Original source · versioned page text
Crilley 2014_Identification of the sources of primary organic aerosols a...
Attached to the e-mail MCU15/0095 & REC15/0022 PM Submission (1) Object - Jennifer Drysdale.
Attached to the e-mail MCU15/0095 & REC15/0022 PM Submissions Object - Petition - Various.
Environmental Pollution 191 (2014) 158e165
Contents lists available at ScienceDirect
Environmental Pollution
journal homepage: www.elsevier.com/locate/envpol
Identification of the sources of primary organic aerosols at urban
schools: A molecular marker approach
Leigh R. Crilley a,1, Raeed M. Qadir b, c, Godwin A. Ayoko a, *, Jürgen Schnelle-Kreis b,
Gülcin Abbaszade b, Jürgen Orasche b, Ralf Zimmermann b, c, Lidia Morawska a
a
International Laboratory for Air Quality and Health, Institute of Health and Biomedical Innovation, Queensland University of Technology, Brisbane, QLD
4001, Australia
b
Joint Mass Spectrometry Centre e Comprehensive Molecular Analytics, Helmholtz Zentrum München, Ingolstädter Landstraße 1, D-85764 Neuherberg,
Germany
c
Institute of Chemistry, University of Rostock, Dr.-Lorenz-Weg 1, D-18051 Rostock, Germany
a r t i c l e i n f o a b s t r a c t
Article history: Children are particularly susceptible to air pollution and schools are examples of urban microenviron-
Received 18 January 2014 ments that can account for a large portion of children’s exposure to airborne particles. Thus this paper
Received in revised form aimed to determine the sources of primary airborne particles that children are exposed to at school by
14 April 2014
analyzing selected organic molecular markers at 11 urban schools in Brisbane, Australia. Positive matrix
Accepted 17 April 2014
factorization analysis identified four sources at the schools: vehicle emissions, biomass burning, meat
Available online 20 May 2014
cooking and plant wax emissions accounting for 45%, 29%, 16% and 7%, of the organic carbon respectively.
Biomass burning peaked in winter due to prescribed burning of bushland around Brisbane. Overall, the
Keywords:
Positive matrix factorization
results indicated that both local (traffic) and regional (biomass burning) sources of primary organic
Sources aerosols influence the levels of ambient particles that children are exposed at the schools. These results
Primary organic aerosols have implications for potential control strategies for mitigating exposure at schools.
Exposure Ó 2014 Elsevier Ltd. All rights reserved.
Schools
1. Introduction characteristic of a specific emission source and crucially also have a
long lifetime in the atmosphere to enable detection at monitoring
Organic aerosols are frequently the largest component of stations (Cass, 1998). Examples of some of the more specific organic
ambient fine particles in urban atmospheres (Jimenez et al., 2009). markers include levoglucosan for biomass burning (Robinson et al.,
These aerosols are a complex mixture of around 105 and 106 2006a) and cholesterol for meat cooking (Kleeman et al., 2008).
compounds (Goldstein and Galbally, 2007) from a number of pri- Hopanes are present in fossil fuels, and the different hopanoid
mary and secondary sources, where sources that directly emit compounds are specific to the different types of fossil fuels such as
particles into the atmosphere can be referred to as primary sources. coal, petroleum and lubricating oil ((Schnelle-Kreis et al., 2005) and
Whereas secondary sources refer to particles formed by reactions references therein)). Thus these organic compounds along with
in the atmosphere of gaseous species in the atmosphere, such as others can be used as molecular markers for identifying the
the photo-oxidation of volatile organic compounds (Seinfeld and contributing sources of primary organics to ambient particles.
Pandis, 2006). In an urban environment primary sources of parti- A quantitative assessment of the contribution of emission
cles can include vehicle emissions, biomass burning (from both sources is available using positive matrix factorization (PMF),
domestic and natural sources), coal combustion, cooking, plant which has been successfully applied previously to organic molec-
abrasion and paved road dust (see e.g. (Qadir et al., 2013; Schnelle- ular datasets (Qadir et al., 2013; Schnelle-Kreis et al., 2007; Shriv-
Kreis et al., 2007; Shrivastava et al., 2007)). For these primary astava et al., 2007). Assessing the contribution of primary sources is
sources there are a number of organic markers that are known to be important as there are numerous detrimental health effects asso-
ciated with exposure to airborne particles (Heal et al., 2012; Rückerl
et al., 2011). In particular, children have been shown to be more
* Corresponding author. susceptible to the health effects associated with exposure to
E-mail address: [email protected] (G.A. Ayoko). airborne particles (Rückerl et al., 2011), and several studies have
1
Present address: School of Geography, Earth and Environmental Science, Uni- linked exposure to vehicle emissions to increased risks of
versity of Birmingham, Birmingham B15 2TT, United Kingdom.
http://dx.doi.org/10.1016/j.envpol.2014.04.023
0269-7491/Ó 2014 Elsevier Ltd. All rights reserved.
See the original
The page text is free to read. Viewing or downloading an original file needs an account.
Log in to see the originalFull text of the other 7 pages
Page 2
L.R. Crilley et al. / Environmental Pollution 191 (2014) 158e165 159
developing asthma and wheezing (Gehring et al., 2010; Ryan et al., organic compounds (e.g. levoglucosan and cholesterol) occurs directly from par-
2009). Children spend a large portion of their day at school. ticulate matter on the filters. Therefore, polar and non-polar compounds are
analyzed in one step. An automated sampling robot exchanged the complete GC
Consequently schools represent a location accounting for a signif- liners placed in the injector which is automatically closed and opened by a liner
icant fraction of their overall daily exposure. Schools are also exchanging unit (Linex, Atas GL, Netherlands). During thermal desorption (16 min,
unique urban environments that are influenced not only by local 300 C) MSTFA enriched carrier gas was passed through the liner. Desorbed and
and regional sources but also by school-related activities such as derivatized molecules were focused at 70 C on the head of a nearly non-polar
capillary column, BPX5, 25 m, 0.22 mm ID, 0.25 mm film (SGE, Australia) which
drop-off/pick-up traffic, which may result in increased exposure to
was installed in an Agilent 6890 gas chromatograph (Agilent, USA). After the
vehicle emissions. However, there is limited information on chil- isothermal run at 70 C, the temperature was increased to 130 C within one minute.
dren’s exposure to vehicle emissions and other sources in the Then the rate was lowered to 8 C min1 until a temperature of 330 C was reached
literature (Mejía et al., 2011). followed by an isothermal time of 30 min. The detection of compounds was carried
Considering the aforementioned gaps in knowledge, the present out on a Pegasus III TOFMS (LECO, USA). The data acquisition frequency was 25
spectra per second within a mass-to-charge range of 35e500. The detection limits
study aimed to determine the primary sources of the ambient fine for the targeted compounds can be found in Tables S1 and S2 (Supporting
particulate matter present at urban schools by quantifying the Information).
levels of selected organic molecular markers. PMF, a receptor model Details of the elemental carbon (EC) and organic carbon (OC) analysis can be
(Paatero and Tapper, 1994), was applied in order to quantitatively found in Crilley (2013). Briefly, PM2.5 samples were collected concurrently at the
same location at each school and were analyzed according to the IMPROVE method
determine the contributions of these sources to levels of ambient
on a thermal/optical transmittance carbon analyser (Sunset Laboratories) (Chow
particles at the schools. This enabled the driving factors to the et al., 2001).
levels of different primary sources to be investigated and we hy-
pothesized that vehicle emissions will be the main primary source
2.4. Data analysis
contributing to children’s exposure at school and this is reflected in
the schools’ selection criteria, outlined in the next section. Positive Matrix Factorization (PMF) was performed in this study using the EPA
PMF v3.0, developed by the U.S EPA (US-EPA and Research Triangle Park N, 2008). A
total of 21 organic compounds were used in the initial analysis and these included n-
2. Method
alkanes from C20e33, four hopanes, cholesterol, dehydroabietic acid and its meth-
2.1. Sampling sites ylester, levoglucosan, mannosan, EC and the OC fraction. Missing values in a certain
school site were replaced with the geometric mean at the same site and the un-
The 11 schools selected for this study, which will be referred to as S15 to S25, are
certainty was set to four times the geometric mean for those values. As about 80% of
in different suburbs in the city of Brisbane, Australia. A map depicting the schools
n-alkanes between C18 and C21 are typically in the gas phase, therefore these
location in Brisbane is given in Figure S1, Supporting Information. This chemical
compounds were excluded from the PMF analysis. Species with more than 50% of
sampling was a part of a larger campaign to study the effect of ultrafine particle from
data below the limit of quantification (LOQ) were also excluded from the inputs.
traffic emissions on children’s health, known as UPTECH (www.ilaqh.qut.edu.au/
Therefore, some alkanes (C20, C21, C26, C28, C32 and C33), dehydroabietic acid and its
Misc/UPTECH%20Home.htm). In this study 25 schools participated but only at the
methylester, and mannosan were excluded from the PMF analyses. In other few
last 11 schools were filters collected for this analysis due to instrument availability.
cases the values were replaced with half of the LOQ and the uncertainties set to 5/6
The sampling campaign for the current paper was conducted from November 2011
of the LOQ. For the other data, the uncertainties are set according to the following
till August 2012.
equation:
The schools that were chosen were not near any other large source of air
pollution, other than road traffic and were also not close to any large infrastructure h i0:5
projects. A site close to the middle of the school which gave the best overall rep- u ¼ ðerror fraction concentrationÞ2 þ ðLOQ Þ2 (1)
resentation was chosen to conduct the measurements and sampling at each school.
Inlets for the sampling were placed on the top of a trailer, which served to house all where error fraction is the percentage of uncertainty, LOQ is the limit of quantifi-
of the instruments at the site including an automatic weather station (Monitor cation. Overall error fractions were estimated from sampling and analytical errors.
Sensors). Data from nearby weather stations was also obtained from the Bureau of Species with low signal to noise ratio (S/R < 2), were assigned as weak. One hundred
Meteorology (BOM) and Queensland Department of Science, Information Technol- bootstrap runs were performed to assess the uncertainty of the factor loadings and
ogy, Innovation and the Arts (DSITIA) as some of the schools would have been profiles. Changing the Fpeak value did not improve the source profiles or Q value and
affected by local winds due to schools buildings. Sampling inlets were approxi- so base model results (Fpeak ¼ 0) are reported. The PMF solutions are discussed in
mately 3 m off the ground. Traffic counts were taken on the busiest road next to the Section 3.3.
school, referred to as the main road throughout. In the traffic count, vehicles that The other data analysis technique employed was one way ANOVA used to
were classified as light were cars, motorbikes and scooters. Trucks with 2, 3 and 4 compare the concentrations between the seasons and this was performed using
axels were classified as medium vehicle and long articulated trucks classified as SPSS v19.
heavy vehicles.
2.2. Sampling methodology 3. Results and discussion
The particle size fraction was selected using a low-volume PM2.5 cyclone and the
required flow rate for the cyclone was maintained using critical orifices. The parti- 3.1. General school characteristics
cles were collected in polycarbonate filter holders on 47 mm, Quartz fiber filters
(Whatman) which was pre-treated at 500 C for 1 h prior to use. The sampling
The sampling was conducted over almost a full year and covered
period was from 08:00 until 08:00 the next day, Monday to Friday for one week to
give a total of four samples at each school. This sampling regime typically gave both winter and summer, as reflected in the meteorological con-
sampling volumes of 6 m3. After sampling each filter was placed in a Petri dish and ditions described in Table 1. Brisbane is a subtropical city with the
covered in aluminum foil. The samples were then sealed in a ziplock bag and frozen mean temperatures ranging from 14 to 25 C (Table 1). October till
until analysis. The flow rate for the cyclones was checked at the beginning of each April is warmer with higher humidity and this period is referred to
sample and always set to within 0.1 L per minute of the desired flow rate. Field
as summer. May till September is cooler with typically more stable
blanks were collected at each school and all results were field-blank corrected.
weather and it is referred to as the winter months. A range of traffic
2.3. Chemical analysis conditions were also observed at the schools (Table 2). Some of the
schools were on busy arterial roads (S19, S20) while others were in
The organic composition of the samples was analyzed using In-situ Derivati-
zation Thermal Desorption (IDTD) followed by Gas Chromatography Time-Of-Flight residential areas with low traffic counts (S16, S18, S21). There were
Mass Spectrometry (GC-TOF-MS) (Orasche et al., 2011; Schnelle-Kreis et al., 2011) at also schools alongside roads that service commercial/light indus-
the Joint Mass Spectrometry Centre, Helmholtz Zentrum München, Neuherberg, trial areas (S23, S25) and such schools recorded a higher proportion
Germany. In brief, filter punches were placed in an injection liner and spiked with of heavy and medium vehicles (Table 2). Based on the organic
internal standards (isotope labeled reference compounds) and 10 ml derivatizing
agent N-Methyl-N-trimethylsilyl-trifluoroacetamide (MSTFA). The method allows
compounds detected and their relative abundance, the major
derivatization using MSTFA reagent of the polar organic fraction of filter samples in- contributing sources of primary organic aerosols were determined
situ on the filter during thermal desorption. Derivatization and desorption of polar at the schools and discussed in the next sections.
Page 3
160 L.R. Crilley et al. / Environmental Pollution 191 (2014) 158e165
Table 1
Average meteorological conditions of the schools studied during the sampling week. Note that temperature and relative humidity readings were taken from the school weather
station while wind speed measurements were taken from the closest BOM or DSTIA station.
S15 S16 S17 S18 S19 S20 S21 S22 S23 S24 S25
Sampling date Nov 2011 Nov 2011 Dec 2012 Mar 2012 Mar 2012 Apr 2012 Jun 2012 Jun 2012 Jul 2012 Aug 2012 Aug 2012
Temperature ( C) 24 25 24 22 23 20 14 14 15 15 15
Relative humidity (%) 65 75 71 71 79 63 58 66 60 40 68
Wind Speed (m s1) 4.1 1.0 0.91 1.0 3.3 0.7 0.8 0.6 0.7 3.0 3.5
3.2. Detected organic compounds at the schools vehicle emissions than plant wax emissions. This was probably
because the schools sampled in summer recorded higher traffic
A number of particulate organic compounds were detected at flows (e.g. S15, S19 and S20). At the schools sampled in winter there
the schools as summarized in Table 3, which demonstrates that the was a higher abundance of plant wax alkanes (C>27) compared to
organic species and their concentrations varied from school to the schools sampled in summer (Fig. 1). Plant waxes alkanes can
school. Of the analyzed species, the most abundant organic com- come from either abrasion of vegetation detritus and incomplete
pound was generally levoglucosan, which was detected at all of the biomass burning (Cass, 1998; Pietrogrande et al., 2011; Simoneit,
schools except for S16 (Table 3). Of the other anhydrous sugars 2002). In Brisbane, most trees are evergreen and winter is the
targeted, mannosan and galactosan were detected only at schools driest time of year, increasing the likelihood of emissions from leaf
sampled in winter, at S21-5 and S22, S24-5, respectively (Table 3). abrasion. In addition, the practice of prescribed burning of bush-
The most abundant resin acids detected were dehydroabietic acid land, described in more detail in Section 3.3.3, would also explain
(DHAA) and dehydroabietic acid methylester (DHAA-ME), and the higher abundance of alkanes from plant wax during winter.
these compounds were observed only at the schools sampled in Further differentiation between vehicle emissions and biogenic
summer (Table 3). Cholesterol, a known tracer molecule for meat sources is possible by determining the Cmax, which refers to the
cooking emissions, was observed at the schools (Table 3) (Kleeman most abundant alkane at each school, from the data in Table 3.
et al., 2008), indicates that cooking was a source. The most abun- Diesel and gasoline vehicle emissions have a different Cmax, at C20
dant hopanes were 17a(H),21b(H)-30-norhopane (29ab) and and C23e5, respectively while plant wax emissions typically have a
17a(H),21b(H)-hopane (30ab) at all of the schools. In addition, Cmax above C27 (Cass, 1998; Rogge et al., 1993a; Schnelle-Kreis et al.,
17a(H),21b(H)-22S-homohopane (31abS) and 17a(H),21b(H)-22R- 2011, 2005). At the schools sampled in summer the Cmax was
homohopane (31abR) were detected at S18-25 in appreciable similar to that expected for gasoline vehicles (C23e5) whereas at
quantities. The n-alkanes analyzed were octadecane (C18) to hex- schools sampled in the winter, the Cmax was within the range for
atriacontane (C36) and their concentrations and distributions varied plant waxes (C>27). The exception was S23, which had a Cmax of 24
across the schools. The average values for each alkane at the schools and this indicates that vehicle emissions were the most likely major
are given in Table S2 (Supporting Information), with the sum of n- contributing source of n-alkanes at this school. Carbon preference
alkanes summarized in Table 3. Throughout this paper, the n-al- index (CPI) value, which is defined as the ratio of the sum con-
kanes are referred to as Cn, where n indicates the number of carbon centration of odd- to even-carbon numbered n-alkanes, has been
atoms. For example, pentacosane is referred to as C25. All abbrevi- widely used to distinguish between sources of n-alkanes (Brown
ations are listed in Table S2. The range of n-alkanes detected at the et al., 2002). Vehicle emissions are expected to have a ratio
schools indicated that fossil fuel combustion was another signifi- around 1 whereas plant waxes are expected to have a ratio >2
cant source at the schools, which is investigated further in the next (Schnelle-Kreis et al., 2007). In Table 3, the schools with a CPI
section. around 1 were S18, S19, S20, S22 and S23, which were generally the
schools beside comparatively higher trafficked roads. However, the
3.3. Sources of primary PM2.5 based on the detected organic CPI at S15e17 would suggest an influence from plant waxes in
compounds addition to that from vehicle emissions. Thus, from the n-alkanes it
was difficult to distinguish between the contributions of sources,
3.3.1. n-alkane concentrations and distribution though it is clear that the two main sources were likely to be vehicle
The concentration of the targeted n-alkanes is presented in emissions and plant waxes. Further investigations of the influence
Fig. 1, averaged for schools sampled during summer and winter. of vehicle emissions were possible by analyzing the concentrations
Vehicles emit alkanes in the range of C19e32 (Rogge et al., 1993a). and distributions of the hopanes.
The majority of the detected alkanes were observed in this range,
suggesting that this was the main source of alkanes at the schools. 3.3.2. Hopane concentrations and distribution
Another possible source which has an overlapping alkane distri- Generally, the same hopanes distribution was found at all of the
butions are plant waxes, which are associated with alkanes greater schools and the average concentration for each hopane is presented
than C27 (Cass, 1998). From Table S2 and Fig. 1 alkanes within this in Fig. 2. The homohopane isomerization index (HHI, defined as
range were also detected, indicating that this was another potential 31abS/31abSþ31abR) is about 0.1e0.4 for coal burning and 0.6 for
source at all schools. From Fig. 1, the schools sampled in summer mineral oil emission sources, such as vehicles (Oros and Simoneit,
had a higher abundance of alkanes and a higher association with 2000; Schnelle-Kreis et al., 2011). For the schools where these
Table 2
Mean traffic conditions from the main at the schools studied over the sampling week.
S15 S16 S17 S18 S19 S20 S21 S22 S23 S24 S25
Traffic (Veh hr1) 858 46 757 35 1093 1121 151 469 657 160 332
Light vehicles (%) 92.1 97.8 95.3 97.9 89.9 96.2 97.4 96.8 71.1 97.1 92.0
Medium vehicles (%) 5.9 1.6 3.0 2.1 3.2 2.5 2.1 2.5 28.0 2.4 6.5
Heavy vehicles (%) 0.9 0.4 0.6 0.0 2.0 0.8 0.4 0.6 0.6 0.5 1.2
Page 4
Table 3 Average concentrations (in ng m~°) of selected hopanes and polar organic compounds along with calculated values for the alkanes at each school with average of EC and OC (in wg m~*). Abbreviat ms of compound names are given in parenthesis. CPI is defined as the ratio of the sum concentration of odd- to even-carbon numbered n-alkanes. Uncertainties are 1 standard deviation. <DL indicates below the detection $25 6.1 S24 5.9 1.6 $23 37 0.9 $22 9.3 $21 7.9 15 S18 S19 $20 3.1 S17 5.6 17 S16 5.6 18 S15 6.2 Molecule 1.14 0.8 10.2 0.8 n-alkanes CPI 19 1.2 0.7 2.8 Cmax 0.2 + 0.2 0.14 0.1 0.1 + 0.09 0.09 + 0.07 0.06 + 0.07 0.05 + 0.06 7.8463 <DL <DL 0.1 + 0.06 0.2 + 0.03 0.2 + 0.03 0.03 + 0.05 0.03 + 0.05 0.03 + 0.04 0.03 + 0.04 41424 <DL <DL 10+ 0.4 1340.6 0.7404 0.8 + 0.4 3441.7 8844.1 1.0 + 0.7 10405 0.7404 0540.3 <DL 0.9408 1141.0 0540.4 04404 0.6 + 0.4 <DL 0.8 + 04 0.5 + 0.2 0.7 +0.1 1.0 + 0.7 0.9 + 0.7 DL DL 170(H),21B(H)-30-norhopane (29ab) 7a(H),21B(H)-hopane (30ab) 0.1 + 0.04 0.07 + 0.03 0.06 + 0.01 3441.0 0.6 + 0.02 0.07 + 0.06 0.06 + 0.06 5.3444 <DL <DL 0.1 + 0.03 0.09 + 0.03 6.9431 <DL <DL DL DL 0.7 + 1.2 44414 5.14 62 <DL DL DL 0.6 + 0.2 2.7438 6.9 + 46 <DL 7a(H),21B(H)-22S-homohopane (31abS) 17a(H),21B(H)-22R-homohopane (31abR) Cholesterol 1.9 + 1.0 <DL <DL 4549.1 34.1 + 25.8 6.2412 <DL Dehydroabietic acid (DHAA) 17 4 15.9 14417 <DL 2.9419 <DL Dehydroabietic acid methylester (DHAAM) Mannosan 3.7423 0.2405 2.7416 43438 0140.14 Crilley et al. / Environmental Pollution 191 (2014) 158—165 98.7 + 84.1 0.73 + 0.13 3.51 + 0.19 130 + 67.9 86.2 + 67.5 0.67 + 0.29 1.81 + 0.66 98.3 + 36.8 1.14 + 0.33 2.25 + 1.10 45.7473 48 +172 12.8441 0.48 + 0.06 2.43 + 0.42 29.5 + 21.2 0.28 + 0.24 19.2 + 18.8 0.40 + 0.37 2.07 + 0.71 DL 0.53 + 0.04 2.04 + 0.52 44.6 + 303 0.38 + 0.11 3.03 + 0.56 Levoglucosan 0.39 + 0.12 3.01 + 0.71 0.20 + 0.10 0.62 + 0.60 0.84 + 0.40 2.18 + 0.86 Elemental carbon (EC) Organic carbon (OC) 1.89 + 0.60 161 25 WE Summer [23 Winter 204 Ee P1584 © S © § 104 8 e & 3 054 0.04 DHOMKNMYTHOKRHOSRO- AMNYTHO S2RRNRRAKKRARABRHKBSABS eoOO OOO OOOO OOOO OOOO oO Fig. 1. Average summer (n = 23) and winter (n = 20) concentrations of the measured n-alkanes. Variability shown is one standard deviation of the average. homohopanes were detected, the HHI ranged from 0.5—0.6, indi- cating vehicle emissions as the source. This was expected as very little coal and oil burning for domestic heating occurs in Brisbane and this may explain why the other targeted hopanes were below the detection limit of majority samples (See Fig. 2). The hopane concentrations were observed to be higher at the schools sampled in summer (S15—20, See Table 3 for selected hopanes) and is the reverse of what was previously observed in Europe (Schnelle-Kreis et al., 2005). This was likely to be to the lower emissions of hopanes from coal burning as well as the observed increased contribution from vehicle emissions at the schools sampled in summer. There- fore, based on the hopanes detected the main source of fossil fuel combustion at the schools was vehicle emissions. 3.3.3. Biomass burning tracers A number of biomass burning tracer compounds were detected in this study, and include the anhydrous sugars; levoglucosan, galactosan, mannosan (Qadir et al., 2013; Robinson et al., 2006a; Simoneit, 2002) and the resin acids DHAA and DHAA-ME (Rogge 1.2 1.0 4 0.85 0.6 4 0.45 Concentration (ng m°) 0.25 0.0 T T T T i Ts Tm AA tb 27b 29ab 29ba 30ab 30ba 31abS 31abR32abS 32abR: Fig. 2. Average concentration of analyzed hopanes across all schools. Variability shown is one standard deviation of the average. Note the following abbreviations used not described in the text: Ts (18a(H)-22,29,30-trisnorneohopane), Tm (17a(H)-22,29,30- trisnorhopane), 27b (17B(H)-22,29,30-trisnorhopane), 30ba (17$(H),21(H)-hopane), 32abS —(17a(H),21P(H)-22S-bishomohopane) and 32abR (170(H),21B(H)-22R- bishomohopane).
Page 5
162 L.R. Crilley et al. / Environmental Pollution 191 (2014) 158e165
% of Species Concentration of species
1000 100
Biomass Burning 90
100 80
10 70
60
1 50
40
0.1 30
0.01 20
10
0.001 0
100 100
Cooking 90
10 80
70
1 60
50
0.1 40
30
Concentration of Species, ng m-3
% of Species total
0.01 20
10
0.001 0
1000 100
Vehicle Emissions 90
100 80
70
10 60
50
1 40
30
0.1 20
10
0.01 0
1000 100
Plant Wax 90
100 80
70
10 60
50
1 40
30
0.1 20
10
0.01 0
Fig. 3. Source factor profiles from the positive matrix factorization (PMF) analysis using EPA-PMF3.0 model. Bars represent mass concentration while the squares indicate the
percentage contribution.
et al., 1998). DHAA is also a constituent of vehicle tires, indicating All of these seasonal differences were statistically significant
the presence of another possible source at the schools (Rogge et al., (p < 0.05). The higher concentrations of the biomass burning
1993b; Sheya, 2002). As levoglucosan was generally the most tracers, levoglucosan and mannosan in winter is expected in Bris-
abundant compound detected at most of the schools (Table 3), it bane due to the influence of prescribed burning of bushland sur-
indicates that biomass burning was a significant source. Schools rounding the city during the winter to reduce the risk of summer
sampled in the summer generally had a lower concentration than bushfires. Prescribed burning typically occurs during JulyeAugust,
the schools sampled in winter, with the averages of 27.6 22.4 and which was when the highest concentration of levoglucosan and
94.6 60.3 ng m3 for summer and winter, respectively. By com- mannosan were observed (Table 3). The levoglucosan detected at
parison, using the same analytical technique, the mean levogluco- the schools sampled in the summer are likely to be from a different
san concentrations were higher in Augsburg, Germany compared to wood burning source as unlike at the schools sampled in winter
the current study at 112 84 and 337 174 ng m3 for summer and (Table 3), these schools recorded concentrations of DHAA above the
winter, respectively (Pietrogrande et al., 2011). The lower concen- detection limit. DHAA is emitted only from softwood combustion
trations in the current study are likely to be due to the warmer (specifically pine) unlike levoglucosan that is emitted by combus-
climate in Brisbane and the resulting low levels of domestic wood tion of all forms of cellulose containing materials (Pietrogrande
burning for heating. et al., 2011; Robinson et al., 2006a; Rogge et al., 1998). This sug-
Distinct seasonal trends were observed between the schools gests that in the summer there is a different source of biomass
sampled in summer and winter in the concentrations and distri- burning, possibly related to domestic wood burning such as from
butions of the biomass burning tracers. Schools sampled in winter barbeques, which in Brisbane predominantly use pine. In the pre-
recorded higher concentrations of levoglucosan and mannosan, vious section, higher contributions from vehicle emission at schools
compared to schools sampled in summer, whereas the schools sampled in summer were observed and another possible source of
sampled in summer had higher concentrations of DHAA (Table 3). DHAA is tire wear (Rogge et al., 1993b; Sheya, 2002). Thus, traffic
Page 6
L.R. Crilley et al. / Environmental Pollution 191 (2014) 158e165 163
emissions may have been a contributing source to the observed Fig. 4) recorded high traffic counts. Unlike the other schools, which
levels of DHAA at the schools along with domestic wood burning were in suburban areas, S18 is in an inner-city location and this may
barbeques. explain the high levels for this factor.
The first factor was assigned to biomass burning as it was
characterized by high contributions of levoglucosan, the molecular
3.4. Source apportionment using PMF
marker for biomass burning (Qadir et al., 2013; Robinson et al.,
2006a; Simoneit, 2002). The OC/EC ratio for this factor was 2.8,
To apportion the contributions of the different sources of pri-
which is lower than the ratio of OC/EC from previous study held in
mary organic aerosols identified at each school to the OC fraction,
Australia (Keywood et al., 2011). However, this could be explained
PMF was applied to the dataset. Interpretability was the key basis
by higher contribution of EC, which is a marker for vehicle emis-
for judging the optimum solution to the PMF analysis. Interpretable
sions (Sonntag et al., 2012), to the PMF profile. The biomass burning
solutions were those that grouped source-class-specific sets of
factor was the second largest source overall during the study, ac-
markers into distinct factors. Models that grouped markers for
counting for 29 23% of the OC (Fig. 4). From Fig. 4, the highest
multiple source classes into the same factor, distributed markers for
contributions were as expected, found during the winter (e.g. S22,
one source class across multiple factors, or contained factors with
S24 and S25) at an average of 48 19%, primarily due to the
no distinctive groupings of compounds were judged less inter-
practice of prescribed burning of bushland surrounding Brisbane.
pretable and rejected. While no molecular marker pattern can be
The next source, which accounted for 16 13% of the OC was
unambiguously associated with a specific source class, this
characterized by cholesterol. As cholesterol is a known tracer for
approach provides a systematic basis for sorting through the
meat cooking (Kleeman et al., 2008), this source was assigned to
possible PMF models (Shrivastava et al., 2007). Solutions with 3e5
meat cooking emissions. Cooking emissions are increasingly being
factors were investigated and the three factor solution was rejected
identified as a significant urban source (Mohr et al., 2012; Robinson
as it could not separate out the cooking and biomass burning fac-
et al., 2006b) and the observed high contributions may be indica-
tors. The five factor solution was not used due to splitting of the
tive of localized emissions from school canteens. This is illustrated
vehicle emissions source factor into unrealistic factors. Therefore
at S24 where the high concentrations (Fig. 4) were likely due to a
the four factor solution was found to best represent the data as it
school barbeque event that occurred on one of the sampling days.
gave realistic source profiles related to distinct primary sources.
In addition, S23 had high concentration of meat cooking emissions
Source profiles are given in Fig. 3 and were assigned to biomass
because it is located near a large number of restaurants. This
burning, cooking, vehicle and plant wax emissions based upon the
probably suggests that apart from the school canteen, there was an
high concentrations of known tracer molecules for these sources.
additional source at the school.
The contributions of each identified source factors to the OC
High contributions of the alkanes C25, C27 and C29 e C31 char-
measured at each individual school are given in Fig. 4 and the
acterized the last factor, which was assigned to plant waxes. Plant
factors will be discussed in order of their contribution to the OC
waxes particles can enter the atmosphere from either leaf abrasion
rather than factor number.
(Rogge et al., 1993c) or incomplete biomass burning (Simoneit,
The major source, vehicle emissions (See Fig. 4), was charac-
2002). Both of these source pathways were likely as the highest
terized by the C20-25 alkanes and the hopanes 29ab, 30ab, 31abS
contributions were observed at schools (S20, S21 and S22) sampled
and 31abR. C20 and C23e5 alkanes have previously been shown to be
during the driest time of the year and schools that recorded strong
associated with diesel and gasoline vehicle emissions (Rogge et al.,
biomass burning influences (Fig. 4). This factor accounted for 7 6%
1993a). This together with the fact that the hopanes determined in
of the OC and was the smallest contributing source of primary
Section 3.3.2 were deemed to be from fossil fuel combustion, points
aerosols at the schools.
to vehicle emissions as the source. The HHI for this factor was 0.85
As the schools selected for this study were not close to any other
and in the range expected for vehicle emissions, further indicating
large source of air pollution other than road traffic, the main source
vehicle emissions as a source. This factor contributed an average of
of primary organic aerosol was hypothesized to be vehicle emis-
45 30% of the OC at the schools and was therefore the main
sions. This hypothesis was confirmed based upon the PMF analysis
source in our study. The factor was only somewhat correlated with
on selected tracer organic molecules (Fig. 4). The lower contribu-
the traffic counts beside the schools (r2 of 0.3), probably due to the
tion from biomass burning source to OC mass at S15eS20 (sites
multiple variables that contribute to the observed levels of vehicle
measured in summer, Fig. 4) was due, in part, to the sampling at
emissions at the schools (Crilley, 2013). However, it was observed
these schools occurring when the contributions of biomass burning
that with the exception of S18 (Table 2), the schools with the
from prescribed burning were at a minimum. Generally, year-round
highest contributions from vehicle emissions (S15, S17, S19 and S20,
vehicle emissions were found to have a largest overall contribution
to the OC mass at the schools. Previous work on PM2.5 in urban
100% environments using an organic molecular marker approach also
90% found that vehicle emissions rather than wood combustion were
Biomass
80% Burning the largest primary source of PM2.5 (Green et al., 2013; Wang et al.,
% Organic Carbon (OC) mass
70%
2012). Unlike in the current work where prescribed fires in sur-
Cooking
rounding bushland were the source of biomass burning, the wood
60%
combustion was attributed primarily to residential heating in those
50%
Vehicle studies (Green et al., 2013; Wang et al., 2012).
40% Emissions
30%
Plant Wax 4. Conclusions
20%
10% Selected organic molecular markers were quantified in order to
0% apportion the contributing sources of primary organic aerosols to
S15 S16 S17 S18 S19 S20 S21 S22 S23 S24 S25
children’s exposure of PM2.5 at urban schools, with vehicle emis-
Fig. 4. Percentage of contributions of the PMF source factors to organic carbon (OC) sions hypothesized to be the largest contributing source. Using
mass at the 11 schools. PMF, four source factors were identified and assigned to vehicle
Page 7
164 L.R. Crilley et al. / Environmental Pollution 191 (2014) 158e165
emissions, biomass burning, meat cooking and plant wax emis- References
sions. Vehicle emissions, which were characterized by C20-25 al-
kanes and the observed hopanes accounted for 45 30% of the Brown, S.G., Herckes, P., Ashbaugh, L., Hannigan, M.P., Kreidenweis, S.M.,
Collett Jr., J.L., 2002. Characterization of organic aerosol in Big Bend National
overall OC, with schools having the highest traffic counts recording Park, Texas. Atmos. Environ. 36, 5807e5818.
higher contributions. Biomass burning recorded high contribution Cass, G.R., 1998. Organic molecular tracers for particulate air pollution sources. TrAC
to the OC at an average of 29 23% and a seasonal trend that Trends Anal. Chem. 17, 356e366.
Chow, J.C., Watson, J.G., Crow, D., Lowenthal, D.H., Merrifield, T., 2001. Comparison
peaked in winter (48 19%) due to prescribed burning in bushland of IMPROVE and NIOSH carbon measurements. Aerosol Sci. Technol. 34, 23e34.
around Brisbane was also observed. For the same reasons, plant Crilley, L.R., 2013. Investigations of the Physical and Chemical Properties of Ambient
wax alkanes demonstrated a similar seasonal trend, though this Fine Particles in Urban Schools (PhD thesis). Faculty of Science and Engineering.
Queensland University of Technology, Brisbane.
source only accounted for a small fraction (7 6%) of the OC. Gehring, U., Wijga, A., Brauer, M., Fischer, P., de Jongste, J., Kerkhof, M.,
Cooking emissions accounted for 16 13% of the OC, with a school Oldenwening, M., Smit, H., Brunekreef, B., 2010. Traffic-related air pollution and
nearby a restaurant precinct recording high contributions. There- the development of asthma and allergies during the first 8 years of life. Am. J.
Respir. Crit. Care Med. 181, 596e603.
fore, based upon the targeted organic molecular markers, vehicle
Goldstein, A.H., Galbally, I.E., 2007. Known and unexplored organic constituents in
emissions were the major source of organic aerosols at the schools the Earth’s atmosphere. Environ. Sci. Technol. 41, 1514e1521.
studied, confirming the hypothesis. The second largest contribution Green, M.C., Chow, J.C., Chang, M.C.O., Chen, L.W.A., Kuhns, H.D., Etyemezian, V.R.,
Watson, J.G., 2013. Source apportionment of atmospheric particulate carbon in
was from a more regional source, in this case biomass burning as a
Las Vegas, Nevada, USA. Particuology 11, 110e118.
result of controlled burning of bushland surrounding Brisbane to Heal, M.R., Kumar, P., Harrison, R.M., 2012. Particles, air quality, policy and health.
reduce the risk of summer bushfires. Chem. Soc. Rev. 41, 6606e6630.
Controlled burning of bushland is common practice near urban Jimenez, J.L., Canagaratna, M.R., Donahue, N.M., Prevot, A.S.H., Zhang, Q., Kroll, J.H.,
DeCarlo, P.F., Allan, J.D., Coe, H., Ng, N.L., Aiken, A.C., Docherty, K.S., Ulbrich, I.M.,
areas in Australia as well as in other countries that are affected by Grieshop, A.P., Robinson, A.L., Duplissy, J., Smith, J.D., Wilson, K.R., Lanz, V.A.,
wild bush or forest fires such as the U.S.A and Mediterranean Hueglin, C., Sun, Y.L., Tian, J., Laaksonen, A., Raatikainen, T., Rautiainen, J.,
countries. Thus the significant influence of biomass burning from Vaattovaara, P., ehn, M., Kulmala, M., Tomlinson, J., Collins, D.R., Cubison, M.J.,
Dunlea, E.J., Huffman, A., Onasch, T.B., Alfarra, M.R., Williams, P.I., Bower, K.N.,
both controlled and wild fires that was observed in the results from Kondo, Y., Schneider, J., Drewnick, F., Borrmann, S., Weimer, S., Demerjian, K.L.,
this study is potentially applicable to other urban areas in Australia Salcedo, D., Cottrell, L., Griffin, R., Takami, A., Miyoshi, T., Hatakeyama, S.,
and other countries. Another notable aspect of the results was the Jayne, J.T., Herndon, S.C., Trimborn, A., Williams, L.R., Wood, E.C.,
Middlebrook, A., Kolb, C.E., Baltensperger, U., Worsnop, D.R., 2009. Evolution of
large contribution from meat cooking emissions, possibly due to organic aerosols in the atmosphere science. Science 326 (5959), 1525e1529.
activities at the schools. This result warrants further investigations Keywood, M., Guyes, H., Selleck, P., Gillett, R., 2011. Quantification of secondary
to determine the full impact of meat cooking emissions at schools. organic aerosol in an Australian urban location. Environ. Chem. 8, 115e126.
Kleeman, M.J., Robert, M.A., Riddle, S.G., Fine, P.M., Hays, M.D., Schauer, J.J.,
Finally, although vehicle emissions were identified as a major source
Hannigan, M.P., 2008. Size distribution of trace organic species emitted from
in this study, it is not necessarily the only major source of pollution biomass combustion and meat charbroiling. Atmos. Environ. 42, 3059e3075.
in urban areas, as only a subset of organic compounds were analyzed Mejía, J.F., Choy, S.L., Mengersen, K., Morawska, L., 2011. Methodology for assessing
exposure and impacts of air pollutants in school children: data collection,
in this work. Overall, the results from this study point to the influ-
analysis and health effects e a literature review. Atmos. Environ. 45, 813e823.
ence of both local (traffic) and regional (biomass burning) sources of Mohr, C., DeCarlo, P.F., Heringa, M.F., Chirico, R., Slowik, J.G., Richter, R., Reche, C.,
primary organic aerosols on the levels of ambient particles that Alastuey, A., Querol, X., Seco, R., Peñuelas, J., Jiménez, J.L., Crippa, M.,
children are exposed at schools. This has implications for potential Zimmermann, R., Baltensperger, U., Prévôt, A.S.H., 2012. Identification and
quantification of organic aerosol from cooking and other sources in Barcelona
control measures for mitigating exposure at schools, as contribu- using aerosol mass spectrometer data. Atmos. Chem. Phys. 12, 1649e1665.
tions from regional biomass burning events can at certain times Orasche, J., Schnelle-Kreis, J., Abbaszade, G., Zimmermann, R., 2011. In-situ deriva-
surpass those of local traffic emissions. Therefore, while efforts to tization thermal desorption GC-TOFMS for direct analysis of particle-bound
non-polar and polar organic species. Atmos. Chem. Phys. 11, 8977e8993.
reduce the levels of primary organic aerosols could focus on Oros, D.R., Simoneit, B.R.T., 2000. Identification and emission rates of molecular
reducing local traffic emissions at schools, consideration should also tracers in coal smoke particulate matter. Fuel 79, 515e536.
be given to regional biomass burning sources, particularly in loca- Paatero, P., Tapper, U., 1994. Positive matrix factorization: a non-negative factor
model with optimal utilization of error estimates of data values. Environmetrics
tions affected by wild and controlled fires. 5, 111e126.
Pietrogrande, M.C., Abbaszade, G., Schnelle-Kreis, J., Bacco, D., Mercuriali, M.,
Zimmermann, R., 2011. Seasonal variation and source estimation of organic
Acknowledgments compounds in urban aerosol of Augsburg, Germany. Environ. Pollut. 159, 1861e
1868.
Qadir, R.M., Abbaszade, G., Schnelle-Kreis, J., Chow, J.C., Zimmermann, R., 2013.
This work was supported by the Australian Research Council
Concentrations and source contributions of particulate organic matter before
(ARC), QLD Department of Transport and Main Roads (DTMR) and and after implementation of a low emission zone in Munich, Germany. Environ.
QLD Department of Education, Training and Employment (DETE) Pollut. 175, 158e167.
through Linkage Grant LP0990134. Our particular thanks go to R. Robinson, A.L., Subramanian, R., Donahue, N.M., Bernardo-Bricker, A., Rogge, W.F.,
2006a. Source apportionment of molecular markers and organic aerosol. 2.
Fletcher (DTMR) and B. Robertson (DETE) for their vision regarding Biomass smoke. Environ. Sci. Technol. 40, 7811e7819.
the importance of this work. We would also like to thank Prof G. Robinson, A.L., Subramanian, R., Donahue, N.M., Bernardo-Bricker, A., Rogge, W.F.,
Marks, Dr P. Robinson, Prof K. Mengersen, Prof Z. Ristovski, Dr. M. 2006b. Source apportionment of molecular markers and organic aerosol. 3.
Food cooking emissions. Environ. Sci. Technol. 40, 7820e7827.
Mazaheri, Dr C. He, Dr G. Johnson, Dr R. Jayaratne, Dr S. Low Choy, Rogge, W.F., Hildemann, L.M., Mazurek, M.A., Cass, G.R., 1998. Sources of fine
Prof G. Williams, W. Ezz, F. Salimi, S. Clifford, M. Mokhtar, N. Mishra, organic aerosol. 9. Pine, oak, and synthetic log combustion in residential fire-
R. Laiman, L. Guo, Prof C. Duchaine, Dr H. Salonen, Dr X. Ling, Dr J. places. Environ. Sci. Technol. 32, 13e22.
Rogge, W.F., Hildemann, L.M., Mazurek, M.A., Cass, G.R., Simoneit, B.R.T., 1993a.
Davies, Dr L. Leontjew Toms, F. Fuoco, Dr A. Cortes, Dr B. Toelle, A. Sources of fine organic aerosol. 2. Noncatalyst and catalyst-equipped automo-
Quinones, P. Kidd, E. Belousova, Dr M. Falk, Dr F. Fatokun, Dr J. Mejia, biles and heavy-duty diesel trucks. Environ. Sci. Technol. 27, 636e651.
Dr D. Keogh, Prof T. Salthammer, R. Appleby and C. Labbe for their Rogge, W.F., Hildemann, L.M., Mazurek, M.A., Cass, G.R., Simoneit, B.R.T., 1993b.
Sources of fine organic aerosol. 3. Road dust, tire debris, and organometallic brake
contribution to the UPTECH project. lining dust: roads as sources and sinks. Environ. Sci. Technol. 27, 1892e1904.
Rogge, W.F., Hildemann, L.M., Mazurek, M.A., Cass, G.R., Simoneit, B.R.T., 1993c.
Sources of fine organic aerosol. 4. Particulate abrasion products from leaf sur-
Appendix A. Supplementary data faces of urban plants. Environ. Sci. Technol. 27, 2700e2711.
Rückerl, R., Schneider, A., Breitner, S., Cyrys, J., Peters, A., 2011. Health effects of
particulate air pollution: a review of epidemiological evidence. Inhal. Toxicol.
Supplementary data related to this article can be found at http://
23, 555e592.
dx.doi.org/10.1016/j.envpol.2014.04.023.
Page 8
L.R. Crilley et al. / Environmental Pollution 191 (2014) 158e165 165
Ryan, P.H., Bernstein, D.I., Lockey, J., Reponen, T., Levin, L., Grinshpun, S., Sheya, S.A., 2002. Development of Thermal Desorption gas Chromatography/mass
Villareal, M., Khurana Hershey, G.K., Burkle, J., LeMasters, G., 2009. Exposure to Spectrometry as a Rapid Method for Ambient Particulate Characterization.
traffic-related particles and endotoxin during infancy is associated with Department of Materials Science and Engineering. University of Utah, USA.
wheezing at age 3 years. Am. J. Respir. Crit. Care Med. 180, 1068e1075. Shrivastava, M.K., Subramanian, R., Rogge, W.F., Robinson, A.L., 2007. Sources of
Schnelle-Kreis, J., Orasche, J., Abbaszade, G., Schäfer, K., Harlos, D., Hansen, A.A., organic aerosol: positive matrix factorization of molecular marker data and
Zimmermann, R., 2011. Application of direct thermal desorption gas chroma- comparison of results from different source apportionment models. Atmos.
tography time-of-flight mass spectrometry for determination of nonpolar or- Environ. 41, 9353e9369.
ganics in low-volume samples from ambient particulate matter and personal Simoneit, B.R.T., 2002. Biomass burning d a review of organic tracers for smoke
samplers. Anal. Bioanal. Chem. 401, 3083e3094. from incomplete combustion. Appl. Geochem. 17, 129e162.
Schnelle-Kreis, J., Sklorz, M., Orasche, J., Stölzel, M., Peters, A., Zimmermann, R., Sonntag, D.B., Bailey, C.R., Fulper, C.R., Baldauf, R.W., 2012. Contribution of lubri-
2007. Semi volatile organic compounds in ambient PM2.5. Seasonal trends and cating oil to particulate matter emissions from light-duty gasoline vehicles in
daily resolved source contributions. Environ. Sci. Technol. 41, 3821e3828. Kansas city. Environ. Sci. Technol. 46, 4191e4199.
Schnelle-Kreis, J., Sklorz, M., Peters, A., Cyrys, J., Zimmermann, R., 2005. Analysis of US-EPA, 2008. Positive Matrix Factorization (PMF) 3.0 Fundamentals & User Guide,
particle-associated semi-volatile aromatic and aliphatic hydrocarbons in urban in: Research Triangle Park, N. (Ed.).
particulate matter on a daily basis. Atmos. Environ. 39, 7702e7714. Wang, Y., Hopke, P.K., Xia, X., Rattigan, O.V., Chalupa, D.C., Utell, M.J., 2012. Source
Seinfeld, J.H., Pandis, S.N., 2006. Atmospheric Chemistry and Physics: from Air apportionment of airborne particulate matter using inorganic and organic
Pollution to Climate Change, second ed. John Wiley and Sons. species as tracers. Atmos. Environ. 55, 525e532.