c# wpf adobe pdf reader : Extract pages from pdf on ipad Library application component .net windows asp.net mvc amt-7-1929-20140-part166

Atmos.Meas.Tech.,7,1929– 1941,2014
www.atmos-meas-tech.net/7/1929/2014/
doi:10.5194/amt-7-1929-2014
©Author(s)2014.CCAttribution3.0License.
IntercomparisonofanAerosolChemicalSpeciationMonitor
(ACSM)withambientfineaerosolmeasurementsindowntown
Atlanta,Georgia
S.H.Budisulistiorini
1
,M.R.Canagaratna
2
,P.L.Croteau
2
,K.Baumann
3
,E.S.Edgerton
3
,M.S.Kollman
4
,
N.L.Ng
4,5
,V.Verma
5
,S.L.Shaw
6
,E.M.Knipping
7
,D.R.Worsnop
2
,J.T.Jayne
2
,R.J.Weber
5
,andJ.D.Surratt
1
1
DepartmentofEnvironmentalSciencesandEngineering,GillingsSchoolofGlobalPublicHealth,TheUniversityofNorth
CarolinaatChapelHill,ChapelHill,NC27599,USA
2
AerodyneResearch,Inc.,Billerica,MA01821,USA
3
AtmosphericResearch&Analysis,Inc.,Cary,NC27513,USA
4
SchoolofChemicalandBiomolecularEngineering,GeorgiaInstituteofTechnology,Atlanta,GA30332,USA
5
SchoolofEarthandAtmosphericSciences,GeorgiaInstituteofTechnology,Atlanta,GA30332,USA
6
ElectricPowerResearchInstitute,PaloAlto,CA94304,USA
7
ElectricPowerResearchInstitute,Washington,D.C.20036,USA
Correspondenceto:J.D.Surratt(surratt@unc.edu)
Received:6November2013–PublishedinAtmos.Meas.Tech.Discuss.:19December2013
Revised:19May2014–Accepted:22May2014–Published:2July2014
Abstract.Currently,therearealimitednumberoffieldstud-
iesthatevaluatethelong-termperformanceoftheAerodyne
AerosolChemicalSpeciationMonitor(ACSM)againstes-
tablishedmonitoringnetworks.Inthisstudy,wepresentsea-
sonalintercomparisons ofthe ACSMwithcollocated d fine
aerosol(PM
2
.
5
)
measurementsattheSoutheasternAerosol
ResearchandCharacterization(SEARCH)JeffersonStreet
(JST)siteneardowntownAtlanta,GA,during2011–2012.
IntercomparisonoftwocollocatedACSMsresultedinstrong
correlations(
r
2
>0.8)forallchemicalspecies,exceptchlo-
ride(
r
2
=
0
.
21)indicatingthatACSMinstrumentsareca-
pableofstableandreproducibleoperation.Ingeneral,spe-
ciatedACSMmass concentrationscorrelatewell(
r
2
>0.7)
withthefilter-adjustedcontinuousmeasurementsfromJST,
althoughthecorrelationfornitrateisweaker(
r
2
=
0
.
55)in
summer.CorrelationsoftheACSMNR-PM
1
(non-refractory
particulate matterwith aerodynamic diameterless than or
equalto1µm)pluselementalcarbon(EC)withtaperedel-
ementoscillatingmicrobalance(TEOM)PM
2
.
5
andFederal
ReferenceMethod(FRM)PM
1
massarestrongwith
r
2
>0.7
and
r
2
>0.8,respectively.Discrepanciesmightbeattributed
toevaporativelossesofsemi-volatilespeciesfrom thefil-
termeasurements usedtoadjustthecollocatedcontinuous
measurements. This suggests that adjusting g the e ambient
aerosol continuous s measurements s with h results s from filter
analysisintroducedadditionalbiastothemeasurements.We
alsorecommendtocalibrate theambientaerosolmonitor-
inginstrumentsusingaerosolstandardsratherthangas-phase
standards.ThefittingapproachforACSMrelativeionization
forsulfatewasshowntoimprovethecomparisonsbetween
ACSMandcollocatedmeasurementsintheabsenceofcal-
ibratedvalues,suggestingtheimportanceofaddingsulfate
calibrationintotheACSMcalibrationroutine.
1 Introduction
Atmosphericfineparticulatematterwithaerodynamicdiam-
eterslessthanorequalto2
.
5µm(PM
2
.
5
)
haveadverseef-
fectsonhumanhealth(Dockeryetal.,1993),reducevisibil-
ity,andplayaroleinEarth’sclimate(IPCC,2013).Asare-
sult,therehasbeenanongoingneedtoresolvethechemical
compositionofPM
2
.
5
inordertoidentifytheirexactsources,
andthus,developeffectivecontrolstrategies.Organicmatter
(OM)contributesamajorfraction(25–70%)ofthesubmi-
cron(PM
1
)
mass inthetroposphere;however, itssources,
PublishedbyCopernicusPublicationsonbehalfoftheEuropeanGeosciencesUnion.
Extract pages from pdf on ipad - software SDK dll:C# PDF Page Extract Library: copy, paste, cut PDF pages in C#.net, ASP.NET, MVC, Ajax, WinForms, WPF
Easy to Use C# Code to Extract PDF Pages, Copy Pages from One PDF File and Paste into Others
www.rasteredge.com
Extract pages from pdf on ipad - software SDK dll:VB.NET PDF Page Extract Library: copy, paste, cut PDF pages in vb.net, ASP.NET, MVC, Ajax, WinForms, WPF
Detailed VB.NET Guide for Extracting Pages from Microsoft PDF Doc
www.rasteredge.com
1930
S.H.Budisulistiorinietal.:IntercomparisonofanACSMwithambientfineaerosolmeasurements
composition,andatmosphericchemicaltransformationsre-
mainunclear(Jimenezetal.,2009).Inorganicaerosolcon-
stituents,suchassulfate(SO
2
4−
)
,nitrate(NO
−3
)
,ammonium
(NH
+4
)
,andchloride(Cl
)
canalsobemajorcomponentsof
PM
2
.
5
,dependingonlocationandtimeofyear.
Numerousmethods formeasuringthemassandchemi-
calcompositionofPMhavebeenputforward,includingin-
tegratedfiltersamplerswithsubsequentlaboratoryanalysis
(e.g.,Baumannetal.,2003;Solomonetal.,2003b),semi-
continuousmethods(e.g.,Weberetal.,2003a,b;Limetal.,
2003),andreal-timeinstruments(e.g.,Gardetal.,1997;Lee
etal.,2002;Jimenezetal.,2003).Differencesbetweensam-
plingtechniquesmayoccurforahostofreasons,including
design,analysismethods,andassumptionsusedindatare-
duction.Hence,comparisonofnewsamplingmethodswith
establishedtechniquesallowsonetodetermineitssuitability
forlong-termairqualitymonitoring.
The Aerodyne Aerosol l Chemical Speciation Monitor
(ACSM,Ngetal.,2011)isdesignedforreliablelong-term
operationwith minimal userintervention. . The keydiffer-
encesbetweentheACSMandtheaerosolmassspectrom-
eter (AMS, , Jayne e et t al., 2000) is s that the former lacks
a particle beam chopper and uses a relatively lowersen-
sitivity quadrupole and, therefore, , data mustbe averaged
overalongerperiodtoobtainsufficientsignal-to-noisefor
quantification.RecentstudiesshowedthattheACSMdata
arestronglycorrelated(
r
2
>0.8)withthe Aerodyne high-
resolution time-of-flight aerosol l mass s spectrometer r (HR-
ToF-AMS)(Ng et al., , 2011), , time-of-flight ACSM (ToF-
ACSM),andcompacttime-of-flightAMS(Fröhlichetal.,
2013).ComparisonsofSO
2
4−
aerosolshowedgoodcorrela-
tionsbetweentheACSMandtheparticle-into-liquidsampler
coupledtoanionchromatograph(PILS-IC),andtheThermo
ScientificSulfateParticulateAnalyzer(model5020i),where
theACSMmeasured31%lowerforSO2
4−
thanthesetwo
instruments. ForNO
−3
aerosol, theACSMmeasured25%
lowerthanthePILS-IC(Ngetal.,2011).Arecentdeploy-
mentoftheACSMinBeijing,China,reportedagoodcorre-
lationbetweenthetotalnon-refractoryPM
1
(NR-PM
1
)
esti-
matedfromthesumofallspeciesmeasuredbytheACSM
withthePM
2
.
5
measuredbythetaperedelementoscillating
microbalance(TEOM),wheretheACSMNR-PM
1
reported
64%oftheTEOMPM
2
.
5
mass(Sunetal.,2012).
ThepresentstudycomparesambientNR-PM
1
measured
bytheACSMwithasuiteofcollocatedparticlemeasure-
mentsinAtlanta,Georgia.Thecollocatedparticlemeasure-
mentsincludeanotherACSMoperatedbytheGeorgiaInsti-
tuteofTechnology(GIT),continuousSO2
4−
,NO
−3
,andNH
+4
measurementsoperatedbyAtmosphericResearch&Anal-
ysisInc.(ARA),semi-continuousorganiccarbon/elemental
carbon(OC/EC)measurements,totalPM
2
.
5
massmeasured
by TEOM, integrated SO
2
4−
, NO
−3
, and NH
+4
by parti-
cle composition monitor (PCM)developedby ARA, and
integratedPM
2
.
5
andPM
1
massmeasurementsbasedonthe
FederalReferenceMethod(FRM).
Inthediscussionthatfollows,wefirstcompareindivid-
ualspecies(i.e.,OM,SO
2
4−
,NO
−3
,NH
+4
,andCl
)
andtotal
NR-PM
1
massmeasuredfromcollocatedACSMsduringa
shortperiodbetweenJanuaryandFebruary2012.Secondly,
wecomparespeciesmeasurements(minuschloride)andto-
talmassfromtheACSMwithorganiccarbon(OC),SO2
4−
,
NO
−3
,NH
+4
,andPM
2
.
5
fromcontinuousandfiltermeasure-
mentsattheJeffersonStreet(JST)siteduringsummerand
fall2011.WecomparemassfromtheACSMwithtotalmass
fromintegratedFRMmeasurementsinthreeshortperiods
ofJanuary–February,April–May,andJuly2012.Lastly,we
estimateaerosoldensityfromcontinuousmeasurementsbe-
tween17Octoberto20November2012.Fromthisintercom-
parison,wehavegainedmoreknowledgeoncontinuousam-
bientaerosolmeasurements,includingtheimportanceofcal-
ibratingtheroutinemonitoringaerosolinstrumentswithtrue
aerosolstandardsratherthangas-phasestandards,aswellas
sulfate calibrationas additional routine calibrationforthe
ACSM.
2 Experimentalsection
2.1 Sitedescription
Ambient aerosol from Atlanta, , Georgia, , was s collected d at
the Jefferson n Street (JST)site (33.7775
N, 84.4166
W),
whichislocatedinamixedindustrial–residentialareaabout
4.2kmnorthwestofdowntownAtlanta(Hansenetal.,2003;
Solomonetal.,2003a).TheJSTsiteisoneoftheresearch
sitesofSoutheasternAerosolResearchandCharacterization
(SEARCH)networkthatis equippedwith h a suiteofgas,
particle,andmeteorologicalmeasurements.Detailsofthese
measurements are described in subsequent sections. . The
UniversityofNorthCarolinaatChapelHill(UNC)ACSM
wasoperatedcontinuouslyatJSTfrom27July2011through
21September2012,whiletheGITACSMwasdeployedat
thissitefrom31Januarythrough29February2012.Thepe-
riodwhenbothACSMs werecollocatedatJST is usedto
evaluatetheACSMperformance,andtheextendedperiods
in2011and2012areusedtoevaluatetheaccuracyofACSM
measurementsagainstestablishedmonitoringnetworkmea-
surements.
2.2 NR-PM
1
andchemicalmeasurementsbytheACSM
DuringFebruary2012,NR-PM
1
wasmeasuredbytwoAC-
SMs that belong g toUNC and GIT, , andplaced d in anair-
conditionedtraileratJST.SamplingconditionsforbothAC-
SMsaredescribedinTable1.BothACSMswereoperatedto
scan150mass-to-charge(
m/z)
ratiosoffragmentedionsata
rateof500msamu
1
.Vaporizerandheaterbiasesweresetat
600
Cand100.30V,respectively,withthebiasvoltagecho-
sentomaximizetheN
2
(
m/z
28)signal.Particle-ladenand
Atmos.Meas.Tech.,7,1929–1941,2014
www.atmos-meas-tech.net/7/1929/2014/
software SDK dll:VB.NET PDF: Create PDF Mobile Viewer in VB.NET Doc Image Program
you will not need to view and browse your PDF document on your computers for it is available for you to read PDF document in your iPhone or iPad devices at any
www.rasteredge.com
S.H.Budisulistiorinietal.:IntercomparisonofanACSMwithambientfineaerosolmeasurements
1931
Table1.UNCandGITACSMssamplingsetupattheJSTsiteforashortperiodbetweenJanuary–February2012.
UNC
GIT
Samplinginlet
PM
2
.
5
cyclone
PM
2
.
5
cyclone
Samplinglinelength
5.00m
5.00m
Samplinglinediameter
0.64cmODand0.46cmID
stainlesssteeltube
1.27cmIDfor1moflength
0.95cmIDfor4moflength
Sampledrying
50-tubeNafiondryer(PermaPurePD-
50T-24SS)with7.00Lmin
1ofsheath
aircomingfromdry/zeroairsystem
200-tubeNafiondryer(PermaPure
PD-200T-12MPS)runningwith
0.50Lmin
1
sheathairflow(under
vacuum)
ACSMsamplingflowrate
3.00Lmin
1
3.00Lmin
1
RF
NO
3
calibration
3
.
79
×
10
11
3
.
97
×
10
11
RIE
NH
4
calibration
6.00
4.30
RIE
SO
4
fitting
0.79
0.54
RIE
NO
3
default
1.10
1.10
RIE
Cl
default
1.30
1.30
RIE
Organic
default
1.40
1.40
Referenceflow(
Q
cal
incm3s
1
)
1.39
1.35
Dataacquisitionsoftware
ACSMDAQv1.4.2.2
ACSMDAQv1.4.2.5
Dataanalysisprocedure
ACSMLocalv1.5.2.0
ACSMLocalv1.5.2.0
particle-freeairweresampledinterchangeablyandaveraged
over
30minintervalsforeachmeasurement.Wecalibrated
theACSMonsite.TheACSMswerecalibratedforresponse
factor(RF)andrelativeionizationefficiency(RIE)usinga
separatecalibrationsystemforUNCandGIT.Theresulting
valuesforeachinstrumentarereportedinTable1,andfor
UNCACSM,differentcalibrationvalueswereusedfordif-
ferentseasons.
DataacquisitionsoftwareprovidedbyARIwasusedto
processthemeasurementstoobtaintotalorganicandinor-
ganic(i.e.,SO2
4−
,NO
−3
,NH
+4
,andCl
)
aerosolmasscon-
centrations.Furtherdetailsoftheconcentrationcalculation
arediscussedbyNgetal.(2011)andshowninEq.(1).
C
s
=
CE
s
T
m/z
×
10
12
RIE
s
×
Q
cal
×G
cal
RF
NO
3
×
1
Q×G
all
i
IC
s
,i
(1)
Speciesmassconcentration(
C
s
)
iscalculatedbasedonmea-
suredioncurrent(ICinamps)atfragmention
i
.CE
s
iscol-
lectionefficiencyforspeciess,andRF
NO
3
isinstrumentre-
sponsefactorfromcalibration.
T
m/z
iscorrectionforthe
m/z
dependent ion n transmission efficiency of the quadrupole.
Q
cal
and
G
cal
arethevolumetricsampleflowrateandmul-
tipliergain,respectively,andweredeterminedfromcalibra-
tion,while
Q
and
G
areobtainedduringthemeasurements.
Duringdataprocessing,calibratedandmeasured
Q
and
G
canceleachotheroutaspartofairbeamcorrectionfactor
(Eq.2),andnoseparatecorrectionisappliedforflowrate.
Theairbeamcorrectionisappliedasitisuncertainwhether
airbeamsignalchangesduetogainorflowchanges.
Airbeamcorrectionfactor
=
Q
cal
×G
cal
Q×G
(2)
Anair beam signal (i.e.,
m/z
28)was used d to normalize
themeasurementswithrespecttoinstrumentmeasurement
sensitivity(i.e., secondary y electronmultiplier(SEM)gain
decay) and sampling flow rate. . The effusive e naphthalene
sourcewasnotusedduetolowersignal-to-noisecompared
to
m/z
28anditsdependencyoneffusionflowand/orback-
groundcontamination. Moreover, thechangesinflowrate
needtobeaccountedforbyusingthefilteredairbeam.The
ACSMusesafilteredairmassspectrumtoaccountforback-
grounds(e.g.,N
2
andCO).Thesesignalswillvarywithflow
rateorslowlydesorbingmaterial.Contributionoftheslowly
desorbingmaterial,however,isgenerallysmallcomparedto
theN
2
signalat
m/z
28.
RIE
s
forspecies swasdeterminedfrom calibrationsof
laboratory-generatedaerosolsofeachspecies usingAero-
dyneAMS(Alfarraetal.,2004;Canagaratnaetal.,2007).
SincetheACSMparticlevaporizationandionizationsource
aresimilarbutnotidenticalindesigntothatoftheAMS,
theremaybedifferencesinRIEvaluescomparedtothose
referencedabove.ThevaporizerisidenticalbetweenACSM
andAMSsystems.TheionformationchamberintheACSM
issomewhatsmallerthanintheAMS.Theionsourcevol-
umeintheACSMiscalculatedtobe370mm
3
andthatof
theAMSis580mm
3
.Wenote,however,thattheeffective
volumeisreallydefinedbytheelectricfieldsanditisnot
easily calculated. . Inbothsystemsthediameterofthe ex-
tractionintotheionopticlensregionis3mm.Thesmaller
ionsourcevolume(withtighterspatiallydistributedelectric
fields)intheACSMcouldresultinlargervariabilityofthe
relativeionizationefficiencieswithrespecttopreciseparticle
beamalignment,whichiscurrentlybeinginvestigated.
www.atmos-meas-tech.net/7/1929/2014/
Atmos.Meas.Tech.,7,1929–1941,2014
1932
S.H.Budisulistiorinietal.:IntercomparisonofanACSMwithambientfineaerosolmeasurements
Table2.StatisticsofcalibrationvaluesobtainedfromUNCandGITACSMssincemid2011toearly2013.
UNCACSM
GITACSM
Date
RF
NO
3
RIE
NH
4
RIE
SO
4
RF
NO
3
RIE
NH
4
RIE
SO
4
Mean
4.17
×
10
11
5.71
0.67
3.26
×
10
11
4.40
0.59
1-stddeviation
1.53
×
10
11
1.01
0.09
1.26
×
10
11
0.38
0.04
%uncertainty
37%
18%
14%
39%
9%
7%
Sulfateaerosolcalibrationswerenotconducteduntilearly2013.
ThedefaultRIEvalueforammonium(RIE
NH
4
)
was3.5;
thevalue obtainedfrom ACSM calibrations was approxi-
mately5.71(Table2).ThedefaultRIEofsulfatewas1.2,
whichtherealvaluecouldbeestimatedbyfittingmeasured
sulfateandpredictedsulfatevalues, derivedfromNH
4
,
pred
equation(Eq.3).Measuredsulfate(SO
4
,
meas
)
issulfatethat
ismeasuredbytheACSM,whilepredictedsulfate(SO
4
,
pred
)
istheestimatedvalueofsulfatefromionbalanceapproach
(Eq.4).
NH
4
,
pred
=
2
MWNH
4
MWSO
4
SO
4
,
meas
(3)
+
MWNH
4
MWNO
3
NO
3
,
meas
+
MWNH
4
MWChl
Cl
meas
SO
4
,
pred
=
(4)
NH
4
,
meas
MWNH
4
MWNO
3
NO
3
,
meas
MWNH
4
MWChl
Cl
meas
2
MWNH
4
MWSO
4
ThepreviousvalueofRIE
SO
4
1.2isthenmultipliedbyslope
obtainedfrom fittingSO
4
,
pred
versusSO
4
,
meas
andusedas
theRIE
SO
4
valueofthisstudy.UNCACSMappliedfitted
RIE
SO
4
valuesof0.95,0.77,0.79, 1.1,0.73, and0.44for
summerandfall2011,winter,spring,summer,andfall2012
datasets,respectively.ExplicitcalibrationofRIE
SO
4
byat-
omizing(NH
4
)
2SO
4
usingthesamecalibrationsystemfrom
UNCduringwinter2013yieldedavalueof0.67
±
0.09indi-
catingthatthefittingapproachvalue(0
.
79
±
0
.
22)isconsis-
tentwiththecalibrations,withalargeruncertainty(Table2).
WefoundthatSO2
4−
percentdifferencebetweenACSMand
collocated measurementatJSTwas s improved d from m about
50%tolessthan30%.Therefore,inadditiontoregularcal-
ibrationusingNH
4
NO
3
,werecommendadditionalcalibra-
tionusing(NH
4
)
2
SO
4
toobtainanRIE
SO
4
valuespecificfor
theACSM.
ACEof0.5was usedtocalculatemass concentration.
WeusedaNafiondryertodryambientairsamples;inves-
tigationofspecies-dependentCE(Middlebrooketal.,2012)
suggestedthatCEisnotinfluencedbyhighlyacidicaerosol
(Fig.S1intheSupplement)orammoniumnitrate(Fig.S2
intheSupplement)as providedinthesupplementalinfor-
mation.Somemeasurementperiodswereexcludedfromthe
dataanalysisduetooperationalandmaintenanceissues,such
asshutdownduringcalibrations.Aerosolmassspectrometer
uncertaintywas estimated20–35%(Bahreinietal., 2009)
whichincludedCEuncertaintyof30%.Arecentstudyof
compositiondependentCEparameterization(Middlebrook
etal.,2012)hassubstantiallycontributedtonarrowtheun-
certaintyofAMS,whichcouldbeusedasaguidelinefor
ACSMaccuracy(
30%).
2.3 Chemicalconstituentsmeasuredbyintegratedand
continuousparticlemeasurementsatJSTsite
DetailsoftheJSTsitemeasurementsareprovidedelsewhere
(Hansenetal.,2003;Edgertonetal.,2005,2006).Inletsfor
particlesamplersaremountedontherooftopofthesampling
trailerabout5mabovegroundlevel.Theparticlemeasure-
mentsconsistof24hfiltersamplingconductedeverythird
day(dailyforPM
2
.
5
andPM
1
mass),andofcontinuousand
semi-continuousmeasurementsbyinstrumentsplacedinan
air-conditionedtrailer.Integrated,semi-continuous,andcon-
tinuousPM
2
.
5
measurementsarelistedinTable3,andde-
scribedbrieflybelow.FieldblankloadingsofJSTsitemea-
surements aregenerally y insignificant forSO 2
4−
, NH
+4
and
OC,butcanbesignificantforNO
−3
andECmostlydueto
loadings atorbelowdetection n limitofthose components
(Edgertonet al., , 2005). . We e emphasize here thatthe JST
siteaerosolinstrumentsarebasedongasphasedetectionof
aerosolconversionproducts(e.g.,SO
2
fromSO2
4−
andNO
fromNO
−3
)
,therefore,arecalibratedwithstandardgasesin-
steadofdirectlybyparticlemassgeneratedfromanatomizer
combinedwith scanningelectrical mobility sizer (SEMS)
mixingcondensationparticlecounter(MCPC)asdonefor
theACSM.
Particlecomponentsmeasurements
Detailsofthesemi-continuousandcontinuousPM
2
.
5
sam-
plingandanalysisareprovidedinEdgertonetal.(2006)and
inthesupportinginformation.Briefly,PM
2
.
5
massismea-
suredcontinuouslyusinganR&PModel1400a/bTEOM
operated at 30
C to o reduce losses s of f semi-volatile com-
poundsandwithmainflowrateof3Lmin
1
.Sampleairwas
pulledthroughaPM
10
inletfollowedbyaPM
2
.
5
VerySharp
CutCyclone(BGIIncorporated)thatgoesinsidethetrailer
wherea multi-tube Nafiondrier(Perma Pure)is installed
Atmos.Meas.Tech.,7,1929–1941,2014
www.atmos-meas-tech.net/7/1929/2014/
S.H.Budisulistiorinietal.:IntercomparisonofanACSMwithambientfineaerosolmeasurements
1933
Table3.Summaryofintegrated,semi-continuous,andcontinuousPM
2
.
5
analysesatJST.
Analyte
Instrument
Analyticalmethod
DetectionLimit(mgm
3)
Frequency/TimeResolution
Integratedsamples
Mass
FRM(Teflon,47mm)
Gravimetry
0.2
daily
SO
2
4−
PCM1(Teflon,47mm)
IC
0.05
3-day
NO
−3
PCM1(Teflon,47mm)
IC
0.01
3-day
NH
+4
PCM1(Teflon,47mm)
AC
0.03
3-day
Volatile-NO
−3
PCM1(Nylon,47mm)
IC
0.02
3-day
Volatile-NH
+4
PCM1(Citricacid-coatedcellulose,47mm) AC
0.04
3-day
OC
PCM3(Quartz,37mm)
TOR
0.08
3-day
Continuoussamples
Mass
R&P1400a/bTEOM(modified)
Oscillatingmicrobalance
2.0
5min
SO
2
4−
HSPH(modified)
ReductiontoSO
2
/PF
0.4
1min
NO
−3
ThermoScientific
ReductiontoNO/CL
0.25
1min
NH
+4
ThermoScientific
OxidationtoNO/CL
0.07
1min
OC/TC
SunsetOC/ECAnalyzer
CombustiontoCO
2
/NDIR
0.5
60min
Notes:Volatile-NO
3
andVolatile-NH
+
4
arecollectedonbackfiltersasHNO
3
andNH
3
dissociationonthefrontfilter;ICrepresentsionchromatographytechnique;ACrepresentsautomated
colorimetrymethod;TORindicatesthermal/opticalreflectancemethod;PFrepresentspulsedfluorescencetechnique;CLindicatesozone-NOchemiluminescencemethod;HSPHstandsforHarvard
SchoolofPublicHealth.
todrythesample.SO2
4−
ismeasuredcontinuouslyusinga
modifiedHarvardSchoolofPublicHealth(HSPH)Sulfate
ParticulateAnalyzer.NH
+4
andNO
−3
weremeasuredusinga
three-channelcontinuousdifferencingmethoddevelopedby
ARA, Inc.(Edgertonetal., 2006).Totalcarbon(TC)was
semi-continuouslymeasuredbyaSunsetOC/ECinstrument
(model3),whichcollectsparticlesonafilter.Oncecollection
iscomplete(after
50min),theovenispurgedwith10%O
2
inHe,andthenrampeduptoasetpointof850
Caccording
totheNIOSH5040analysisprotocol.
Inorganics, OC,andtotalmassconcentrations fromthe
continuousanalyzerswereadjustedtomatchthefilter-based
datavialinearregressionsincethecontinuousanalyzershave
beenshowntodriftovertime.Newadjustmentsareapplied
every1–2months,dependingonthestabilityoftheindivid-
ualanalyzer.Withrespecttocarbonmeasurements,OCis
calculatedasthedifferencebetweenfilter-adjustedTCand
filter-adjustedEC,andOMis estimatedfrom applyingan
OM/OCratioof1.4(Edgertonetal.,2006).
Thecomponentmassloadingsfromeachfilterwereblank-
corrected using SEARCH network-wide average e loadings
from fieldblanks,thenthecorrectedloadingwasnormal-
izedbysamplingvolume.Detailsoftheintegratedmeasure-
mentsattheJSTsiteareprovidedinEdgertonetal.(2005).
Thisstudywillfocus oncomparisonbetweenACSMand
JSTfilter-adjustedcontinuousmeasurements(Figs. S3,S4
andS5intheSupplement).Resultsofintercomparisonbe-
tweenACSMandfiltermeasurementsarepresentedinthe
supportinginformation(Figs.S6andS7intheSupplement).
2.4 Totalparticlemassmeasurements
PM
2
.
5
massconcentrationswereobtainedbyseveralmeth-
ods duringthis campaign. Continuous totalmass concen-
trationswereobtainedwiththeTEOM(afteradjustmentto
matchtheintegratedPCM-basedPM
2
.
5
)
.TheJST-integrated
PM
2
.
5
valueswereobtainedbyaddingblank-correctedPCM
measurementstogetherwithvolatileNO
−3
fromPCMnylon,
volatileNH
+4
andvolatileOMfromPCMbackfilter.
FRM filtersamples were collected for 24h using dual
R&PModel2025sequentialFRMmonitors todetermine
bothPM
2
.
5
andPM
1
mass. 47mm diameterTeflon n filters
(2µmporesize)wereusedforthesemeasurements,andthe
collection,processing,andanalysisofthesefiltersfollowed
FRMprotocol(CodeofFederalRegulations,2001).PM
1
fil-
tersweresampledduringthreeseparatesamplingperiods:
JanuarytoFebruary,April,andJuly2012,representingwin-
ter,spring,andsummerseasons,respectively.
2.5 Aerosoldensityestimation
Total PM
1
volume measurements were obtained using
the Brechtel l Manufacturing g Incorporated (BMI) SEMS
equippedwith acylindrical-geometrydifferentialmobility
analyzer(DMA)andcoupledtoanMCPC(Sorooshianet
al.,2008).TheDMAwassettosizeparticlesbetween10–
1000nmindiameterforbothupanddownscans.Differen-
tialmobilityanalyzersheathairflowratewassetto5Lmin
1
andparticlesweresampledat0.5Lmin
1
.Particlevolume
concentrationfromeachscanwascollectedevery120s,and
bothupanddownscanswereaveragedtogetonedatapoint
every4minand30s,whichincludesthescanningdelaytime.
3 Results
TheACSMmeasuredabout11.6µgm
3
ofOM,3.2µgm
3
ofSO
2
4−
,and0.61µgm
3
ofNO
−3
duringsummer2011.The
numbersdecreasedinthefall2011,exceptfornitrate(Ta-
bleS1intheSupplement).TheACSMmeasuredchlorideon
averageof0.02to0.04µgm
3
insummerandfall,respec-
tively.
www.atmos-meas-tech.net/7/1929/2014/
Atmos.Meas.Tech.,7,1929–1941,2014
1934
S.H.Budisulistiorinietal.:IntercomparisonofanACSMwithambientfineaerosolmeasurements
3.1 IntercomparisonbetweentheUNCand
GITACSMs
TheUNCandGITACSMswerecollocatedfrom10January
to23February2012.Intercomparisonsofchemicalspecies
betweenthetwoACSMsshowninFig.1indicatestrongcor-
relations(
r
2
>0.8), exceptforchloride(
r
2
=
0
.
21).Slopes
andintercepts ofthe linearregressionareprovidedinTa-
ble4.Weakercorrelationsofchloridemightbeduetoitslow
concentrationinAtlanta.
3.2 IntercomparisonofACSMwithcollocated
JSTmeasurements
Intercomparisons ofspecies andtotalmass measurements
bytheACSM,continuousparticlemeasurementsfromJST,
SunsetOCanalyzer(model3), andTEOMPM
2
.
5
(model
1400a/b)atthe JST site aregiven inTable4forsummer
(8Augustto14September)andfall(17Octoberto21De-
cember)2011samplingperiods.Collocatedmassandchem-
icalconstituentmeasurementswereaveragedtotheACSM
samplingtimestoallowforadirectintercomparison.Pre-
vious intercomparison n studies conducted at the same site
have beenlimited to the summerseason(Solomonetal.,
2003a);therefore,resultsfromthisstudycouldrevealpos-
sibleaerosolmeasurementsvariationacrossseasonsandin-
strumentationdifferencesinaerosolmeasurements.
3.2.1 Speciescomparison
ACSMOMisstronglycorrelatedwithOCfromtheSunset
OC/ECanalyzer(
r
2
valuesare0.86and0.92forsummer
andfall, respectively);theresultingratios(fromthelinear
regressionslopesinTable4)ofOM/OCare4
.
85
±
0
.
05and
3.85
±
0.02insummerandfall,respectively.AerosolChem-
icalSpeciationMonitorOMversusSunsetOCcorrelations
arelikelyhighersincetheyarebothreal-timeandnotaf-
fectedbystoragerelatedlosses,suchasthatfromthefilter
measurements.
ACSMSO2
4−
is stronglycorrelatedwiththatfrom JST
continuousmeasurementsinthesummer(
r
2
=
0
.
84)andfor
someperiodsinthefall(
r
2
=
0
.
83;September–November);
however,thecorrelationisweakerforsomeperiodsinDe-
cember(
r
2
=
0
.
22)when JST measured several instances
of very y high h SO2
4−
aerosol. Percent t differences between
themeasurementsare4and44%forsummerandfall,re-
spectively.Theseresultsareclosetoprevioussulfateinter-
comparisonsbetweenACSMandcollocatedmeasurements
(slope=0.95, 0.69, and0.69, forHR-ToF-AMS, PILS-IC,
and sulfate e particulate analyzer, respectively) ) (Ng g et t al.,
2011).
ForNH
+4
comparison, correlationsarehigh(
r
2
=∼
0
.
8)
andinterceptsforbothsummerandfallareinsignificant.Dif-
ferences betweenACSMandJSTmeasurements are20%
(
r
2
=
0
.
79)forsummerand51%(
r
2
=
0
.
76)forfall.
Table4.CorrelationsbetweentheACSMandthecollocatedmea-
surementsatJSTsite.Slopeandintercept
±
1standarddeviation
fromeachlinearregressioncorrelationsarepresented.
JSTContinuousc
Summer2011
Fall2011
Massa
r
2
0.71
0.83
Slope
1.50
±
0.02
2.10
±
0.02
Intercept
2.89
±
0.31
4.36
±
0.20
OMvs.OC
b
r
2
0.86
0.93
Slope
4.85
±
0.05
3.85
±
0.02
Intercept
7.34
±
0.19
2.99
±
0.09
SO
2
4−
r
2
0.84
0.83
Slope
1.04
±
0.01
1.44
±
0.02
Intercept
0.73
±
0.04
0.54
±
0.03
NO
−3
r
2
0.55
0.81
Slope
2.14
±
0.04
1.77
±
0.02
Intercept
0.06
±
0.01
0.08
±
0.02
NH
+4
r
2
0.79
0.76
Slope
1.20
±
0.02
1.51
±
0.02
Intercept
0.19
±
0.02
0.61
±
0.01
ACSMPM
1
iscalculatedfromsumofACSMspeciesand
SunsetEC.bForACSM-to-ACSMcomparison,itisOMvs.
OM.
c
JSTmeasuresPM
2
.
5
massandchemicalconstituents.
IntercomparisonsbetweenACSMNO
−3
andJSTcontinu-
ousNO
−3
resultinpercentdifferencesof114%(
r
2
=
0
.
55)
and77%(
r
2
=
0
.
81)inthesummerandfall,respectively.
Theweakercorrelationandlargerdiscrepancyinthesum-
mermightbeduetothelowNO
−3
loadingsandevaporative
lossesfromfiltersthatwillbediscussedlater.
3.2.2 Totalmasscomparison
ACSMPM
1
masswasdeterminedfromthesumofACSM
OM, SO2
4−
, NO
−3
,NH
+4
,andCl
aswellasECfromthe
SunsetOC/ECanalyzer.TheintercomparisonoftheACSM
PM
1
andTEOMPM
2
.
5
showsagoodcorrelation, with
r
2
values of0.71and0.83, respectively, anddiscrepanciesof
50and110%forsummerandfall, respectively(Table4).
AsinthespeciatedACSMandPCMmeasurementcompar-
isons,discrepanciesinthefallmighthaveresultedfromposi-
tivebiasesofspeciesmeasurementsbytheACSM.Sincethe
TEOMmeasurementsareadjustedtomatchfiltermasscon-
centrations,itisalsopossiblethattheadjustedTEOMvalues
arelowerthantheACSMPM
1
valuesbecauseofevaporation
Atmos.Meas.Tech.,7,1929–1941,2014
www.atmos-meas-tech.net/7/1929/2014/
S.H.Budisulistiorinietal.:IntercomparisonofanACSMwithambientfineaerosolmeasurements
1935
50
40
30
20
10
0
U NC NR-PM1
40
30
20
10
0
GIT NR-PM
1
0.30
0.25
0.20
0.15
0.10
0.05
0.00
U- NC Cl
0.20
0.10
0.00
GIT Cl
-
3.0
2.0
1.0
0.0
U+NC NH4
2.0
1.0
0.0
GIT NH
4
+
6
4
2
0
U- NC NO3
6
4
2
0
GIT NO
3
-
12
10
8
6
4
2
0
U2NC SO4
16
12
8
4
0
GIT SO
4
2-
40
30
20
10
0
U NC Org
40
30
20
10
0
GIT Org
50
40
30
20
10
0
2/10/2012
2/20/2012
Date and Time (Local)
4
3
2
1
0
3.0
2.0
1.0
0.0
A-CSM Mass Concentration (µg m)
6
4
2
0
16
12
8
4
0
40
30
20
10
0
UNC Org GIT Org
UNC SO
4
2-
GIT SO
4
2-
UNC NO
3
-
GIT NO
3
-
UNC NH
4
+
GIT NH
4
+
UNC Cl
-
GIT Cl
-
UNC NR-PM
1
GIT NR-PM
1
(a)
(b)
r
2
=0.95
f(x)=(-0.06±0.07)
+(1.14±0.01)x
r
2
=0.95
f(x)=(0.20±0.01)
+(0.73±0.01)x
r
2
=0.89
f(x)=(0.13±0.02)
+(0.98±0.01)x
r
2
=0.82
f(x)=(0.20±0.01)
+(1.21±0.02)x
r
2
=0.92
f(x)=(0.08±0.12)
+(1.09±0.01)x
r
2
=0.21
f(x)=(0.01±0.00)
+(0.60±0.04)x
Figure 1. (a)Linearregressioncorrelationand (b)timeseriesplotsoforganicandinorganicconstituentsmeasuredbytheUNCandGIT
ACSMs. ACSM measurements from UNC are colored by specieswhile those from GIT are colored in black.
of semi-volatile organics and nitrates from the filters during
storage.
The ACSM data were averagedto the FRM filtersampling
times, which was 24h (midnight to midnight) during each
sampling period. Comparison between the ACSM NR-PM
1
and FRM PM
1
in winter, spring, and summer 2012 shows
a good correlation, with
r
2
values of >0.80 (Fig. 2), and
the mass concentrations differences vary from 10% in sum-
mer to 73% in winter. For the same period, comparison of
ACSM NR-PM
1
and FRM-PM
2
.
5
shows a good correlation
r
2
>0.80). The tightercomparisons during summer(
r
2
>0.8)
compared to winter (
r
2
=∼
0
.
6) might suggest meteorologi-
cal influence on totalmass measurements duetopositive bias
from filter measurement during colder seasons (Solomon et
al., 2003a, b).
4 Discussion
4.1 IntercomparisonbetweenACSM instruments
Slopes of the linear regression from UNC ACSM vs. GIT
ACSM (Table 4) suggest percentage differences of speci-
atedmass concentrations are 4to38% between twoindepen-
dentACSM measurements. The SO2
4−
differenceof25% can
be attributed to uncertainty in the instrument RIE fitting re-
sults. The percent uncertaintyofthe fitting approach is larger
(28%) than calibration results (7–14%) recently conducted
at both ACSMs. Larger differences of Cl
measurements
(79%) are due to its significantly lower concentration in
Atlanta during the entire sampling period. This resulted in
weaker correlation between the two instruments although
both instruments capture similar large peaks of Cl
forsome
periods.
4.2 OM/OC ratio
The OM/OC ratios derived from the regression linear slopes
are larger than mostOM/OC ratios previously reported inthe
literature. These values are significantly higher than the tra-
ditionally used values of 1.6 for urban aerosol and 2.1 for
non-urban aerosol (Turpin and Lim, 2001; Lim and Turpin,
2002; Russell, 2003). They are also larger than those found
from recent HR-ToF-AMS intercomparisons withthe Sunset
OC/EC analyzer that report
1.8 from September in Pitts-
burgh(Zhangetal., 2005a), 1.8and1.6from summerand fall
in Tokyo (Takegawa et al., 2005), 1.41–2.15 from March in
Mexico (Aiken et al., 2008), 2.59 from August in New York
City (Sun et al., 2011) and 3.3 from summer in Pasadena
(Hayes et al., 2013). Studies in Atlanta also reported a high
variability of OM/OC ratio, from 1.23–3.44 in August 1999
(Baumannet al., 2003) and 1.77 in December1999 to 2.39 in
July 1999 (El-Zanan et al., 2009). These suggest variability
in OM/OC ratios based on location, time and meteorological
conditions, and/or that the ACSM is measuring organic mass
much higherthan it should since it is using AMS-based RIE
values fororganic (i.e., RIE =1.4)rather than those that have
been explicitly measured for ACSM instruments.
www.atmos-meas-tech.net/7/1929/2014/
Atmos. Meas. Tech., 7, 1929–1941, 2014
1936
S. H. Budisulistiorini et al.: Intercomparisonof an ACSM with ambient fine aerosol measurements
30
20
10
0
20
15
10
5
0
12
8
4
0
A-CSM NR-PM1 (µg m)
12
8
4
0
25
20
15
10
5
0
16
12
8
4
0
FRM PM
1
(µg m
-3
)
r
2
=0.89
f(x)=(-1.44±1.35)+(1.73±0.14)x
(a)
r
2
=0.86
f(x)=(-0.06±0.54)+(0.81±0.06)x
(b)
r
2
=0.91
f(x)=(0.40±0.65)+(1.10±0.08)x
(c)
30
20
10
0
20
15
10
5
0
12
8
4
0
A-CSM NR-PM1
12
8
4
0
25
20
15
10
5
0
16
12
8
4
0
FRM PM
2.5
(µg m
-3
)
r
2
=0.88
f(x)=(-1.55±1.42)+(1.53±0.13)x
r
2
=0.83
f(x)=(0.05±0.60)+(0.70±0.06)x
r
2
=0.76
f(x)=(-1.07±1.30)+(1.07±0.13)x
Figure 2.CorrelationofACSMNR-PM1measurementswiththoseofFRMPM
1
and PM
2
.
5
methods during(a) winter,(b) spring, and(c)
summer 2012, respectively.
The large OM/OC ratios might alsobe attributed to under-
estimation of OC due toevaporation of semi-volatile organic
compounds (SVOCs) from the Sunset OC analyzer, and/or
overestimation of OC due to condensation of SVOC or ad-
sorption of VOC on the filter (Couvidat et al., 2013). This is
reflected in a large offset at the Sunset OC (Figs. S4 and S5
in the Supplement). The presence of a denuder on the inlet
of Sunset OC/EC analyzer, for example, might cause evapo-
ration of particulate OC from the collection filter due to re-
partitioning of SVOC after removal of gaseous organics by
the denuder (Grover et al., 2008). Also, 20% of Sunset OC
uncertainty (Peltier et al., 2007) together with ACSM uncer-
tainty might propagate the OM/OC ratio.
Overestimation of OM by the ACSM could arise from un-
derestimation of the RIE value of organic species. The RIE
values used in this study are based on experiments examin-
ing a suite of organic standards using the AMS instrument
(Jimenez et al., 2003; Alfarra et al., 2004). Since the two
instruments rely on the same vaporizer and ionization con-
ditions (i.e., electron ionization), it was assumed that the
RIE values for organics should be similar. However, based
on the high OM/OC ratios observed from our intercompar-
ison study, sets of authentic organic standards covering a
wide range of chemical classes as well as secondary or-
ganic aerosol generatedfrom laboratory experiments,such as
isoprene-derived secondary organic aerosol (SOA) (Kleindi-
enst et al., 2006; Lin et al., 2012), need to be systematically
analyzed in future work in order to determine the RIE value
for organics in the ACSM.
The large OM/OC ratios might also suggest photochemi-
cally, well-aged, and well-mixed air masses contain particle-
phase organics that are more oxygenated and less-volatile
compared to more stagnant air masses where less polar
and more volatile organics can be found possibly due to
incomplete photochemical oxidation leading to more labile
functional groups and intermediates. An offline polarity-
based analysis suggested values of 1.9 to 2.1 for OM/OC
ratios due to aging and oligomerization processes in the at-
mosphere(Polidori, 2008). Inaddition, water-soluble organic
aerosol was observed to have higher OM/OC ratios than
that of less water-soluble organics, ranging from 2.1–2.3 in
the Great Smoky Mountains to 3.3 in downtown Los Ange-
les (Turpin and Lim, 2001). Furthermore, ratios of 2–3.12
were observed from organic fractions that could not be ex-
tractedusing organicsolvent (Polidori, 2008), indicatingthat
compound-specific polarity might be related tosources of or-
ganic aerosol. Therefore, besides overestimation of OM by
ACSM as noted above, high OM/OC ratios might indicate
that the organic aerosol is more water-soluble in nature.
4.3 SO
2
4
andNH
+4
measurements variations
Sulfatemeasurements from ACSMandthe filtershow agood
trend (
r
2
>0.7, see Fig. S7 in the Supplement) for the De-
cember period, suggesting that the large discrepancies ob-
served between the ACSM and JST data might be caused
by some unknown issues with either the JST continuous
measurements or ACSM during this sampling period. Both
ACSM and continuous measurements show that the slopes
of NH
+4
measured versus NH
+4
predicted (neutralized) are
slightly less than 1 (Fig. S8 in the Supplement). This sug-
gests during both summer and fall 2011, the aerosol was
slightly acidic. Investigation of the period where correla-
tion between ACSM and collocated measurement is low in
fallseasonsuggestsome organic interferences (hydrocarbon-
like organic aerosol/HOA)in sulfate fragments, in particular
m/z
81 (Fig. S9 in the Supplement).
Atmos. Meas. Tech., 7, 1929–1941, 2014
www.atmos-meas-tech.net/7/1929/2014/
S. H. Budisulistiorini et al.: Intercomparison of an ACSM with ambientfine aerosol measurements
1937
40
30
20
10
0
C- oncentration (µg m; µm cm)
10/21/2012
10/31/2012
11/10/2012
11/20/2012
Date and Time (Local)
40
30
20
10
0
A-CSM NR-PM1 (µg m)
25
20
15
10
5
0
SEMS PM
1
(µm
3
cm
-3
)
ACSM NR-PM
1
mass conc.
SEMS PM
1
volume conc.
(a)
(b)
r
2
=0.89
f(x)=(0.31±0.12)
+(1.59±0.01)x
Figure 3. (a)Timeseriesand(b)correlationoftotalaerosolmassmeasuredbyACSM(NR-PM
1
)
and SEMS DMA/MCPC during periodof
17 October to 20 November, 2012. Aerosol density wasestimated from the linear regression slope of 1.59 multiplied by 1.10 to account for
the 10% of elemental carbon (EC) component that is not measured by ACSM. This results in estimated aerosol density of 1.75gcm
3.
Previous comparison of SO2
4−
measurements from the
Thermo Electron 5020 Sulfate Particulate Analyzer with
filter-based methods from laboratory and field studies ob-
served good correlations (i.e., slope derived from field study
was closer to 1 than that of laboratory study)(Schwab et al.,
2006). It should benoted that Schwab et al. (2006)suggested
that the slope differences are due to ambient SO
2
4−
from
the field study being catalytically converted to SO
2
faster
than the laboratory-generated SO
2
4−
.During this study, the
ACSM SO2
4−
measurements discrepancies are 4–44% com-
pared to that of the continuous modified HSPH sulfate an-
alyzer, with the largest difference occurring during colder
months (fall season). This difference is within the expected
accuracyoftheACSMmeasurements, butsincetheJST con-
tinuous SO
2
4−
values are obtained after adjusting to the filter
data, the bias could be due to artifacts from the filter data.
4.4 Discrepancies of NO
3
measurements
ACSM NO
−3
measurements are based on the measured
m/z
30 and
m/z
46 ion signals. Positive biases at
m/z
30
are possibly due to contributions to this ion from NO
+
frag-
ments of organic nitrates and/or contributions from organic
CH
2
O
+
ions. A detailed investigation of the interference of
m/z
30is providedinthesupplementalsection. The relation-
ship of estimated excess signal of
m/z
30 linked to organic
and oxygenated organic aerosol is found to be heteroscedas-
tic. Thus, oxygenated organic species could not be suggested
to directly influence nitrate fragments.
The continuous NO
−3
data are adjusted to the integrated
NO
−3
data, which can impose measurement biases, espe-
cially forsemi-volatile compounds such as NO
−3
.Heringand
Cass (1999) reported lower aerosol NO
−3
mass from Teflon
filters compared to that from denuded nylon filters. For this
study, the PCM filter samples utilized both Teflon and ny-
lon filters downstream of a denuder in order to account for
NO
−3
losses. Previous SEARCHresults have compared NO
−3
measurements with parallel systems: one with a Teflon pre-
filterand nylon backup filter (PCM1) and the other with just
anylon filter (PCM2) (Edgerton et al., 2005). Both systems
were denuded to remove artifacts of HNO
3
and NH
3
,thus
thermodynamics should favor metathesis of NH
4
NO
3
.Sum-
mer results showed that PCM1 agreed with PCM2 within
5% and that >95% of the NO
3
from PCM1 was on the ny-
lon backup filter. Fall results showed agreement within 10%
and with >90% on the nylon filter (Edgerton et al., 2005).
While the use of nylon backup filters likely minimized NO
−3
losses duringsampling, additional lossesduringfilter storage
andconditioning before off-line chemical analysis cannot be
ruledoutandcould have contributedto the observed discrep-
ancy.
Changes inmeteorological conditions from summerto fall
might influence the equilibrium partitioning behavior of ni-
trogenous compounds. Low temperatures and high relative
humidity (RH) in the fall could create thermodynamic con-
ditions that favor the partitioning of gaseous NO
−3
to the
aerosol phase (Hennigan et al., 2008; Rastogi et al., 2011).
The fact that the observed NO
−3
discrepancies are larger in
the fall than the summer is consistent with evaporative loss
of NO
−3
from the filter samples and reflected in the filter-
adjusted continuous data.
In summary, it is unclear if the higher ACSM NO
−3
load-
ings reflect true NO
−3
levels which include contributions
from organic nitrate notcaptured by JST NO
−3
,orifit isfrom
inaccurate subtraction of
m/z
30 originating from oxidized
organic aerosol. Also, it is possible the discrepancy may be
due to the underestimation of JST NO
−3
due to volatility
losses from the filters which are used to scale the JST NO
−3
data. It is likelysome combinationof all ofthe above, which
cannot be clearly determined from this data set, explains the
differences between NO
−3
measurements.
4.5 Total mass measurements variations
ACSM PM
1
is sum of ACSM NR-PM
1
(i.e., organic and in-
organics) plus EC measurements from JST site. This study
shows that total mass differences between ACSM PM
1
and
TEOM PM
2
.
5
are 50–110%. Previous intercomparisons of
www.atmos-meas-tech.net/7/1929/2014/
Atmos. Meas. Tech., 7, 1929–1941, 2014
1938
S. H. Budisulistiorini et al.: Intercomparisonof an ACSM with ambient fine aerosol measurements
the same instruments in summer in Beijing suggested that
ACSM NR-PM
1
measured
30% less than TEOM PM
2
.
5
(Sun et al., 2012). Since the ACSM PM
1
mass is a sum of
species concentrations, the discrepancies in species specific
intercomparisons described above result in high discrepan-
cies of PM
1
mass. Uncertainties in RIE values, particularly
for organic species, may be partly responsible for overesti-
mation of certain species resulting in overestimation of NR-
PM
1
mass. On the other hand, loss of semi-volatile species
from the filters (which are used together to adjust TEOM
loadings) couldalso result in lower TEOM PM
2
.
5
concentra-
tion. This is supportedby the fact that infall, when the mete-
orological conditions favor semi-volatile organic aerosol en-
hancement, the slope of the ACSM PM
1
to TEOM PM
2
.
5
is
much higher than that insummer(i.e., slope of 1.80 in fall to
1.19 in summer).
Differences between NR-PM
1
masses measured by the
ACSM and PM
1
mass measured by the FRM method are
about 10–73%, with the lowest difference observed in the
summer data set (Fig. 2; Table S2 in the Supplement).
Discrepancies between the ACSM and FRM methods are
larger during winter and spring compared to that of sum-
mer, and the direction of the discrepancy is different in
spring (ACSM<FRM) as compared to winter and summer
(ACSM>FRM). Thismightbeduetopositiveartifacts ofthe
filter sampling method, which are likely enhanced in colder
months (Solomon et al., 2003a, b). On the other hand, uncer-
tainties in RIE values may also result in inaccurate ACSM
chemical constituent measurements leading to over- or un-
derestimation ofACSM NR-PM
1
mass.
The slope resulting from the intercomparison of ACSM
NR-PM
1
mass concentration and SEMS PM
1
volume con-
centration can be used to estimate aerosol density. Compar-
ison suggests a slope of 1.59 (Fig. 3); however, this num-
ber will be larger when the refractory components (i.e., EC)
are added to NR-PM
1
.Since the EC measurement for this
period (October–November 2012) are not available, we es-
timated that EC contributes about 10% to total PM based
on available data (i.e., October–November 2011). Hence,
the estimated aerosol density in Atlanta is 1.75gcm
3
for
fall 2012. In addition, we estimated typical dry aerosol den-
sity based on average particle composition of 60.1% of or-
ganics, 30.8% of inorganics, plus 10% of EC, and the as-
sumption of organic, inorganics, and EC densities are 1.2,
1.77, and 1.77gcm
3
(Zhang et al., 2005b, and references
therein). This approached resulted in typical dry density of
1.61gcm
3
.These numbers are consistent withrecent ambi-
ent aerosol density estimations, such as 1.61gcm
3
in Bei-
jing (Hu et al., 2012) and1.46gcm
3
in Pasadena (Hayes et
al., 2013).
5 Conclusions
This study aims to compare species and total mass measure-
ments from the ACSM tothe collocatedmeasurements at the
JSTsite (i.e., ACSM, JST continuous andfiltersamplers, and
FRM filters) overdifferent seasons. Mass concentrations ob-
tained from the two ACSMs agree within 4–38%, except for
Cl
.Overall, the percentage differences ofACSM speciated
mass concentrations agree within4–51% from the SEARCH
network measurements, except for NO
−3
(77–114%). Com-
parison of ACSM OM to JST Sunset OC yielded OM/OC
ratios of 4.85 and 3.85 for summer and fall periods, respec-
tively. Discrepancies betweenACSM PM
1
andTEOM PM
2
.
5
are 50–110%, while discrepancies betweenACSM PM1and
FRM PM
1
are 10–73%. Estimated aerosol density based on
ratio of mass to volume concentration is 1.75gcm
3
.
Discrepancies found in the intercomparisons ofthe ACSM
and the collocated measurements might be explained by the
following: (1) RIE values of organics might have dependen-
cies on sources oforganic aerosol; (2) possible interferences
from organic and organic-nitrate-specific fragments to the
m/z
30 ion signal that constitute ACSM inorganic NO
−3
sig-
nal; and (3) evaporative losses of semi-volatile species from
the filter measurements used in the collocated continuous
measurement adjustment. Future work shouldsystematically
examine all of the possibilities. Additionally, calibration of
the continuous instruments used at monitoring sites should
also be routinely checked with a standard aerosol in addition
to the standard gas calibration that is typically performed.
The Supplement related to this article is available online
atdoi:10.5194/amt-7-1929-2014-supplement.
Acknowledgements. WethanktheElectricPowerResearchInsti-
tute (EPRI)for theirsupport. We thankJerry Brown and Mike Boaz
of Atmospheric Research and Analysis (ARA) for their mainte-
nance of the ACSM at the JST site. We thank Fred Brechtel for
his input on SEM-MCPC operation and data analysis. S. H. Bud-
isulistiorini is supported by a Fulbright Presidential Fellowship
(2010–2013) for attending the University of North Carolina at
Chapel Hill. We also thank Wendy Marth for her assistance in
setting up the ACSM at the JST site and Sriram Suresh for his
assistance in ACSM fitting RIE formula derivation. GIT ACSM
measurements were supported by US EPA grant # RD83479901
as part of the Emory/Georgia Tech Clean Air Center (SCAPE).
Regardingthe GIT ACSMdata; the contents ofthispaperare solely
the responsibility of the grantee and do not necessarily represent
the official views of the US EPA. These agencies do not endorse
the purchase of any commercial products or services mentioned in
the publication.
Edited by: P. Herckes
Atmos. Meas. Tech., 7, 1929–1941, 2014
www.atmos-meas-tech.net/7/1929/2014/
Documents you may be interested
Documents you may be interested