+ All Categories
Home > Documents > Seasonal effect and source apportionment of polycyclic aromatic hydrocarbons in PM2.5

Seasonal effect and source apportionment of polycyclic aromatic hydrocarbons in PM2.5

Date post: 04-Mar-2023
Category:
Upload: ukm
View: 0 times
Download: 0 times
Share this document with a friend
13
Seasonal effect and source apportionment of polycyclic aromatic hydrocarbons in PM 2.5 Md Firoz Khan a, * , Mohd Talib Latif a, b , Chee Hou Lim b , Norhaniza Amil b, c , Shofan Amin Jaafar b , Doreena Dominick a, b , Mohd Shahrul Mohd Nadzir a, b , Mazrura Sahani d , Norhayati Mohd Tahir e a Centre for Tropical Climate Change System (IKLIM), Institute for Climate Change, Universiti Kebangsaan Malaysia, 43600 Bangi, Selangor, Malaysia b School of Environmental and Natural Resource Sciences, Faculty of Science and Technology, Universiti Kebangsaan Malaysia, 43600 Bangi, Selangor, Malaysia c School of Industrial Technology (Environmental Division), Universiti Sains Malaysia,11800 Penang, Malaysia d Environmental Health and Industrial Safety Program, School of Diagnostic Science and Applied Health, Faculty of Health Sciences, Universiti Kebangsaan Malaysia, Jalan Raja Muda Abdul Aziz, 50300 Kuala Lumpur, Malaysia e Environmental Research Group, School of Marine Science and Environment, Universiti Malaysia Terengganu, 21030 Kuala Terengganu, Terengganu, Malaysia highlights graphical abstract Sixteen USEPA priority PAHs deter- mined in PM 2.5 at a tropical semi- urban site. High molecular weight PAHs are signicantly higher in PM 2.5 . The combustion of gasoline, diesel and heavy oil are dominant sources of PAHs. No potential carcinogenic risk of the airborne BaP eq was found at current site. Monsoon effect inuences the PAHs distributions as well as health risk. article info Article history: Received 6 August 2014 Received in revised form 28 January 2015 Accepted 30 January 2015 Available online 2 February 2015 Keywords: Monsoon effect PAH diagnostic ratio Positive matrix factorization Health risk abstract This study aims to investigate distribution and sources of 16 polycyclic aromatic hydrocarbons (PAHs) bound to ne particulate matter (PM 2.5 ) captured in a semi-urban area in Malaysia during different seasons, and to assess their health risks. PM 2.5 samples were collected using a high volume air sampler on quartz lter paper at a ow rate of 1 m 3 min 1 for 24 h. PAHs on the lter paper were extracted with dichloromethane (DCM) using an ultrasonic centrifuge solid-phase extraction method and measured by gas chromatographyemass spectroscopy. The results showed that the range of PAHs concentrations in the study period was between 0.21 and 12.08 ng m 3 . The concentrations of PAHs were higher during the south-west monsoon (0.21e12.08 ng m 3 ) compared to the north-east monsoon (0.68e3.80 ng m 3 ). The high molecular weight (HMW) PAHs (5 ring) are signicantly prominent (>70%) compared to the low molecular weight (LMW) PAHs (4 ring) in PM 2.5 . The Spearman correlation indicates that the LMW and HMW PAHs correlate strongly among themselves. The diagnostic ratios (DRs) of I[c]P/I[c]P þ BgP and B[a]P/B[g]P suggest that the HMW PAHs originated from fuel combustion sources. The source appor- tionment analysis of PAHs was resolved using DRs-positive matrix factorization (PMF)-multiple linear regression (MLR). The main sources identied were (a) gasoline combustion (65%), (b) diesel and heavy * Corresponding author. E-mail addresses: md[email protected], md[email protected] (M.F. Khan). Contents lists available at ScienceDirect Atmospheric Environment journal homepage: www.elsevier.com/locate/atmosenv http://dx.doi.org/10.1016/j.atmosenv.2015.01.077 1352-2310/© 2015 Elsevier Ltd. All rights reserved. Atmospheric Environment 106 (2015) 178e190
Transcript

lable at ScienceDirect

Atmospheric Environment 106 (2015) 178e190

Contents lists avai

Atmospheric Environment

journal homepage: www.elsevier .com/locate/atmosenv

Seasonal effect and source apportionment of polycyclic aromatichydrocarbons in PM2.5

Md Firoz Khan a, *, Mohd Talib Latif a, b, Chee Hou Lim b, Norhaniza Amil b, c,Shoffian Amin Jaafar b, Doreena Dominick a, b, Mohd Shahrul Mohd Nadzir a, b,Mazrura Sahani d, Norhayati Mohd Tahir e

a Centre for Tropical Climate Change System (IKLIM), Institute for Climate Change, Universiti Kebangsaan Malaysia, 43600 Bangi, Selangor, Malaysiab School of Environmental and Natural Resource Sciences, Faculty of Science and Technology, Universiti Kebangsaan Malaysia, 43600 Bangi,Selangor, Malaysiac School of Industrial Technology (Environmental Division), Universiti Sains Malaysia, 11800 Penang, Malaysiad Environmental Health and Industrial Safety Program, School of Diagnostic Science and Applied Health, Faculty of Health Sciences,Universiti Kebangsaan Malaysia, Jalan Raja Muda Abdul Aziz, 50300 Kuala Lumpur, Malaysiae Environmental Research Group, School of Marine Science and Environment, Universiti Malaysia Terengganu, 21030 Kuala Terengganu,Terengganu, Malaysia

h i g h l i g h t s

* Corresponding author.E-mail addresses: [email protected], md

http://dx.doi.org/10.1016/j.atmosenv.2015.01.0771352-2310/© 2015 Elsevier Ltd. All rights reserved.

g r a p h i c a l a b s t r a c t

� Sixteen USEPA priority PAHs deter-mined in PM2.5 at a tropical semi-urban site.

� High molecular weight PAHs aresignificantly higher in PM2.5.

� The combustion of gasoline, dieseland heavy oil are dominant sourcesof PAHs.

� No potential carcinogenic risk of theairborne BaPeq was found at currentsite.

� Monsoon effect influences the PAHsdistributions as well as health risk.

a r t i c l e i n f o

Article history:Received 6 August 2014Received in revised form28 January 2015Accepted 30 January 2015Available online 2 February 2015

Keywords:Monsoon effectPAH diagnostic ratioPositive matrix factorizationHealth risk

a b s t r a c t

This study aims to investigate distribution and sources of 16 polycyclic aromatic hydrocarbons (PAHs)bound to fine particulate matter (PM2.5) captured in a semi-urban area in Malaysia during differentseasons, and to assess their health risks. PM2.5 samples were collected using a high volume air sampleron quartz filter paper at a flow rate of 1 m3 min�1 for 24 h. PAHs on the filter paper were extracted withdichloromethane (DCM) using an ultrasonic centrifuge solid-phase extraction method and measured bygas chromatographyemass spectroscopy. The results showed that the range of PAHs concentrations inthe study period was between 0.21 and 12.08 ng m�3. The concentrations of PAHs were higher during thesouth-west monsoon (0.21e12.08 ng m�3) compared to the north-east monsoon (0.68e3.80 ng m�3).The high molecular weight (HMW) PAHs (�5 ring) are significantly prominent (>70%) compared to thelow molecular weight (LMW) PAHs (�4 ring) in PM2.5. The Spearman correlation indicates that the LMWand HMW PAHs correlate strongly among themselves. The diagnostic ratios (DRs) of I[c]P/I[c]Pþ BgP andB[a]P/B[g]P suggest that the HMW PAHs originated from fuel combustion sources. The source appor-tionment analysis of PAHs was resolved using DRs-positive matrix factorization (PMF)-multiple linearregression (MLR). The main sources identified were (a) gasoline combustion (65%), (b) diesel and heavy

[email protected] (M.F. Khan).

M.F. Khan et al. / Atmospheric Environment 106 (2015) 178e190 179

oil combustion (19%) and (c) natural gas and coal burning (15%). The health risk evaluation, by means ofthe lifetime lung cancer risk (LLCR), showed no potential carcinogenic risk from the airborne BaPeq(which represents total PAHs at the present study area in Malaysia). The seasonal LLCR showed that thecarcinogenic risk of total PAHs were two fold higher during south-westerly monsoon compared to north-easterly monsoon.

© 2015 Elsevier Ltd. All rights reserved.

1. Introduction

Polycyclic aromatic hydrocarbons (PAHs) are ubiquitous, semi-volatile, persistent organic pollutants (POPs) and also environ-mental carcinogens. PAHs are released into ambient air through theincomplete combustion of organic materials, with major anthro-pogenic sources including fossil fuel burning, motor vehicle emis-sions, waste incineration, oil refining, the coke and steel industryand coal combustion (Call�en et al., 2014; Harrison et al., 1996;Hedberg et al., 2005; Jang et al., 2013). In urban, industrial-urbanand semi-urban areas, emissions from motor vehicles are recog-nized as the main source of PAHs in ambient air (Fang et al., 2006;Jiang et al., 2009; Sarkar and Khillare, 2013; Sharma et al., 2007). Ingeneral, sources of PAHs can be broadly classified into two maingroups: pyrogenic and petrogenic, where the pyrogenic are knownas the dominant sources. The pyrogenic sources include oil de-rivatives, coal combustion, natural gas, and traffic-related pollution,whereas petrogenic sources consist of direct contamination such asthe spillage of oil products (Liu et al., 2009).

Particle-bound PAHs are considered to be very hazardous tohuman health. The inhalation of PAHs has been associated with anincrease in cancer risk (Armstrong et al., 2004). A study onoccupation-related PAH exposure has reported increased lung andbladder cancer risks (Bosetti et al., 2007). Due to their toxicokineticeffects, seven PAHs congeners have been classified as carcinogenicby the United States Environmental Protection Agency (US EPA) (USEPA, 2010). These are benzo(a)anthracene, benzo(a)pyrene, ben-zo(b)fluoranthene, benzo(k)fluoranthene, chrysene, dibenzo(a,h)anthracene and indeno(1,2,3-cd)pyrene. At levels above 1.0 ng m�3,benzo(a)pyrene has been predicted to cause a greater genomicfrequency of translocation, micronuclei and DNA fragmentation.PAHs are also of great concern due to their non-carcinogenic effects,i.e. intrauterine growth restriction, bronchitis, asthma and asthma-like symptoms and fatal ischaemic heart disease (Choi et al., 2010).Based on the long list of health effects on humans, 16 PAHs havebeen selected by the United States (US) Environmental ProtectionAgency (EPA) and China EPA to be monitored by their regulatorybodies, while the European Union has selected 15þ1 (the addi-tional was PAH highlighted by the joint Food and Agriculture Or-ganization/World Health Organization (FAO/WHO) expertcommittee on food additives (JECFA)) priority PAHs; eight of themare also listed in the US EPA 16 PAHs (EU, 2005).

As an industrial developing country, the emission of PAHs to theambient air cannot be avoided in Malaysia. An initial study by Abasand Simoneit (1996) has suggested that organic matter in urbanareas was derived from biogenic sources and the anthropogenicutilization of fossil fuel products, with greater PAH concentrationsfound during haze episodes. Omar et al. (2002) first presentedvehicular emissions as the dominant source of PAHs in PM10 inKuala Lumpur atmospheric particles, with benzo(g,h,i)perylene andcoronene reported as the most abundant PAHs. A detailed analysisof the PAHs in total suspended particulatematter (TSP), during hazeand non-haze episodes at two urban areas, revealed that biomassburning, vehicular emissions, urban activities and natural sources

are contributors to PAH concentrations back in 1997 (Abas et al.,2004). A further study by Omar et al. (2006) revealed that theambient and street level distribution of PAHswere similar andwereattributed to vehicular emissions. In addition, they also reported adifferent pattern of selected PAHs during haze episodes and thatsmoke haze particles had a potential health risk four times higherthan that of street level particles. A study at a very busy highwaytoll station has shown that PAHs with car exhaust characteristicswere present in plant leaves, which further indicates the influenceof traffic on the concentrations of PAHs (Azhari et al., 2011). Arecent study conducted by Jamhari et al. (2014) reintroducedBenzo(g,h,i)perylene as an abundant PAH in PM10 and furthersuggested traffic emissions are the main source of PAHs in Malay-sian city areas. Overall, PAHs in urban and sub-urban areas inMalaysia have been associated with traffic and haze.

Multivariate receptor models are very useful tools in the studiesof source apportionment of pollutants, especially in environmentalstudies. This established method is also used in the studies ofsources of PAHs. The most commonly and widely used receptormodels are: a) chemical mass balance models (CMB) (Watson et al.,1990), b) positive matrix factorization (PMF) (Paatero and Tapper,1994), c) UNMIX (Henry, 1987), d) principal component analysiscoupled with absolute principal component score (PCA/APCS)(Thurston and Spengler, 1985). Among these multivariate receptormodelling techniques, PMF is the most preferred and trusted one.The first and foremost advantage of this procedure is that the priorsource information, or priori knowledge, of pollutants is notnecessary. This model uses a weighted least-squares fit, includeserrors as an input and can impose non-negativity constraints,weighing each data point individually (Paatero, 1997; Paatero andTapper, 1994). Missing values, noisy data, outliers, and valuesbelow detection limit can be treated and made use of in the PMFprocedure (Baumann et al., 2008; Khan et al., 2012; Polissar et al.,1998a, 1998b). Application of PMF in the apportionment ofparticle-bound PAHs can provide robust and accurate resultscompared to the PCA-multiple linear regression (MLR).

Studies of the source apportionment and health impact of PAHsare very important in Malaysia due to various sources of combus-tion, particularly from motor vehicles, industrial activities andbiomass burning. Identification of the sources of PAHs in PM2.5 by arobust and accurate receptor model has yet to be performed inMalaysia. With the advancement of receptor modelling as well asanalytical analysis, together with health risk assessment, accurateand detailed analysis of the distribution of PAHs is now possible.Therefore, the main objectives of this study are to investigate theseasonal distribution of 16 PAHs in PM2.5 ambient aerosol and todetermine the source of 16 PAHs by means of a diagnostic ratio(DR)-PMF-MLR model. This study also aims to assess the healtheffects of PAHs using a health risk assessment approach whichdetermines the lifetime lung cancer risk (LLCR) based on airborneBaPeq that represents the 16 primary PAHs in PM2.5.

M.F. Khan et al. / Atmospheric Environment 106 (2015) 178e190180

2. Material and methods

2.1. Description of sampling site

This sampling campaign was stationed on the rooftop of theBiology Building (5 storeys) at the Faculty of Science and Technol-ogy, Universiti Kebangsaan Malaysia (UKM), about 65 m above sealevel. Located in a semi-urban area, UKM Bangi is about 20 km tothe south of central Kuala Lumpur (Fig. 1). The nearest main roadsare about 300 m from the site. Predominantly light duty vehiclestravel on these roads. The two nearest highways (PLUS Highwayand SILK Highways) are about 1e2 km away from the monitoringsite, used by both heavy duty and light vehicles. In addition, the siteitself is surrounded by dense forest. Sampling period covers twomajor monsoons of Malaysia i.e. the south-west monsoon (SW;June 21, 2013 to September 5, 2013) and the north-east monsoon(NE; January 29, 2014 to February 18, 2014). The total number ofsamples taken was 34 for the entire sampling period. The yearlyambient temperature and relative humidity in this area were 29 �C(25 �Ce36 �C) and 80% (33%e100%), respectively. During the SWmonsoon, the temperature and relative humidity were 29 �C(25 �Ce36 �C) and 77% (33%e100%), respectively. However, in theNE monsoon, the prevailing temperature and relative humiditywere 28 �C (21 �Ce38 �C) and 74% (20%e100%), respectively (http://www.wunderground.com). The backward trajectories calculated atthe sampling location are shown in Fig. 2. The Hybrid Single-Particle Lagrangian Integrated Trajectory Model (HYSPLIT 4.9) andfire hotspot data from the Moderate Resolution Imaging Spectror-adiometer (MODIS) were employed for biomass fire hotspots and

Fig. 1. Sampling location of PM

the backward trajectories. Overall, the air mass was transportedfrom the south-westerly direction in the month of June, 2013 toSeptember, 2013 and the north-easterly wind was predominant inthe month of December, 2013 to March, 2014.

2.2. PM2.5 sampling

The PM2.5 samples were collected on quartz microfibre filters(203 mm� 254 mm, Whatman™, UK) using a PM2.5 Tisch HighVolume Sampler at a flow rate of 1.13 m3 h�1. As part of the prep-aration, the filters were prebaked at 500 �C for 3 h to remove anydeposited organic compounds. Prior to weighing, the blank filterswere conditioned in a desiccator for 24 h to ensure the equilibriumof mass concentration. Likewise, after sampling, the exposed filterpapers were left in a desiccator for 24 h prior to weighing. The filterpapers were weighed using a five-digit high resolution electronicbalance (A&D, GR-202, Japan) with a 0.01 mg detection limit. Thefilter samples were then refrigerated at 4 �C until the extraction ofPAHs was carried out.

2.3. Extraction of PAHs using solid phase extraction-silica column

The filter samples were cut into pieces which were placeddirectly into 50 mL centrifuge tubes. Dichloromethane (DCM) (R &M Chemicals, UK) was used as extraction solvent. 20 mL of DCMwere added into the centrifuge tube with filter samples. In theextraction procedure, ultrasonic vibration, centrifuge and me-chanical shaking were applied as described by Sun et al. (1998). Thesamples were then sonicated in an ultrasonic bath (Elmasonic

2.5 at Bangi area, Malaysia.

Fig. 2. The cluster of backtrajectories and biomass burning fire hotspot during (a) south-westerly (SW) and (b) north-easterly (NE) monsoon in 2013e2014.

M.F. Khan et al. / Atmospheric Environment 106 (2015) 178e190 181

M.F. Khan et al. / Atmospheric Environment 106 (2015) 178e190182

S70H, Elma, Germany) for 20 min in total. The extraction solutionswere then centrifuged at 2500 rpm (Kubota 5100, Japan) for 10 minbefore shaken using a vortexmixer for 10 min. The sonication andcentrifuged steps were repeated for three times before the extractwas filtered using glass microfibre filters (Whatman™, UK). Thesolution volume was then reduced to about 200 mL under a gentlestream of nitrogen gas (N2) before 800 mL of n-hexanewas added toreconstitute the residue. The volume suppression process wasrepeated for two more times. Next, silica SPE cartridges (Lichrolut®RP-18, Merck, Germany) were used for the clean-up process andpre-concentration of the PAH samples and RP-18 was subjected toconditioning by 10 mL of n-hexane (Friendemann Schmidt, Ger-many) before the extraction solutions were loaded and passedthrough the cartridge under gentle vacuum. The RP-18 cartridgeswere then eluted by DCM:n-hexane (1:9) at a flow rate of1 mL min�1. The n-hexane was used as an eluent at this stage. Theeluates were collected into 20 mL centrifuge tubes. The volume ofeluates were further reduced under a gentle stream of N2 gas to500 mL and reconstituted with n-hexane up to 1.5 mL into vial to beused in the GCeMS analysis.

2.4. Analysis of PAHs using GCeMS

The samples were analysed using gas chromatographyemassspectroscopy (GCeMS) (Agilent, 5975C, USA). A capillary column(HP-5MS) of internal diameter (id) 0.25 mm, length 30 m andthickness 0.2 5 mm was used with the GCeMS. The selected ionmonitoring (SIM) mode was used to collect the data which givesmore sensitivity than the full scan mode. External calibration wasused for the quantification of each of 16 PAHs with the standardmixtures of PAHs (SS EPA 610 PAHMix, Supelco, USA). The list of theUS EPA 16 PAHs is: naphthalene (NAP), acenaphthene (ACP), ace-naphthylene (ACY), anthracene (ANT), fluorene (FLR), phenan-threne (PHE), fluoranthene (FLT), pyrene (PYR), benzo(a)anthracene (B[a]A), chrycene (CHY), benzo(b)fluoranthene (B[b]F),benzo(k)fluoranthene (B[k]F), benzo(a)pyrene (B[a]P), indeno[1,2,3-cd]pyrene (I[c]P), dibenzo[h]anthracene (D[h]A), and benzo[g,h,i]perylene (B[g]P). In addition, at the beginning of the extrac-tion, we added 500 ppb each of Chrysene-D12 (Supelco, USA) andPerylene-D12 (Supelco, USA) as surrogate standards to therandomly selected samples. The average recoveries (%) were 92 and84 for Chrysene-D12 and Perylene-D12, respectively. The respectiveranges of recovery were 76e106% and 76e90% for Chrysene-D12and Perylene-D12. The overall average recovery (%) of surrogatestandards of Chrysene-D12 and Perylene-D12 was used to makecorrection of the concentration of each of 16 PAHs. The overall re-covery of each PAH ranges from 59% for NAP to 175% for B[a]P,determined from the standardmixtures of 16 PAHs (SS EPA 610 PAHMix, Supelco, USA) by external calibration. The limit of detection(LOD) of each PAH was calculated as three times the standard de-viation of the eight replicates. The estimated LOD (ng m�3) of eachPAH were: NAP e 0.08, ACP e 0.08, ACY e 0.07, ANT e 0.10, FLR e

0.12, PHE e 0.08, FLT e 0.10, PYR e 0.04, B[a]A e 0.13, CHY e 0.10, B[b]Fe 0.10, B[k]Fe 0.08, B[a]Pe 0.08, I[c]Pe 0.13, D[h]Ae 0.11, andB[g]P e 0.05.

2.5. Source apportionment techniques

2.5.1. Diagnostic ratio (DR)The use of DRs is a common conventional method for deter-

mining the potential sources of PAH congeners. The biggestadvantage of this method is that it is less complicated and easier tobe interpreted than other methods. DRs of two PAH congeners orone PAH and the total PAHs are used as an indicator of particularsource. The DR of higher molecular weight (HMW) (�5-ring) PAHs

are used with confidence as these molecules are quite stable (Alamet al., 2013). However, like every other analysis tool or method,there is a shortcoming (Yunker et al., 2002; Jamhari et al., 2014).The DR of low molecular weight (LMW) compounds (�4-ring),which are less stable and more susceptible to atmospheric pro-cesses, requires well-defined samples to establish the thresholdvalues (Alam et al., 2013). For example, ANT, B[a]A, B[a]P anddibenzo[a]pyrene are the most reactive compounds towards ozoneoxidation, and due to rapid photodegradation, the atmosphericlifetime of these compounds is relatively short (Perraudin et al.,2007). Since well-defined samples are necessary to establish thethreshold values (Alam et al., 2013), LMW compounds are thereforean issue in source apportionment analysis. To avoid these short-comings, numerical i.e. the PMF has been used as an alternativeand/or complementary tool to identify the sources of PAHs. Asmentioned earlier, PMF imposes non-negativity constraints andactually weighs each data point individually. The factors in PMF arenot necessarily orthogonal to each other which resembles theobservation of real-source signatures that are also not orthogonalto each other therefore overcoming the limitations of DR (Aydinet al., 2014; Galarneau, 2008).

2.5.2. Positive matrix factorization (PMF) and multiple linearregression (MLR)

PAH source apportionment analyses were conducted using theUS EPA PMF 5.0 model of the United States Environmental Pro-tection Agency (US EPA) as suggested by Norris et al. (2014). PMF isa factor-based receptor model that decomposes a matrix of sampledata into two matrices, i.e. chemical compositions and the contri-bution of each factor to each sample. Mathematically, PMF can bedefined as (Eq. (1))

XIJ ¼XPK¼1

gikf kj þ eij (1)

where, XIJ is the concentration of jth species on ith day, p is thenumber of factors, gik is the contribution of kth factor on the ith day,fkj is the factor profile of jth species on kth factor and eij is the re-sidual matrix for the jth species measured in the ith sample.

The PMF solution minimizes the object function Q with theadjusted value of g, f and p. Q is defined by (Eq. (2))

Q ðEÞ ¼Xmi¼1

Xnj¼1

"eijSij

#2¼

Xmi¼1

Xnj¼1

"Xij �

Ppk¼1gikf kjSij

#2(2)

where Sij is an estimate of the uncertainty in the jth species in thejth sample. PMF 5.0 operates in a robust mode by default, whichdownweighs the outliers affecting the fitting of the contributionsand profiles. To run PMF, two data files are needed as input for eachsample: 1) concentration and 2) uncertainty. The concentration ofeach PAH was pretreated and validated based on the noisy oroutliers, missing and/or values below method detection limit(MDL). The variables with outliers are excluded if there were any.The missing values were replaced with half of the mean value andthe species with concentrations belowMDL were replaced with thehalf of the MDL (Baumann et al., 2008; Polissar et al., 1998a, 1998b).The second data file is the uncertainty value of each variable byeach sample. As we could not have any measurement or method-ological data to calculate errors, the following empirical equation(Eq. (3)) was used to estimate the error of the speciesconcentration:

M.F. Khan et al. / Atmospheric Environment 106 (2015) 178e190 183

sij ¼ 0:01�Xij þ Xj

�(3)

where sij is the estimated measurement error for jth species in theith sample, Xij is the observed PAHs concentration and Xj is themean value of each PAH. The factor 0.01 was determined throughtrial and error procedures. Ogulei et al. (2006a; 2006b) uses thismethod in estimation of uncertainty. Thus, the measurement ofuncertainty (Sij) can be computed with the following (Eq. (4)):

Sij ¼ sij þ C3Xij (4)

where sij is the estimation of measurement error (Eq. (4)) and C3 isa constant. This empirical procedure was used to estimate the un-certainty of variables if there were measurements or methodo-logical data to estimate errors (Ogulei et al., 2006a,b). We alsoapplied similar a step for the values below detection limit. Harrisonet al. (2011) and Hedberg et al. (2005) applied this procedure ofuncertainty estimation to their work. Our observations of each PMFrun using the variability of C factor (Eq. (4) in the methodology ofPMF 5.0) value are summarized in Supplementary Table 1. In theprocedure of PMF 5.0, we selected the value of 0.2 for C as the endcalculation was optimized with lower error (%) and the Q true/Qexp was 1.17. An additional 5% uncertainty was added to account formethodological errors in the preparation of filter papers, gravi-metric mass measurements and preparing the calibration curves.

The model output of source contribution is provided asnormalized or dimensionless (average of each factor contribution isone). Therefore, the mass concentrations of the identified sourceswere scaled by using the following MLR analysis (Eq. (5))

Mi ¼ S0 þXpk¼1

Skgik (5)

where, Mi is the concentration of total concentration of PAHs in ithsample, Sk is the scaling constant, and gik the source contribution(average¼ 1) found in the result of PMF modelling. Several otherresearchers have successfully applied this MLR approach to expressthe output of PMF (Call�en et al., 2014; Hedberg et al., 2005; Khanet al., 2012).

Table 1Summary results of PAHs in PM2.5 samples 2013e2014.

PAHs(ng m�3)

PAHs

Overall mean(range ofconcentration)

aSW(range ofconcentration)

bNE(range ofconcentration)

NAP 0.09 (n.d.e0.25) 0.10 (0.01e0.22) 0.08 (n.d.e0.25)ACY 0.03 (n.d.e0.16) 0.04 (n.d.e0.16) 0.03 (n.d.e0.12)ACP 0.12 (0.02e0.65) 0.18 (0.02e0.65) 0.07 (0.03e0.21)FLR 0.06 (n.d.e0.35) 0.09 (n.d.e0.35) 0.04 (n.d.e0.10)ANT 0.04 (n.d.e0.16) 0.06 (n.d.e0.16) 0.02 (n.d.e0.05)PHE 0.04 (n.d.e0.19) 0.06 (n.d.e0.19) 0.02 (n.d.e0.08)FLT 0.09 (n.d.e0.37) 0.16 (n.d.e0.37) 0.04 (0.01e0.09)PYR 0.07 (n.d.e0.28) 0.12 (n.d.e0.28) 0.04 (0.01e0.07)B[a]A 0.04 (n.d.e0.16) 0.06 (n.d.e0.16) 0.03 (n.d.e0.06)CYR 0.09 (n.d.e0.43) 0.13 (n.d.e0.43) 0.05 (0.01e0.09)B[k]F 0.25 (n.d.e0.97) 0.33 (n.d.e0.97) 0.18 (0.05e0.40)B[a]P 0.30 (0.04e1.08) 0.37 (0.04e1.08) 0.23 (0.09e0.47)B[b]F 0.57(0.01e2.67) 0.75 (0.01e2.67) 0.42 (0.13e0.89)I[c]P 0.36 (0.04e1.72) 0.51 (0.04e1.72) 0.22 (0.06e0.53)D[h]A 0.16 (0.01e0.64) 0.22 (0.01e0.64) 0.11 (0.04e0.27)B[g]P 0.54 (0.06e3.16) 0.81 (0.06e3.16) 0.31 (0.07e0.80)Total PAHs 2.79 (0.21e12.08) 3.85 (0.21e12.08) 1.85 (0.68e3.80)

n.d.: not detected.a SW: south-westerly.b NE: north-easterly.

2.6. Health risk assessment

The most appropriate indicator to assess the carcinogenic po-tential of PAHs in air is B[a]P, as recommended by theWorld HealthOrganization (WHO) and it has been often used in relevant studiesas a reference compound (WHO, 1987). B[a]P is the most widelystudied PAH, and much of the information on the toxicity andoccurrence of PAHs is based on this particular compound. However,the use of B[a]P alone might underestimate the carcinogenic po-tential of airborne PAH mixtures, since co-occurring substances arealso carcinogenic (WHO, 1987). Therefore, the Benzo[a]pyreneequivalent concentration (BaPeq) was estimated using (Eq. (6))

XBaPeq ¼

Xn¼1

i

Ci � TEFi (6)

Where Ci is the concentration of the ith target PAH, TEFi is the toxicequivalency factor of the ith target compound. The toxic equiva-lency factor (TEF) of the PAHs was taken from a study by Nisbet andLaGoy (1992). The carcinogenic risk of each PAH as a lifetime lungcancer risk (LLCR) was calculated using: (Eq. (7))

LLCR ¼X

BaPeq � URBaP (7)

BaPeq is estimated by Eq. (6). Recommended by the WHO, theinhalation cancer unit risk (URBaP) for PAHs is 8.7 10�5 which sig-nifies that there are 8.7 cases per 100,000 people with chronicinhalation exposure to 1 ng m�3 BaP over a 70 year lifetime (WHO,2000).

3. Results and discussions

3.1. Level of PAHs associated with PM2.5

Table 1 shows a summary of the concentrations of PAHs inPM2.5. The individual and total PAHs data is shown as overall meanvalues and values for the SWand NEmonsoon seasons. The averageand ranges of total PAHs were 2.79 (0.21e12.08), 3.85 (0.21e12.08)and 1.85 (0.68e3.80) ng m�3 for the samples collected during thetotal sampling campaign, the SW monsoon and NE monsoonrespectively. The concentrations of PAHs during the SW monsoonare almost double the NE monsoon results which indicates morecarbonaceous aerosol in the surrounding area during that timeperiod. The high concentrations of particles during the SWmonsoon, due to biomass burning and haze episodes in SoutheastAsia, is expected to contribute to the amount of PAHs in ambient air.According to Anwar et al. (2010), the high amount of particles inambient air during these episodes has the ability to trap localemissions, particularly from motor vehicles and industrialactivities.

The individual PAH concentrations in PM2.5 are shown in Table 1.The average concentrations of PAHs have the decreasing order of B[b]F> B[g]P> I[c]P> B[a]P> B[k]F>D[h]A>ACP> FLT>NAP> CHY> PYR> FLR> PHE>ANT> B[a]A>ACY.Overall, the 5- and 6-ring PAHs, i.e. B[k]F, B[a]P, B[b]F, I[c]P, D[h]A,and B[g]P, are dominant. The same results are reported for both theSW and NE monsoons (Table 1). However, the concentrations of 6-ring PAHs in the overall dataset (40%) are higher than in the 5-ringPAH (38%). Furthermore, the 6-ring PAHs were more dominant inthe SW monsoon than the NE monsoon (Fig. 3b and c). Lowerconcentrations are represented by the 2, 3- and 4-ring PAHs (NAP,

Fig. 3. Contributions of PAHs by number of rings (%) to total PAHs at (a) overall period, (b) south-westerly (SW) and (c) north-easterly (NE) monsoon.

M.F. Khan et al. / Atmospheric Environment 106 (2015) 178e190184

ACY, ACP, FLR, ANT, PHE, FLT, PYR, B[a]A and CHY) which wererelatively lower in the NE monsoon compared to the SW monsoon.The seasonal variability of the paired samples t-test shows a sig-nificant difference (t¼ 2.23, p< 0.05) between the SW and NEmonsoons. The Klang Valley in Peninsular Malaysia was heavilyaffected by transboundary haze pollution from Sumatra ofIndonesia during the SW monsoon (Koe et al., 2001). A study byJuneng et al. (2011) found that local meteorological conditionsgoverned the largest day-to-day variations of PM10 concentrationsduring the dry SW monsoon. However, during the wet NEmonsoon, the transboundary pollution is thought to have an effecton the concentration of aerosol particles from the Chinese region.Thus, a strong seasonal dependency of the total PAHs was also

Table 2Comparison of ambient PAHs concentrations (ng m�3) with other studies.

City Location type PM fractions

Bangi, Malaysia Semi-urban PM2.5

Bangi, Malaysia Semi-urban PM10

Kuala Lumpur, Malaysia Urban PM10

Kuala Lumpur, Malaysia Urban PM10

Qingdao, China Urban roadside PM2.5

Huaniao Island, China Island PM2.5

Mount Taishan, China Mountain PM2.5

Guangzhou, China Urban PM2.5

Hung Hom, Hong Kong Roadside PM2.5

Saitama, Japan Roadside PM2.5

Seoul, South Korea Urban PM2.5

Barcelona, Spain Urban PM1

Montseny, Spain Rural PM1

Atlanta, USA Urban PM2.5

North Carolina, USA Urban PM2.5

Houston, USA Urban PM2.5

Rio de Janeiro, Brazil Urban PM2.5

Sao Paulo, Brazil Urban PM2.5

observed in this study.Direct comparison of PAHs in PM2.5 found in this study with

those found in the surrounding area inMalaysia is not possible as todate all reported work have focused on PAHs in PM10. For example,studies by Omar et al. (2002, 2006) reported that the concentra-tions of PAHs in PM10 in the capital city of Kuala Lumpur were6.28± 4.35 ng m�3 and 3.10± 2.92 ng m�3, respectively. Morerecently, Jamhari et al. (2014) conducted a study to determinesources of PAHs in PM10 at similar location to the present study andfound that total concentration of PAHs ranged from 1.64 to3.45 ng m�3. All these earlier studies found that vehicular emissionsources are the main contributor of total PAHs in the particulatematter.

Concentrations (ng m�3) References

2.79 This study2.54 Jamhari et al., 20146.28± 4.35 Omar et al., 20023.10± 2.92 Omar et al., 200689 Guo et al., 20095.24± 5.81 Wang et al., 20146.88 Li et al., 201015.46 Gao et al., 201233.96 Guo et al., 20033.37± 2.90 Naser et al., 200826.3± 29.4 Park et al., 20024.31 van Drooge et al., 20120.85 van Drooge et al., 20123.16 Li et al., 20091.91 Pleil et al., 20040.78 Fraser et al., 20023.80± 2.88 Oliveira et al., 201410.8 Bourotte et al., 2005

M.F. Khan et al. / Atmospheric Environment 106 (2015) 178e190 185

Table 2 presents a summary of results from selected studiesworldwide on PAHs in PM2.5. In general, the total PAHs concen-trations at the current study location are comparatively lower thanthe most of the locations reviewed in Table 2. Concentrations oftotal PAHs in PM2.5 varied with locations due to various factors suchas variations in population density, vehicle volume, extent of ur-banization and industrialization in the area, meteorological con-ditions and geographical locations. For example mega urban citieslike those in China, Hong Kong (Guo et al., 2003), Seoul, South Korea(Park et al., 2002) and Rio de Janeiro and Sao Paulo, Brazil (Oliveiraet al., 2014; Bourotte et al., 2005) exhibited much higher PAHsconcentration than the present study which has a much lowerpopulation density. However, two locations in USA (Pleil et al.,2004; Fraser et al., 2002) and a location in Spain (van Droogeet al., 2012) that showed lower concentration of PAHs comparedto all other sites.

3.2. Spearman correlations and types of traffic dependency amongthe PAH compounds

To understand the correlation among the PAHs, we conducted aSpearman correlation analysis as shown in Supplementary Table 2.We focused only on the correlation coefficients (r) of �0.80 wheresignificant values were highlighted for p< 0.05 and p< 0.01. Theresults for the datasets as a whole show that the lighter or smallerPAH compounds correlate strongly among themselves and the

Fig. 4. Sensitivity of PAHs to the traffic frequency by (a) Lig

heavier or bigger compounds also correlate strongly amongthemselves. There is an obvious separation of low and high mo-lecular weight molecules within the individual PAHs. The correla-tion between these two groupswas found to be poor with positive rvalues. However, the correlation results differ between the samplesin the SWand NEmonsoons. The NEmonsoon samples in particularshow no significant correlation among the LMW PAHs. The pair ofeach PAH consisting of HMW PAHs shows significantly strongcorrelations at each of three situations (the whole dataset, the SWmonsoon dataset and the NE monsoon dataset) providing the evi-dence that the PAHs of this group might share a similar emissionsource.

Light vehicles, heavy vehicles and total number of vehicles datawas applied to the model to study the dependency of motor vehicletype on the release of PM2.5 PAH compounds. The traffic data wasmade available by the Malaysian Public Works Department (PWD)under the Malaysian Ministry of Works (MOW). According to PWD,light vehicles include motor cars and motor cycles which are lessthan 3000 kg while heavy vehicles weighed more than that andinclude lorries, trucks, vans, and buses. The latest 2 years availabletraffic data (2010 and 2011) recorded by PWD for 16 h daily(6:00e22:00) were used to represent the traffic data for the studyperiod. This is base on the patterns of motor vehicles surroundingthe sampling area are almost consistent for the past 5 years.

For a clearer understanding, the 16 PAHs have been classifiedinto five groups based on their chemical characteristics (i.e.: 2-ring,

ht vehicle, (b) Heavy duty vehicle and (c) Total vehicle.

M.F. Khan et al. / Atmospheric Environment 106 (2015) 178e190186

3-ring, 4-ring, 5-ring and 6-ring). Sensitivity analysis, using theone-factor-at-a-time (OAT) technique and standardized regressiondetermination coefficient (R2) values, was used to examine thelevel of importance of each of the five groups of PAH compoundstowards the different type of vehicles. The results of the de-pendency of motor vehicle type on the concentrations of PAHs areshown in Fig. 4 (a, b and c). Each type of vehicle gives a unique set ofresults for the five PAH groups. Overall, the groups least affected bylight, heavy and total numbers of vehicles are 4-ring and 3-ringwith percentage ranges between 1% and 4%. The 5-ring are 18%,25% and 18% for LV, HV and total vehicles, respectively. In contrast,light vehicles influence the 4-ring group the most with a percent-age of 32%. Heavy vehicles influence the 3-ring group the most(30%) while the total number of vehicles influences the concen-trations of the 6-ring group the most (45%). A strong dependencywas found for HMW PAHs on the total number of vehicles.

3.3. Source apportionment of PAHs

3.3.1. Diagnostic ratios (DRs)One way to identify sources of PAHs is to employ a qualitative

identification. Table 3 shows the DRs of the selected PAHs related totheir respective sources while Fig. 5 shows bi-variate plots of PAHsratios illustrating the different source types. The DR of ANT/ANTþ PHE is about 0.46, indicating the dominance of a pyrogenicsource. An indication of a fuel combustion source was revealedbased on the DR of FLT/FLTþ PYR. The DR of BaA/BaAþCHY sug-gests that the combustion of coal is predominant to the pyrogenicsources of these PAHs. The characteristics of DRs for larger mole-cules of PAHs (�5-ring) e.g. I[c]P, B[g]P and B[a]P were also reported

Table 3Diagnostic ratios (DRs) of PAHs associated with PM2.5.

DRs Indicator sources Thisstudy

Factor 1 Factor 2 Factor 3

ANT/ANTþ PHE

<0.1: Petrogenica,b,c

>0.1: Pyrogenica,b,c0.46 0.54 0.46 0.44

FLT/FLTþ PYR

<0.4: Petrogenicb,d

>0.4: Pyrogenicb,d

0.4e0.5: fuel oilb,d

>0.5: Grass, wood, coalb,d

0.6e0.7: Diesele,f

0.4: Gasolinee,f

0.53 0.55 0.69 e

BaA/BaAþ CHY

<0.2: Petrogenicg,b,h

>0.35: Pyrogenicg,b,h

0.2e0.35: Coalg,b,h

>0.5: Wood burningg,b,h

0.33 0.31 0.50 0.32

IcP/IcPþ BgP

<0.2: Petrogenicb,c,i,j

>0.2: Pyrogenicb,c,i,j

0.2e0.5: Petroleum/gasolineb,c,i,j

>0.5: Grass, wood and coalb,c,i,j

0.82: Oil combustionb,c,i,j

0.35e0.70: Dieselb,c,i,j

0.41 0.42 0.39 0.47

BaP/BgP

<0.6: Nontrafficb,c

>0.6: Trafficb,c0.76 0.49 0.39 4.69

BaA/CHY

0.66e0.92: Woodk

0.54e0.66: Industryk0.50 0.46 0.99 0.46

a Pies et al. (2008).b Yunker et al. (2002).c Br€andli et al. (2008).d De La Torre-Roche et al. (2009).e Sicre et al. (1987).f Rogge et al. (1993).g Manoli et al. (2004).h Akyüz and Çabuk (2010).i Manoli et al. (2004).j Khalili et al. (1995).k Dickhut et al. (2000).

in Table 3 as well as in the bi-variate plots (Fig. 5a). The DRs valuesof I[c]P/I[c]Pþ BgP and B[a]P/B[g]P are about 0.41 and 0.76respectively, which confirms that these representative HMW PAHsare from fuel combustion sources. The average ratio value of BaA/CHY is 0.50 which highlights the strong influence of industrialemissions. Thus, the main source of PAHs was from pyrogenicoriginwhere fuel combustion, coal burning and industrial emissionidentified based on their respective threshold values published inthe literature (Akyüz and Çabuk, 2010; Br€andli et al., 2008; De LaTorre-Roche et al., 2009; Dickhut et al., 2000; Khalili et al., 1995;Manoli et al., 2004; Pies et al., 2008; Rogge et al., 1993; Sicre et al.,1987; Yunker et al., 2002).

3.3.2. PMF modelIntroducing a 34�17 matrix (sample number� 17 PAHs

(including total PAHs)) data set to US EPA PMF 5.0, three factorswere obtained as presented in Fig. 6a. Before further discussion, wewould like to highlight the steps undertaken during the model runthat led to the presented results. The concentrations of each vari-able were scanned using the inbuilt time series function to deter-mine whether the expected temporal patterns are present in thedata, and if there are any unusual events. Based on the temporalpattern, the extreme events were noted for possible exclusion fromthe dataset. Then, the base model was run using the PAH data witha specific seed. In this study, 20 runs and a seed of 25 were selected.The numbers of factors were chosen depending on the under-standing of the sources. The lowest or optimized goodness-of-fit byQ value was selected as Q (robust) (calculated excluding the out-liers). On the other hand, Q (true) was calculated including allpoints. The lowest values of Q (robust) and Q (true) was 498.7 forboth, for 34 samples and 172 computational steps. The Q (theo-retical) can also be calculated using the formula as nm-p � (nþm),where n is the number of species as 17 (including total PAHs), m isthe number of samples as 34, and p is the number of factors as 3.Thus, the value of Q (theoretical) was estimated as 425. Boot-strapping was also performed using a selected base run number,seed number, number of bootstraps, minimum correlation, andblock size which were 8, 9, 100, 0.6 and 4, respectively. The opti-mized values of Q (robust) and Q (true) were generated as 498.7and 498.7, respectively and 172 computational steps wereconverged. The stability and uncertainty of the solution wascompared to the output of the base run as shown in SupplementaryFig. S1 by bootstrapping procedure. Fpeak rotations were madebased on�1 toþ1. The output results without rotationwere chosento describe the data as the results were interpretable and showedclear physical meaning (Norris et al., 2014).

Factor 1: The individual PAH tracers heavily contributed toFactor 1 were B[k]F, B[a]P, B[b]Fth, I[c]P, D[h]A and B[g]P (Fig. 6a).This factor reflects the influence of the molecules with�5-ring. The5-ring and larger PAHs were released into atmosphere entirelyfrom a vehicle source (Venkataraman and Friedlander, 1994). Somestudies described I[c]P and B[g]P as related to a gasoline source(Miguel et al., 1998). B[k]F, B[b]F, B[a]P, I[c]P and D[h]A wereindicative of gasoline exhaust according to Gupta et al. (2011).Okuda et al. (2010) also described a gasoline source due to thedominance of HMW PAHs. Furthermore, the molecular DR value ofB[a]P/B[g]P in this study confirms that these larger molecule PAHsoriginate from a traffic source (Br€andli et al., 2008; Yunker et al.,2002). The HMW PAHs are generated significantly from pyrogenicsources. As evidenced by the DR value of I[c]P/I[c]Pþ B[g]P foroverall data (0.41) and Factor 1 (0.42), the combustion of petroleumoil or gasoline produces the HMW PAHs in most cases (Br€andliet al., 2008; Khalili et al., 1995; Manoli et al., 2004; Yunker et al.,2002). Thus, the predominance of B[k]F, B[a]P, B[b]Fth, I[c]P, D[h]A and B[g]P in Factor 1 is an indication that the source is gasoline or

Fig. 5. Bivariate plots for the ratios of (a) BaA/BaAþ CHY vs. IcP/IcPþ BgP and (b) BaA/CHY vs. FLT/FLTþ PYR.

Fig. 6. (a) Source profiles of PAHs by PMF 3.0, and (b) comparison of the estimated PAHs modelled by PMF and the observed PAHs in this study.

M.F. Khan et al. / Atmospheric Environment 106 (2015) 178e190 187

fuel oil combustion.Factor 2: In this factor, the dominant individual PAHs were ANT,

PHE, FLT, PYR, B[g]P, and I[c]P. Themass fraction contribution of PYRalong with PHE and FLT in this factor is higher compared to othervariables (Fig. 6a). Several studies have shown that PYR is an in-dicator tracer for emissions from diesel combustion (Guo et al.,2003; Ho et al., 2002; Khalili et al., 1995; Miguel et al., 1998;Sarkar and Khillare, 2013). The DR ratios of FLT/FLTþ PYR foroverall data and Factor 2 show the value of 0.53 and 69, respec-tively, which imply that these compounds are the atmosphericsignatures of diesel combustion (Sicre et al., 1987; Rogge et al.,1993). However, the combination of FLR, FLT and PYR with a fewHMW PAHs, e.g. B[b]F and I[c]P, are typically indicative of oilcombustion (Harrison et al., 1996). Thus, Factor 2 is associated witha diesel and heavy oil combustion source.

Factor 3: NAP, ACY, ACP, FLR, and ANTare 2- and 3-ring PAHs andare abundant in Factor 3 (Fig. 6a). Yunker et al. (2002) suggestedthat the generic source of LMW PAHs (2- and 3-ring) is of

petrogenic origin. In contrast to LMW PAHs, the HMW PAHs (�4-ring) are emitted into ambient air from a pyro-synthesis or pyrol-ysis source. However, the DR values of ANT/ANTþ PHE for overalldata and Factor 3 by factors of 0.46 and 0.44, respectively, show thatthese compounds are associated with a pyrogenic source comparedto the studies performed by Pies et al. (2008), Yunker et al. (2002),and Br€andli et al. (2008). However, a study by Alam et al. (2013)suggested that ANT/ANTþ PHE should not be used as an indicatorto differentiate source as the reactivity and volatility of LMW PAHsare faster than HMW PAHs. Lighter PAHs were dominant in a nat-ural gas source. For example, the LMW PAHs, e.g. ANT and FLR,evaporate from the combustion of natural gas (Hanedar et al.,2013). FLR and ANT are tracers representing a coal combustionsource (Liu et al., 2009). Thus, Factor 3 is related to natural gas andcoal burning sources.

The source contribution by each factor was scaled using thePMF-MLR procedure. The PMF outputs of three source contribu-tions were regressed against total PAHs (p< 0.01, R2¼ 0.99, n¼ 34).

Fig. 7. Contribution of various sources (%) of PAHs at (a) overall period, (b) south-westerly (SW) and (c) north-easterly (NE) monsoon.

M.F. Khan et al. / Atmospheric Environment 106 (2015) 178e190188

The correlation of the predicted and the measured PAHs shows astrong and significant correlation (R2¼ 0.97, p< 0.01, conf.int.¼ 0.95) (Fig. 5b). Fig. 7 shows the contribution of each source inthe overall data, the SW monsoon data and the NE monsoon data.For overall data, gasoline, diesel and heavy oil combustion, naturalgas and coal burning sources contributed 65%, 19%, and 15%respectively, leaving 1% unaccounted or determined for within totalPAHs (Fig. 7). Total PAHs originating from the gasoline source wasslightly higher in the north-east than the south-west monsoon data

Table 4The BaPequivalents (BaPeq) and lifetime lung cancer risk (LLCR) of individual and total P

PAHs TEFa BaPeq (ng m�3)

Overall SW NE

NAP 0.001 9.6� 10�5 9.9� 10�5 8.0� 1ACY 0.001 3.4� 10�5 4.3� 10�5 2.8� 1ACP 0.001 1.3� 10�4 1.8� 10�4 6.8� 1FLR 0.001 6.3� 10�5 9.2� 10�5 3.7� 1ANT 0.01 3.5� 10�4 5.5� 10�4 1.8� 1PHE 0.001 4.0� 10�5 6.4� 10�5 2.2� 1FLT 0.001 7.2� 10�5 1.6� 10�4 4.1� 1PYR 0.001 9.0� 10�5 1.2� 10�4 3.7� 1B[a]A 0.1 4.1� 10�3 60.3� 10�4 26.6� 1CYR 0.01 8.5� 10�4 13.4� 10�4 4.7� 1B[k]F 0.1 24.6� 10�4 330.1� 10�4 175.8� 1B[a]P 1 289.1� 10�3 373.3� 10�3 227.7� 1B[b]Fth 0.1 542.7� 10�4 747.2� 10�4 420.5� 1D[h]A 0.1 358.5� 10�4 509.5� 10�4 223.4� 1B[g]P 1 1574.3� 10�4 2176.6� 10�4 1118.4� 1I[c]P 0.01 54.3� 10�4 80.5� 10�4 31.0� 1P

Total PAHs e 572.6� 10�3 766.3� 10�3 428.2� 1

a Nisbet and LaGoy (1992).b WHO (2000).

sets, followed by the diesel and heavy oil combustion sources. Thelocal circulation of air pollutants from the Kuala Lumpur city centreto the southern part of the sub-urban areas during north-east asmentioned by Latif et al. (2012) might pronounce the gasolinesource at this location.

3.4. LLCR of PAHs

Table 4 shows the toxic equivalency factor (TEF), or relative

AHs.

Unit risk (URBaP)b LLCR

Overall SW NE

0�5 8.7 10�5 8.3� 10�9 8.6� 10�9 7.0� 10�9

0�5 3.0� 10�9 3.7� 10�9 2.4� 10�9

0�5 1.1� 10�8 1.6� 10�8 5.9� 10�9

0�5 5.5� 10�9 8.0� 10�9 3.2� 10�9

0�4 3.1� 10�8 4.8� 10�8 1.6� 10�8

0�5 3.5� 10�9 5.6� 10�9 1.9� 10�9

0�5 6.3� 10�9 1.4� 10�8 3.6� 10�9

0�5 7.8� 10�9 1.1� 10�8 3.2� 10�9

0�4 3.6� 10�7 5.3� 10�7 2.3� 10�7

0�4 7.4� 10�8 1.2� 10�7 4.1� 10�8

0�4 2.1� 10�6 2.9� 10�6 1.5� 10�6

0�3 2.5� 10�5 3.3� 10�5 2.0� 10�5

0�4 4.7� 10�6 6.5� 10�6 3.7� 10�6

0�4 3.1� 10�6 4.4� 10�6 1.9� 10�6

0�4 1.4� 10�5 1.9� 10�5 9.7� 10�6

0�4 4.7� 10�7 7.0� 10�7 2.7� 10�7

0�3 5.0� 10�5 6.7� 10�5 3.7� 10�5

M.F. Khan et al. / Atmospheric Environment 106 (2015) 178e190 189

potency, BaPeq of individual PAHs and the respective LLCR. Thecarcinogenic activity of the integrated PAHs was 572.6 10�3 ng m�3,which is two fold higher in the SW monsoon compared to that inthe NE monsoon. The BaPeq of individual B[a]P was more than halfof that shown by the total PAHs. Other studies also showed that thecarcinogenic potency of B[a]P is in the range of 27e67% of thecarcinogenic activity of total PAHs (Petry et al., 1996; Castellanoet al., 2003). Thus, B[a]P has been regulated in many countriesand has a set guideline value. For example, theWHO recommendedunit risk of B[a]P is 8.7 10�5 (ngm�3) per year (WHO, 2000) and thetarget value of B[a]P set by the European Union (EU) is 1 ng m�3 peryear in PM10 (EU, 2008). The secondmost potent BaPeq PAHwas BgPwhich poses a carcinogenicity risk of 27% compared to total PAHs.The BaPeq of BgP was also two fold higher in the SW monsoonsamples when compared the NE monsoon samples. The LLCR wasalso estimated based on the BaPeq of each PAH and the inhalationcancer unit risk (UR) of B[a]P. The value of UR advised by WHO is8.7 10�5, i.e. 8.7 cases per 100,000 people who experience chronicinhalation exposure to 1 ngm�3 B[a]P over a 70 year lifetime (WHO,2000). The results of LLCR were calculated as 5.0 10�5, 6.7 10�5 and3.7 10�5 for the overall data, the SW monsoon data and the NEmonsoon data respectively. The upper limit of the LLCR valuesshould be less than 10�6 to 10�4 per year maximal risk level(European Commission, 2001). Thus, the carcinogenic risk of thetotal PAHs has shown an acceptable risk level in the present studyarea in Malaysia.

4. Conclusions

This study reports the concentrations of PM2.5-bound PAHs; theaverages of the total PAHs were 2.79 (0.21e12.08), 3.85(0.21e12.08) and 1.85 (0.68e3.80) ngm�3 for the overall data, SWand NE monsoon samples, respectively. In general, the total PAHsconcentrations at the current study location are comparativelylower than the most of the locations being reviewed. Strong sea-sonal dependency of the total PAHs was observed in this study,where the SWmonsoon showed higher concentrations than the NEmonsoon. Molecule-wise, HMW PAHs (�5-ring) are significantlypredominant in the fine particle-bound aerosol. The LMW PAHscorrelate strongly with other LMW PAHs while the HMW PAHscorrelate strongly with other HMW PAHs.

The sources identified by PMF are consistent to the PAH sourcesidentified by the DRs. Source apportionment analysis of the PAHsusing the PMF 5.0 model identified three main sources: gasolinecombustion (65%), diesel and heavy oil combustion (19%), naturalgas and coal burning (15%). Seasonally, there were almost similarratios and percentage of PAHs sources, indicating weaker seasonalinfluence. The DRs of I[c]P/I[c]Pþ B[g]P and B[a]P/B[g]P confirmthat the HMW PAHs originate from fuel combustion sources. Inaddition, fuel combustion sourcewas identified based on the DRs ofFLT/FLTþ PYR. Thus, the DRs-PMF-MLR, an integration of the threemodels, was successfully applied and would be chosen in futurestudies. The carcinogenic activity of the integrated PAHs was572.6 10�3 ng m�3 which poses a two fold-higher risk in the SWmonsoon compared to the NE monsoon. The B[a]Peq of individual B[a]P was more than half of that shown by total PAHs. The study ofthe LLCR suggests that the carcinogenic risk of the total PAHs is ofan acceptable risk level (10�6 to 10�4 per year) in the present studyarea in Malaysia.

Acknowledgements

The authors would like to thank Universiti KebangsaanMalaysiafor the Iconic Grant (ICONIC-2013-004) and the Ministry of Edu-cation for the Fundamental Research Grant (FRGS/1/2013/STWN01/

UKM/02/2). Special thanks to Dr Rose Norman for proofreading thismanuscript.

Appendix A. Supplementary data

Supplementary data related to this article can be found at http://dx.doi.org/10.1016/j.atmosenv.2015.01.077.

References

Abas, M.R., Oros, D.R., Simoneit, B.R.T., 2004. Biomass burning as the main source oforganic aerosol particulate matter in Malaysia during haze episodes. Chemo-sphere 55, 1089e1095.

Abas, M.R., Simoneit, B.R.T., 1996. Composition of extractable organic matter of airparticles from Malaysia: initial study. Atmos. Environ. 30, 2779e2793.

Akyüz, M., Çabuk, H., 2010. Gaseparticle partitioning and seasonal variation ofpolycyclic aromatic hydrocarbons in the atmosphere of Zonguldak, Turkey. Sci.Total Environ. 408, 5550e5558.

Alam, M.S., Delgado-Saborit, J.M., Stark, C., Harrison, R.M., 2013. Using atmosphericmeasurements of PAH and quinone compounds at roadside and urban back-ground sites to assess sources and reactivity. Atmos. Environ. 77, 24e35.

Anwar, A., Juneng, L., Othman, M.R., Latif, M.T., 2010. Correlation between hotspotsand air quality in Pekanbaru, Riau, Indonesia in 2006-2007. Sains Malays. 39,169e174.

Armstrong, B., Hutchinson, E., Unwin, J., Fletcher, T., 2004. Lung cancer risk afterexposure to polycyclic aromatic hydrocarbons: a review and meta-analysis.Environ. Health Persp. 112, 970e978.

Aydin, Y.M., Kara, M., Dumanoglu, Y., Odabasi, M., Elbir, T., 2014. Source appor-tionment of polycyclic aromatic hydrocarbons (PAHs) and polychlorinated bi-phenyls (PCBs) in ambient air of an industrial region in Turkey. Atmos. Environ.97, 271e285.

Azhari, A., Dalimin, M.N., Wee, S.T., 2011. Polycyclic aromatic hydrocarbons (PAHs)from vehicle emission in the vegetation of highway roadside in Johor, Malaysia.Int. J. Environ. Sci. Dev. 2 (6), 465e468.

Baumann, K., Jayanty, R.K.M., Flanagan, J.B., 2008. Fine particulate matter sourceapportionment for the chemical speciation trends network site at Birmingham,Alabama, using positive matrix factorization. J. Air Waste Manage. Assoc. 58,27e44.

Bosetti, C., Boffetta, P., La Vecchia, C., 2007. Occupational exposures to polycyclicaromatic hydrocarbons, and respiratory and urinary tract cancers: a quantita-tive review to 2005. Ann. Oncol. 18, 431e446.

Bourotte, C., Forti, M.-C., Taniguchi, S., Bícego, M.C., Lotufo, P.A., 2005. A wintertimestudy of PAHs in fine and coarse aerosols in S~ao Paulo city, Brazil. Atmos. En-viron. 39, 3799e3811.

Br€andli, R.C., Bucheli, T.D., Ammann, S., Desaules, A., Keller, A., Blum, F., et al., 2008.Critical evaluation of PAH source apportionment tools using data from theSwiss soil monitoring network. J. Environ. Monit. 10, 1278e1286.

Call�en, M.S., Iturmendi, A., L�opez, J.M., Mastral, A.M., 2014. Source apportionment ofthe carcinogenic potential of polycyclic aromatic hydrocarbons (PAH) associ-ated to airborne PM10 by a PMF model. Environ. Sci. Pollut. Res. 21, 2064e2076.

Castellano, A.V., Cancio, J.L., Alem�an, P.S., Rodríguez, J.S., 2003. Polycyclic aromatichydrocarbons in ambient air particles in the city of Las Palmas de Gran Canaria.Environ. Int. 29, 475e480.

Choi, H., Harrison, R., Komulainen, H., et al., 2010. Polycyclic aromatic hydrocarbons.In: WHO Guidelines for Indoor Air Quality: Selected Pollutants. World HealthOrganization, Geneva, p. 6. Available from: http://www.ncbi.nlm.nih.gov/books/NBK138709/.

De La Torre-Roche, R.J., Lee, W.Y., Campos-Díaz, S.I., 2009. Soil-borne polycyclicaromatic hydrocarbons in El Paso, Texas: analysis of a potential problem in theUnited States/Mexico border region. J. Hazard. Mater. 163, 946e958.

Dickhut, R.M., Canuel, E.A., Gustafson, K.E., Liu, K., Arzayus, K.M., Walker, S.E., et al.,2000. Automotive sources of carcinogenic polycyclic aromatic hydrocarbonsassociated with particulate matter in the Chesapeake Bay region. Environ. Sci.Technol. 34, 4635e4640.

European Union (EU), 2008. Directive 2008/50/EC of the European Parliament andof the Council of 21 May 2008 on Ambient Air Quality and Cleaner Air forEurope (OJ L 152, 11.6.2008), pp. 1e44.

EU, 2005. Commission recommendation 2005/108/EC. Off. J. Eur. Comm. L34, 43.European Commission, 2001. Polycyclic Aromatic Hydrocarbons (PAH) Position

Paper (July 2001) (Prepared by the Working Group on Polycyclic AromaticHydrocarbons).

Fang, G.-C., Wu, Y.-S., Chang, C.-N., Ho, T.-T., 2006. Retracted: a study of polycyclicaromatic hydrocarbons concentrations and source identifications by methods ofdiagnostic ratio and principal component analysis at Taichung chemical Harbornear Taiwan Strait. Chemosphere 64, 1233e1242.

Fraser, M.P., Yue, Z.W., Tropp, R.J., Kohl, S.D., Chow, J.C., 2002. Molecular compositionof organic fine particulate matter in Houston, TX. Atmos. Environ. 36,5751e5758.

Galarneau, E., 2008. Source specificity and atmospheric processing of airbornePAHs: implications for source apportionment. Atmos. Environ. 42, 8139e8149.

Gao, B., Guo, H., Wang, X.-M., Zhao, X.-Y., Ling, Z.-H., Zhang, Z., Liu, T.-Y., 2012.Polycyclic aromatic hydrocarbons in PM2.5 in Guangzhou, southern China:

M.F. Khan et al. / Atmospheric Environment 106 (2015) 178e190190

spatiotemporal patterns and emission sources. J. Hazard. Mater. 239e240,78e87.

Guo, H., Lee, S.C., Ho, K.F., Wang, X.M., Zou, S.C., 2003. Particle-associated polycyclicaromatic hydrocarbons in urban air of Hong Kong. Atmos. Environ. 37,5307e5317.

Guo, Z., Lin, T., Zhang, G., Hu, L., Zheng, M., 2009. Occurrence and sources ofpolycyclic aromatic hydrocarbons and n-alkanes in PM2.5 in the roadsideenvironment of a major city in China. J. Hazard. Mater. 170, 888e894.

Gupta, S., Kumar, K., Srivastava, A., Srivastava, A., Jain, V.K., 2011. Size distributionand source apportionment of polycyclic aromatic hydrocarbons (PAHs) inaerosol particle samples from the atmospheric environment of Delhi, India. Sci.Total Environ. 409, 4674e4680.

Hanedar, A., Alp, K., Kaynak, B., Avsar, E., 2013. Toxicity evaluation and sourceapportionment of Polycyclic Aromatic Hydrocarbons (PAHs) at three stations inIstanbul, Turkey. Sci. Total Environ. 488e489, 437e446.

Harrison, R.M., Smith, D.J.T., Luhana, L., 1996. Source apportionment of atmosphericpolycyclic aromatic hydrocarbons collected from an urban location in Bir-mingham, U.K. Environ. Sci. Technol. 30, 825e832.

Harrison, R.M., Beddows, D.C.S., Dall'Osto, M., 2011. PMF analysis of wide-rangeparticle size spectra collected on a major highway. Environ. Sci. Technol. 45,5522e5528.

Hedberg, E., Gidhagen, L., Johansson, C., 2005. Source contributions to PM10 andarsenic concentrations in Central Chile using positive matrix factorization.Atmos. Environ. 39, 549e561.

Henry, R.C., 1987. Current factor analysis receptor models are ill-posed. Atmos.Environ. 21, 1815e1820.

Ho, K.F., Lee, S.C., Chiu, G.M.Y., 2002. Characterization of selected volatile organiccompounds, polycyclic aromatic hydrocarbons and carbonyl compounds at aroadside monitoring station. Atmos. Environ. 36, 57e65.

Jamhari, A.A., Sahani, M., Latif, M.T., Chan, K.M., Tan, H.S., Khan, M.F., et al., 2014.Concentration and source identification of polycyclic aromatic hydrocarbons(PAHs) in PM10 of urban, industrial and semi-urban areas in Malaysia. Atmos.Environ. 86, 16e27.

Jang, E., Alam, M.S., Harrison, R.M., 2013. Source apportionment of polycyclic aro-matic hydrocarbons in urban air using positive matrix factorization and spatialdistribution analysis. Atmos. Environ. 79, 271e285.

Jiang, Y.-F., Wang, X.-T., Wang, F., Jia, Y., Wu, M.-H., Sheng, G.-Y., et al., 2009. Levels,composition profiles and sources of polycyclic aromatic hydrocarbons in urbansoil of Shanghai, China. Chemosphere 75, 1112e1118.

Juneng, L., Latif, M.T., Tangang, F., 2011. Factors influencing the variations of PM10aerosol dust in Klang Valley, Malaysia during the summer. Atmos. Environ. 45,4370e4378.

Khalili, N.R., Scheff, P.A., Holsen, T.M., 1995. PAH source fingerprints for coke ovens,diesel and gasoline engines, highway tunnels, and wood combustion emissions.Atmos. Environ. 29, 533e542.

Khan, M.F., Hirano, K., Masunaga, S., 2012. Assessment of the sources of suspendedparticulate matter aerosol using US EPA PMF 3.0. Environ. Monit. Assess. 184,1063e1083.

Koe, L.C.C., Arellano Jr., A.F., McGregor, J.L., 2001. Investigating the haze transportfrom 1997 biomass burning in Southeast Asia: its impact upon Singapore.Atmos. Environ. 35, 2723e2734.

Latif, M.T., Hey, L.S., Juneng, L., 2012. Variations of surface ozone concentrationacross the Klang Valley, Malaysia. Atmos. Environ. 61, 434e445.

Li, P.-H., Wang, Y., Li, Y.-H., Wang, Z.-F., Zhang, H.-Y., Xu, P.-J., Wang, W.-X., 2010.Characterization of polycyclic aromatic hydrocarbons deposition in PM2.5 andcloud/fog water at Mount Taishan (China). Atmos. Environ. 44, 1996e2003.

Li, Z., Porter, E.N., Sj€odin, A., Needham, L.L., Lee, S., Russell, A.G., Mulholland, J.A.,2009. Characterization of PM2.5-bound polycyclic aromatic hydrocarbons inAtlantadseasonal variations at urban, suburban, and rural ambient air moni-toring sites. Atmos. Environ. 43, 4187e4193.

Liu, Y., Chen, L., Huang, Q.-H., Li, W.-Y., Tang, Y.-J., Zhao, J.-F., 2009. Source appor-tionment of polycyclic aromatic hydrocarbons (PAHs) in surface sediments ofthe Huangpu River, Shanghai, China. Sci. Total Environ. 407, 2931e2938.

Manoli, E., Kouras, A., Samara, C., 2004. Profile analysis of ambient and sourceemitted particle-bound polycyclic aromatic hydrocarbons from three sites innorthern Greece. Chemosphere 56, 867e878.

Miguel, A.H., Kirchstetter, T.W., Harley, R.A., Hering, S.V., 1998. On-road emissions ofparticulate polycyclic aromatic hydrocarbons and black carbon from gasolineand diesel vehicles. Environ. Sci. Technol. 32, 450e455.

Naser, T.M., Yoshimura, Y., Sekiguchi, K., Wang, Q., Sakamoto, K., 2008. Chemicalcomposition of PM2.5 and PM10 and associated polycyclic aromatic hydrocar-bons at a roadside and an urban background area in Saitama, Japan. Asian J.Atmos. Environ. 2, 90e101.

Nisbet, I.C.T., LaGoy, P.K., 1992. Toxic equivalency factors (TEFs) for polycyclic aro-matic hydrocarbons (PAHs). Regul. Toxicol. Pharm. 16, 290e300.

Norris, G., Duvall, R., Brown, S., Bai, S., 2014. EPA Positive Matrix Factorization (PMF)5.0 Fundamentals & User Guide. Prepared for the US Environmental ProtectionAgency, Washington, DC. National Exposure Research Laboratory, ResearchTriangle Park, USA.

Ogulei, D., Hopke, P.K., Wallace, L.A., 2006a. Analysis of indoor particle size distri-butions in an occupied townhouse using positive matrix factorization. IndoorAir 16, 204e215.

Ogulei, D., Hopke, P.K., Zhou, L., Patrick Pancras, J., Nair, N., Ondov, J.M., 2006b.Source apportionment of Baltimore aerosol from combined size distribution

and chemical composition data. Atmos. Environ. 40 (Suppl. 2), 396e410.Okuda, T., Okamoto, K., Tanaka, S., Shen, Z., Han, Y., Huo, Z., 2010. Measurement and

source identification of polycyclic aromatic hydrocarbons (PAHs) in the aerosolin Xi'an, China, by using automated column chromatography and applyingpositive matrix factorization (PMF). Sci. Total Environ. 408, 1909e1914.

Oliveira, R., Loyola, J., Minho, A., Quiterio, S., de Almeida Azevedo, D., Arbilla, G.,2014. PM2.5-bound polycyclic aromatic hydrocarbons in an area of Rio deJaneiro, Brazil impacted by emissions of light-duty vehicles fueled by ethanol-blended gasoline. Bull. Environ. Contam. Toxicol. 93, 781e786.

Omar, N.Y.M.J., Abas, M.R.B., Ketuly, K.A., Tahir, N.M., 2002. Concentrations of PAHsin atmospheric particles (PM10) and roadside soil particles collected in KualaLumpur, Malaysia. Atmos. Environ. 36, 247e254.

Omar, N.Y.M.J., Mon, T.C., Rahman, N.A., Abas, M.R.B., 2006. Distributions and healthrisks of polycyclic aromatic hydrocarbons (PAHs) in atmospheric aerosols ofKuala Lumpur, Malaysia. Sci. Total Environ. 369, 76e81.

Paatero, P., 1997. Least squares formulation of robust non-negative factor analysis.Chemom. Intell. Lab. 37, 23e35.

Paatero, P., Tapper, U., 1994. Positive matrix factorization: a non-negative factormodel with optimal utilization of error estimates of data values. Environmetrics5, 111e126.

Park, S.S., Kim, Y.J., Kang, C.H., 2002. Atmospheric polycyclic aromatic hydrocarbonsin Seoul, Korea. Atmos. Environ. 36, 2917e2924.

Perraudin, E., Budzinski, H., Villenave, E., 2007. Kinetic study of the reactions ofozone with polycyclic aromatic hydrocarbons adsorbed on atmospheric modelparticles. J. Atmos. Chem. 56, 57e82.

Petry, T., Schmid, P., Schlatter, C., 1996. The use of toxic equivalency factors inassessing occupational and environmental health risk associated with exposureto airborne mixtures of polycyclic aromatic hydrocarbons (PAHs). Chemosphere32, 639e648.

Pies, C., Hoffmann, B., Petrowsky, J., Yang, Y., Ternes, T.A., Hofmann, T., 2008.Characterization and source identification of polycyclic aromatic hydrocarbons(PAHs) in river bank soils. Chemosphere 72, 1594e1601.

Pleil, J.D., Vette, A.F., Rappaport, S.M., 2004. Assaying particle-bound polycyclic ar-omatic hydrocarbons from archived PM2.5 filters. J. Chromatogr. A 1033, 9e17.

Polissar, A.V., Hopke, P.K., Malm, W.C., Sisler, J.F., 1998a. Atmospheric aerosol overAlaska: 1. Spatial and seasonal variability. J. Geophys. Res. 103, 19035e19044.

Polissar, A.V., Hopke, P.K., Paatero, P., Malm, W.C., Sisler, J.F., 1998b. Atmosphericaerosol over Alaska: 2. Elemental composition and sources. J. Geophys. Res. 103,19045e19057.

Rogge, W.F., Hildemann, L.M., Mazurek, M.A., Cass, G.R., Simonelt, B.R.T., 1993.Sources of fine organic aerosol. 3. Road dust, tire debris, and organometallicbrake lining dust: roads as sources and sinks. Environ. Sci. Technol. 27,1892e1904.

Sarkar, S., Khillare, P.S., 2013. Profile of PAHs in the inhalable particulate fraction:source apportionment and associated health risks in a tropical megacity. En-viron. Monit. Assess. 185, 1199e1213.

Sharma, H., Jain, V.K., Khan, Z.H., 2007. Characterization and source identification ofpolycyclic aromatic hydrocarbons (PAHs) in the urban environment of Delhi.Chemosphere 66, 302e310.

Sicre, M.A., Marty, J.C., Saliot, A., Aparicio, X., Grimalt, J., Albaiges, J., 1987. Aliphaticand aromatic hydrocarbons in different sized aerosols over the MediterraneanSea: occurrence and origin. Atmos. Environ. 21, 2247e2259.

Sun, F., Littlejohn, D., Gibson, M.D., 1998. Ultrasonication extraction and solid phaseextraction clean-up for determination of US EPA 16 priority pollutant polycyclicaromatic hydrocarbons in soils by reversed-phase liquid chromatography withultraviolet absorption detection. Anal. Chim. Acta 364, 1e11.

Thurston, G.D., Spengler, J.D., 1985. A quantitative assessment of source contribu-tions to inhalable particulate matter pollution in metropolitan Boston. Atmos.Environ. 19, 9e25.

United States Environmental Protection Agency priority list (US EPA), 2010. Fed.Regist.

van Drooge, B.L., Crusack, M., Reche, C., Mohr, C., Alastuey, A., Querol, X., Prevot, A.,Day, D.A., Jimenez, J.L., Grimalt, J.O., 2012. Molecular marker characterization ofthe organic composition of submicron aerosols from Mediterranean urban andrural environments under contrasting meteorological. Atmos. Environ. 61,482e489.

Venkataraman, C., Friedlander, S.K., 1994. Size distributions of polycyclic aromatichydrocarbons and elemental carbon. 2. Ambient measurements and effects ofatmospheric processes. Environ. Sci. Technol. 28, 563e572.

Wang, F., Lin, T., Li, Y., Ji, T., Ma, C., Guo, Z., 2014. Sources of polycyclic aromatichydrocarbons in PM2.5 over the East China Sea, a downwind domain of EastAsian continental outflow. Atmos. Environ. 92, 484e492.

Watson, J.G., Robinson, N.F., Chow, J.C., Henry, R.C., Kim, B.M., Pace, T.G., et al., 1990.The sUSEPA/DRI chemical mass balance receptor model, CMB 7.0. Environ.Softw. 5, 38e49.

World Health Organization (WHO), 2000. Polycyclic Aromatic Hydrocarbons, sec-ond ed. In: Air Quality Guidelines for Europe WHO Regional Office for Europe(WHO Regional Publications, European Series, No. 91), Copenhagen.

WHO, 1987. Air Quality Guidelines for Europe e Copenhagen. WHO Regional Officefor Europe, Publication No. 23.

Yunker, M.B., Macdonald, R.W., Vingarzan, R., Mitchell, R.H., Goyette, D.,Sylvestre, S., 2002. PAHs in the Fraser River basin: a critical appraisal of PAHratios as indicators of PAH source and composition. Org. Geochem. 33, 489e515.


Recommended