Nov 27, 2000 - Physics Department, University of Manchester Institute of Science and Technology, Manchester, England, U.K.. Abstract. Particle size ...
JOURNAL OF GEOPHYSICALRESEARCH, VOL. 105,NO. D22, PAGES26,891-26,905, NOVEMBER 27, 2000
Behavior of ultrafine particles in continental and marine air massesat a rural site in the United Kingdom Hugh Coe andPaul I. Williams PhysicsDepartment,Universityof ManchesterInstituteof ScienceandTechnology,Manchester,England,U.K.
Gordon McFiggans 1 Schoolof EnvironmentalSciences,Universityof East Anglia, Norwich, England,U.K.
Martin W. Gallagher,Karl M. Beswick,Keith N. Bower,andTom W. Choularton PhysicsDepartment,Universityof ManchesterInstituteof ScienceandTechnology,Manchester,England,U.K. Abstract.
Particle size distribution measurements were made at a coastal site in the United
Kingdom. Thesearepresented,andthe behaviorof recentlyformedultrafineparticlesis discussed.No ultrafineparticleswere observedin maritimeair masses;however,3 to 7 nm particleswere frequentlyobservedat enhancedconcentrations whenthe wind directionwas from the land. Their formationwas favoredat lower temperatures, when 1 ppbv or more of SO2waspresentandin air massesthat hadnot beenagedextensively.On dayswhenenhancedultrafineparticleconcentrations were observed,3 nm particlesincreasedsharplyin the morning,approximately30 to 90 min afterthe UV solarflux first increased.By early afternoonthe ultrafineparticleconcentration hadreturnedto backgroundlevels. Rapid measurementsof 5 nm particlesshowedno correlationwith turbulenceparameters,althoughthe boundarylayer mixing scaleswere similarto growthtimesof freshlynucleatedparticlesto 5 nm diameter.However,ultrafineparticleconcentrations do correlatewith the availabilityof sulphuricacid vapor. A delayof approximatelyan hourbetweenthe increaseof H2SO4in the morningand a largeincreasein ultrafineparticleconcentrations is dueto the growthof particlesto observablesizes,not the nucleationprocessitself. An analysisof the timescalesfor growthshowedthat coagulationmay be importantimmediatelyafterthe particleshavenucleatedbut its effectivenessreducesasnumberconcentrationfalls. Conversely,growthby condensationis initially slow due to the Kelvin effect but increasesin importanceasthe particles reach observable
sizes.
1. Introduction
correlatedwith UV flux and low tides. A postulatedsource of precursorsis the macroalgaein the intertidal zone; howUltrafine aerosolparticles,between3 and 10 nm diameter, ever, the formationprocessesand the speciesinvolved remain provide an importantsourceof new aerosolparticlesin the unknown. Grennfell et al. [1999] used a massbalance apatmosphere;however, much uncertaintysurroundstheir forproachto showthatthe hydroxylradicalinitiatedoxidationof mation and growth. Observationsof theseparticleshave now di-methylsulphide,emittedfrom biota in the intertidalzone, been made in a number of different environmentsincluding couldnot explainthe amountof materialpresent. They were the remote marine boundary layer [Covert et al., 1992], coastal locations [Allen et al., 1999; O'Dowd et al., 1999], unableto explain the phenomenon. However, this approach assumesthat all of the observedultrafine particle masswas forests[(M•ikel•i et al., 1997], remotecontinentalsites[Weber producedin the intertidalzone. O'Dowd et al. [1999] calcuet al., 1997] and the free troposphere[Clarke, 1992]. In all lated the gas phasesulphuricacid concentrationand showed casesthe presenceof theseparticlesis sporadic,occurringin that it displays a coherencewith ultrafine particles. The eventsor burstslasting from severalminutesto a few hours, concentrationswere not, however, sufficient to explain nuand only occursduringdaylight,implyingthe gasprecursors cleation by binary H2SO4-H20, and a ternary mechanism, are photochemicallyproduced. The data from the coastal possiblyinvolvingNH3, is proposed.Even so, O'Dowd et al. sites show the periodsof new particle productionare often argue a fourth species,emitted from the intertidal zone, is requiredto grow the particlesto observablesizes. Recentexperimentsin the backgroundmarineenvironment •Nowat Physics Department, UMIST, P.O. Box 88, Manchester [Weber et al., 1995], at a clean continental site [Weber et al., M60 1QD, England,U.K. 1997], and over the remote Pacific Ocean [Clarke et al., 1998; Weber et al., 1998] have combinedultrafine particle Copyright2000 by the AmericanGeophysicalUnion. measurementsand direct measurementsof gas phase sulphuric acid and the hydroxyl radical. These papers have Papernumber2000JD900234. 0148-0227/00/2000JD900234509.00 recentlybeencomparedandreviewedby Weberet al. [1999]. 26,891
26,892
COE ET AL.: BEHAVIOR
OF ULTRAFINE
PARTICLES
roleof turbulence andmixingis investigated by comparing They concludethat large nucleationeventsare found exclusivelyin regionsof enhancedsulphuricacidvaporconcentra- rapid measurementsof ultrafine particles with turbulence
tions(upto 5x107molecules cm-3).Theycompared their results with model predictions of binary nucleation and showed that at higher altitudes in the troposphere,colder temperaturesenhancethe model predictionsof H2SO4-H20 nucleationand result in broad agreementwith the measured rates. Close to the surface,model predictionsof nucleation ratesare many ordersof magnitudeslowerthanthoseinferred from measurements.They provide someevidencethat ammonia is involved in a ternary processthat can achieve nucleationat muchlower levelsof sulphuricacid vapor. In the urban backgroundof Helsinki, H•imeri et al. [1996] observedlarge concentrationsof new particleswith diameters lessthan 10 nm. They were unable to link theseobservations to a source. Williams et al. [1998] made ultrafine particle measurementsin Manchester,United Kingdom, and observed large concentrationsof 3 to 5 nm particles and growth through the mode. These particles correlatedstronglywith SO2, implicating sulphuricacid vapor is involved in the formationof theseparticles. Easter and Peters [1994] have arguedthat as the nucleation rate is extremely sensitiveto changesin temperature, relative humidity and the relative acidity of sulphuric acid vapor, mixing processes,and turbulence greatly enhance ultrafine particle formation. They showed that large scale variations,causedby temperaturefluctuationsof 1 to 2 K over 1 hour inducedby large eddiesor waves,could enhance thebinarynucleationrateby over an orderof magnitude.The rapid ascent of air under convective conditions may also induce similar effects. Smaller-scalefluctuationscausedby turbulencewere also modeled by Easter and Peters [1994], and increasesin the nucleationrate of up to a factor of 70 were predictedfor a well-mixed boundarylayer. Nilssonand Kulmala [1998] discussthis previouswork and outline cases wherelarger-scalemixing may alsoenhancenucleation. In our work, ultrafine particle measurementswere made at a coastallocation,and their presencewas linked to precursor species. In contrast to the measurementsof O'Dowd et al. [1999], made at Mace Head, Ireland, there are no exposed macroalgaein the intertidalzone at the site. The presenceof ultrafine particles in maritime conditionswould therefore
indicatethat gas-to-particleconversionprocesses are taking placein the marineboundarylayer over open water.In addition, new particle processesin rural conditionstypical of much of the United Kingdom were also investigated. In southerlyand southwesterlydirectionsthe site is representative of much of the rural United Kingdom and also experienceslong-rangetransportof aged polluted air from central Europe. These different scenarioswere used to investigate the formation of new particles from the gas phasein a wide rangeof conditions. Previous measurementsof ultrafine particles have used either two particle counters with different size cuts [e.g., O'Dowd et al., 1999] or pulse height analysiswith a single counter [e.g., Saros et al., 1996]. In the former case the countedparticlescover a rangeof particlesizesand so represent particleswith different time histories. Here we use a differentialmobility analyzer,fixed at one size, as an inlet to a particlecounter. This methodallowsus to observeparticles of a singlesize at high temporalresolution. In additionto the measurementsof ultrafine particlesand their precursors,the
parameters.
2. Methodology The experimenttook place at the WeybourneAtmospheric Observatoryon the north coastof Norfolk, United Kingdom (52.951øN, 1.125øE) from June 14 to 30, 1998 (Figure 1). A full descriptionof the site is given by Penkettet al. [1999]. The observatorylies within 100 m of the coastline to the northat a heightof 15 m abovesealevel on grasslandsloping down to a pebble sea defence. In northerlywind directions air, often from north of the Arctic Circle, advects over the
North Sea directly to the station. The shorelinehas no significant tidal zone, and incomingmarine air is unlikely to be affectedby emissionsfrom the near field. To the south and west, Weybourne is surroundedby flat arable land and is
well-removedfrom immediatelocalsources.Whensampling air from these sectors the location is characteristic
of much of
the rural United Kingdom and is several hundredkilometers downwindof the large conurbationsof London (190 km) and Birmingham(250 km). Most of the largestpoint sourcesof SO2in the United Kingdom are locatedapproximately150 km to the west and northwestof the site. Trace gas analyzersare locatedin the main building and are sampledfrom a 10 m tall Pyrex glasssamplingstack. The particle sizing instrumentation was housedin a mobile laboratoryimmediatelyadjacent to the main laboratory,and was sampledthrougha 3 m tall V2 inch diameter inlet.
Particle size distributionsin the range 3 nm to 450 nm diameterwere measuredusing a Differential Mobility Particle Sizing (DMPS) systemthat hasbeen constructedby the Physics departmentat University of ManchesterInstitute of Science and Technology and is based on the Vienna design
[Winkelmayret al., 1991]. The DMPS wascomposed of two differential mobility analyzers(DMAs). The largest had a centralrod lengthof 28 cm, was coupledto Thermo Systems Inc. (TSI) 3010 condensationparticle counter (CPC), and measuredparticlesbetween 20 and 450 nm diameter. The secondDMA had a centralrod lengthof 11 cm, was coupled to a TSI 3025 condensationparticle counter,and measured particlesbetween3 and 20 nm diameter. The sampleflows of
the shortandlongDMAs weresetto 1.5 and1 1 min'J, respectively,and the sheathair flows to 10 times the sample flow in each case. The voltageson the DMAs were incrementedso that in total 52 size channelswere retrievedevery 8 min. The DMA had a triangulartransferfunction, whosebase width was approximately5% of the selecteddiameter, and that had a 5% uncertaintyin midpoint diameter. The inversion procedurewas a Newton-Raphsoniteration of Stokes' law in termsof the diameterfor a given voltageand set parametersof the DMA [Williams, 1999]. The efficiency of DMAs of this type have been estimatedempiricallyby Birmilli et al. [1997]. Number concentrations of largeraerosol particles (100 nm to 3 gm diameter) were measuredas a function of their size using an optical'scattering probe (ASASP-X). The ASASP-X sizesparticlesinto 31 channels andhasa total countaccuracyof 10%. Full size distributions were retrieved every 10 min throughoutthe experiment. However, ultrafineparticleconcentrationscan vary rapidlyover very shorttimescalesas their
COE ET AL.:
BEHAVIOR
OF ULTRAFINE
PARTICLES
26,893
.... ':::'"'::•:c.:•i..:..:.,....: ..... ,onhS,• I •h
-.•,,:2:..'..•,...::i::.:...i•
EYBO
WEYBOURNE
I
/i
o,
Figure 1. (top) WesternEurope and the back trajectorysectorsarrivingat Weybournethroughoutthe durationof the experiment.(bottom)The locationof the observatorycloseto the coastand over 1 km distantfrom the small villagesalongthe coastroad. The locationof Weybourneon the northNorfolk coastis shownin the inset.
lifetimesare very short. To investigatethesefluctuations,two additional Differential Mobility Analyzers were set up to continuouslymeasure5 nm and 20 nm diameterparticleswith a time responseof 1 s. This approachwas usedin preference to the differencingmethodbecauseunlike the latter it providesinformationat a singlesize with a narrowtransferfunction of typicallyonly0.25 nm full width at 5 nm. The differencingtechniquecountsthe numberof particlesbetweenthe lowest size cuts of two different CPCs, usually 3 nm and 7 nm in the case of the most commonly used TSI 3025 and 3022 instruments.However, the growth times of 3 and 7 nm particlesare considerablydifferent. Clearly, the presenceof a largenumberof 7 nm diameterparticlesand no 3 nm particles is not an indicationof immediatenew particle formation,yet this will be indistinguishable from suchan event by the differencingmethod. Furthermore,retrievinginformationon the 3 to 7 nm rangeby differencingbetweentwo largenumbersis not sensitivewhen backgroundconcentrationsare high relative to the ultrafine numberconcentrations.The rapid measurementof 5 nm particlesafter classifyingwith a DMA in combinationwith completesize distributionmeasurements at lower temporalresolutioninformationyields more information about the behavior of small particlesthan differencing alone. However, the data from the singlesizedDMA needto
be averagedto 1 min to overcomethe poor countingstatistics at these sizes.
The importantchemicalprecursorsand physicalparameters relevantto new particle formationwere also measured. SO2 was monitoredusing a Teco 42S to provide an estimateof the sulphuricacid gas concentration,a likely precursordue to its extremelylow saturatedvapor pressure. The Teco 42S has a lower detectionlimit of 0.3 ppbv and a precisionof 1%. The influence of photolyric processeson the formation of new particleswas investigatedusing radiometricmeasurements of the photolysisrate of NO2, j•o2. The hydroxylradical is the main oxidant involved in the formation of aerosolprecursor gases,and its primary productionrate can be found from the ozone concentrationand its rate of photolysisto the excited
O(•D)state.Thiscanbeestimated by scaling j•o2usingthe clear-skyratiojo(•t>l:j•o2for any given time of day. This approximation hasbeenvalidatedduringperiodswhenajo(m) radiometer
was available
and was shown to be accurate to
25%. Measurementsof ozone were made using a Teco 49 with a precisionof 1 ppbv. All of the above data were recordedat a time resolutionof at most 1 min so that processes influencingthe behaviorof new particlescould be correlated with the fast 5 nm particle measurements.In addition to the chemical measurementsa full range of meteorologicalpa-
26,894
COE ET AL.: BEHAVIOR OF ULTRAFINE PARTICLES
rameterswas measuredusing an automatic weather station
nearlyconstantair temperature of 12øCand very low SO2
mountedat the top of the 10 m samplingtower. Lastly,a sonic anemometer(Gill Solent model N1012R) and a fast responseUV hygrometer(Campbell Model KH20) were
ment. The aerosolnumberconcentrations duringthisperiod
mountedon a separate10 m mast to measurethe three wind components,temperature,and water vaporat 20 Hz. Sensible
experiment;however,the aerosolsurfacearea was greater
concentrations at or below the detection limit of the instru-
weretypically 2400particles cm'3, thelowestduringthe than later in the experimentdue largelyto an increasednum-
and latent heat fluxes were calculated from these measure-
ber of accumulation and coarsemodeparticles.The measmentsto investigatethe role of turbulencein new particle urementsprovidedno evidencefor new particleformationin formation. The sonicanemometerresolvesthe 3 orthogonal contrast to the experiments conducted at Mace Head
windcomponents witha resolution of +/-0.01cms-• andhas [O'Dowdet al., 1999]wherethereis a largesourceof precuran uncertaintyof 1%.
sor from marine biota in the intertidal zone.
3. Results
air advectedover land from the southand west. However,it is clearfromboththebacktrajectoryanalysisin Figure1 and the tracegas and meteorological data in Figure2 that there
For the remainderof the experimentthe site experienced 3.1.
General
was considerable variabilityin the type and composition of the air arrivingat Weybourne.In particular,thebacktrajecadvecting Atlanof a contour plotof dN/dlogDp.Thegeneral characteristics of toryanalysisshowsthatcyclonicconditions, on mostdaysfrom the aerosolsize distributionschangemarkedlythroughoutthe tic air overtheUnitedKingdom,prevailed experiment. Figure 2 showsthe temperature,radiation,and June17 to 30 (sectorsA andD). The exceptionto thiswasa conditions SO2time seriesthroughoutthe experimentalperiod,together 2 dayperiodon June20 and21 whenanticyclonic with the aerosol surface area and the transfer coefficient to deliveredair from the southover continental Europe(secThe hourly averaged aerosol size distributions taken throughoutthe campaignare presentedin Plate 1 in the form
preexistingaerosol surfaces,k•o.•s (see equation (2)). Also shownin Figure 2 is the time seriesof hourly averaged5 nm diameter particle concentrationsfor comparisonwith the other parameters. There is an increaseduncertaintyin the
tor C).
DuringJune20 and21 the air arrivingat the measurement
sitewasmuchwarmerthanthat experienced on otherdays
duringthe experiment,varyingbetween16ø and25øC. There wasconsiderable cumulusactivityon June20 but verylittle were not usedin any subsequentanalyses.Back trajectories cloudcoveron June21. The air arrivingat Weyboumehad Europe,andasa were calculatedusing EuropeanCentre for Medium Range spentat leastthelast5 daysovercontinental wereenhanced for prolonged periWeatherForecasts(ECMWF) analysesto arriveat Weybourne resultSO2concentrations at 1200 for each day during the experimentand have been ods. High solar flux and SO2 concentrationswill lead to an in therateof production of sulphuric acidandpotengroupedinto four broadcategories.The spatialextentof each increase of thesegroupingsis shownin Figure 1, and periodsduring tially a sourceof materialfor ultrafineparticleformation. which Weybourneexperiencedair of thesetypes are identi- However, large increasesin the aerosol surface area and fied in Figure 2. Each of the parametersshownin Figure2 hencein theuptakecoefficient k•o.•werealsoobserved during thesinkfor sulphuric acidontopreexcan affect the formationof new particlesin the atmosphere; thisperiod,increasing however, becausemany processesare occurringsimultane- istingsurfaces. Enhancedaerosolconcentrations duringthis ously, there is unlikely to be a straightforwardrelationship periodcanbe identifiedeasilyin Plate 1 by the considerably betweenany individualparameterand the measurements of 5 enhancedpeak aerosolnumberconcentrations occurringat nm particles. To assessthe extent to which a relationship largersizes(up to 100 nm) than at othertimesduringthe betweentheseparametersand 5 nm diameterparticlenumber experiment. Thesepropertiesof the aerosolare as a resultof processing duringseveraldays'transport from concentration Ns exists,the Spearmanrank correlationcoeffi- considerable Europe.Thisis compounded by reports cient R.•was calculatedfor each variableand Ns for daytime centralandsouthern periodsonly. The ranksof variables,ratherthanthe variables of desertdustfallsin partsof theUnitedKingdom,implying calibration of the SO2 data before June 20, and these data
themselves,are used to calculate R.•.as the rank of a distribution is normally distributedeven though the variable itself may not be. The R,. values for temperature,radiation, SO2,
k•os., andaerosol surface areaare5x10"•,0.4,0.37,-0.3,and0.35. These data show a link between radiation and SO2, and
Ns, mostlikely from their role in the photochemical production of H2SO4precursormaterial. Aerosol surfacearea and Ns are negativelyrelated as a result of preexistingsurfaces competingfor precursorwith nucleationand growthof new particles.Similaranalyseshavebeenperformedby Weberet
that long-range transportfrom the Saharawas experienced duringthese2 days[WeatherLog, 1998]. Both the increased particlenumberandalsotheirlargesizeenhanced the surface areaof the aerosolpopulationand so provideda moreefficient sink for precursormaterial,reducingthe potentialfor new particleformation.
For the remainderof the experiment,air was arriving at Weybournefrom sectorsA and D. In both thesecasesthe air masseswerecyclonicandhad crossedthe United Kingdomin less than a day from the Atlantic Ocean. However, recent al. [1997] at a remote continental site in the United States. formation of particles,indicatedin Figure 2 by the 5 nm Our results show strong similarities with this earlier work, diameternumberconcentration time series,doesnot appearto which was performedat a much cleanersite and which sam- occuron all of thesedaysbut is clearly evidenton June 17, pledpredominately free tropospheric air. 22, 25, and 27-30.. A comparisonof the backtrajectorieson June 15 and 16 were the only daysduringthe experiment thesedayswith thosewhen new particleswere not seen,June when marine air from the north was advected over the North
18, 19, 23, 24 and 26, shows a distinct difference between the
Sea to the observatoryand are identified as sectorB in Figure 1. Both days were overcastand are characterized by a
days when new particleswere and were not observed,identified as sectorsA and D, respectively.Althoughall the back
COE ET AL.: BEHAVIOR
(mu)
OF ULTRAFINE
PARTICLES
26,895
26,896
COE ET AL.: BEHAVIOR
A
B
D
C
OF ULTRAb2NE PARTICLES
A
D
A
D
A 200
30
25 150
20 ,
15
100
o Z
10 50 5 0
800 t500
i
200 • 0
-
-,--
150 •'•
ß
'•'
'+
"
•....
,....
•'
50
0
200
150 •,
so
•
I 200
•,4.E-05 t 3.E-05
• 2.E-05 o
1.E-05
50
• 0.E+00 15/06
17/06
19/06
21/06
23/06
25/06
27/06
29/06
01/07
Figure2. Temperature, radiation,sulphurdioxide,uptakecoefficient, andaerosolsurfaceareatimeseriesat the Weybourne Atmospheric Observatory fromJune14 to July1, 1998. Thetransfer coefficient k•oss wascalculated
fromthesizedistribution measurements using therelation givenbySchwartz [1986,equation (2)]. Thelightgrey timeseries shows thenumber concentration of 5 nmparticles. Periods whenthesiteexperienced air fromthe different sectors shown in Figure1 arealsoindicated: A, westerly newparticle production; B, marine; C, continen-
tal; D, westerlyno new particleproduction.
COE ET AL.: BEHAVIOR OF ULTRAFINE PARTICLES
trajectoriesin both groupsare cyclonic,thosegroupedinto sectorA are generallyfrom the west and crossthe northof Englandand Scotland;thosegroupedinto sectorD traverse the southof Englandand are generallyfrom moresoutherly latitudes in the Atlantic
Ocean.
26,897
that showa sharpchangefrom stableto unstableconditions aroundthis time on each of thesedays. Although this mixing
may enhancenew particleformationalso, the data cannot identifywhetherthismaybe the causeof the largeincreases of 5 nm particlesaround60 min later. By early afternoonthe numberof 3 nm particleshad greatlydiminished, andin mostcasesthisverysmallparticle
The meteorological,gas, and aerosolcharacteristics of thesetwo groupsare also significantlydifferent. In both casesthe air temperatureincreasedthroughoutthe morning modereducedin both size and numberand no longercontribandearlyafternoon,althoughdaysin groupD increased from uted to the Aitken mode above 10 nm. The peak in the Ait15ø to 20øC, whereas those in group A were somewhat ken mode fraction steadilyincreasedin size throughoutthe colder, 10ø to 15øC. June 26, a type D day, was, however, day,reaching50 nm by the evening,andthe numberconcensimilarin temperature to thoseof typeA. SO2mixingratios tration diminishedthrough coagulationand dilution. The on groupD dayswere low, usuallylessthan0.5 ppbvand Aitken mode continued to reduce in number, but the mode increasedin diameterthroughthe night. Althoughthe general often below the detection limit of the instrument. In contrast, SO2mixing ratioson groupA dayswere very variablebut behaviorof the ultrafineparticleswas similar on all of these generally above1 ppbvin themorningperiods.Thevariation days,there are distinctdifferencesbetweenthem. Sincethe in the SO2concentrationon timescalesof 1 to 2 hourscorre- experimentwas conductedin a Eulerian framework, the lates with changesin the local wind directionand suggests variations in the ultrafine mode aerosol throughoutthe day may only be interpretedas growth from the freshlynucleated thatthe sitewasin the plumeof a largesourceseveraltensof kilometersdistant. The exceptionto this was June27 when aerosolif the air masswas spatiallyhomogeneousthroughout the SO2concentration was between0.5 and 1 ppbv. Aerosol the period. This may be the case in this study as on many surfaceareason dayswhennew particleswereobserved(type days the upper Aitken and accumulationmode aerosolconA) were the lowest experiencedduring the experimentand centrationsdo not change substantiallythroughout the day twice as low as thoseobservedon type D days. Again, June (Plate 1 and Figure 3). 26 is the exceptionand displayssurfaceareacharacteristics Figure3 showsfour setsof five size distributionspectra similarto thoseon dayswhennew particleswere observed. measuredusing the DMPS systemduring periodsof high The differencesin temperature,SO2, and aerosolsurface ultrafineparticleconcentration on June 17, 22, 28, and 30. areabetweengroupsA andD all favor the formationof new Different behavior is observedon each of these days. On particlesin the formergroup. A reductionin temperature of June 17 very large concentrations of 3 nm particlesare ob5øC will increase the rate of binary nucleation by over an servedto occuras early as 0630, and the modeincreasesin orderof magnitude[Easterand Peters, 1994] and will also size and 30 min to 1 hour later is combined with the Aitken increase the condensationrate of material onto recently mode. The concentrationof 3 nm particlesdiminishesby formedparticles(Figure7), thoughthis effect is only slight. midday, but the modal diameterincreasesthroughoutthe The increasedconcentrations of SO2on type A daysprovide afternoon(Plate 1). On June22, similarlargeconcentrations an increased source of H2SO4, and reduced aerosol surface of 3 nm particlesareobserved to occurby 0700, yet,although areasprovidea reducedsink. Note alsothat severalperiods the numberof theseparticlesincreasesthroughoutthe mornof increasedSO2 above 2 ppbv occurredeither at night or ing, thereis little evidencefor theseparticlesaffectingthe duringperiodsof very high aerosolsurfaceareaandno new Aitken mode. The concentrationof 3 nm particlesdeclines particleswereobserved.Clearlya combination of factorsis aftermidday,andthereappearsto be no increasein particle required. The largestpoint sourcesof SO2 in the United concentrationat sizesabove5 nm. However, a separatemode Kingdomare locatedin an areato the west andnorthwestof largerthan 10 nm, peakingat around20 nm (Plate 1), was Weybourneapproximately100 km distant. Both the back observedshortlybefore midday and whosemodal diameter trajectoryanalysisand the variabilityin the SO2suggestthat increasedthroughoutthe afternoon.The appearance of these thisareais the likely sourceof precursormaterial. particlesappearsto be unrelatedto the 3 nm particlesobservedearlierin the day, indicatingthat theseparticleswere formedseparately somedistanceupwind. A rapidincreaseof 3.2. Formation and Behavior of Ultrafine Particles 3 nm diameterparticleswas againobservedat 0800 on June When ultrafineparticleswere observed(type A), similar 28, but on this day 10 nm particleswere seento increaseat the sametime. More 3 nm particleswereobservedthaneither behaviorwas seenon all of the days(Plate 1 andFigure2). A large increasein 3 nm diameterparticleswas observedat 5 or 10 nm particles,but like June22 therewere more 10 nm theday. The around0900, throughoutthe morningthe modaldiameterof particlesthanthoseof 5 nm diameterthroughout of the 3 nm particlesis not directlyrelatedto the theseparticlesincreased rapidly,andby 1100it wascontrib- appearance utingto the peakof the Aitken modeat 20 nm. On mostdays large increasein 10 nm particlesand implies that there are occurring. On June30 a muchlower concenwhennew particleproductionwas observed,largeincreases two processes in ultrafinemode aerosolwere precededby reductionsin the trationof 3 nm particleswas observedthanon otherdays,but preexisting particlepopulation.This canbe seenbothin the a largemodecenteredat 10 nm wasobservedsuddenlyafter contourplot andalsoin thetimeseriesof aerosolsurfacearea 0800. The modal diameterof theseparticlesappearsto inand 5 nm diameterparticlenumberconcentration.One ex- creasewith time and mergewith the Aitken modethroughout planationof thereductionin preexisting aerosolconcentration the afternoon(Plate 1). Again, the 3 nm particlesare no of a new is the breakdownof a low-level nocturnalinversionearly in longerpresentafter midday.The suddenappearance the morning,mixing cleanerair from aloft with the aerosol- mode, centeredat 10 nm, without a precedingincreasein richair trappednearthe surfaceovernight.This hypothesis is smallerparticlessuggeststhat ultrafineparticleson June 30 supported by turbulenceandsensibleheatflux measurements appearedto have been formed away from the measuring
26,898
COE ET AL.:
BEHAVIOR
June
PARTICLES 22 June
06:17
100000
OF ULTRAFINE
100000
•6:17
............ 06:33
..... 07:04 ........07:28
10000
.,••
•7:12 .....
8:07
........
8:23
10000
10:13
08:07
1
o
1000
1000
100
100
i
i
10
100
1000
i
i
i
10
100
1000
100000
100000
28 June
30 June
07:45
•6:11
•
07:53
•
7:06
.....
08:09
.....
7:53
........
08:25
•:
10000
10000
,.
........ 8:09
08:40
8:17
;:
.,
o
1000
'•'•'1.
1000
' ii:3'
..
lOO
t.?
100 x,
1
i
i
i
10
100
1000
diameter (nm)
'
10
i
1
10
100
1000
diameter (nm)
Figure 3. Examplesfrom four separatedaysof particlenumberdistributions measuredusingthe DMPS system duringthe periodswhenenhancedultrafineparticleconcentrations wereobserved.On eachday,five spectraare shownto illustratethe developmentof the ultrafineparticlepopulationthroughthe morning. The 4 daysshow distinctlydifferentbehavior. stationand transportedto the site. No new particleformation appearsto be occurringin situ. Measuringnew particleconcentrationsusing either the differencebetweentwo CPCs, or a single,fixed size DMA alone will not discriminatebetween these different events.
Size distributions
must also be meas-
ured.
3.3.
The Role of Turbulence
Clouds have been implicatedin new particle formationin the remote marine environment [Clarke et al., 1998; Radke
and Hobbs, 1991], and freshlynucleatedparticleshave often been observedin subsidingair masses[Covert et al., 1996]. Easter and Peters [1994] and Nilsson and Kulmala [1998] have shown in modeling studiesthat the mixing and turbu-
lencemay well increasebinarynucleationratesby an orderof magnitudeor more. To investigatethe role of mixing in the formation of new particles, a sonic anemometerwas used throughoutthe Weybourneexperimentto retrieverapid measurementsof wind velocity in three dimensionsand temperature. The presenceof large concentrations of 5 nm particles followed the onsetof convectionin the morningby between 30 min and an hour, and their disappearance in midafternoon coincided with a marked reduction in heat flux, the covari-
ance of vertical wind speedand temperatureon every day. Formationof particlesdirectly from the gas phaseis likely to be a very strongfunction of temperature,and enhancedmixing in the morning may promote new particle formation by rapid lifting and coolingof air, rich in precursormaterial,that is close to the surface [Easter and Peters, 1994; Nilsson and
COE ET AL.:
5O
BEHAVIOR
OF ULTRAFINE
eß
ß
40
• e•e ß ß
PARTICLES
26,899
ß
ß
ß
30
ß
20
ß
ß ß
ß e el ß
.
ß
. • ß ...
ß
ß .:
ß ..'•. ß ' ß ß ../ ;:.'". ...' .• ßß ee.,... eeee eee•a • Ieeß e• ..&
ß
ß ß
10
ß
ß
ß
..
..
..
ß ee• ß •
ße
0 eß eee
-10
ee
ß
ß
ß
©©
eee ß ß
-20 ß
ß
ß
ß
ß
-30 -40
-50
I
-0.3
-0.2
-0.1
0
