View metadata, citation and similar papers at core.ac.uk brought to you by CORE
provided by Elsevier - Publisher Connector
AtmosphericPollutionResearch5(2014)759Ͳ768
AAtm spheric PPollution RResearch www.atmospolres.com
The 2013 severe haze over the Southern Hebei, China: PM2.5 composition and source apportionment
ZheWei1,Li–TaoWang1,Ming–ZhangChen2,YanZheng3
1DepartmentofEnvironmentalEngineering,SchoolofCityConstruction,HebeiUniversityofEngineering,Handan,Hebei056038,China 2SchoolofCivilEngineering,ShijiazhuangRailwayInstitute,No.17North2nd–RingEastRoad,Shijiazhuang,Hebei050043,China 3TheEnvironmentalMonitoringCenterofHandan,Handan,Hebei056002,China
ABSTRACT
PM2.5sampleswerecollectedandanalyzedforthefirsttimeinHandanCity,whichwaslistedinthetop4pollutedcities inChina,duringDecember2012toJanuary2013whentherecord–breakingseverehazepollutionhappened.Positive MatrixFactorizationmethod(PMF)wasappliedtounderstandmajorsourcestotheseverehazepollutionoverthiscity. –3 The daily average concentration of PM2.5 was 160.1ʅgm , which was 2.1 times of the National Ambient Air Quality –3 2– StandardsofChina(ClassII,AnnualAverageLevel)fordailyaveragePM2.5of75ʅgm .SO4 wasthemostabundantion + – (15.4%),followedbyNH4 andNO3 .Theyaccountedfor39.5%ofPM2.5.Eightfactorswereidentifiedbypositivematrix factorization(PMF)model.Themajorsourceswerecoalcombustionsource(25.9%),secondarysource(21.8%),industry source (16.2%), Ba, Mn and Zn source (12.7%), motor vehicle source (7.7%), road dust source (10.9%), K+, As and V –1 source(6.3%)andfueloilcombustionsource(2.5%).Themeanvalueofextinctioncoefficient(Bext)was682.1Mm and –1 CorrespondingAuthor: thelargestcontributortoBextwasammoniumsulfatewiththemeanvalueof221.0Mm ,accountedfor32.4%ofthe Li–Tao Wang B . ext :+86Ͳ310Ͳ8578755 :+86Ͳ310Ͳ8578755 Keywords:PM2.5,OC,Bext,PMF,sourceapportionment :[email protected] ArticleHistory: Received:07February2014 Revised:06June2014 Accepted:08June2014 doi:10.5094/APR.2014.085 1.Introduction 2.ExperimentalandAnalysisMethodology Extremely severe, widespread haze occurred in Central and 2.1.Sampling Eastern China during January 2013. The haze event attracted a wide attention locally and worldwide, because the haze in this Handan is located in the intersection area of Hebei, Shanxi, month was the most serious pollution episode since 1961. Henan,andShandong.Thesamplingsiteislocatedonthetopofa According to the statistics published by the Ministry of 12–mhighbuildingatSchoolofCityConstruction,HebeiUniversity Environmental Protection for January 2013, the top ten polluted of Engineering (Figure 1). The sampling period was from DecemͲ cities are Xingtai, Shijiazhuang, Baoding, Handan, Langfang, ber1, 2012 to January31, 2013. Daily PM2.5 samples were Hengshui,Jinan,Tangshan,BeijingandZhengzhou,sevenofwhich collected continuously from 8:00 am to 7:30 am of the next day arewithinHebei. usingaHighVolumePM2.5airsampler(ThermoScientificCo.)with 20.3×25.4cmquartzfilters(0.2ʅmofporesize).Flowratewas Veryfewstudieswerepursuedtostudythecharacteristicsof 1.13m3min–1andatotalof55sampleswerecollected. particulate matter in Handan, especially for fine particles. In the recentstudies,Wangetal.(2012),Weietal.(2013)andWanget 2.2.Chemicalcomponentsanalysis al.(2014)appliedtheCMAQmodelandusedtheso–called“Brute Force” method (BFM) (Dunker et al., 1996) to simulate the air Ionic analyses. A circular portion of the quartz filter (1cm2) was pollution over the southern Hebei cities and address the major extracted into 10mL ultrapure water (18.2Mɏcm) to determine sourcesofPM2.5.Wangetal.(2012)andWangetal.(2014)found the concentration of water–soluble inorganic ions. The extracted that local sources contributed 64.2% to PM2.5 in Handan in solution was filtered through a microporous membrane filter December 2007, 69.2% in January and 63.0% in February, 2013, (0.45ʅmofporesize),andstoredinarefrigeratorat–18°Cuntil respectively. However, to the best of our knowledge, there is no chemical analysis. An ion chromatograph (Dionex, DX–600, USA) researchfocusingonthechemicalcomponentsofPM2.5inHandan was used to measure the water–soluble cations. Another ion City.Inthisstudy,wesampledandanalyzedPM2.5duringthemost chromatoͲgraph (Dionex, ICS–2100, USA) was used to measure polluted period in Handan and applied PMF, a widely implied water–solubleanions. receptor model, to indentify source apportionment of PM2.5, to supportthepolicy–makinginairpollutioncontrolinthiscity. Carbon analysis. The total carbon, organic carbon and element carbonwereanalyzedinTsinghuaUniversitythroughmethodology ofTORusedThermal/OpticalCarbonAnalyzer(Model2001A)that
©Author(s)2014.ThisworkisdistributedundertheCreativeCommonsAttribution3.0License. Wei et al. – Atmospheric Pollution Research (APR) 760 was produced by Desert Research Institute (DRI). A piece of 0.01ʅgmL–1,respectively.Theprecisionwas<5%forTCand<10% processed sample filter (0.53cm2) was placed in an environment forOCandEC.TheMDLwas0.82ʅgcm–2forTOCand0.20ʅgcm–2 with pure He gas without O2, and was heated progressively at forTEC,respectively. 140°C, 280°C, 480°C, and 580°C, first to determine organic carboncontentsOC1,OC2,OC3,andOC4,respectively.Then,inan Elemental analyses. ICP–MS was calibrated by standard injection 2 environment with 2% O2 and 98% He, the sample was further (r >0.99) and the samples were analyzed in triplicate. Relative heatedprogressivelyat580°C,740°C,and840°Ctodeterminethe standarddeviationwaskeptbelow5%.Internalstandardrecovery elementalcarboncontentsEC1,EC2,andEC3. wascontrolledattherangefrom80%to120%. Elemental analyses. This study used ICP–MS to analyze the 2.4.PositiveMatrixFactorization(PMF) concentrationsoftraceelements(i.e.,Ti,V,Cr,Mn,Ni,Cu,Zn,As, Se,, Sr Cd, Ba and Pb). A piece of filter (1cm2) was digested for PMF is a multivariate factor analysis tool that decomposes a 25min at 190 °C with 8mL concentrated nitric acid (BV–III) and matrix of speciated sample data into two matrices, factor 0.5mL H2O2 (30%) in a Teflon microwave digestion tank (CEM, contributions and factor profiles, which then need to be MARS). Then, the bulk solution was moved into a 100mL flask. interpretedbyananalystastowhatsourcetypesarerepresented High–puritywaterwasusedtomakeupthevolume.Finally,10mL usingmeasuredsourceprofileinformation,winddirectionanalysis, solution was centrifuged for 10min to remove any suspended andemissioninventories(U.S.EPA,2008).Themethodisdescribed solidsbeforeinstrumentalanalyses. ingreaterdetailbyPaateroandTapper(1994)andPaatero(1997). PMFhasbeensuccessfullyappliedtoidentifythepollutantsources 2.3.QualityAssurance/QualityControl)(QA/QC (Leeetal.,1999;Chueintaetal.,2000;Qinetal.,2006;Brownet al.,2007;KimandHopke,2008;Yangetal.,2013). Sampling and weighing. There were several low–rise buildings aroundthesamplingsite.Therewasnotabigemissionsourcein The brief principle is described in the following: A speciated thisarea.Toensuretheflowof1.13m3min–1,flowofthesampling datasetcanbeviewedasadat matrixXofnbymdimensions,in equipmentwascheckedeverymonth.Thesiliconeoilwasreplaced whichnisthenumberofsamplesandmisthenumberofchemical every15daystoensurethesamplingrateofPM2.5.Thesequartz speciesmeasured.SoX=GF+E,G=n×pandF=p×m,wherepisthe filterswerebakedat450°Cfor4hours.Then,filterswereputina numberofpollutantsources,Gisthematrixof sourcecontribution, chamber at 25°C and 50% of relatively humidity for 24h until F is the matrix of species profile, and E is theresidual matrix for sampling.Thesesampledfilterswereputinthesamechamberfor eachsample/species.ThegoalofPMFistoidentifythenumberof 24h until the second weigh. Then, filters were preserved in a pollutantsourcesFandG[seeEquations(1)–(3)].ThePMFsolution refrigeratorthatsetat–20°C.Allprocedureswerestrictlyquality minimizes the object function Q [Equation (3)], based upon the controlledtoavoidanypossiblecontaminationofthesamples. uncertainties (u). Then PMF could calculate the G and F. The productofGandFcanexplainthesystematicvariationsinX. The sampling method was found to be effective since the measured concentrations agreed well with those measured using TEOM 1405D. In this period, three blank filters were preserved ܺ ൌ݃ ݂ ݁ (1) under the same environment were analyzed to guarantee the ୀଵ accuracyandreliabilityofsampling.Thecalculatedconcentrations fortheblankfilterswereusedtoadjusttheconcentrationsofthe measuredcompositions. (2) ݁ ൌܺ െ݃ ݂ Ionicanalysesandcarbonanalysis.Glasscontainerswerewashed, ୀଵ and soaked in ultrapure water for over 24 h. Then, they were cleaned with supersonic wavesfor 45min andwere washed two (times with ultrapure water (18.2Mɏcm). The method detection ܳൌൣ݁Ȁݑ൧ (3 + + + 2+ 2+ 2– – – – limit(MDL)ofNH4 ,Na ,K ,Mg ,Ca ,SO4 ,Cl ,NO2 andNO3 ୀଵ ୀଵ were 0.01, 0.001, 0.001, 0.004, 0.005, 0.01, 0.005, 0.01 and
Figure1.LocationofthesamplingsiteinHandancity(Ÿ indicatesHebeiUniversityofEngineering,and*indicatesthesamplingsite). Wei et al. – Atmospheric Pollution Research (APR) 761 Inthisstudy,USEPAPMF3.0wasusedtoanalyzethesource average concentrations of PM2.5 in December and January were apportionment. The 10% of concentrations was used as the 135.1ʅgm–3and190.1ʅgm–3,whichare1.8and2.5timesofthe –3 uncertainty of each species in the data set. If the concentration dailyPM2.5limitvalue(75ʅgm )setbytheNationalAmbientAir was less than or equal to MDL provided, 1/2 MDL was used to QualityStandardinChina(CNAAQS)(MEP,2012),respectively. replace the concentration, the uncertainty was calculated using 5/6 MDL. Signal to noise (S/N) [see Equation (4), Xij is the InWeietal.(2014)study,therewasatypicalpollutionepisode th th concentrationofthej componentinthei samplingday,Sijisthe during January 2013. The present study was conducted to better standard deviation of Xij] was used to review the dataset. The understand the different characteristics and detailed components dataset was used directly when S/N exceeded 2. The results of particles before and after the episode. Table 1 presents the showedthatthesmallestvalueofS/Nwas6.90ofTi.Therefore,we detailed concentrations measured between January 1–13 and consideredthatthedatasetwasreliable. January14–31. Ǥହ Water–soluble ions in PM2.5. In this period, water–soluble ions ଶ ଶ ܵȀܰ ൌ ͲǤͷ ൫ܺ൯ Ȁ൫ܵ൯ ൩ (4) contributed48.3%toPM2.5;almostahalfofPM2.5suggestingthat ୀଵ ୀଵ water–soluble ions are the major fraction in PM2.5. Additionally, water–soluble ions contributed 54.6% and 42.9% to PM2.5 in Among these measured compositions, 23 species were DecemberandJanuary,respectively.TheproportionofDecember – selected to run the model except that NO2 and Ti. We run the washigherthanJanuary.Itimpliedthattheproportionofwater– modelwith5to12factorstoexplorethebestsolution.Aneight– soluble ions didn’t increase along with increase of absolute factor solution gave the most interpretable and reproducible concentration of PM2.5. The proportions of January 1–13 and results within the iterations tested. Scaled residuals were inͲ January 14–31 were 29.1% and 65.3%, respectively. It indicated spected,and95.6%werebetween–3and3forallofthespecies. thattheproportionofwater–solubleionsincreasedafterthepeak 2– + – of the episode. The sum of SO4 , NH4 and NO3 accounted for 3.ResultsandDiscussion 39.5%,42.3%and37.2%ofPM2.5duringthisperiod,Decemberand January,respectively.Theywerethemajorcontributorstowater– 3.1.ChemicalcompositionsofPM2.5 solubleions.TheproportionofJanuary1–13was23.3%whilethe proportion of January 14–31 was 58.6%. That was to say, in the Table 1 illustrated the concentrations and chemical compoͲ earlystageofepisode,therewerelesswater–solublepollutantsin nents of PM2.5 in different periods. The daily average concenͲ theatmosphere. –3 tration of PM2.5 was 160.1ʅgm during this period. The daily
Table1.AverageconcentrationsofchemicalcomponentsinPM2.5duringDecember2012toJanuary2013 MeanValue(ʅgm–3) S.D.a(ʅgm–3) Whole Whole December January 1–13 14–31 December January 1–13 14–31 Period Period
PM2.5 160.1 135.1 190.1 217.7 139.5 77.9 57.2 89.4 139.5 41.1 OC 26.3 22.7 30.6 38.6 26.1 12.0 8.9 13.9 19.5 6.8 2– SO4 24.6 23.1 26.4 17.0 31.6 14.6 15.0 14.2 12.1 12.8 + NH4 19.9 17.4 22.9 18.9 25.1 9.1 8.5 9.0 10.0 7.8 – NO3 18.8 16.6 21.4 14.9 25.0 9.4 8.1 10.3 9.2 9.3 EC 15.8 16.8 14.7 20.1 11.6 7.3 6.8 7.9 10.5 3.6 Cl– 9.2 4.8 10.9 9.0 6.0 7.1 5.1 3.6 4.2 2.7 K+ 2.4 1.1 2.6 2.4 2.1 2.2 1.2 1.0 1.4 0.8 Na+ 1.3 1.7 0.9 0.9 1.0 0.7 0.6 0.5 0.3 0.6 Ca2+ 0.9 1.3 0.4 0.8 0.2 0.9 1.1 0.4 0.5 0.1 Pb 0.3 0.2 0.3 0.3 0.3 0.2 0.2 0.2 0.2 0.1 Zn 0.3 0.3 0.3 0.3 0.3 0.2 0.2 0.2 0.3 0.1 – –3 –3 –3 –3 –3 –3 –3 –3 –3 –3 NO2 88.0x10 142.3x10 40.4x10 0.6x10 226.5x10 153.1x10 102.6x10 154.8x10 50.9x10 146.9x10 Mg2+ 125.2x10–3 153.5x10–3 91.2x10–3 120.0x10–3 75.1x10–3 82.7x10–3 98.1x10–3 39.6x10–3 35.5x10–3 32.7x10–3 As 21.4x10–3 16.4x10–3 27.5x10–3 26.0x10–3 28.3x10–3 17.7x10–3 13.8x10–3 20.1x10–3 24.4x10–3 18.1x10–3 Ba 10.1x10–3 10.1x10–3 10.1x10–3 16.7x10–3 6.3x10–3 11.3x10–3 12.2x10–3 10.5x10–3 14.5x10–3 4.8x10–3 Cd 4.3x10–3 3.4x10–3 5.3x10–3 4.9x10–3 5.5x10–3 2.7x10–3 2.1x10–3 3.0x10–3 4.1x10–3 2.2x10–3 Cr 8.2x10–3 7.2x10–3 9.4x10–3 8.3x10–3 9.9x10–3 4.8x10–3 4.5x10–3 4.9x10–3 6.7x10–3 3.7x10–3 Cu 18.8x10–3 13.2x10–3 25.4x10–3 18.7x10–3 29.2x10–3 13.6x10–3 10.8x10–3 13.9x10–3 15.8x10–3 11.5x10–3 Mn 67.6x10–3 62.6x10–3 73.4x10–3 96.4x10–3 60.6x10–3 39.0x10–3 1.7x10–3 45.3x10–3 65.7x10–3 22.2x10–3 Ni 3.4x10–3 2.7x10–3 4.2x10–3 3.5x10–3 4.6x10–3 3.5x10–3 1.7x10–3 4.7x10–3 2.4x10–3 5.6x10–3 Se 10.6x10–3 9.0x10–3 12.6x10–3 14.8x10–3 11.4x10–3 6.3x10–3 4.8x10–3 7.4x10–3 11.2x10–3 4.1x10–3 Sr 6.4x10–3 7.4x10–3 5.2x10–3 6.9x10–3 4.2x10–3 6.3x10–3 7.5x10–3 4.1x10–3 5.4x10–3 2.9x10–3 Ti 15.1x10–3 65.7x10–3 11.0x10–3 8.9x10–3 12.2x10–3 20.3x10–3 0.2x10–3 14.6x10–3 19.0x10–3 12.1x10–3 V 5.4x10–3 5.7x10–3 5.0x10–3 5.3x10–3 4.8x10–3 5.9x10–3 4.7x10–3 7.1x10–3 7.1x10–3 7.4x10–3 aStandarddeviation Wei et al. – Atmospheric Pollution Research (APR) 762 2– – + –3 –3 SO4 wasthemostabundantioninthisperiod,itsconcentraͲ concentrationsofNO3 andNH4 were18.8ʅgm and19.9ʅgm , –3 –3 tion ranged from 5.3ʅgm to 73.3ʅgm with an average of respectively.Theyaccountedfor11.7%and12.4%ofPM2.5 24.3% –3 24.6ʅgm asshowninFigure2.Itaccountedfor15.4%ofPM2.5 and25.7%oftheallmeasuredions,respectively. and approximately 31.8% of the all measured ions. The average
Figure2.TheaveragedconcentrationsforchemicalcomponentsinPM2.5 duringDec.2012toJan.2013. Thebottomandtopoftheboxrepresentthe25thand75thpercentiledistributionforthecomponent, respectively;theblacklineintheboxrepresentsthe50thpercentilevalue;theendsofthewhiskers representthe10thand90thpercentiledistributionforthecorrespondingcomponent;thesolidround representminimumandthemaximumvalue,respectively;thesquareintheboxrepresentsmeanvalue ofthecomponent. Wei et al. – Atmospheric Pollution Research (APR) 763 OC and EC. The daily average concentrations of OC and EC were elements between January 1–13 and January 14–31, the results 26.3ʅgm–3and15.8ʅgm–3,whichaccountedfor16.4%and9.9% showedthattheconcentrationsofBa,Mn,Pb,Se,Sr,VandZnin of PM2.5 during this period, respectively. In January 1–13, the January 1–13 were higher than in January 14–31 period. The average concentrations of OC and EC were 38.6ʅgm–3 and concentrationsofotherelementswerehigherduringJanuary14– 20.1ʅgm–3,respectively.InJanuary14–31,theaverageconcentraͲ 31. tionsofOCandECwere26.1ʅgm–3and11.6ʅgm–3,respectively. It was found that water–soluble ions and OC, EC presented 3.2.TheresultsofPMFanalysis differentvariationsbeforeandaftertheepisode. AsshowninFigure3,thefirstfactorwascharacterizedbyNi, ElementsinPM2.5.Theconcentrationsofmeasuredtraceelements CrandV,whichwereusuallyregardedas themarkercomponents are presented in Table 1. The order of concentrations for the offuel–oilcombustion(KarnaeandJohn,2011;Yangetal.,2013). elements were Zn>Pb>Mn>As>Cu>Ti>Ba>Se>Cr>Sr>V>Cd>Ni durͲ Therefore,thisfactorcouldbeindentifiedasfuel–oilcombustion. –3 ing this period. In comparison with the concentrations of these Thissourcecontributed3.7ʅgm (2.5%)toPM2.5. (a) )DFWRU±)XHO2LO&RPEXVWLRQ )DFWRU)XHO2LO&RPEXVWLRQ
3HUFHQWRI6SHFLHV 9 . 6U 6H &O &U 1L &X $V &G %D 3E &D =Q 1D (& 0J 2& 0Q 62 12 1+
(b) )DFWRU±&RDO&RPEXVWLRQ)DFWRU&RDO&RPEXVWLRQ 3HUFHQWRI6SHFLHV 9 . 6U 6H &O &U 1L &D %D 3E &X =Q $V 1D &G (& 2& 0J 0Q 62 1+ 12 (c)
)DFWRU,QGXVWU\)DFWRU±,QGXVWU\
3HUFHQWRI6SHFLHV 9 . 6U &O 6H &U 1L &D &X $V &G %D 3E 1D =Q (& 0J 2& 0Q 62 12 1+
(d)
)DFWRU±%D0QDQG=Q)FWRU%D0QDQG=Q
3HUFHQWRI6SHFLHV
9 . 6U &O 6H &U 1L &G %D 3E &X $V &D 1D =Q (& 0J 2& 0Q 62 12 1+ Figure3.SourceprofilesinEVvalues.(a)Factor1–Fueloilcombustion,(b)Factor2–Coalcombustion, (c)Factor3–Industry,(d)Factor4–Ba,MnandZn,(e)Factor5–K+,AsandV,(f)Factor6–Secondary,(g)Factor 7–Motorvehicle,(h) Factor8–Roaddust. Wei et al. – Atmospheric Pollution Research (APR) 764 (e) )DFWRU.)DFWRU±. $VDQG9$VDQG9
3HUFHQWRI6SHFLHV 9 . 6U &O 6H &U 1L &X $V &G %D 3E &D =Q 1D (& 0J 2& 0Q 62 12 1+ (f)
)DFWRU±6HFRQGDU\)DFWRU6HFRQGDU\
3HUFHQWRI6SHFLHV
9 . 6U 6H &O &U 1L &X $V &G %D 3E &D =Q 1D (& 2& 0J 0Q 62 1+ 12 (g)
)DFWRU0RWRU9HKLFOH)DFWRU±0RWRU9HKLFOH
3HUFHQWRI6SHFLHV 9 . 6U 6H &O &U 1L &X $V &G %D 3E &D =Q 1D (& 0J 2& 0Q 62 12 1+ (h) )DFWRU±5RDG'XVW )DFWRU5RDG'XVW
3HUFHQWRI6SHFLHV
9 . 6U &O 6H &U 1L &D &X $V &G %D 3E 1D =Q (& 0J 2& 0Q 62 1+ 12 Figure3. Continued. Thesecondfactormightbeidentifiedasthecoalcombustion factor could be regarded as industry source, which contributed + 2– + + –3 source.BecausethisfactorhadhighloadingsofNH4 ,SO4 ,Na ,K 24.1ʅgm (16.2%)toPM2.5. – – andNO3 combinedwiththemedianloadingsofCl ,OC,EC,Cr,Se –3 – + and Cd. This source contributed 38.5ʅgm (25.9%) to PM2.5. It Thefactor4hadtracerspeciesofCl ,K ,Mn,Zn,Se,Sr,Cd,Ba was noted that the coal combustion was a major contributor to andPb.Theseelementswerequitecommoninnumberofsource PM2.5. categories, including metal industry, biomass burning, waste incineratorsandsoon(Brownetal.,2007).Therefore,inthisstudy, ThethirdfactorwascharacterizedbyhighloadingsofCr,Mn, this factor was indentified as the Ba, Mn and Zn source, which –3 Ni,Cu,Zn,As,Se,CdandPb.V,CrandMnarecloselyassociated contributed18.9ʅgm (12.7%)toPM2.5. withmetalprocessing(Yuetal.,2013).Moreover,steelmanufacͲ turinglocatedatwesternofHandanisapillarindustry.ZnandPb Thefifthfactorwassimilartothefourthfactor,anditshould areemittedprimarilyfromthesteelmills(Yangetal.,2013).This also be identified as a mixed source. Zn may be from waste Wei et al. – Atmospheric Pollution Research (APR) 765 incinerator,WhileK+couldbeseenasaveryimportantmarkerof ammoniumsulfaterangedfrom26.5Mm–1to635.2Mm–1withthe biomass burning. oil combustion for industry and utilities was mean value of221.0Mm–1 that accounted for 32.4% to the total –1 characterized by V and Ni, and As could be emitted from coal Bext, followed by 23.2% of EC (ranged from 58.2Mm to combustion.Eventhoughthesecompositionscomefromdifferent 352.0Mm–1),22.8%ofammoniumnitrate(rangedfrom14.5Mm–1 sources, they are produced by combustion. The fifth factor was to 492.9Mm–1) and 21.6% of OC (ranged from 46.8Mm–1 to named as K+, As and V source. This factor contributed 9.3ʅgm–3 356.5Mm–1). The similar conclusion was that the biggest contriͲ (6.3%)toPM2.5. butorofBextisammoniumsulfatereportedinJinan,inGuangzhou and in eastern United States (Sisler and Malm, 2000; Yang et al., + 2– – The sixth factor had high loadings of NH4 , SO4 and NO3 . 2007;Taoetal.,2009).Additionally,themeanvaluesofBextwere These ions, in the atmosphere, generated by the photochemical 582.2Mm–1 and 802.0Mm–1 in December and January, reaction of precursor gases (such as SO2 and NOX) emitted from respectively. Comparison with December, all of Bext of OC, coal combustion and motor vehicles. The sixth factor was thus ammonium nitrate and ammonium sulfate increased in January, –3 regarded as the secondary source that contributed 32.5ʅgm while Bext of EC decreased. A cause was that the absolute (21.8%)toPM2.5onaverage. concentrations of OC, ammonium nitrate and ammonium sulfate increase, yet the concentration of EC decreased in January. In TheseventhfactorwascharacterizedbyhighloadingofECand addition, for ammonium nitrate and ammonium sulfate, high – 2+ moderate loadings, of Cl , Ca , OC, Mn, Zn, Se, Sr and Pb. EC is relativelyhumiditycouldbeauxiliaryforenhancingBext.Themean emitted from the vehicle engines. Mn, Zn and Pb are typically relatively humidity was 76.2% in January, which was higher than relatedtothebrakeofthemotor(Yangetal.,2013).Inaddition,Zn 60.1%ofDecember. was a component of a common fuel detergent and anti–wear additiveandcouldpresentthedieselemissions,aswellasintires, OC was the biggest contributor to Bext, followed by EC, brakeliningsandpads(Brownetal.,2007).Ca2+couldbefromthe ammonium sulfate and ammonium nitrate in January 1–13. road dust. So the factor 7 was indentified as the motor source, However, in January 14–31, ammonium sulfate ranked number –3 whichcontributed11.5ʅgm (7.7%)toPM2.5. one, followed by ammonium nitrate, OC and EC. That was also likely due to the increased concentrations of water–soluble ions The eighth factor was represented by not only Ca2+ but also andthehigherRHinJanuary14–31.TheRHincreasedfrom 54.3% thehighloadingsofSr,Ba,Mg2+andthemedianloadingsofCr,Mn, of January 1–13 to 88.5% of January 14–31, respectively. NiandCu.Metalsulfates(e.g.,CaSO4andMgSO4)canbeformed Meanwhile,thef(RH)increasedfrom1.33to2.46. whenatmosphericH2SO4/SO2reactswithcrustalaerosols.Cr,Mn, NiandCuwereassociatedwithtailpipeemissions,suchasbrake 4.Conclusion andtirewear(Yuetal.,2013).Sothefactorisregardedastheroad –3 –3 dust source, which contributed 10.4ʅgm (6.9%) to PM2.5 on ThedailyaverageconcentrationofPM2.5was160.1ʅgm in average. HandanduringDecember1,2012toJanuary31,2013.Thewater– soluble ions contributed 48.3% of PM2.5. Almost a half of PM2.5 As shown in Table 2, the average predicted concentration of suggested water–soluble ions were a major fraction in PM2.5 in –3 PM2.5 by PMF was 149.0ʅgm , which was slightly lower than this period. The daily average concentration of January was 160.1ʅgm–3ofthemeasuredconcentrationasshowninTable1. 190.1ʅgm–3, which was higher than 135.1ʅgm–3 of December. ThehigherproportionofDecemberimpliedthattheproportionof 3.3.ChemicalcompositionforvisibilitydegradationBext water–solubleiondidn’tincreasealongwithincreaseofabsolute concentrationofPM2.5.ThedailyaverageconcentrationsofOCand –1 –3 –3 Visibility could be calculated by Bext (unit in Mm ) with ECwere26.3ʅgm and15.8ʅgm ,whichaccountedfor16.4% Bext=1.9/visibility(Schichteletal.,2001),whereBextwasextinction and9.9%ofPM2.5duringthisperiod,respectively. coefficient. Bext was calculated by the concentration of chemical componentsassomeresearchesmentioned,aIMPROVEEquation In the extremely polluted period of January 1–13, the daily –3 (5) was accepted and was used by other researches (Byun and average concentration of PM2.5 was 217.7ʅgm . However, the Ching, 1999; Sister and Malm, 2000). In this Equation, Malm and daily concentration of January 14–31 was only 139.5ʅgm–3. The 2– + – Day (2001) proposed a relative humidity (RH) growth function proportionofthesumofSO4 ,NH4 andNO3 inJanuary1–13was 2– – f(RH), indicated how efficiencies increase for SO4 and NO3 as 23.3%; yet the proportion of January 14–31 was 58.6%. It was 2– + – they absorbed liquid water. (Table 3 gives the detailed value of foundthatSO4 ,NH4 andNO3 presentedlowerconcentrationsin – f(RH)). Where, (NH4)2SO4=4.125 [S]; NH4NO3=1.29 [NO3 ]; POM theearlystageoftheepisode.Thedailyaverageconcentrationsof (particulate organic matter)=1.4 [OC]; fine soil=2.2 [Al]+2.19 [Si] OC and EC were 38.6ʅgm–3 and 20.1ʅgm–3 in January 1–13, +1.63 [Ca]+2.42 [Fe]+1.94 [Ti]; coarse mass=[PM10]–[PM2.5]; 10 respectively. In January 14–31, the average concentrations of OC representsclearairscattering. and EC were 26.1ʅgm–3 and 11.6ʅgm–3, respectively. It was noted that the concentrations of OC and EC presented higher B |3f(RH)[(NH ) SO +NH NO ]+4[POM]+10[EC] concentrationsintheearlystageoftheepisodethaninthelaterof ext 4 2 4 4 3 (5) +1[finesoil]+0.6[coarsemass]+10 episode. –3 Considering that the fine soil, coarse mass and clear air The predicted concentration of PM2.5 was 149.0ʅgm applyingPMFmodel,whichwasslightlylowerthan160.1ʅgm–3of scatteringhadlittleeffectontheBext(Cheungetal.,2005).Sothis studyusedtheequationasfollows: themeasuredconcentration.EightfactorswereidentifiedbyPMF method. They were coal combustion source, secondary source, industrysource,Ba,MnandZnsource,motorvehiclesource,road Bext|3f(RH)[(NH4)2SO4+NH4NO3]+4[POM]+10[EC] (6) + dustsource,K ,AsandVsource,fueloilcombustionsource,which contributed 25.9%, 21.8%, 16.2%, 12.7%, 10.9%, 7.7%, 6.3% and Figure 4 illustrated Bext of four major chemical components 2.5%ofPM2.5,respectively. during the different periods. The mean value of Bext was –1 682.1Mm in this period. The highest contributor to Bext was Wei et al. – Atmospheric Pollution Research (APR) 766 (a) (b) 800 700 700 600 600
500
500
) ) 1 1 Ͳ 400 Ͳ 400 0P Mm Mm
( 300 ( 0P 300 ext ext B 200 B 200 %H[W 100%H[W 100 0 0 Ͳ100 Ͳ100 2& (& 1+12 1+ 62 OC2& EC (& NH 1+124NO3 1+ (NH4 62)2SO4 OC EC NH4NO3 (NH4)2SO4
800 (c) 500 (d) 700 400 600
) 500 ) 1 1
Ͳ Ͳ 300 400 P Mm Mm 0 ( ( 0P 200
ext 300 ext B B 200 H[W % %H[W 100 100 0 0 Ͳ100 2& (& 1+12 1+ 62 2& (& 1+12 1+ 62 OC EC NH4NO3 (NH4)2SO4 OC EC NH4NO3 (NH4)2SO4
800 (e) 700 600 500 ) 1 Ͳ 400 Mm ( 0P 300 ext B 200 %H[W 100 0 Ͳ100 2& (& 1+12 1+ 62 OC EC NH4NO3 (NH4)2SO4
Figure4.ThemonthlyBextforkeycomponentsofPM2.5.(a),(b),(c),(d)and(e)figurerepresentsthewholeperiod,December,January, 1–13Januaryand14–31January,respectively. –1 The daily meanvalue of Bext was 682.1Mm during this peͲ Acknowledgment riod. The daily mean value of Bext ammonium sulfate was –1 221.0Mm whichaccountedfor32.4%ofBext,followedby24.9% This study was sponsored by the National Natural Science ofEC,22.8%%ofammoniumnitrateand21.6%ofOC.InJanuary FoundationofChina(No.41105105),theNaturalScienceFoundation 1–13, OC was the biggest contributor for Bext, followed by EC, of Hebei Province (No. D2011402019), the State Environmental ammonium sulfate and ammonium nitrate. However, in January Protection Key Laboratory of Sources and Control of Air Pollution 14–31, ammonium sulfate ranked number one, followed by Complex(No.SCAPC201307),theExcellentYoungScientistFoundation ammoniumnitrate,OCandEC.ThatwaslikelyduetohigherRHin ofHebeiEducationDepartment(No.YQ2013031),the Programforthe January14–31. OutstandingYoungScholarsofHebeiProvinceatHEBEU,China. Wei et al. – Atmospheric Pollution Research (APR) 767
Wei et al. – Atmospheric Pollution Research (APR) 768 Table 3. Statistical summary of mean f(RH) values in selected relative Malm, W.C., Day, D.E., 2001. Estimates of aerosol species scattering humidityranges characteristics as a function of relative humidity. Atmospheric Environment35,2845–2860. RH(%) f(RH) Paatero,P.,1997.Leastsquaresformulationofrobustnon–negativefactor 30a35 1.16 analysis.ChemometricsandIntelligentLaboratorySystems37,23–35. 35a40 1.21 Paatero, P., Tapper, U. 1994. Positive matrix factorization: a non–negative 40a45 1.22 factormodelwithoptimalutilizationoferrorestimatesofdatavalues. 45a50 1.27 Environments5,111–126. 50a55 1.33 Qin, Y.J., Kim, E., Hopke, P.K., 2006. The concentrations and sources of 55a60 1.38 PM2.5 in metropolitan New York City. Atmospheric Environment 40, S312–S332. 60a65 1.45 Schichtel,B.A.,Husar,R.B.,Falke,S.R.andWilson,W.E.,2001.Hazetrends a 65 70 1.55 over the United States, 1980–1995. Atmospheric Environment 35, 70a75 1.65 5205–5210.
75a80 1.83 Sister, J.F., Malm, W.C., 2000. Interpretation of trends of PM2.5 and 80a85 2.10 reconstructedvisibilityfromtheimprovenetwork.JournaloftheAir& WasteManagementAssociation50,775–789. 85a90 2.46 >90 3.17 Tao, J., Ho, K.F., Chen, L.G., Zhu, L.H., Han, J.L., Xu, Z.C., 2009. Effect of chemicalcompositionofPM2.5 onvisibilityinGuangzhou,China,2007 spring.Particuology7,68–75. References U.S.EPA(U.S.EnvironmentalProtectionAgency),2008.EPAPositiveMatrix Factorization(PMF)3.0Fundamentals&UserGuide. Brown, S.G., Frankel, A., Raffuse, S.M., Roberts, P.T., Hafner, H.R., Wang,L.T.,Wei,Z.,Yang,J.,Zhang,Y.,Zhang,F.F.,Su,J.,Meng,C.C.,Zhang, Anderson,D.J.,2007.Sourceapportionmentoffineparticulatematter Q., 2014. The 2013 severe haze over southern Hebei, China: Model inPhoenix,AZ,usingpositivematrixfactorization.JournaloftheAir& evaluation, source apportionment, and policy implications. WasteManagementAssociation57,741–752. AtmosphericChemistryandPhysics14,3151–3173. Byun,D.W.,ChingJ.K.S.,1999.ScienceAlgorithmsoftheEPAModels–3 Wang,L.T.,Xu,J.,Yang,J.,Zhao,X.J.,Wei,W.,Cheng,D.D.,Pan,X.M.,Su,J., Community Multiscale Air Quality (CMAQ) Modeling System, 2012. Understanding haze pollution over the southern Hebei area of EPA/600/R–99/030. chinausingtheCMAQmodel.AtmosphericEnvironment56,69–79. Cheung,H.C.,Wang,T.,Baumann,K.,Guo,H.,2005.Influenceofregional Wei, Z., Yang, J., Wang, L T., Wei, W., Zhang, F F. and Su, J., 2014. pollutionoutflowontheconcentrationsoffineparticulatematterand Characteristics of the severe haze episode in Handan city in January, visibility in the coastal area of Southern China. Atmospheric 2013. Environmental Science & Technology (In Chinese) (Article in Environment39,6463–6474. Press). Chueinta, W., Hopke, P.K., Paatero, P., 2000. Investigation of sources of Wei,W.,Wang,L.T.,Pan,X.M.,Cheng,D.D.andSu,J.,2013.Simulation atmospheric aerosol at urban and suburban residential areas in study of haze pollution sources in cities of Hebei Province. Thailandbypositivematrixfactorization.AtmosphericEnvironment34, EnvironmentalScience&Technology36,116–119(InChinese). 3319–3329. Yang, L.X., Cheng, S.H., Wang, X.F., Nie, W., Xu, P.J., Gao, X.M., Yuan, C., Dunker,A.M.,Morris,R.E.,Pollack,A.K.,Schleyer,C.H.,Yarwood,G.,1996. Wang,W.X.,2013.SourceidentificationandhealthimpactofPM ina Photochemicalmodelingoftheimpactoffuelsandvehiclesonurban 2.5 heavilypollutedurbanatmosphereinChina.AtmosphericEnvironment ozone using auto oil program data. Environmental Science & 75,265–269. Technology30,787–801. Yang,L.X.,Wang,D.C.,Cheng,S.H.,Wang,Z.,Zhou,Y.,Zhou,X.H.,Wang, Karnae,S.,John,K.,2011.Sourceapportionmentoffineparticulatematter W.X., 2007. Influence of meteorological conditions and particulate measured in an industrialized coastal urban area of south Texas. matteronvisualrangeimpairmentinJinan,China.ScienceoftheTotal AtmosphericEnvironment45,3769–3776. Environment383,164–173. Kim,E.,Hopke,P.,2008.SourcecharacterizationofambientĮneparticles Yu,L.D.,Wang, G.F.,Zhang,R.J.,Zhang,L.M.,Song,Y.,Wu,B.B.,Li,X.F.,An, at multiple sites in the Seattle area. Atmospheric Environment 42, K.,Chu,J.H.,2013.CharacterizationandsourceapportionmentofPM 6047–6056. 2.5 inanurbanenvironmentinBeijing.AerosolandAirQualityResearch Lee, E., Chan, C.K., Paatero, P., 1999. Application of positive matrix 13,574–583. factorizationinsourceapportionmentofparticulatepollutantsinHong Kong.AtmosphericEnvironment33,3201–3212. MEP(MinistryofEnvironmentalProtection),2012.ChinaNationalAmbient AirQualityStandards,GB3095–2012.