficientsof

pH≈2asafunctionofionicstrength,indicatingapositivecorrelationbetween

ionicstrengthsandreactionratecoefficients.Whentheionicstrengthsincrease

from0.6Mto1.6M,whicharecontributedbyenrichedconcentrationsofNaClin

thetrappeddroplets,thereactionratecoefficientscanincreaseabouttentimes.

AccordingtoLaidler, theratecoefficientkofareactionbetweenan ionandan

neutralmolecule,orbetweentwoneutralmolecules,ationicstrengthImayhave

anapproximaterelationwiththeratecoefficientatzeroionicstrengthk0:39

log!"𝑘 = 𝑙𝑜𝑔!"𝑘! + 𝑏!𝐼 (4)

where b' is an empirical constant. Figure 4 also shows the fitting based on

equation(4)todataofpH≈2atdropletionicstrength0.5Mto1.7M,whichwas

dominatedbyenrichedNaCl,andthefittedvaluesofk0andb'are(1.7±1.1)×

105M−1s−1 and1.2±0.2M−1, respectively.Figure5plots all themeasured rate

coefficientsofpH≈6atdropletionicstrength1.7Mto5.3M,whichweremostly

contributed by enriched AH−, Na+ and phosphates in the trapped droplets.

Similar todataofpH≈2, thereactionratecoefficientsofpH≈6alsoexhibita

weakpositivedependencewithionicstrengths.Wealsoappliedequation(4)to

fitthesedataofpH≈6,asshowninFigure5.Thefittedvaluesofk0andb'are

(1.1±0.6)×107M−1s−1and0.23±0.04M−1,respectively.

Figure 5. Bimolecular reaction rate coefficientk2 (symbols) of pH≈6plotted as a functionof

dropletionicstrengthandthefit(line)usingequation(4).Thepressureofozoneappliedtoeach

data is below 1.5 ppm. Each plotted error bar includes all experimental errors (see the

SupportingInformation).

The activity coefficients of solutes in liquid phase can be affected by the ionic

strength,andsucheffectforreactionscontainingneutralmoleculesisconsidered

in Equation (4) by means of Debye-McAulay approach.39 Base on such

framework, the positive ionic-strength dependence could be ascribed to

increases in activity coefficients of reactants or/and the stabilization of the

activatedcomplexwhichareapproximatedby theb'I term inEquation (4).40,41

ThereactionofozoneandAH2involvesanattackofozonetothecarbon-carbon

doublebondofAH2,thecorrespondingpre-reactivecomplexcouldhavealarger

dipolemoment than reactants. Thus, under conditions of high ionic strengths,

theactivatedcomplexcouldbemorestabilized thanreactantsdue to its larger

ion-dipoleinteractions.41Sucheffectcanfurtherresultinalowerenergybarrier

andthusafasterreactionrate.ForbothcasesofpH≈2and6,weassumethat

thiseffectmayplayaroleintheirpositiveionic-strengthdependences.However,

forthecaseofpH≈2,theincreaseintheionicstrengthcouldalsoincreasethe

activitycoefficientofAH2inthefreeacidformandthusdecreaseitssolubility,so

called salting-out effect.42 When increasing the ionic strength, the water

0 1 2 3 4 5107

108

k 2 /

M−1

s-1

Ionic strength / M

moleculesinthesolvationshellofAH2willbedisplacedbysaltions,reducingthe

availablevolumeoftheaqueoussolutiontodissolveAH2.Inaerosolphase,such

effectcouldfurthercauseAH2moleculestoberepelledtothewater-airinterface,

and the concentration of AH2 near the surface would be higher than in bulk

phase,enhancingthereactionsofAH2withozonenearthesurface.Suchsurface

enrichmentduetoincreasingtheionicstrengthhasbeeninvestigatedbyseveral

previous studies.40,43–45 For the case of pH ≈ 6, this salting-out effect probably

doesnotplayarole,astheincreaseoftheionicstrengthcandecreasetheactivity

coefficient and thus increase the solubility of AH− as charged species instead.

Thismaypartiallyexplainwhythefittedvalueofb'forpH≈2islargerthanthat

forpH≈6.

Figure6.Bimolecularreactionratecoefficientsk2(symbols)ofpH≈2anddropletionicstrength

≈1Masafunctionofgas-phaseozoneconcentrationandthefit(line)utilizingEquation(5).The

plottederrorbarofeachratecoefficientcorrespondsto1σstatisticerrors.

Effectofgaseousozonepressureonbimolecularratecoefficients

Wealsomeasuredthereactionratecoefficientsk2atdifferentgas-phaseozone

pressures.Figure6plotstheratecoefficientsofthesameionicstrengthatpH≈2

andfromP≈2ppmto100ppm.Thehigherozonepressuresyieldthesmaller

valuesofk2.Suchozonepressuredependencemaybeattributedtothesurface

saturation effects for reactants, so called Langmuir-Hinshelwood mechanism,

which could cause the decrease of reactive uptake coefficients and the

underestimationofthemeasuredk2.46,47Therelationshipbetweenthemeasured

rate coefficient k2 and the gas-phase ozone pressure P for the

Langmuir-Hinshelwood-typemechanismcanbemodelledusing:39

WhereKozone is the ozone gas-to-surface equilibrium constant and k2,max is the

maximumbimolecularratecoefficientmeasuredatthelowozonepressurelimit.

Figure6showsthatthefitbasedonEquation(5)hasagoodagreementwiththe

data.Thefittedvaluesfork2,maxandKozoneare(3.1±0.3)×106M−1s−1and(3.2±

0.3) × 10−15 cm3molecule−1, respectively. The fitted value ofKozone for aqueous

AH2dropletsissimilartothosepreviouslyreportedonaqueousaerosolparticles

ofmaleicacid((9±4)×10−15cm3molecule−1)andfumaricacid((5±2)×10−15

cm3molecule−1).38 The results of fitting also verify that the ozone pressures

appliedforthedatainFigures4and5aresmallenough,sothatthissaturation

effectshouldnotappearinourkineticsmeasurements.

Experimentaluncertainty

The error reported for each bimolecular reaction rate coefficient k2 obtained

fromEquation (3) includes theerrorsofmeasuring theozonepressure, [AH2]0

andtheradiusandfittingthegradientfromtheRamanintensitytimeprofile.The

percentageerrorofmeasuringtheozonepressureisfixedto2.6%,whichisthe

assumedmaximumerrorofthemeasuredUVabsorptioncrosssectionofozone

at250nmestimatedbyliterature.48Thepercentageerrorofmeasuring[AH2]0is

about4%,whichisattributedtofittingerrorofcalibrationcurvesandstatistics

ofRamanintensities.Theerrorofthedropletradiusisassumedtobe0.15μm,

whichisanaveragedchangeofmeasuredradiiduringthereactionprogress,as

thefittingerrorfromthesimulation(fewnanometers)andstatisticsofdata(~

20nm,seeFigure2)aresignificantlysmaller.ForthecaseofpH≈2,thetypical

percentageerrorof fittedgradients is about1%.Suchsmall fittingerror could

probably be ascribed to the removal of CERS interferences and small Raman

intensity fluctuations gained from concentrated AH2 (~3.3 M on average) in

aerosols.ForthecaseofpH≈6,thetypicalpercentageerroroffittedgradientsis

about24%.Thisrelatively large fittingerrormaybeattributedto largeRaman

intensityfluctuationsduetotherelativelylowconcentrationsofAH2(~0.6Mon

k2 =k2,max

1+KozoneP (5)

average)inaerosolsandapotentiallyinsufficientremovalofCERinterferences,

whichwillbediscussedlater.

Overall, theseerrorswerepropagated intothedeterminationof theerrorofk2

byEquationsS4andS5showninESI.Asaresult,thetypicalpercentageerrorof

thederivedk2atpH≈2isabout12%,whichisdominatedbytheuncertaintiesof

reactantconcentrationsandradius.ThisvalueatpH≈6isabout36%,whichis

dominatedbythefittingerrorofgradient.Itshouldbenotedthattheerrorsofk2

reportedheredonot include the systematic errors potentially associatedwith

sizechangesofdroplets.Thetime-varyingdropletvolumeduetosizechangecan

alterthespeciesconcentrationsandalsotheionicstrengthduringtheprogress

of reaction, causing additional deviations to the value of k2 determined by

Equation(3),whichassumesafixeddropletvolume.Forthecaseof3μmdroplet

radius,thecorrespondingdecreaseofdropletvolumeduetoradiusdropof0.15

μmis14%.AccordingtoEquation(4),anincreaseof14%inionicstrengthcan

lead to an increase of about 32% or 7% in k2 for pH ≈ 2 or 6, respectively,

causinganoverestimationofk2.Ontheotherhand,theenrichmentofreactants,

suchasAH2,duringthereactioncouldprolongthereactiontimeandleadtoan

underestimation of k2. Based on Equation (3), we could roughly estimate that

suchunderestimationofk2issimilartothechangeofvolume,suchas14%.Asa

result,themaximumsystemicdeviationofk2duetoshrinkingdropletvolumeis

estimatedtobeabout18%overestimationor7%underestimationforpH≈2or

6,respectively.

Finally, we would like to discuss the potential contributions of CERS

interferencestothefittingerrorofgradientinEquation(3)andhowtodealwith

them.Asdescribedbefore,wedealtwiththeCERSinterferencesviatracingthe

CERSpeakssuperposedonthemolecularRamanbandofinterestandexcluding

thedatapointsassociatedwithsuchinterference.Withoutsuchtreatment,thefit

inFigure3cansignificantlydeviatefromthetime-dependenttrendinintensity

definedbyEquation(3).However,wewerenotabletoassignweakCERSsignals

which resemble the noise of spontaneous Raman spectra, and they could just

manifest themselves as an intensity fluctuation in the Raman intensity time

profile.Thisisprobablyoneofmainsourcesoftherelativelylargeuncertaintyof

k2 at pH ≈ 6. To further reduce such interference noise, we propose that it is

actuallydesiredtoenhanceCERSsignals,sothattheycanbeeasilyassignedand

deconvolutedindataanalysis.AstheintensityoftheCERSpeakisproportional

to the spontaneous Raman intensity at the same wavelength,35 one feasible

optionistoenhancethespontaneousRamansignalsviaincreasingthepoweror

lowering the wavelength of the Raman excitation laser. In future works, we

prefer to upgrade the single-beam AOT utilized in the present work to a

dual-beam AOT with a counter-propagating geometry which has been

demonstratedbyseveralresearchgroups,49–51assuchdual-beamgeometrycan

facilitateastabletrappingwitharelativelylargelaserpower.

Table 1. Reaction rate coefficient k2 determined from the present work and

previousmeasurements.

Study I/M freeacida/M−1s−1 monoanionb/M−1s−1

Thiswork 0.0c (1.7±1.1)×105 (1.1±0.6)×107

0.2d (3.1±2.0)×105 (1.2±0.6)×107

Giamalvaetal.6 (6.9±2.3)×105 (5.6±2.6)×107

KanofskyandSima7 ～0.2 (5.6±0.4)×105 (4.8±0.4)×107

Kermanietal.9 ～0.2 (5.5±0.4)×104

a.pH=1.9forthiswork,and2.0forGiamalvaetal.andKanofskyandSima.

b.pH=6forthiswork,4.8forGiamalvaetal.,7.0forKanofskyandSimaand7.4

forKermanietal.

c.k0fromthefittingresultsofFigures4and5.

d.estimatedviaequation(4)andthefittedk0andb'.

Comparingtheresultswithliterature

Table1summarizesthebimolecularratecoefficientsk2determinedbythisstudy

andpreviousmeasurements.Tocomparewiththeliteraturevaluesmeasuredat

ionicstrength0.2M,weestimatedthevaluesofk2atthisionicstrengthfromour

resultsfittedwithequation(4).Forthecaseoffreeacid(pH<pKa),ourresult,

(3.1±2.0)×105M−1s−1,hasthesimilarmagnitudesasthoseofpreviousworks,

such as Giamalva et al.6 and Kanofsky and Sima.7 Such agreement with the

literature valuesmeasured in bulk solution (Giamalva et al.6) implies that our

observedreactionkinetics isdominatedbydiffusion-limitedkinetics inside the

droplets.Forthecaseofmonoanion(pH>pKa),ourestimatedvalueofk2ationic

strength0.2M,(1.8±0.8)×107M−1s−1,hasthesameordersofmagnitudewith

thosemeasured by Kanofsky and Sima7 or Giamalva et al.,6while our value is

about three times less their values. Finally, our results exhibit the significant

dependenceofthereactivitywithpH,agreeingwiththeresultsoftheseprevious

works. Such pH dependence can be attributed to the different reaction

mechanismsoftheAH2freeacid(pH<pKa)andmonoanion(pH>pKa)forms.

Ontheotherhand,theagreementwithliteraturevaluesalsojustifiestheaerosol

pHmeasurementsofthisstudy.

Table2.Estimateddiffuso-reactivelengths,l,forozoneundervariousconditions

ofdifferentpHandconcentrationsofAH2.

Condition k2/M−1s−1 [AH2]/M l/nm

Thiswork(pH≈6) 1.2×107 0.4-1.1 12-19

Thiswork(pH≈2) 3.1×105 1.6-5.6 32-60

ELF(pH>pKa) 1.2×107 4×10−3 ~190

ELF(pH<pKa) 3.1×105 4×10−3 ~1200

Comparisonwithsurfacereactionkinetics

The kinetics of the AH2 + O3 reaction at the air-water interface has been

investigated, and the correspondingsurface reaction rateswere found tobeat

least twoordersofmagnitude larger thanthose fromthebulkmeasurements.1

To observe the surface kinetics in micro-droplets by means of AOT, this may

ideallyrequirethediffuso-reactivelengthstobewithinfewnanometers.26Table

2 lists thepredicteddiffuso-reactive lengthswith theexperimentalparameters

used in thisstudy.For theconditionof thisworkatpH≈2, thecorresponding

diffuso-reactive lengths are about few tens of nm, implying that any interface

effect should not be observable in this study. Thus, the observations of the

enhanced reactivities at pH ≈ 2 in this study could bemainly ascribed to the

effectofionicstrengthinstead.

Fortheconditionof thisworkatpH≈6, theestimateddiffuso-reactive lengths

are reduced to about 12-19 nm, because of the significantly large value of

bimolecular reaction rate coefficient k2. However, even such small lengthmay

still be too large to observe the desired interface enhancement. Indeed, the

maximumvalueofk2atpH≈6observedinthisstudyis(2.4±0.7)×108M−1s−1,

which is only about three times larger than the literature value, but not the

expectedtwoordersofmagnitudeenhancementduetointerfacereaction.Ifwe

wanttofurtherreducethediffuso-reactivelengthtofewnanometers,wehaveto

increasetheconcentrationofAH2tooverfewmolarities.Ontheotherhand,the

relativelyhighconcentrationsofsodiumphosphaterequiredformaintainingpH

≈6alsoresultinlargeionicstrengthsintheaqueousAH2droplets.Theeffectof

ionic strength to the reactive uptake of liquid phase aerosols have been

investigated by several aerosol studies, such as the reactions of SO2 with

peroxidesandthereactionsofmethoxyphenolswithozone.52,53Asthedataofk2

canbereasonablymodelledbyequation(4),asshowninFigure5,theobserved

enhancementofk2atpH≈6 in thisstudy isalsoascribedto theeffectof ionic

strength.

The potential role of interfacial chemistry in aerosol reaction kinetics

investigatedherecouldalsobeexaminedbyvaryingthedropletsizetochange

thesurfacearea-to-volumeratio.Thisapproachhasbeenoftenusedinprevious

studiesofaerosolkineticstounderstandwhethertheheterogeneousreactionis

dominated by volume- or surface-limited case, such as the studies on

heterogeneous reactions of N2O5 on organic or inorganic aerosol particles.54,55

These studies demonstrated that the volume-limited case can manifest the

dependence of reactive uptake coefficients on particle size. In contrast, the

surface-limitedcasedoesnotexhibitsuchdependence.54,55Forthepresentwork,

we plotted the bimolecular rate coefficients corrected to zero ionic strength

(ke−b'I,accordingtoEquation(4))atpH≈2and6asafunctionofdropletradius,

asshowninFiguresS6andS7,respectively.Theseresultsexhibitnoobservable

dependence of size, verifying that the observed kinetics of this work is

dominatedbythesurface-limitedcase(diffuso-reactivelength<<radius).54,55

ImplicationsforELFandbioaerosol

Several previous studies have suggested that the diffusions of potentially

unreacted ozone or secondary oxidants generated from ozonolysis of

antioxidants across the ELF could induce the oxidation damages and even the

cellinjuryofbiomembranes,whilethetypeofsecondaryspecies,suchassinglet

O2orascorbateozonide,coulddependontheacidityofELF.1,3Thefindingsofthe

present study indicate that the acidity can also affect the kinetics of the AH2

oxidation reaction and the diffusion lengths of ozone. Table 2 summarizes the

estimateddiffuso-reactivelengthsofozoneinELFattwodifferentpH,assuming

thattheionicstrengthofELFis0.2M.56Normally,thepHofELFisabout6.9,57

andtheconcentrationofAH2iswithinfewmM.58Thus,accordingtoTable2,the

large reaction rage coefficient of the AH2 + O3 reaction at this pH can yield a

relatively small diffuso-reactive length for ozone which could avoid the full

penetration of ozone through the ELF and the direct contacts of ozone to the

biomembranes. On the other hand, at higher acidities the significantly smaller

reaction rate coefficients of the AH2 + O3 reaction can yield larger

diffuso-reactivelengths,whichcanbeevenmuchlargerthanthedepthofELF.As

a result, the higher acidity of ELF provided by various pathologies or inhaled

particularmatterscannotonlygeneratemoreAOZ,1butalsoresultinthedeeper

penetrationsofAOZandozonethroughtheELFandthusthehigherchancesof

theirdirectcontactstotheaerialbiosurfaces.

TheaqueousAH2aerosolsstudied inthepresentworkcanalsoberegardedas

prototypebioaerosols.Humanscanexhalesmalldropletsofairway-liningfluids,

such as ELF, during normal breathing, coughing or sneezing. These exhaled

bioaerosolsmaycarryairbornepathogens,andadetailedunderstandingabout

the influences of aerosol chemistry and microphysics to these pathogens in

aerosolsbecomescrucialrecently.59–61Theairway-liningfluidstypicallycontain

trace amountsof antioxidants, suchasAH2, uric acid and reducedGlutathione,

besides salts and proteins. The findings of this study suggest that the

concentrations of above species and the ionic strength inside the bioaerosol

dropletscouldincreaseatleastaboutafewtimesduetoevaporation,whilethe

bioaerosolsmaystillremainthesimilarpHat6.9.Asaresult,theconcentrated

AH2inbioaerosolremainsthehighreactivitiesandyieldssmalldiffuso-reactive

lengths forozone (~120nm,assumingenriched [AH2]=10mM), compared to

the typical sizeofbioaerosol (<1-10μm).Thus, theantioxidants, suchasAH2,

actuallybecometoprotect thepathogens insidebioaerosolagainstozone from

environments for a period of time. The reaction time required to scavenge

enrichedAH2(~10mM)inbioaerosols(~5μminsize)withindoorozone(~10

ppb)couldtakeaboutafewhours.Forthemore"toxic"atmosphericozone(~50

ppb),thescavengetimeisthenreducedtoaboutahalfhour.

Finally,wewouldliketojustifytheusageoftherelativelyhighpressureofozone

inthiswork(~2ppmonaverage),whencomparingtoambientozone(~50ppb).

Firstly,theexperimentalresultsofpressuredependenceinthisworkjustifythat

the typical ozone pressure used here is sufficiently low, so that the

correspondingvalueofk2,3.0×106M−1s−1(atpH≈2,ozonepressure=1.4ppm

and ionicstrength=1M,seeFigure6), isalreadyveryclose to thatat the low

pressurelimit,3.1×106M−1s−1.ThismeansthattheozonolysisofAH2dropletsat

both~2ppmand~50ppbozonepressures canbe approximated to the same

special case of heterogeneous reaction kinetics, i.e., diffusion-limited case

characterized byEquation (3) anddiffuso-reactive length,without the need to

include any ozone-pressure dependent kinetics, such as

Langmuir-Hinshelwood-type mechanism. As a result, the main features of

heterogeneousreactionkineticsinvestigatedbythisworkshouldbethesameas

thoseforambientozone.Ontheotherhand,theprimarypurposeofhigherozone

pressuresusedinthis laboratory investigation is toshortentheentirereaction

time, allowing for building statistics via repeating more measurements on

reasonableexperimentaltimescales.

ConclusionsIn this work, we investigated the kinetics of aqueous AH2 reaction in

micron-sizeddropletswithgaseousozonebymeansofaerosolopticaltweezers,

at different ozone pressures, ionic strengths and pH. Particularly, this study

demonstrates that the kinetics and pH of single aerosol droplets can be

determinedsimultaneously.Themeasuredbimolecularreactionratecoefficients

atlowozonepressuresandlowionicstrengthsagreewiththosefromprevious

bulkmeasurements,indicatingthattheobservedaerosolreactionkineticscanbe

solelyexplainedintermsofliquidphasediffusionandAH2+ozonereaction,and

no necessary to include any surface effect. We found that the measured

bimolecular reaction rate coefficients exhibit a Langmuir-Hinshelwood

dependenceonozonepressures, ascribed to theheterogeneousnatureof such

aerosol reaction. We also found that the measured bimolecular reaction rate

coefficients have positive correlations with ion strengths. This study also

confirmsthatthereactionratecoefficientsoftheAH2+ozonereactionatpH<

pKa,suchaspH≈2,canbeabouttwoordersofmagnitudesmallerthanthatat

pH > pKa, such as pH ≈ 6, agreeing with previous bulkmeasurements. These

resultsimplythatthehigheracidityinairway-liningfluidscouldcausethelower

reactivityofAH2toozone.Ontheotherhand,AH2inexhaledbioaerosolscould

actually protect pathogens within against the oxidative damages caused by

atmosphericozoneforaperiodoftime,suchasafewhours.

ConflictsofinterestTherearenoconflictsofinteresttodeclare.

AcknowledgementsThis work was supported by the Ministry of Science and Technology, Taiwan

(MOST107-2113-M-110-004-MY3 and MOST109-2113-M-110-010-) and

NSYSU-KMUjointresearchproject(NSYSUKMU108-I002).Wealsothankforthe

financialsupportsfromAerosolScienceResearchCenter,NSYSU,Taiwan.Finally,

wethankProf.AdamJ.Trevittforthetutorialofmanipulatingaerosols,andwe

also thank Prof. Chia C. Wang and referees of this manuscript for their very

helpfulcomments.

Notesandreferences1 S.Enami,M.R.HoffmannandA.J.Colussi,Proc.Natl.Acad.Sci.,2008,105,

7365–7369.

2 J.R.KanofskyandP.Sima,J.Biol.Chem.,1991,266,9039–9042.

3 C.A.Ballinger,R.Cueto,G.Squadrito, J.F.Coffin,L.W.Velsor,W.A.Pryor

andE.M.Postlethwait,FreeRadic.Biol.Med.,2005,38,515–526.

4 M.J.Davies,Biochem.Biophys.Res.Commun.,2003,305,761–770.

5 L.F.Agnez-Lima,J.T.A.Melo,A.E.Silva,A.H.S.Oliveira,A.R.S.Timoteo,K.

M.Lima-Bessa,G.R.Martinez,M.H.G.Medeiros,P.DiMascio,R.S.Galhardoand

C.F.M.Menck,Mutat.Res.Mutat.Res.,2012,751,15–28.

6 D.Giamalva,D.F.ChurchandW.A.Pryor,Biochem.Biophys.Res.Commun.,

1985,133,773–779.

7 J.R.KanofskyandP.D.Sima,Arch.Biochem.Biophys.,1995,316,52–62.

8 I.S.MudwayandF.J.Kelly,Toxicol.Appl.Pharmacol.,1998,148,91–100.

9 S.Kermani,A.Ben-JebriaandJ.S.Ultman,Arch.Biochem.Biophys.,2006,451,

8–16.

10 G.D.Smith,E.Woods,C.L.DeForest,T.BaerandR.E.Miller,J.Phys.Chem.A,

2002,106,8085–8095.

11 G.D.Smith,E.Woods,T.BaerandR.E.Miller, J.Phys.Chem.A,2003,107,

9582–9587.

12 O.R.Hunt,A.D.WardandM.D.King,RSCAdv.,2013,3,19448.

13 T.C.PrestonandJ.P.Reid,J.Opt.Soc.Am.A,2015,32,2210.

14 C. Fitzgerald, N. A. Hosny, H. Tong, P. C. Seville, P. J. Gallimore, N. M.

Davidson, A. Athanasiadis, S. W. Botchway, A. D. Ward, M. Kalberer, M. K.

KuimovaandF.D.Pope,Phys.Chem.Chem.Phys.,2016,18,21710–21719.

15 J.F.DaviesandK.R.Wilson,Anal.Chem.,2016,88,2361–2366.

16 T.C.Preston,J.F.DaviesandK.R.Wilson,Phys.Chem.Chem.Phys.,2017,19,

3922–3931.

17 K.Gorkowski,N.M.DonahueandR.C.Sullivan,Environ.Sci.Technol.,2017,

51,12154–12163.

18 T.Endo,K. Ishikawa,M.Fukuyama,M.Uraoka,S. IshizakaandA.Hibara, J.

Phys.Chem.C,2018,122,20684–20690.

19 J.P.Reid,A.K.Bertram,D.O.Topping,A.Laskin,S.T.Martin,M.D.Petters,F.

D.PopeandG.Rovelli,Nat.Commun.,2018,9,956.

20 C.CaiandC.Zhao,Atmos.Environ.,2018,189,50–60.

21 X.-J.Lv,Y.Wang,C.Cai,S.-F.Pang,J.-B.MaandY.-H.Zhang,Spectrochim.Acta.

A.Mol.Biomol.Spectrosc.,2018,200,179–185.

22 A.Rafferty,K.Gorkowski,A.ZuendandT.C.Preston,Proc.Natl.Acad.Sci.,

2019,116,19880–19886.

23 E.M.Coddens,K. J.AngleandV.H.Grassian, J.Phys.Chem.Lett.,2019,10,

4476–4483.

24 G.Rovelli, Y.-C. Song,A.M.Maclean,D.O.Topping,A.K.Bertramand J. P.

Reid,Anal.Chem.,2019,91,5074–5082.

25 H. C.Boyer,K.Gorkowski andR. C. Sullivan,Anal. Chem., 2020,92, 1089–

1096.

26 O. R. Hunt, A. D.Ward andM. D. King,Phys. Chem. Chem. Phys., 2015,17,

2734–2741.

27 M.D.King,K.C.Thompson,A.D.Ward,C.PfrangandB.R.Hughes,Faraday

Discuss,2008,137,173–192.

28 B.J.Dennis-Smither,F.H.Marshall,R.E.H.Miles,T.C.PrestonandJ.P.Reid,

J.Phys.Chem.A,2014,118,5680–5691.

29 B. J. Dennis-Smither, R. E. H. Miles and J. P. Reid, J. Geophys. Res.

Atmospheres,,DOI:10.1029/2012JD018163.

30 B.J.Dennis-Smither,K.L.Hanford,N.-O.A.Kwamena,R.E.H.MilesandJ.P.

Reid,J.Phys.Chem.A,2012,116,6159–6168.

31 M. D. King, K. C. Thompson and A. D.Ward, J. Am. Chem. Soc., 2004,126,

16710–16711.

32 R.J.Hopkins,L.Mitchem,A.D.WardandJ.P.Reid,Phys.Chem.Chem.Phys.,

2004,6,4924.

33 R.W.Berg,Appl.Spectrosc.Rev.,2015,50,193–239.

34 R.Criegee,Angew.Chem.Int.Ed.Engl.,1975,14,745–752.

35 R.Symes,R.M.SayerandJ.P.Reid,Phys.Chem.Chem.Phys.,2004,6,474.

36 J. P. Reid, H. Meresman, L. Mitchem and R. Symes†, Int. Rev. Phys. Chem.,

2007,26,139–192.

37 S.E.SchwartzandJ.E.Freiberg,AtmosphericEnviron.1967,1981,15,1129–

1144.

38 S.WeisenbergerandA.Schumpe,AIChEJ.,1996,42,298–300.

39 K.J.Laidler,ChemicalKinetics,HarperCollinsInc.,1987.

40 M. Mekic, M. Brigante, D. Vione and S. Gligorovski,Atmos. Environ., 2018,

185,237–242.

41 H.Herrmann,Chem.Rev.,2003,103,4691–4716.

42 S. Endo, A. Pfennigsdorff and K.-U. Goss, Environ. Sci. Technol., 2012, 46,

1496–1503.

43 E.DemouandD.J.Donaldson,J.Phys.Chem.A,2002,106,982–987.

44 M.C.K.Soule,P.G.BlowerandG.L.Richmond,J.Phys.Chem.B,2007,111,

13703–13713.

45 K.Sahu,V.F.McNeillandK.B.Eisenthal,J.Phys.Chem.C,2010,114,18258–

18262.

46 J. J.Nájera,C. J.PercivalandA.B.Horn,Phys.Chem.Chem.Phys.,2010,12,

11417.

47 M.Ammann,U.PöschlandY.Rudich,PhysChemChemPhys,2003,5,351–

356.

48 J.P.Burrows,A.Richter,A.Dehn,B.Deters,S.Himmelmann,S.VoigtandJ.

Orphal,J.Quant.Spectrosc.Radiat.Transf.,1999,61,509–517.

49 L.Rkiouak,M. J.Tang, J.C. J.Camp, J.McGregor, I.M.Watson,R.A.Cox,M.

Kalberer, A. D.Ward and F. D. Pope,Phys Chem Chem Phys, 2014,16, 11426–

11434.

50 Z.Gong,Y.-L.PanandC.Wang,Rev.Sci.Instrum.,2016,87,103104.

51 A. Rafferty and T. C. Preston, Phys. Chem. Chem. Phys., 2018, 20, 17038–

17047.

52 T.Liu,S.L.CleggandJ.P.D.Abbatt,Proc.Natl.Acad.Sci.,2020,117,1354–

1359.

53 M. Mekic, Y. Wang, G. Loisel, D. Vione and S. Gligorovski, Environ. Sci.

Technol.,2020,54,12898–12907.

54 J.A.Thornton,C.F.BrabanandJ.P.D.Abbatt,Phys.Chem.Chem.Phys.,2003,

5,4593–4603.

55 C.J.GastonandJ.A.Thornton,J.Phys.Chem.A,2016,120,1039–1045.

56 A.Pelfrêne,M.Cave,J.WraggandF.Douay,Int.J.Environ.Res.Public.Health,

2017,14,112.

57 A.W.Ng,A.BidaniandT.A.Heming,Lung,2004,182,297–317.

58 A.Snyder,L.SkozaandY.Kikkawa,Lung,1983,161,111–121.

59 D.A.Edwards,J.C.Man,P.Brand,J.P.Katstra,K.Sommerer,H.A.Stone,E.

NardellandG.Scheuch,Proc.Natl.Acad.Sci.,2004,101,17383–17388.

60 B.R.Bzdek,J.P.ReidandM.I.Cotterell,Commun.Chem.,2020,3,105.

61 K.A.Prather,C.C.WangandR.T.Schooley,Science,2020,368,1422–1424.