JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 104,NO. D17, PAGES 21,343-21,354,SEPTEMBER 20, 1999
A new measurement technique of peroxyacetyl nitrate at parts per trillion by volume levels: Gas chromatography/negative ion chemical ionization mass spectrometry Hiroshi Tanimoto, JunHirokawa, YoshizumiKajii, andHajime Akimoto ResearchCenterfor AdvancedScienceandTechnology,Universityof Tokyo,Tokyo
Abstract. A newtechniqueformeasuring peroxyacetyl nitrate(PAN) at partspertrillion by volume(pptv)levelshasbeendeveloped usinggaschromatography with negativeion chemicalionizationmassspectrometry (GC/NICI MS). Four fi'agment ions, CH3C(O)O- (mass-to-charge ratios (rn/z)=59),NO3- (rn/z=62),CH3C(O)OO- (rn/z=75),andNO2- (rn/z=46),wereidentifiedby NICI massspectra.NO3- (rn/z=62)wasusedin selectedion monitoring(SIM) modeamong thesefi'agment ions,sinceit gavethebestsignal-to-noise (S/N) ratio,anda potentialinterference fromisopropylnitrate(IPN) wasexcluded.Optimizationof massspectrometry anduseof capillary gaschromatography enabledus to detectPAN at a retentiontime of 2.6 + 0.1 min in ambient air mixingratiorange.The presentdetection limit is 15 + 4 pptv(S/N=3), whichis comparable to thoseof previouslypublishedGC/ECD methods,with goodlinearityat pptv levels.The instrumental accuracy andprecisionareestimated to be +20% and+ 15%,respectively. OtherPAN analogs(i.e., peroxypropionyl nitrateandperoxymethacryloyl nitrate)werealsomeasured within 10 min. Exploratoryambientair measurements weremadeat our laboratoryin a suburban areaof Tokyo, Japan.Theseresultsdemonstrated the performance of GC/NICI MS forambientair measurements.
1. Introduction
Peroxyacetylnitrate (PAN) [CH3C(O)OONO2] is a photochemicaloxidation productofhydrocarbonsin the presenceof NO x,whichwasfirst discoveredin the Los Angelesatmosphere by Stephenset al. [ 1956]. One ofthe importantroles of PAN in the chemistryofthe atmosphere is that it works asa NOx reservoir.PAN is highly stableatlow temperature, henceit is capableof transportingNO•in organicformto the middleand upper troposphereand to remoteregions [Singh, 1987]. In contrast, PAN canbethermallydecomposedto supplyNO• for netozone productionin the warmerboundarylayer afterlong-rangetransport. A numberof airborne measurementshave revealed that PAN is ubiquitous and an importantconstituentin the atmosphere.Its mixingratioscanevenexceedthoseof NO•in clean, remoteregions [e.g., Singh et al., 1985, 1986; Ridley et al., 1998]. PAN hasbeenextensivelymeasuredin ground-basedobservations in clean backgroundand regionally polluted atmosphere [Roberts, 1990; Altshuller, 1993; Roberts, 1995]. Peroxyacetylnitratesandits homologuescanbeexcellentindicatorsofphotochemicalactivity; henceboth types of measure-
mentindicatea considerable associationwith photochemical formationof ozone. Observationsin remoteregionshave revealedthat PAN accumulationis relatedto springtimeozone maximum in the backgroundatmosphere[Penkettand Brice, 1986; Jaffe, 1992; Bottenheim et al., 1994; Beine et al., 1996, 1997]. Simultaneous measurements of ozone, PAN, peroxypropionylnitrate (PPN) (CH3CH2C(O)OONO2) , and Copyright 1999bytheAmericanGeophysical Union. Papernumber 1999JD900345. 0148-0227/99/1999JD900345509.00
peroxymethacryloyl nitrate (MPAN) (CH2=C(CH3)C (O)OONO2)havebeenusedto estimatethe relativeimportance of anthropogenicand biogenichydrocarbonemissionsfor the photochemical formationofozoneovertheeasternUnited States [Williamset a/., 1997; Robertset a/., 1998a,b; Nouaimeet al., 1998]. Recentglobal chemicaltransportmodel(GCTM) studies, whichsimulatedglobal distributionof NO• andPAN andtheir rolesinthetroposphere, reportthatPAN redistributesNOxand thus affectsphotochemistryin remoteregions[Moxim et al., 1996].
Mixing ratiosof PAN in theatmosphere havebeenmeasured usingvarioustechniquesincluding Fouriertransform-infrared spectroscopy (FT-IR), gas chromatography with electroncapturedetection(GC/ECD) [Gaffneyetal. , 1989; Roberts,1990], andgaschromatography with luminol-chemiluminescent detection (GC/LCD) [Gaffney et al., 1998]. GC/ECD is the most widelyusedmethodforambientairmeasurements ofPAN, especially at remoteandrural sitesbecauseof its highersensitivity. Recentstudiesreportvery low detectionlimits (equivalentto or lower than 15 pptv), which allow measurement of PANs at remotesiteswithout preconcentration[e.g.,Bottenheimet al., 1994; Schrimpf et al., 1995; Robertset al., 1996; Beine et al., 1997].
However,otherhigh-electronaffinitycompounds suchashalogenatedcompoundsandorganicnitratesmaybeeluted closely with PAN, possibly causingpoor accuracy.In addition to thesecompounds,oxygenand watervapor significantlyaffect measurements of PAN by GC/ECD [Miiller and Rudolph, 1989; Schrimpfet al., 1995]. To minimizethese effectsin the GC/ECD system,a preseparationcolumncanbeusedto remove oxygenbeforeit reachesthe main column combinedwith a backflushtechnique,which preventswatervapor •omentering the maincolumn[Miiller and Rudolph, 1989; Schrimpf et al., 1995].Reportedtimeresolutionby GC/ECD measurements is 5
21,343
21,344
TANIMOTO ET AL.: PEROXYACETYL NITRATE MEASUREMENTTECHNIQUE
sufficientsensitivity. Somestudies using GC/NiCI MShave re-
to 30 min [e.g.,Grosjean et al., 1993a; Schrimpfet al., 1996; Robertset al., 1996] for ground-basedmeasurements, whereas fastertimeconstants(1.5 to 6 min) aredescribedfor airborneinstruments[Singhet al., 1994; Schrimpfet al., 1995; Williams et al., 1997; Gaffhey et al., 1998].
portedtracegas measurements other than PAN in ambientair. Measurements of >C3alkyl nitrateshavebeenreportedby Atlas [ 1988] at a rural site and over the remotePacific Ocean and following observationsby BeRneetal. [ 1996] in Alaska.Theidentiffcation of nitroaromaticsin diesel exhaustparticulatehas beendescribedby Newtonet al. [1982]. Specificdetectionsof halocarbons(e.g.,CFCs,hydrochlorofluorocarbons (HCFCs), hydrofluorocarbons (HFCs), halons,and methylhalides,etc.) andorganicnitrates(e.g.,alkyl nitratesand multifunctionalnitrates)using GC/NICI MS are being investigatedby the Na-
Radioactive materials (commonly 63Ni foil)used foranion sourcein ECD are under strict regulation regardingfield deploymentin somecountries,includingJapan.Therefore, to date, a rather limited number of field measurements of PAN
and its
homologues havebeenmadein the East Asian Pacificrim and over westernPacific regions(for moredetails,seeSingh et al. [ 1996],Kondoet al. [ 1997], Singhet al. [ 1998], and Watanabe
tionalCenterforAtmosphericResearch (NCAR) (E. Atlas,per-
et al. [ 1998]). We usedgaschromatography/negative ion chemicalionization massspectrometry (GC/NICI MS) to measurePAN in ambientair. Negativeion productionby thermalelectronattachment in the NICI MS employedhereprovidesthe possibility ofhigh sensitivity for PAN. Potential interferencesfromoxygen,water vapor,otherCFC-type compounds,and organicnitratescanbe avoidedin the selectedion monitoring(SIM) technique.ThereforeNICI MS in the SIM mode is regardedas a moreselective techniquethan ECD with high sensitivity, thoughNICI MS is a morefragiledetector,which mayneedmoreintensiveand frequentmaintenance (e.g.,filamentreplacement) than ECD. NICIMS had not been applied to PAN detectionuntil very recently.Spicer et al. [1998] and Srinivasan et al. [1998] reportedambientairmeasurements andlaboratoryinvestigationof PAN usingnegativeion chemicalionization massspectrometry, respectively.Measurementsof PAN by massspectrometryusing other ionization methods(i.e., electron impact ionization [e.g.,BertmanandRoberts, 1991], photoionization[Kacmarek et al., 1978], and positive ion chemicalionization [Pate et al., 1976]) have been reported previously; however, these techniqueshave beenused in laboratoryexperimentsto investigate the structureand chemical properties of PANs but have not beenusedforambientair measurements, probablybecauseof in-
sonal communication,1999). In this study, we optimized the GC/NICI MS instrumentto achievehigh sensitivityand selectivityfor PAN. The GC/NICI MS instrumentwasevaluatedusingourcalibrationmethod;linear responseat pptv levels, detectionlimit, overall accuracy, andprecisionoftheinstrumentwere investigated.Potentialinterferencefromisopropylnitrate (IPN) was evaluatedand then eliminated. A detection scheme was utilized to determine PAN
analogs(PPN andMPAN). Ambient measurements were made in the universitysettingto testthe instrument's performance as a new measurement technique for PANs in ambientair. The resultsandinterpretationof theseobservationsarereported.
2. Experimental Setup A schematic diagramofthe experimental systemusedherein is shownin Figure 1. A Hewlett Packard5980/5989A model
gaschromatograph/negative ionchemical ionizationmassspectrometer was used for the detection of PAN. 2.1. PAN Source
PAN wassynthesizedin liquid ntridecaneby thenitration of peracetic acid[Gaffheyet al., 1984].No furtherpurification
Vent D'• F........... I•l--1--• Personal Computer •G
Vent Vent
Pe,sona, Compute, ..........................
,
I
,
50SCCM
IMFCl
Cb_NOx I0112OSOOM ,02(99.9999%) Zero gas generator
•
I
!. QP IS • ---• •BG [-1(• oll••--•hhhhhh• IW
,I I
I I '
•
I I,
I•
II
I GC-MS
I MMFC•
"P I Iinto"øo •
I..
. .
• I I
I ;•
BPV
•
B•vent [ PG •
1•',I•I•-5SLM II
I JL•I• II
- R N,,on
½Mp,ng%ai•oI I filter II ........ III ....."......... II Thermostat f
' • valve • I Needle Reagent gas
(CH4;99.995%)
• gas • Carrier (He;99.9999%)• •
- - '
•_• •
1 Ambient air : Stopvalve
MSP: Metal bellows pump PG:Pressure gauge BG'' Baratron gauge
•: •ar r pump u•: umus•onpump
DPP: Diaphragmpump
PAN/ n tridecane MFC:Mass flow controller Figure1. Schematic diagram oftheexperimental setup.Thesystem consists offivecomponents: aperoxyacetyl nitrate(PAN) source, a chemiluminescent NOranalyzer, asix-portgassampling valve,a gaschromatograph, anda quadrupolemassspectrometer.
TANIMOTO ET AL.' PEROXYACETYL NITRATE MEASUREMENT TECHNIQUE
21,345
ofPAN wasperformed. Purity ofthe PAN standardwas evaluat- dardtemperatureand pressure)maintainedby a massflow controller, and the rest was exhaustedby a metal bellows pump ed by gas chromatography/mass spectrometryanalysis describedin detail in section 2.4. PPN and MPAN were syntheplacedbeforethe six-port gassamplingvalve (seeFigure 1). Air sizedby similarmethodsto PAN but startingwith correspond- sampleswere directedto a gas chromatographby whole air inwasdirectly connected ing acid anhydrides.PPN was synthesizedaccordingto the jection. A quadrupolemassspectrometer to the capillary columnvia a temperature controlled GC-MS inmethoddescribedby Nielsen et al. [1982]. The solution was terface. washedby ice/water,anhydrousMgSO4was added,and isolaTwo typesof capillarycolumns,which havesimilarphysical tion was not performed,similar to PAN. MPAN was syntheproperties(i.e., 0.25 rrrnID, 30 mcolumn length, and 0.25 sizedby the methodreportedby Bertmanand Roberts [1991]. film thickness)but differentkinds of liquid stationary phases, No isolation was madeto obtain a pure standardfor MPAN for the presentpurposes.Each flask containing PAN, PPN, and weretested:DB-210 (J&W ScientificCompany,Inc.) and SPB1701 (SupelcoCompany,Inc.). DB-210 andSPB- 1701 arefused MPAN was storedfrozenin a refrigeratorat-20øC until used. silica capillary columns coated with trifruoropropyl and A capillary diffusionsourceusedin this study is similar to cyanopropyl-phenylbased liquid phase, respectively.Both thosereportedelsewhere[Roberts et al., 1988; Gregory et al. types of liquid phaseshave been usedsuccessfullyfor PANs 1990; Fehsenfeldetal., 1987; Ridleyet al., 1990]. A capillary andalkyl nitrate measurements [Schrimpfet al., 1995; Roberts diffusiontube (Gastec Corporation Company,Inc.) typically et al., 1996, 1998a, b; Nouaime et al., 1998]. Both types were usedwas5 cmlong with a volumeof approximately5mLand ID of 5.5 mm The diffusiontube was placed in a stainlesssteel foundto providealmostthe sameresultsforseparationand sensitivity for PAN but not forretention time.Sincethe SPB- 1701 flask (300 mL volume),the inside of which was coated with Halocarbonwax(HalocarbonProductsCompany,Inc., number. columngave shorterretention time for PAN than the DB-210 column,the formerwas preferredfor the purposeof allowing a 15-00). This wax was usedin orderto reduceloss of PAN on shorter time resolution of measurements in ambient air. Therefore the stainlesssteel surface.The flask was kept in an ice/water only resultsobtainedby theSPB-1701 aredescribedin this padewerat 0øC,which was controlled by an electric thermostat. per. PAN-freeair was suppliedby a zero-gasgenerator (ThermoEnvironmentalInstruments,TECO model 111) with an oil-free air Temperaturesofthe samplingloop,the capillarycolumn,and compressor. Initial mixing ratio of watervapor in the zero-gas theGC-MS interfacewerekept at 30øCin orderto minimizetherstreamwas---0.7%, andCO/CO 2wasremoved.Thismoisturewas mal decompositionof PAN before it was introduced into the massspectrometer.Ultrahigh purity (UHP) grade He (Taiyo removedupstreamofthe PAN sourcewithamolecularsieve4A Toyo SansoCompany,Inc., 99.9999%) was usedas carriergas to preventwater vaporfromcondensingin the source.The flow rateoverthe capillarydiffusiontube wascontrolledto 1-5 stan- with no furtherpurification. dardliters per minute(sLpm)by a massflow controller(Kofloc Company, Inc.). Pressureinsidethe sourcewascontrolledby a 2.3. Negative Ion Chemical Ionization Mass Spectrometry back-pressure valve (Kofloc Company,Inc.) andmeasured by a The massspectrometer consistsofthree components:an ion pressure transducer(MKS Company,Inc., Baratrongauge622). source,a quadrupolemassanalyzer,and anion detector.ChemiThissourceprovided astableoutput ofPAN foroptimizationof calionizationin negativeion modeis performed typically atan the GC/NICI MS. ion sourcepressureof 1.0 to 2.0 torr as shown below. The mixingratio of PAN in the gas streamwas determined by a NO-O3 chemiluminescent NO,• analyzer (TECO model e•,+ CH4 -*e• + e• + CH2',CH;•,CH•' (1) 42CTL) with a molybdenumconverterheld at 320øC. Fluctuations ofPAN signals determinedbythe NO,•analyzerwere•-5% e• +M--.e 2+M (2) for 12 hours at the level of 20-50 ppbv. PAN solution, which X+e; -* X-' (3) had been storedat -20øC for severalweeks in the refrigerator, wasusedin the experiments for optimization.Although the soX-' + M -* X- + M (4) lution usedcould include impurities,this did not causeserious uncertaintyfor the optimization. where, e-p,e-•e th,andM denote accelerated primary electrons, To obtain massfragmentationpatternsand retentiontimes, secondaryelectrons,thermal electrons,and third bodies (reindividual samplesof PAN, PPN, MPAN, methyl nitrate, and agent gasandcarrier gasmolecules), respectively. X,X-*,andXisopropylnitrate,a portion (•-10 [tL)was injectedby a syringe indicatemoleculeswith high electroncapturerate,excitedions into a 1-L tedlarbag (Aldrich ChemicalCompanyInc.) thathad aftercapturingthermalelectrons,andstabilizednegativeions, beenfilled with pureair.In this manner,samplesoftypicalppmv respectively. mixingratioswere obtained. Samplesin the tedlar bag werediElectrons, which are emitted from a filament and accelerated rectly introduced to the GC/NICI MS via a six-port gas samin an electricfield, collidewith reagentgasmoleculesto propling valve. Mass spectrawere obtained in the total ion moniducesecondaryelectrons(see(1)). The secondaryelectronsare toring(TIM) modewith ascanrangeofgenerally m/z=O to 200, thermalized via collisionswith reagentgasandpossibly with wherem/z is mass-to-charge ratio. carriergasmolecules(see(2)). Whenmolecules(X) with a high electroncapturerate are introducedinto the ion source,they 2.2. Gas Chromatography capturethermalelectronsto formnegativeions (electroncapAn air streamcontaining PAN was introduced into a gas ture-negativeion chemicalionization) (see(3) and (4)). Bechromatographby means of a six-port gas sampling valve tweentheion sourceandthe quadrupolemassanalyzer,two ion equippedwith a samplingloop of 0.25 mL volume (Hewlett lensesareinserted.Forthedetectionofnegativeions,ahighenPackardCompany,Inc.). A small part of the gas flow fromthe ergyconversiondynodewasused to convertnegativeions to PAN sourcewas sampledby adiaphragmpumpat a flow rateof positiveones,followedbydetectionwith an electronmultipli-
50cm3 min -1STP (standard cubic centimeter perminute atstan-
er.
21,346
TANIMOTO ET AL.: PEROXYACETYL NITRATE MEASUREMENT TECHNIQUE
In this work, methane(NipponSansoCompany,Inc.,99.995 %) and isobutane (Takachiho ChemicalIndustrial Company, Inc.,99.5%) weretestedasreagentgases,describedin detail in thenextsection.Ion sourcetemperature,impactelectronenergy, emissioncurrent,repeller voltage, and voltage applied to two ion lenseswere optimized.Briefly, the repeller,which is positively charged,pushesnegative ions toward the quadrupole massanalyzer.Thetwo ion lensesaredesignedto draw negative ions, which areproduced in the ion source,into the massanalyzer moreeffectively.In the quadrupolemassanalyzer,tempera ture,massresolution,and appliedvoltagesto the electronmultiplier andthe high-energyconversiondynodewereoptimized. In the SIM mode,ion integrationtime (massscanrate) was adjustedto obtain,simultaneouslybetterS/N ratioandchromatographicpeak shape.
2.4. Calibration Technique and Its Evaluation
To evaluatethe potential of the GC/NICI MS techniquefor measuringPAN at ambientair mixing ratio levels, characterization of the instrumentbelow 1 ppbv was performed.PAN at pptv mixingratioswas producedby adynamicdilution method
thatwascomposed oftwomass flowcontrollers (200cm 3min-• STPand 20 sLpmfull scale).The initial mixingratio of PAN in thegasstreamwas 10-20ppbv, and it wasdiluted by a factorof 5 to 1000 to produce pptv PAN levels. Polyfluoroalkoxy (PFA) teflon was used for tubing, tube fittings, and valves downstreamof the PAN sourcein the calibration system.In order forthe mixingratios of PAN at pptv levels to be measured, thesensitivityof aTECO modelchemiluminescent NO xanalyzer(TECOmodel42CTL)(hereinafterreferred to theCL-NO xanalyzer) was improved as follows. Oxygen (Taiyo Toyo Sanso Company,Inc.,99.9999%) was usedfor O3generation.Theflow rateforbothoxygenandsampleairwas controlledby massflow controllers(Kofloc Company,Inc.). The sampleflow was 0.8
sLpm, sligh3tly less 1than theusual and theoxygen flowwas kept
at 120 cm min- STP. A more capable diaphragmpump (VacuubrandGmbH,type MD4) wasusedto reducethechamber pressure.The pressurein the reactionchamberwas 27 + 1 torr. Temperature of the molybdenumconverterwas kept at 320øC. By thesemodifications,the detectionlimit for NO and NO• was improvedto 50 pptv (30) for 1 min average.In this manner, PAN at pptv levels in the calibrationgasstreamwas directly measured by the CL-NO• analyzer.However,only dataabove 100 pptv were used for calibration of PAN in order to obtain
NO 2 maybe presentin the headspace abovethe diffusion tubecell because of its lowersolubility in ntridecaneandlargerdiffusionratethanPAN [Robertset al., 1988; Singhet al., 1990].The quantity of NO 2 in the exit PAN streamwas measuredby a NO-O3chemiluminescent NO• analyzer(EcoPhysics,modelCLD 770ALppt) afterphotolytic conversion(Eco Physics,modelPLC 760). Photolyticconverters providemuch moreselectivemeasurement of NO 2thanheatedmetalconverters. The interference fi'om P AN deduced fi'om thermal PAN de-
composition in the photolysissystemin this type of instrumentswas described by Beineet al. [ 1997];hencetheNO2signalswereconsideredto be an upperlimit. The resultshowed that lessthan 1% of NO 2 that was presentin the gas stream flowed out ofthe PAN source.
CH3ON02,whichcouldbealwayspresentin thePAN solution,wasmeasured by GC/NICI MS analysis.Theretentiontime andthemassspectrum ofCH3ONO2will be describedin detail in section3.5 and in Figures7 and 8. Themonitoring of both m/z=46and62 in the SIM modecoulddetectpotentialimpurities in the gas streamemitted •om the PAN source,which wouldbe detectedas"NO• signals"andcausedan overestimate of PAN by the CL-NOx measurements. A small amount of CH3ONO2was oftendetectedin the PAN stream.The solution was used for the calibration source when the total abundance
wasconfirmed to be lessthan 2% ofthatofPAN assumingthe sameresponsefactor.In this mannerthe total abundanceof impuritieswas estimatedto be, at most,3%.
Theotherfactorsignificantly contributingto theaccuracy in thecalibrationis theconversion efficiency ofPAN onamolybdenurn converter.Wineret al. [1974] reportedthe conversion efficiencyof PAN toNO with a molybdenumconverter as92%. Fehsenfeld et al. [ 1987]andSinghet al. [ 1990]described the efficiencies of 91% and 100%, respectively. More recently, Nouaimeet al. [ 1998]useda TECOmodelNO• analyzer(model 42S)forthefieldcalibration with nocorrection. In thepresent work,weexamined the dependence ofPAN signalsonthetemperatureof the molybdenumconverter.Almost constantPAN signalswereobtainedat a temperature above300øC.Therefore,
PAN wasassumed to be converted to NO at an efficiency of nearly100%at320øC.Anadditionalexperimentwas performed ontheconversion efficiency. Theexhaust gasoftheCL-NO•analyzerwasmeasured by GC/NICI MS in theSIM mode.Theseex-
perimentsverified theconversion efficiency ofPAN onthemolybdenumconverterto be >99%. No correctionwasmadeto the
PANmixingratiosasdetermined by theCL-NO• analyzer used in the calibration.
significantlyaccurate andprecisesignals.TheN O andN O • data wereprocessed on a personalcomputerviaan analogto digital 3. Results and Discussion (A/D) converter.Calibrationof the improvedTECO NO• analyzer was madebasedon the gasphasetitration (GPT) tech- 3.1. MassSpectrumof PAN in NegativeIon Chemical niqueusinga TECO calibrator(TECO model 146) anda stan- Ionization dardNO gascylinder(NipponSansoCompany,Inc., 1.8 ppmv) Themassspectrumof PAN obtainedin the full scanmodeis forbothNO andNO2.NO2wasproducedby thegasphasetitraions,CH3C(O)O-(m/z=59), tion of NO with ozone.The calibratedrange forboth NO and shownin Figure2. Fourfragment NO3-(m/z=62), CH3C(O)OO-(m/z=75 ), andNO2-(m/z=46)wer NOx was 0 to 10 ppbv. Thisfragmentation patterncanbeexplainedbyasSystematic errorsassociated with accuracycanbe causedby eidentified. PAN is dissociated by twopathwayswiththecapimpuritiesin the PAN solution in the diffusionsource,which sumingthat arealsodetectedasNO•by the CL-NO• analyzer.Thepossible ture of thermal electrons in the ion source: impuritiesreportedpreviouslywereCH 3ONO2(methylnitrate), CH3C(O)OONO2 + e; -• CH3C(O)O- + NO 3 (5) NO2,andHNO 3[Robertsetal., 1988; Singhetal., 1990].Fresh PAN solutionswere usedto minimizethe impurityeffectsin '-' CH3C(O)O + NO• (6) calibration.A nylon filter(Fuji filterCompany,Inc.)wasplaced CH3C(O)OONO2+ eT•-. CH3C(O)OO- + NO 2 (7) downstream of the PAN sourceto removeHNO 3 in the gas --'CH3C(O)OO + NO• (8) stream.
TANIMOTO ET AL.' PEROXYACETYL NITRATE MEASUREMENT TECHNIQUE 100
a
80
•
80
•
40
•
20
Figures3a and3b showthe dependenceof S/N ratios onthe ion sourcepressureandthe carriergasflow rate,respectively.In Figure 3a, the S/N ratio increased substantially as the ion sourcepressurewas raisedto -2 torr. This resultwasprobably dueto moreeffectivethermalizationof secondaryelectronsand excitedions (see(2) and (4)). As for the carriergas flow rate,the
PAN
62
21,347
46 75
S/Nratioincreased theflowrateincreased upto5.0mLmin-].
i _!
Much ofthis result was likely causedby decreasedthermaldecompositionof PAN at 30øCin the GC ovenand surfacedecomposition on the active sites in the capillary column due to reducedretention time [Roumelis and Glavas, 1989]. Owing to the limitation of the evacuatingcapacityof the presentsystem,
•_|_ i•1_1 |___| _i_ J___ ,__ , L,___J 100
150
200
Mass Number / amu.
Figure2. Mass spectrum of PAN atpartspermillion by volume (ppmv)levels in the full scanoperationup to mass-to-charge ratio m/z=200. The reagentgas used is methane.
Theparention (PAN-; m/z-121)andits homologues (m/
wefixed thecarrier gasflowrate at5.0mLmin-],which gavethe total ion sourcepressureof 2.0 torr. Optimization was also madefor electronenergyat 250 eV, emissioncurrentat 100 •tA, repellervoltageat 11.50 V, andthe voltagesappliedto two ion lensesat 35.0 and 30.0 V, respectively. Forthequadrupo1e massanalyzer,temperature dependenceof the signal was investigatedin the range•om 40 øto 140øC.The S/N ratio was foundto be higher asthe temperaturedecreased. However,to avoid contaminationfi'omsubstances formedby the self reactionsof fragmentions of methane( CH2 , CH3, and -t-•
-t-
z=120and/or122)werenot observed in thebothcaseofmeth-
CH4+'), whichmaycause contamination by adsorbing onthe
ane and isobutane used as reagentgases.
surface insidethemassanalyzer,thetemperature was maintained at 70øC,which is the sametemperature asthe ion source.The deviation of the massassignmentin the SIM mode could cause sensitivity changesin the SIM measurements. Thereforerelatively low massresolutionis preferablein the SIM modeto reduce
3.2. Optimization of the Instrument
Themassspectrometric conditionswere optimizedby measuringconstantmixingratiosof PAN in theppbvrangeandadjusting the GC/NICI MS fora maximum signal-to-noiseratio. "Noise" is estimated to be baseline
fluctuations
before PAN
peak,"peakheight"isthe top signalof PAN peakaftersubtraction ofthebaselinesignal,and"S/Nratio"isthe peakheightdivided by noise in the chromatogram. Amongthefourfragment ions,NO 3-(m/z=62)hasexhibited the best S/N ratio. Therefore m/z=62was
5OO
(a)
400
used in the SIM mode
300 ...............................
to obtain high sensitivity. In addition to the sensitivity,the monitoring of m/z=62 provided specific detection for PAN, which hasthe advantageof avoidingthe potential interference by isopropylnitrate (seesection3.4.).The integrationtimeof m/z=62was 300 ms,which corresponds to themassscanrateof 3.02 scans/s,in orderto balancebetweensensitivityand peak
200
..................................................
1 O0
..................................................
ß
i ! !ß i • , ß , i , , i i i i i i i i i i i i
shapeof the chromatogram. Thebehaviorof masssignalsof PAN with m/z=62monitor-
0.0
0.5
ing SIMis described below.ThePAN peakheightwasfoundto decrease withtheion source temperature, probablyowingto in-
sourcetemperaturewas below 60øC. Thereforethe ion source
temperature was setat 70øCin this study. GregorandGuilhaus [ 1984]reportedtheefficiencies ofsev-
1.0
1.5
2.0
2.5
Ion Source Pressure / T orr 1500
creasingthermaldecompositionof PAN. Noise increasedwith
the increase ofthetemperature up to approximately 200øC.Accordingly,theS/N ratiowasfoundtobehigheratlowertemperature;however,thetemperatures increased graduallyduringthe operationbecause ofheatproducedby the filamentwhenthe ion
-*--•
(b) ._
o 1000
Z 03
ß 500
...................................................
eraldifferent reagentgasesforelectronattachment-negative ion chemicalionization.Theyreportedthat isobutanewasmoreefI I I I I I I I I I I fectivein the degreeof electronattachment to anthraquinone 0.0 1.0 2.0 3.0 4.0 5.0 6.0 andbis( N,N-di-e thyldithiocarbamato) nickel(II) thanmethane. Isobutane hasbeenthusexpected to give a highersensitivity CarrierGasFlowRate/ mLmin-1 forPAN.In contrastto theirresults, wefoundthatmethane gave Figure3. Theobserveddependence ofsignal-to-noise (S/N) abetterS/N ratiothanisobutaneforPAN. A possiblereasonis ratiosfor(a)theionsource pressure byvaryingthereagentgas that isobutaneis likely to inducedissociationof PAN by flow rate(50 ppbv PAN with the carriergasflow rateof 2.0 collisional energytransferbeforean electroncanbe transferred mL/min)and(b) thecarriergasflowrate(40ppbvPAN with the to PAN, becauseisobutaneis a largermoleculethan methane. optimumreagentgasflow rate).
21,348
TANIMOTO ET AL.' PEROXYACETYL NITRATE MEASUREMENT TECHNIQUE
5.0X106
r- 4.0X106 o
•
•
•
3.OXLO 6
6
2.OXLO
1.ox1o6
o.o
0
200
400
600
800
1000
1200
1400
1600
PAN mixing ratio / pptv Figure 4. The responseof gaschromatography with negative ion chemicalionization massspectrometry(GC /NICI MS) in the selectedion monitoring (SIM) modefor PAN at pptv levels as comparedwith the NO-O3 chemiluminescentNO,,analyzer. Thedata above 100 pptv areusedforcalculation ofthe slope, the intercept,and the correlationcoefficient.Thick line indicatesthe least linear squarefitting. Dashedlines areupperand lower limitsof 95% confidencelevel obtainedby the t-test.Shownarethe equationsofthe fitting line. The slopeand
theintercept aregivenwiththeir95%confidence levels. R2denotes thecorrelation coefficient.
thiseffect. Itwasoptimized to obtainthebestS/Nratioandwas + 1.5 amuasa peakhalf width aroundm/z=62. In the ion detection scheme,sensitivity was improvedregardingelectronmultiplier voltage and conversiondynode voltage.The S/N ratioswerefoundto be higherasthevoltages increased.Since the lifetime of the electronmultiplier can be considerably shortenedwhenusedathighervoltages,the voltageforthe electronmultiplierwasdetermined to be 2.5 kV. The voltagefortheconversiondynodewassetto 10kV ofmaximum value.
3.3. Characterization of the GC/NICI MS Technique Figure4 shows the responseof GC/NICI MS for PAN at mixingratiosrangedfrom100to 1300pptv,comparedwith CLNO,,.Theregressionofthesedatashowedgoodcorrelation.A chromatogram of PAN at 122pptv is shownin Figure5. Thede-
20000
tectionlimit was estimatedto be 15 + 4 pptv (S/N=3) basedon the peak height and the noise level fromthe chromatogram. The reproducibility and the sensitivity changesof the instrumentwere estimatedas follows. Repeatedmeasurements of PAN ofknown mixing ratios at a measurement frequencyof 5 min for 50 timeshave shown the reproducibility of + 15% (20) at or above 100 pptv. This value includes variations of the PAN mixing ratios in the gas streamfromthe PAN sourceand the instrumentsensitivity changes.Repeatedmeasurements of isopropyl nitrate have been made for 1 day. IPN was used sinceit is stable, while PAN could undergo gradualthermal decompositionevenwhen kept at 0øC.IPN hasbeen measured by monitoringm/z-46 (NO2-) in the SIM mode,andthe signals showed no significant trend but a considerablefluctuation within 1 day. The 2o precisionof measurements wasthus estimatedtobe+ 15%, and the sensitivity hasnot changed,at least in the courseof aday. Thereforehalfdaily or daily basiscalibra-
'
18000
PAN (122 pptv)
16000
......................................
14000
12000
....................................
10000 8OOO 6000
,
i
I
4000 2.2
2.3
,
I
2.4
,,
i
2.5
2.6
2.7
, 2.8
i 2.9
,
,
i 3.0
,
i 3.1
, 3.2
Time / min
Figure5. Thegaschromatogram ofPAN at 122 pptv obtainedby integratingm/z=62 in the SIM mode.
TANIMOTO ET AL.' PEROXYACETYL NITRATE MEASUREMENT TECHNIQUE tion is sufficientfor continuousfield measurements using NICI MS, though this meansmorefrequentthan GC/ECD methods. A NICI MS instrumentcould showsensitivity changesduring an individual day and overa longer scale(E. Atlas, personal communication,1999). Our laboratorytests describedabove haveshown that sensitivity hasnotchangedsignificantly during 1 day for IPN (m/z=46); however, longer-termsensitivity drifts have not been tested here.Also, water vapor and CO2 presentat largemixingratios in ambientair canhave a significantimpacton the sensitivity ofthe NICI MS detectorandlead sensitivity changes.It should be certainly well characterized and correctedif the NICI MS exhibits long-term sensitivity changeswhen GC/NICI MS is used for continuous measurements ofambient air. Therefore future work will include the eval-
uationof instrumentsensitivitychangeson alonger scale(e.g., weekly or monthly) when the GC/NICI MS continuesto measureambientair containingmanytracegases.An automatedcalibration method will be developed to correctthe sensitivity changesof the instrument. Overall accuracywas determinedby taking into considerationthe accuracyofthe calibrationsystemandthe precisionof GC/NICI MS measurement. The uncertaintyin the CL-NO,•calibrationtechniquewas estimatedto be+ 10% basedon the combined uncertaintiesin NO standardgas,calibrationofthe CLNO,•by gasphasetitration, gasflow rates,impurity effects, and conversionefficiencyon the molybdenumconverter.The combined uncertaintiesof CL-NO,• and GC/NICI MS instruments
21,349
lOO
•
80
•
60
03
40
:3
20
(a)
IPN 46
57
0
0
20
40
60
80
100
Mass Number / amu. 10000
5000
0000
'
'
•
i
!
i
,
i
5000
resultedin anoveralluncertainty of+20%forPAN. ThesensitivityofPPN is moreinfluenced by decomposition 2.0 2.2 2.4 2.6 2.8 3.0 in the capillarycolumnthanPAN owingto its longerretention Time / min time.In the presentwork, calibrationfor PPN wasperformed of isopropylnitrate(IPN) at basedoncomparison with PAN, whichgavearelativeresponse Figure6. (a) Themassspectrum factorof 1.72+ 0.10 (PAN/PPN). Accordingly,the detection ppmvlevelsobtainedby GC/NICI MS in thefullscanoperation up to m/z= 100. (b) The chromatogram obtainedin the selected limitforPPN wasestimated to be26 + 5 pptv (S/N=3), with an ion monitoringmodewith the monitoringion of(top) m/z=62 overall uncertaintyof +25%. and (bottom)m/z=46. The mixing ratios of PAN and IPN, whichweredetermined by the CL-NO,•analyzer,were500 and 3.4. Interference From Isopropyl Nitrate (IPN)
200 pptv, respectively.
SomeCFC-typecompounds andorganicnitratesarecoeluted andmaybedetectedin thePAN measurement system. Although InFigure 6b, thechromatograms areshownwhen PAN (500 capillarycolumnshavegreaterseparationowing to increased resolutioncompared to packedcolumns,CFC-typecompounds pptv) and IPN (200 pptv) were simultaneouslymeasured.IPN andorganicnitratescouldstill interferewith thePAN measure- was significantly detectedwith good separationfromPAN in thechromatogram when m/z=46 wasmonitored(Figure 6b,top). mentsin GC/ECD systems. Interferences fromCFC-typecompounds areclearlyexpected TheIPN peakwaseliminatedby monitoringm/z=62 alone(Figto be eliminatedby monitoringm/z=62 in GC/NICI MS sys- ure 6b, bottom), and thus specific detection for PAN was fromlPN is avoidedby integrattems,while organicnitratescouldproduceNO3-(m/z=62) as achieved.Thereforeinterference sinceIPN mixingraoneoffragment ionsbecause theyhavenitrooxy(-ONO2)group ing m/z=62 in ambientair measurements, astheir functionalgroup.Of them,IPN, particularly,canaffect tios are usually lower than those of PAN. Moreover, using m thePAN peakif it produces muchNO 3-asa fragment ion, since /z=62 in the SIM modewouldprovidecompleteseparationfrom IPN elutesverycloseto PAN in theECD chromatogramusing coelutedorganicnitratesfor allPAN-type compounds,leading selective measurements for PANs. severalkinds of capillary columns [Roberts et al., 1989; To furtherexaminethe existenceof interferingcompoundsin Bertmanet al., 1993]. In fact,it wasfoundthat IPN elutedvery ambientair,"which was obtained by closeto PAN in the TIC chromatogram using the SPB-1701 ambientair, "decomposed capillary column. passingthe air through heated stainlesssteel tubing (65 cm long) at 150øC[Roberts et al., 1998b; Nouaime et al., 1998], Figure6a showsthemassspectrumofIPN,whichclearlyillustratesthat the mostabundant ion isNO 2-(m/z=46).Theless wasmeasuredin the SIM modeby m/z=62 (for moredetails, see
abundant ion at m/z=57is identifiedas [CH(CH3)20- H 2]-, which is the counterpart fragment ion of NO2. Similarmass spectraof alkyl nitrate(RONO2)havebeenreportedby Atlas [1988]. This patternwas alsosuggested in a previousstudy thatreportedalkoxideions werestabilizedby the lossofH2 [Hayeset al., 1985].NO3-(m/z=62)wascertainlyrecognized, but its abundance is lessthan10%of NO2-(m/z=46).
Figure 10 and section 3.6). Only PAN-type compoundswere
thusexpected to bethermallydecomposed with containingambientlevelsofothertracegasessuchaswatervapor,CO, CO2, nonmethane and/oroxygenated hydrocarbons, andhalocarbons, etc.No significantpeakhasbeenobserved in the chromatogram asseenin Figure 10. This resultsupportsthe premisethat the IPN of ambientmixingratiolevelsdoesnot interferewith PAN
21,350
TANIMOTO ET AL.' PEROXYACETYL NITRATE MEASUREMENT TECHNIQUE lOO
andalso indicatesthe absenceof interferingcompounds with
(a) 46
c
8O
other PANs
CH3ONO 2
in ambient air.
3.5. PAN Analog Detection(PPN and MPAN) 6O
c
NICI massspectraofCH3ONO2,PPN, andMPAN areshown in Figures7a - 7c, respectively. CH3ONO2,PAN, anda small
4O
amount ofP PN were contained as contaminants in the crude so-
"o c
lutionof synthesizedMPAN. Thespectrawereobtainedin the total ion chromatogram modeup to m/z=200. PPN andMPAN wereidentifiedby theirmassfragmentation patterns. Identifica-
2O
0
50
100
150
tionofCH3ON02wassomewhat criticalbymassfragmentation only,sinceNO2- alonewas significantlyobservedhere,contrarytothecaseof IPN described previously. Thisismainlybecausetheorganicfragment of m/z=31(CH30-) wasmaskedby
200
Mass Number / amu. lOO
(b)
c
8O
PPN
the oxygensignal,which is closeto it. However,the retention
timeof synthesized pureCH3ONO2[Luxenhofer et al., 1994] provided theevidence thatthiscompound wasCH3ONO2. PPN and MPAN show fragmentation patternssimilarto those of PAN. Differences areobserved in the abundance ratioof fragment ions. CH3CH2C(O)OO-(m/z=89) and CH2=C(CH3)C (O)00-(m/z=101) ion signalsgavethebestS/N ratiosin the SIMmodeforPPN andMPAN, respectively. Althoughthemeasurement by m/z=62hasbeenshownto provideslightlyless reduced sensitivitythanthatby thebestions,it gavesufficient sensitivityformeasurements of PPN andMPAN atpptvlevels. Thisisagreatadvantage forspecific detection ofotherPAN-typ
89 6O
c
46 62
4O
c
2O
o
50
100
150
200
Mass Number / amu.
e compoundsin ambientair, hencem/z=62 is the best monitoring ion in the SIMmode for the measurementsof PANs in ambi-
100
=
•
80
_(c) 6285
MPAN
ent air.
Figure8 showsthe SIM chromatogram ofCH3ONO2, PAN,
60
•-
40
•
20
PPN, andMPAN. Themonitoringionsshownarem/z=46for CH3ONO2andrn/z=62forPAN, PPN, andMPAN. As illustrat-
46
edin Figure8, detectionof CH3ONO2, PPN, andMPAN was accomplished with retentiontimesof 1.3+0.1, 5.0+ 0.1,and7.9 +0.1 min, respectively.Thus PAN, PPN, and MPAN canbe measured in less than 10 min.
i
0
i
i
i_ •1J
50
100
J.._LI
-J -J• ,--
150
200
3.6. Ambient Air Measurements
Mass Number / amu.
To elucidateinstrumentperformance for continuousfield
Figure 7. The massspectraof(a) CH3ONO2, (b) PPN, and (c) MPAN atppmv levelsobtained by GC/NICI MSin the full scan operationup to m/z=200.
measurementsof PANs, ambient air measurementswere madeat
theuniversitycampus inthesuburbs oftheTokyometropolitan area,Japan.The campusis located-•8 km west fi'omthe centerof
2.0X105
CH3CNO2
c1.5X10 5
o
o •1
c
PAN
1.0X10
MPAN
5
PPN
._•
:•0.5X10 5 • t•I. 0.0
: - I
0.0
1.0
I
2.0
;
[
3.0
I
[
4.0
I
5.0
6.0
[
I
7.0
[
t
8.0
9.0
10.0
Time / min
Figure8. Thegaschromatogram ofCH3ONO2,PAN, PPN, andMPAN obtainedinthe selectedion monitoring modewith monitoringion of m/z=46 (CH3ONO2),and m/z=62 (PAN, PPN, and MPAN). The vertical dashed line shown at 1.8 min denotesthe time for switching monitoring ions.
TANIMOTO ET AL.' PEROXYACETYL
NITRATE MEASUREMENT TECHNIQUE
21,351
50000
ß
40000
PAN
30000
20000
10000
0
0.0
1.0
2.0
3.0
4.0
5.0
6.0
7.0
8.0
9.0
10.0
11.0
12.0
Time / min
Figure9. A typicalgaschromatogram obtained by ambient air measurements conducted at theuniversityat 1300LSTonNovember 19,1998.Themixing ratiosforPAN andPPNwere456 and48 pptv,respectively. The measurements weremadein the selected ion monitoringmodewith a monitoringion ofm/z=62.
Tokyo at an elevation of-30 m above sea level. The weather
conditionsduringthe measurement period in November1998 werefairlygood, with clearskiesand norain. Ozonewassimultaneouslymeasured by UV absorptiontechnique(TECOmodel 49)with a timeconstantofl min.TheO3analyzerwascalibratedby a commercial 03 calibrator(TECOmodel49PS). Ambientairwassampled forbothO3andPANs throughTeflon tubing at 15 m abovethe groundsurface.Sampleair for PANs measurement was continuously drawn at a controlled
A typical gaschromatogram obtainedduringthis period is shownin Figure 9. PAN as well asPPN aredetectedwith rathergoodresolution.No interferingpeakwasobservedin all cases.MPAN hasnot beendetectedin any chromatogram; however,two peaksother thanPAN andPPN havebeen detectedseveral times around the detection limit level at retention times of
4.3 ñ0.1 and 6.6 ñ0.1 min,respectively.Figure 10 showsthe chromatogram obtainedduringa high-pollution episodein Tokyo (not obtainedin the period of Figure 11). The peaksat 6.6 flowrate of50 cm3min-• STP.Measurements ofPAN weremade ñ 0.1 and 9.7 ñ 0.1 min were assignedto peroxyisobutyryl nievery15 minin the SIM modewith a monitoringionofm/z=62, trate ((CH3)2CHC(O)OONO2;PiBN) and peroxy n butyryl niextendingmeasurementtime to 12 minwith considerationofthe trate (CH3CH2CH2C(O)OONO2;PnBN), respectively,which possibilityof detectionof other PAN analogs.Calibrations havebeenreportedpreviously [Grosjeanet al., 1993a, b], fi'om wereperformed on adaily basisto checkinstrumentsensitivity the synthesizedstandards.The otherwas a puzzle. This peak
by injectingknownmixingratiosofPAN (-1 ppbv) into the GC/NICI MS. Significantsensitivitychangeswere not ob-
was foundto be eliminated'concurrently with PAN, PPN, PiBN, and PnBN in a "decomposed ambientair" measurement (seeFigure10), atleastindicatingthatthisisnot anartifactsig-
servedduring the period.
7000
e-
PAN[ unidentified
6000
PPN
o
-• .
5ooo
-
=..,.
PiBN
I
c/)
4000
"''. "i••:•.•..•,.....,•, ,.,;'-.' ' ,'....
3000
2.0
3.0
4.0
5.0
6.0
7.0
8.0
9.0
10.0
11.0
Time / min
Figure10. Theextended SIMgaschromatogram (m/z=62)obtainedduringhigh-pollutionepisode, illustrating detection ofPiBN,PnBN,andanunidentified compound. Theupperchromatogram (solidcurve)represents ambientairsampled atroomtemperature (25øC).Thestippledcurverepresents "decomposed ambientair"(fordetails,seetext),whichis downshifted by 300 countsfi'omtheraw signals.
21,352
TANIMOTO ET AL.' PEROXYACETYLNITRATE MEASUREMENTTECHNIQUE
lOOO 8oo !
600
,
,
400
forsuburbanlocations where PAN mixing ratios oftenexceeda fewppbv ormore[Roberts,1990; Altshuller, 1993]. Mixing ratios ofO 3,whicharesimultaneouslyproducedby photochemical reactionsin polluted air masses,also appearedto be low. This would result mainly fromdecreasedsolar insolation in wintertime.
Ozone as well as PAN
and PPN
200
lations were rather common in observations
100f • • 7
•.
showed distinct
diurnal variationsfor 3 days.There was a generaltendencyfor O3tobe correlatedwithPAN andPPN. Theambientmixingratiosof PAN andPPN wereclearly well correlated.Thesecorre-
6O
40 20 I
•
30
•
20
I
I
o• •o o
18
19
20
Date / November, 1 998
Figure11. Timeseriesofmixingratiosof(a)PAN,(b) PPN, and(c)03 observed attheuniversity, suburban sitein Tokyo, Japan, from November 18 to 20, 1998.Thethicklinesindicate 0000(LST),anddashed linesareshownforevery6 hours.The mixingratiosof anunidentified compound (pluses) andPiBN (triangles) areplottedin Figure1lb andarescaled to theresponsefactorof PPN.
nal producedby detectorsaturation,which maybe causedby halocarbonsand/oroxygenatedcompoundspresentin ambient air at largemixing ratios. Sincethe unidentifiedpeak was detectedconcurrentlywith PiBN in early afternoonwhen PAN andPPN were at amaximum(seeFigure 1lb), it was probably a photochemicaloxidation product. The reportedperoxyacyl nitrate other than PAN, PPN, PiBN/PnBN, and MPAN, which canbe detectedin urban or regionally polluted air masses,is
peroxybenzoylnitrate (C6HsC(O)OONO2; PBzN) [Roberts, 1990; Altshuller, 1993; Roberts, 1995], but nevertheless, PBzN is expectedto have a longer retention time than PnBN. Owing to the retention time ofthis unidentified compoundbetween PAN and PPN, this appearsto be a low molecular(i.e., C2-C3)compound. Sincethepositive identificationhasnot been accomplishedin this work, it is not clearwhetherthe unidentified compound is an unknown PAN homologue or a multifunctionalnitrate (for example,dinitrooxy and nitrooxy carbonyl compounds)showing a m/z-62 fragment(E. Atlas, personalcommunication,1999). Atime seriesof PAN, PPN, and O3is shownin Figure 1la11c. Mean mixing ratios for PAN and P PN were 291 + 145 and 34 + 16 pptv, respectively.This mixing ratio appearedto be low
made at urban or
evenrural siteswherepolluted air massesarepresent[Roberts, 1990; Altshuller, 1993]. Nighttime mixing ratios of O 3,PAN, and P PN were much lowerthanthoseofdaytime.PAN and PPN appearedto bemore persistentthanO3duringnight. Mixingratios of O3weremostly nearzeroduring night, becauseof NO titration anddry depositionin the strongnocturnalboundarylayer (NBL), which limits vertical mixing between the near-surfaceair and the air aloft. Mixing ratios of PAN and PPN gradually decreasedduring the entirenight until sunrise.Steadily decreasingmixing ratiosof PAN and PPN atnight arecausedby combinedeffects of dry deposition on surface[Garland and Penkett, 1976; Shepsonetal., 1992; Schrimpfet al., 1996] andthermaldecomposition followed by the reaction of peroxyacetyl(PA) and peroxypropionyl(PP) radicalswith NO and otherRO2radicals. Ozone,PAN, and PPN rapidly increasedon sunrise.This could result fromtwo factors:photochemicalactivity initiated by increasingsolarinsolation and OH radicalsand downward mixingof the relatively higher mixing ratios above the inversion causedby dissipation of the nocturnal boundary layer. Maximummixingratioswere observedto occurin theearly afternoon(1200- 1500 local standardtime (LST))with thevaluesof 600 and 70 pptv, respectively,whereasminimummixingratios were observedbefore sunrisewith values of approximately80 pptv and below the detectionlimit, respectively.That the daily maximaof PAN and PPN coincidedwith those ofO3 wereascribedto the commonphotochemicalorigin of all compounds. Theonly earlyafternoonmaximumofeachcompoundin daytime indicated the chemistryhere was not dominated by transport fromothersourcesbut by local photochemistrythatoccurredin the Tokyo area.In the late afternoon,the mixing ratios of PAN andPP1'T were constantor lower, dependingon the balancebetween production and destruction. The scatterplotsof PPN to PAN are shown in Figure 12. PAN
and PPN
were not scattered
but well
correlated.
A least
squaresfit gave a slope of 0.10 + 0.01. Previous studies have shown that PPN/PAN ratios varied considerablyin the range of 0.05 to 1.0, dependingon the observation site, air massage, and its origin [Roberts, 1990; Ridley et al. , 1990; Altshuller, 1993; Grosjean et al., 1993a; Williams et al., 1993; Roberts, 1995; Williams et al., 1997; Roberts et al., 1998a,b]. Thevalue of 0.10 + 0.01 obtained in thesemeasurementswas in the range of 0.1-0.4 for urban areasas summarizedby Altshuller [ 1993]. PAN is producedby the reactionofNO 2with PA radicalsderived fromboth anthropogenicand biogenic sources.While PPN is producedpredominantlyfromphotochemistryofanthropogenichydrocarbons(e.g.,propane,propanal,1-butene,methyl ethyl ketone,and largeralkanes),MPAN is predominantly fromthat ofbiogenichydrocarbons(e.g.,isoprene,methacrolein) [Williams et al., 1997; Nouaime et al., 1998]. A biogenic source of PPN
has been discussed
but was found to be an in-
significant contributor to PPN production [Grosjean et al., 1993c]. Therefore the existence of P PN and the absence of
TANIMOTO ET AL.: PEROXYACETYL
NITRATE MEASUREMENT
TECHNIQUE
21,353
lOO 90
I:..... ..... [..... L..... L..... L_T,f..... L..... ':...... i.....
,o ................... 40
......................
30
I',', ', ';', ', lO .....;,i• i"' -._,+ ',.... ..... :" ..... ;'...... .... :...... ;.....
20 o
0
100
200
300
400
500
600
700
800
900
1000
PAN / pptv Figure 12. Thecorrelationof PPN with PAN observedat the universityfi'omNovember 18 to 20, 1998.
MPAN and thegood correlationbetweenPAN and PPN in am-
are appreciativelyacknowledgedfor their beneficialcommentsand sug-
bientairobserved wereattributedto the factthatanthropogenic gestionsbasedon the knowledgeof the "PAN community."This work by CoreResearchfor EvolutionalScienceand hydrocarbons arethemajorprecursors forlocalphotochemistry wasfinanciallysupported in Tokyo. 4.
Technology(CREST)of the JapanScienceandTechnologyCorporation. Thefirst author(H.T.) is supportedby a researchfellowshipof theJapan Societyfor the Promotion of Science(JSPS)for YoungScientists.
Conclusions References
Thegaschromatography with negativeion chemicalionization massspectrometry(GC/NICIMS)techniquepresentedhere Altshuller,A. P., PANsin the atmosphere, J. Air WasteManage.Assoc., is a new method for on-site field measurements of PAN and its
43, 1221-1230, 1993.
alkyl nitratesin rural andremoteatmosphere, homologues. Theoptimum operatingconditionsformonitoring Atlas,E., Evidencefor _>C3 Nature, 331,426-428, 1988. ionofNO3-(m/z=62)in theSIM modeprovidedhigh sensitiviBeine, H. J., D. A. Jaffe, D. R. Blake, E. Atlas, and J. Harris, Measurety, similarto GC/ECD andmuchhigherselectivitythan GC mentsof PAN, alkyl nitrates,ozone,andhydrocarbons duringspring /ECD to measure PANs at pptv levelswith directair injection. in interiorAlaska,J. Geophys.Res.,101, 12,613-12,619,1996. Thepresentdetectionlimit is 15 + 4 pptv (S/N=3)with an Beine,H. J., D. A. Jaffe, J. A. Herring, J. A. Kelley, T. Krognes,andF. Stordal,High-latitudespringtimephotochemistry, I, NO x, PAN and overallaccuracy of +20% anda precisionof + 15% (20). Interozonerelationships, J. Atmos.Chem.,27, 127-153, 1997. Bertman, S. B., and J. M. Roberts, A PAN analog from isoprene systems, canbecompletely eliminatedin this methodby moniphotooxidation, Geophys.Res.Lett., 18, 1461-1464, 1991. toring m/z=62. Time resolution for ambient measurements of Bertman,S. B., M.P. Buhr,andJ. M. Roberts,AutomatedcryogenictrapPAN, PPN, and MPAN is 15 min. Preliminarymeasurements pingtechniquefor capillary GC analysisof atmospherictrace compoundsrequiringno expendablecryogens:Applicationto the meahavedemonstrated thesuitabilityofthistechniqueforambient surementof organicnitrates,Anal. Chem.,65, 2944-2946, 1993. air measurements ofPAN andits homologues. Bottenheim, J. W., A. Sirois,K. A. Brice, andA. J. Gallant,Five yearsof Future work includes the evaluation of the instrument sensicontinuous observations of PAN and ozone at a rural location incast-
ferencefi'omIPN, which is sometimesencounteredin GC/ECD
tivity changes forPAN ona longerscale(e.g., weeklyormonthly) whenthe NICI MS detectorcontinuesto measureambientair
ern Canada,J. Geophys.Res., 99, 5333-5352, 1994. Fehsenfeld,F. C. et al., A ground-basedintercomparison of NO, NOx, andNOy measurement techniques,J. Geophys.Res., 92, 14,710-
containing manytrace gases suchaswatervaporandCO2.Also, 14,722, 1987. anautomated calibrationmethodwill be developedto correct GaffneLl S,R. Fajer,andG.I. Senurn, An improvedprocedurefor high thesensitivity changes oftheinstrument. Additionally,workis puritygaseous peroxyacylnitrateproduction: Use of heavylipid solbeing exploredon the identification of the unidentified com-
poundreportedin this paper. Acknowledgments. We wishtothankH. Takada,M. Takino,andT.
vents,Atmos.Environ., 18, 215-218, 1984.
Gaffney,J. S., N. A. Marley, andE. W. Prestbo,Peroxyacylnitrates (PANs):Theirphysicalandchemicalproperties, in TheHandbookof Environmental Chemistry,vol. 4, partB, editedby O. Hutzinger,pp. 1-38, Springer-Verlag,New York, 1989.
Miyajima(Yokogawa Analytical Systems Company, Inc.,Japan)fortheir Gaffney,J. S.,R.M. Bomick,Y.-H. Chen,andN. A. Marley,Capillary technical support. Thanksare extended to J. Matsumoto (RCAST,Unigas chromatographic analysisof nitrogendioxideand PANs with versity of Tokyo,Japan) for hishelpin calibration of theimproved CLluminolchemiluminescent detection, Atmos.Environ.,32, 1445-1454, 1998. NOxanalyzer andevaluation ofNO2impurity using anEcoPhysics NOx analyzerandS.Hatakeyama (NationalInstitute for Environmental Stud- Garland, J.A., andS.A. Penkett, Absorption of peroxyacetylnitrateand ies,Japan) forproviding a NOxanalyzer usedintheexperiment. I grateozoneby naturalsurfaces,Atmos.Environ.,10, 1127-1131,1976. fullyacknowledge J.Gaffney(Argonne National Laboratory, Chicago, Gregor,I. K., andM. Guilhaus,Comparativestudyof the influenceof Illinois), Y. Fukui(NASAAmesResearch Center,MoffettField,Califortheelectron-energy moderating gasin electron-attachment reactions
nia),andD. Jaffe(University of Washington-Bothell, Seattle, Washington)fortheirvaluable discussions andsuggestions. Anonymous reviewers
underNCI conditions, Int. J. MassSpectrom. Ion Processes, 56, 167176, 1984.
21,354
TANIMOTOET AL.: PEROXYACETYLNITRATEMEASUREMENTTECHNIQUE
Gregory,G. L., J.M. HoellJr.,B.A. Ridley,H. B. Singh, B.Gandrud, L. J. Salas,and J. Shetter,An intercomparison of airbornePAN measurements, d. Geophys. Res.,95, 10,077-10,087,1990.
Grosjean, D., E. L. WilliamsII, andE. Grosjean, Peroxyacyl nitrates at southernCaliforniamountainforestlocations,Environ.Sci. Technol., 27, 110-121, 1993a.
boundary layeroverthe NorthAtlantic,' J. Geophys. Res.,101, 28,947-28,960, 1996. Roberts,J. M., S. B. Bertman, D. D. Parrish,F. C. Fehsenfeld, B. T. Jobson, and H. Niki, Measurementof alkylnitratesat CheboguePoint, Nova Scotia duringthe 1993 North Atlantic RegionalExperiment (NARE) intensive,J. Geophys.Res.,103, 13569-13580,1998a. Roberts,J. M. et al, Measurementsof PAN, PPN, and MPAN made dur-
Grosjean, D., E. L. WilliamsII, and E. Grosjean, Ambientlevelsof ingthe 1994and 1995 NashvilleIntensivesof the SouthernOxidant peroxy-n-butyryl nitrateat a southern Californiamountain forest Study:Implicationsfor regionalozoneproductionfrom biogenichysmogreceptorlocation, Environ.Sci.Technol., 27, 326-331,1993b. drocarbons, J. Geophys.Res.,103,22,473-22,490,1998b. Grosjean, D., E. L. Williams II, andE. Grosjean, A biogenic precursor of N., andS. Glavas,Decomposition of peroxyacetylnitrateand peroxypropionyl nitrate:Atmospheric oxidation of cis-3-hexen-1-ol, Roumelis,
peroxypropionyl nitrate duringgas chromatographic determination Environ. Sci. Technol.,27, 979-981, 1993c. witha wide-bore capillaryandtwopackedcolumns,Anal.Chem.,61, Hayes,R. N., J. C. Sheldon, J.H. Bowie,andD. E. Lewis,Elimination of 2731-2734, 1989. molecularhydrogenandmethanefrom collision-activated alkoxide negative ionsin thegasphase.An ab initioandisotope effectstudy, Schrimpf,W., K. P. Mtiller,F. J. Johnen,K. Lienaerts,andJ. Rudolph,An optimizedmethodfor airborneperoxyacetylnitrate(PAN) measureAust.d. Chem.,38, 1197-1208, 1985. ments,J. Atmos.Chem.,22, 303-317, 1995. Jaffe,D. A., Therelationship betweenanthropogenic nitrogenoxidesand ozonetrendsin theArctic troposphere, in TheTropospheric Chemis- Schrimpf,W., K. Lienaerts,K. P. Mallet, J. Rudolph,R. Neubert, W. Schti151er, andI. Levin, Dry deposition of peroxyacetylnitrate(PAN): try ofOzonein thePolarRegions,NATOASI Set.I, vol. 7, editedby Determinationof itsdepositionvelocity at night from measurements H. Niki andK. H. Becker,pp. 105-115,Springer-Verlag, New York, 1992.
of the atmospheric PAN and 222Radon concentration gradient,
Geophys.Res.Lett.,23, 3599-3602, 1996. Kacmarek,A. J.,I. J. Solomon, andM. Lustig,Preparation andproperties Shepson, P. B., J. W. Bottenheim, D. R. Hastie,andA. Venkatram,Deterof peroxyacetyl nitrate,d. Inorg.Nucl.Chem.,40, 574-576,1978. minationof the relativeozoneandPAN deposition velocitiesat night, Kondo,Y., M. Koike, S. Kawakami,H. B. Singh,H. Nakajima, G. L. Geophys.Res.Lett., 19, 1121-1124,1992. Gregory,D. R. Blake,G. W. Sachse,J. T. Merrill,andR. E. Newell, Profilesandpartitioning of reactivenitrogenover thePacificOcean Singh,H. B., Reactivenitrogeninthetroposphere,Environ.Sci. Technol., 21,320-327, 1987. in winter and early spring,d. Geophys.Res.,102, 28,405-28,424, Singh,H. B. et al., Relationshipbetweenperoxyacetylnitrateand nitro1997. genoxidesin the cleantroposphere, Nature,318, 347-349, 1985. Luxenhofer,O., E. Schneider,and K. Ballschmiter,Separation,detection Singh, H. B., L. J. Salas, and W. Viezee, Global distributionof andoccurrenceof (C2-C8)-alkyl- andphenyl-alkylnitratesas trace peroxyacetylnitrate,Nature,321, 588-591, 1986. compounds in clean andpollutedair, Fresenius d. Anal. Chem.,350, Singh,H. B. et al., Peroxyacetylnitratemeasurements duringCITE 2: At384-394, 1994. mosphericdistribution andprecursorrelationships, J. Geophys.Res., Moxim,W. J.,H. Levy II, andP.S. Kasibhatla,Simulated globaltropo95, 10,163-10,178, 1990. sphericPAN: Its transport andimpactonNO•, d. Geophys.Res.,101, Singh,H. B. et al., Summertimedistribution of PAN andotherreactiveni12,621-12,638, 1996. trogenspeciesin the northernhigh-latitudeatmosphereof eastern Mallet, K. P., and J. Rudolph, An automatedtechniquefor themeasureCanada,J. Geophys.Res.,99, 1821-1835,1994. mentof peroxyacetylnitrate in ambientair at ppband pptlevels,Int. Singh,H. B. et al., Reactivenitrogenandozoneover the westernPacific: J. Environ. Anal. Chem., 37, 253-262, 1989. Distribution, partitioning,and sources,J. Geophys.Res., 101, 1793Newton,D. L., M.D. Erickson,K. B. Tomer,E. D. Pellizzari,P. Gentry,
1808, 1996. and R. B. Zweidinger, Identificationof nitroaromaticsin dieselexof reactivenitrogenin the free haustparticulateusingGas Chromatography/Negative Ion Chemical Singh,H. B. et al., Latitudinaldistribution troposphereover the Pacific Ocean in late winter/early spring,J. IonizationMass Spectrometryand other techniques,Environ.Sci. Geophys.Res., 103, 28,237-28,246, 1998. Technol.,16,206-213, 1982. Spicer,C. W., E.G. Chapman,B.J. Finlayson-Pitts, R. A. Plastridge, J. M. Nielsen, T., A.M. Hansen,and E. L. Thomsen,A convenientmethodfor Hubbe,J. D. Fast,andC. M. Berkowitz,Unexpectedlyhigh concenpreparation of purestandards ofperoxyacetylnitratefor atmospheric trationsof molecular chlorine in coastal air, Nature, 394, 353-356, analyses,Atmos.Environ., 16, 2447-2450, 1982. 1998. Nouaime,G., S. B. Bertman,C. Seaver, D. Elyea, H. Huang, P. B. Srinivasan,N., R. G. Cooks, and P. B. Shepson,Protonaffinity of Shepson, T. K. Starn,D. D. Riemer,R. G. Zika, andK. Olszyna,Seperoxyacetylnitrate sampledby membraneintroductionmassspecquentialoxidationproductsfrom troposphericisoprenechemistry: trometry,Rapid Commun.MassSpectrom.,12, 328-334, 1998. MACRand MPAN at a NO x-richforest environmentin the southeastStephens,E. R., P. L. Hanst, R. C. Doerr, and W. E. Scott,Reactionsof ern UnitedStates,J. Geophys.Res.,103, 22,463-22,471,1998. nitrogendioxide and organic compounds in air, Ind. Eng. Chem. Pate,C. T., J. L. Sprung,and J. N. PittsJr., Chemicalionizationmass Fundam.,48, 1498-1503, 1956. spectrum of peroxyacetylnitrate, Org.Mass.Spectrom., 11,552-555, Watanabe,I., M. Nakanishi,J. Tomita, S. Hatakeyama,K. Murano,H. 1976. Mukai, andH. Bandou,Atmospheric peroxyacylnitratesin urban/rePenkett, S.A., andK. A. Brice,The springmaximum in photo-oxidants in motesitesandthe lowertroposphere aroundJapan,in Proceedings of thenorthernhemisphere troposphere, Nature,319,655-657,1986. Nitrogen,the Confer-N-s,First InternationalNitrogen Conference Ridley,B. A. et al.,Thebehaviorof someorganicnitratesat Boulderand 1998,pp. 253-261, ElsevierSci.,New York, 1998. Niwot Ridge,Colorado,./.Geophys.Res.,95, 13,949-13,961,1990. Williams,E. L., II, E. Grosjean,and D. Grosjean,Ambientlevelsof the Ridley,B. et al., Measurements of NOx andPAN andestimatesof 03 properoxyacylnitratesPAN, PPNandMPAN in Atlanta,Georgia,J. Air ductionover the seasons duringMaunaLoa ObservatoryPhotochemWasteManage. Assoc.,43, 873-879, 1993. istryExperiment,2, d. Geophys.Res.,103, 8323-8339,1998. Williams,J. et al., Regionalozonefrom biogenichydrocarbonsdeduced Roberts,J. M., The atmosphericchemistryof organicnitrates,Atmos. from airbornemeasurements of PAN, PPN, and MPAN, Geophys. Environ., Part A, 24A, 243-287, 1990.
Roberts, J. M., Reactiveodd-nitrogen (NOy) in theatmosphere, in Composition,Chemistry,and Climateof the Atmosphere, editedby H. B. Singh,pp. 176-215,Van NostrandReinhold,New York, 1995. Roberts,J. M., R. W. Fajer, and D. M. McIntire, Characterization of a calibrationsourceandgaschromatographic systemfor the measurementof ambientPAN, paperpresentedat AmericanChemicalSociety 196thNationalMeeting,Am. Chem. Soc.,Los Angeles,Calif., Sep.25 to 30, 1988. Roberts, J. M., R. W. Fajer, and S.R. Springston, Capillarygaschromatographicseparation of alkyl nitratesandperoxycarboxylic nitricanhydrides,Anal. Chem., 61,771-772, 1989.
Roberts,J. M. et al., Episodicremovalof NOv speciesfrom the marine
Res. Lett., 24, 1099-1102, 1997.
Winer, A.M., J. W. Peters,J.P. Smith,and J. N. PittsJr., Responseof commercialchemiluminescent NO-NO 2 analyzersto othernitrogencontaining compounds, Environ.Sci.Technol., 8, 1118-1121,1974.
H. Akimoto,J.Hirokawa,Y. Kajii,andH. Tanimoto,ResearchCenter for Advanced Scienceand Technology,Universityof Tokyo, 4-6-1 Komaba, Meguro-ku, Tokyo 153-8904, Japan. (e-mail: akimoto
@atmchem.rcast. u-tokyo. ac.jp;
[email protected]. u-tokyo.ac.jp;
[email protected]. ac.jp;
[email protected]) (ReceivedJanuary14, 1999;revisedMay 14, 1999; acceptedMay 18, 1999)
