Source apportionment of fine particulate matter organic carbon in Shenzhen, China by chemical mass balance and radiocarbon methods

Environmental Pollution 240 (2018) 34e43

Contents lists available at ScienceDirect

Environmental Pollution journal homepage: www.elsevier.com/locate/envpol

Source apportionment of fine particulate matter organic carbon in Shenzhen, China by chemical mass balance and radiocarbon methods* Ibrahim M. Al-Naiema a, Subin Yoon b, Yu-Qin Wang c, d, Yuan-Xun Zhang c, e, Rebecca J. Sheesley b, *, Elizabeth A. Stone a, f, ** a

Department of Chemistry, University of Iowa, Iowa City, IA 52242, USA Department of Environmental Science, Baylor University, Waco, TX 76798, USA College of Resources and Environment, University of Chinese Academy of Sciences, Beijing 100049, China d School of Environmental Science and Engineering, Shaanxi University of Science & Technology, Xi'an, Shaanxi 710021, China e Huairou Eco-Environmental Observatory, Chinese Academy of Sciences, Beijing 101408, China f Chemical and Biochemical Engineering, University of Iowa, Iowa City, IA 52242, USA b c

a r t i c l e i n f o

a b s t r a c t

Article history: Received 15 December 2017 Received in revised form 28 March 2018 Accepted 16 April 2018

Chemical mass balance (CMB) modeling and radiocarbon measurements were combined to evaluate the sources of carbonaceous fine particulate matter (PM2.5) in Shenzhen, China during and after the 2011 summer Universiade games when air pollution control measurements were implemented to achieve air quality targets. Ambient PM2.5 filter samples were collected daily at two sampling sites (Peking University Shenzhen campus and Longgang) over 24 consecutive days, covering the controlled and uncontrolled periods. During the controlled period, the average PM2.5 concentration was less than half of what it was after the controls were lifted. Organic carbon (OC), organic molecular markers (e.g., levoglucosan, hopanes, polycyclic aromatic hydrocarbons), and secondary organic carbon (SOC) tracers were all significantly lower during the controlled period. After pollution controls ended, at Peking University, OC source contributions included gasoline and diesel engines (24%), coal combustion (6%), biomass burning (12.2%), vegetative detritus (2%), biogenic SOC (from isoprene, a-pinene, and b-caryophyllene; 7.1%), aromatic SOC (23%), and other sources not included in the model (25%). At Longgang after the controls ended, similar source contributions were observed: gasoline and diesel engines (23%), coal combustion (7%), biomass burning (17.7%), vegetative detritus (1%), biogenic SOC (from isoprene, apinene, and b-caryophyllene; 5.3%), aromatic SOC (13%), and other sources (33%). The contributions of the following sources were smaller during the pollution controls: biogenic SOC (by a factor of 10e16), aromatic SOC (4e12), coal combustion (1.5e6.8), and biomass burning (2.3e4.9). CMB model results and radiocarbon measurements both indicated that fossil carbon dominated over modern carbon, regardless of pollution controls. However, the CMB model needs further improvement to apportion contemporary carbon (i.e. biomass burning, biogenic SOC) in this region. This work defines the major contributors to carbonaceous PM2.5 in Shenzhen and demonstrates that control measures for primary emissions could significantly reduce secondary organic aerosol (SOA) formation. © 2018 Elsevier Ltd. All rights reserved.

Keywords: Air quality Aerosol Emission control Secondary organic aerosol Universiade

1. Introduction

*

This paper has been recommended for acceptance by Baoshan Xing. * Corresponding author. ** Corresponding author. Department of Chemistry, University of Iowa, Iowa City, IA 52242, USA. E-mail addresses: [email protected] (R.J. Sheesley), betsy-stone@ uiowa.edu (E.A. Stone). https://doi.org/10.1016/j.envpol.2018.04.071 0269-7491/© 2018 Elsevier Ltd. All rights reserved.

Shenzhen is a rapidly developing and heavily urbanized coastal city in the Pearl River Delta (PRD) region of China. The economic growth of PRD cities has been accompanied by increased emission of air pollutants. Elevated levels of particulate matter (PM) have been attributed to primary emissions from industry and transportation and secondary aerosol formation (He et al., 2011; Huang

I.M. Al-Naiema et al. / Environmental Pollution 240 (2018) 34e43

et al., 2014; Yuan et al., 2006). High concentrations of PM pose risk to public health (Huang et al., 2012) and negatively affect visibility (Tan et al., 2009). Assessment of the chemical composition of PM and its contributing emission sources is therefore crucial to the implementation of effective air quality regulations. To estimate source contributions to ambient PM in the PRD region, receptor models including positive matrix factorization (Dai et al., 2013; Kuang et al., 2015) and chemical mass balance (CMB) modeling (Wang et al., 2016; Zheng et al., 2011) have both been used. CMB has been widely used to apportion PM to its primary emission sources when source profiles are available (Kong et al., 2010). In Guangzhou, for instance, secondary organic carbon (SOC), coal combustion, and cooking sources were found to have contributed more than 20%, 14%, and 11% of fine particle (PM2.5) organic carbon (OC), respectively (Wang et al., 2016). These results underscore the importance of combustion and secondary sources to PM2.5 OC in the PRD. To distinguish carbonaceous PM derived from fossil fuels from that devrived from contemporary sources, radiocarbon (14C, t1/ 2 ¼ 5730 years) measurements have been used (Gelencser et al., 2007; Gustafsson et al., 2009). 14C is conserved with respect to emission conditions, atmospheric transport and chemical transformations (Szidat et al., 2004), making it a reliable tracer of modern carbonaceous matter. Using a combination of 14C measurements and organic tracers in an industrial city in the PRD, Liu et al. (2014) demonstrated that fossil sources contributed 71% of the elemental carbon (EC) and 38% of OC detected. In a Chinese regional background site on Hainan Island (550 km from Shenzhen), radiocarbon analysis demonstrated that the contribution of fossil sources to EC was 51% and OC was 30% (Zhang et al., 2014), with fossil fuel emissions transported from regional industrial cities. These findings demonstrate strong influences from both modern and fossil fuel emissions on carbonaceous PM2.5 in the PRD. Secondary organic aerosols (SOA) that are important contributors to PM mass are produced in the atmosphere through the photooxidation of VOCs from biogenic and anthropogenic precursors (Kroll and Seinfeld, 2008). The contributions of some precursors to SOA can be estimated using a SOA tracer approach, developed by Kleindienst et al. (2007), in which SOA is estimated based on the ambient concentration of SOA tracers by way of SOA tracer-to-SOA (or tracer-to-SOC) mass ratios determined in smog chambers. In this way, aromatic precursors (i.e. toluene) have been shown to contribute two-thirds of the total estimated SOC in the PRD region (Ding et al., 2012). This dominance reflects the significance of anthropogenic activities on SOA production in the PRD, with a minor contribution from biogenic VOCs like isoprene and monoterpenes. The effects of short-term pollution control on the concentration and composition of atmospheric PM have been the focus of prior field studies in China. Vehicular and industrial emission controls were enforced during the 2008 Beijing Olympic Games, reducing PM10, nitrogen oxides (NOx), sulfur dioxide (SO2), and non-methane VOC by 55%, 47%, 41%, and 57%, respectively (Wang et al., 2010). Simultaneously, black carbon (45%), OC (31%), and polycyclic aromatic hydrocarbons (PAH) decreased (Wang et al., 2011). Benzene, toluene, ethylbenzene, and xylenes (BTEX) decreased by  47% (Liu et al., 2009). In addition to emission controls, Gao et al. (2011) suggested that wind direction and precipitation also contributed to air pollutant reductions during this period. In another effort at reducing air pollution during sporting events, during the 16th Asian Games in 2010 in Guangzhou, emissions from power plants, industry, mobile sources, and construction activities were restricted and PM2.5 decreased by 26%, while both SO2 and NOx dropped by >40% (Liu et al., 2013). These studies, focused predominantly on primary air pollutants, underscore the importance of controlling

35

emission sources for improving air quality. However, the effect of pollution controls on the relative abundances of SOC from biogenic and anthropogenic origins has not previously been evaluated. In 2011, Shenzhen hosted the 26th summer Universiade, an international sporting event, during which strict controls on emission sources were implemented to improve air quality, including reduction of: (i) emission of NOx from power plants, commercial and industrial boilers, and motor vehicles; (ii) SO2 emission by controlling fuel sulfur content, the flue gas from desulphurization units, and coal-fired power plants (iii) VOC emissions from industries including printing, adhesives, and furniture; (iv) PM and other air pollutants from construction sites, open biomass burning, and on-road vehicles (Dewan et al., 2016; Wang et al., 2014). In addition to the controls on emission sources in Shenzhen, industrial activities in neighboring cities were minimized. These conditions provided a unique opportunity to examine the effect of anthropogenic activities on the absolute and relative levels of primary and secondary PM2.5 sources in Shenzhen. In this study, we assess PM2.5 concentrations, composition, and sources, both under strict emission controls (the “controlled period”), and after Universiade, when the controls were lifted (the “uncontrolled period”). Organic molecular markers and SOA tracers were measured in PM2.5 collected over 24 days. PM2.5 OC was apportioned by a molecular marker-driven CMB model (Schauer et al., 1996) and the SOC tracer method (Kleindienst et al., 2007). Tracers included levoglucosan for biomass burning (Simoneit et al., 1999), PAH and hopanes for fossil fuels including coal combustion (Zhang et al., 2008) and vehicle emissions (Schauer et al., 2002), odd-numbered n-alkanes for vegetative detritus (Rogge et al., 1993), and SOA products identified in chamber studies for biogenic-and aromatic-VOC derived SOA (Kleindienst et al., 2007). The resulting contributions of fossil and modern sources to OC and elemental carbon (EC) were compared to radiocarbon measurements of fossil and modern carbon over the same time period. This study examines differences in PM2.5 and its sources during and after Universiade in 2011, providing new insight to the effect of primary emission controls on SOC.

2. Methods 2.1. Sampling PM2.5 samples were simultaneously collected from two sampling locations in Shenzhen, China from 12 August to 4 September 2011. Teflon and quartz filters (47 mm, Whatman) were used to collect PM2.5 samples for mass and organic speciation, respectively. The Longgang (LG) site is located in the Longgang district of Shenzhen (22.70 N, 114.21 E, 161 m) on top of a 31-floor residential building at a height of 90 m from the ground level, about 500 m north of the main Universiade stadium. The Peking University (PU) site is located at Nanshan district of Shenzhen (22.60 N, 113.97 E, 50 m, 45 hm north of the LG site) atop of a graduate building at a height of 16 m. The samplers' heights provided well-mixed air masses at the point of sample collection. Detailed descriptions for both sampling site and sampling techniques are provided elsewhere (Dewan et al., 2016). Wind and visibility data during this study were obtained from the weather forecast at Shenzhen Bao'an International Airport (SGSZ), approximately 18 and 48 km east of PU and LG sampling sites, respectively. The difference in the altitudes of the two sites could have affected PM collected, so throughout this report we emphasize differences between controlled and uncontrolled periods at each site, rather than comparisons across the two sites.

36

I.M. Al-Naiema et al. / Environmental Pollution 240 (2018) 34e43

2.2. Chemical analysis of organic species

3. Results and discussion

Filter extraction and gas chromatography-mass spectrometry (GC-MS) followed established methods (Al-Naiema et al., 2015; Stone et al., 2012) and are summarized in the supplementary information (SI).

3.1. PM2.5 mass concentrations

2.3. Chemical mass balance source apportionment modeling Chemical mass balance (CMB v8.2) modeling (EPA, 2004) was used to estimate source contributions to organic carbon in PM2.5, using effective variance weighted least squares (Watson et al., 1984). The CMB model relies on prior knowledge of emission profiles and assumes that those profiles are representative of the investigated samples. Source profiles were selected to represent emission sources and conditions in China when possible, i.e. for biomass burning (Zhang et al., 2007) and coal combustion (Zhang et al., 2008). When such profiles were not available, profiles developed elsewhere, but applied previously in source apportionment in China (Guo et al., 2013; Zheng et al., 2005), were used. Input source profiles and chemical species for the model are reported in the SI. The model fit was considered acceptable when the correlation coefficient (R2) was greater than 0.8, and chi-squared (c2) was less than 7. 2.4. Radiocarbon (14C) measurements Four composites and one lab blank were prepared for 14C analysis. The composites included equal mass fractions of filter samples collected during the controlled and uncontrolled periods for both LG and PU sites. This compositing scheme ensured that composites impartially represented source contributions from each sample regardless of varying daily mass concentration. Filter punches for each composite and blank were collected in baked petri dishes, acid-fumigated in a desiccator over 1 N hydrochloric acid for 12 h, and then dried at 60  C for 1 h. Each petri dish was then wrapped in baked aluminum foil, bagged in individual Ziploc bags, and shipped on ice to the National Ocean Science Accelerator Mass Spectrometry (NOSAMS) facility for 14C analysis. At NOSAMS, samples were analyzed using an accelerator mass spectrometry (AMS) to determine the fraction of modern (Fm) carbon. Fm is the deviation of the 14C/12C ratio in a sample from 95% of the reference “Modern”, NBS Oxalic Acid I, which is normalized to d13CVPDB ¼ 19‰ (Olsson, 1970). Apportionment for contemporary (or non-fossil) to fossil fuel sources can be calculated using a mixing model ratio for D14CTOC:

D14CTOC ¼ (D14Ccontemporary) (fM ) þ (D14Cfossil) (1-fM ) In the calculation, known end-member D14C values were included for radiocarbon-dead fossil fuel (1000‰) (Gustafsson et al., 2009) and contemporary (þ67.5‰). The end-member value for contemporary sources was an average of wood burning (þ107.5‰) (Zotter et al., 2014) and fresh biogenic (þ28.1‰) (Widory, 2006) sources. The fM corrected for knownend-members is multiplied by ambient concentration to calculate fossil and contemporary carbon concentrations. 2.5. Statistical analysis Statistical analyses were used to evaluate significant differences in PM2.5 composition and sources during the controlled and uncontrolled periods. A nonparametric t-test (Wilcoxon) was used to assess the statistical differences at the 95% confidence interval using Statistical Package for the Social Sciences (SPSS) software.

PM2.5 concentrations measured at Longgang (LG) and Peking University (PU) sites in Shenzhen were significantly lower during Universiade games when strict emission controls were implemented (12e23 August) than during the uncontrolled period (24 August e 4 September; Table 1). At LG, the average PM2.5 concentration during the controlled period was 24.9 ± 5.0 mg m3 versus 53.8 ± 6.1 mg m3 during the uncontrolled period. At PU, the average PM2.5 concentrations were 12.8 ± 3.5 mg m3 during the controlled period and 48.0 ± 8.1 mg m3 during the uncontrolled period. Pollution controls were not the only influence on PM2.5 concentrations over these time periods: wind directions shifted between the controlled and uncontrolled periods. Southerly winds that bring relatively clean ocean air to Shenzhen were predominant during the controlled period, while northwesterly winds transporting relatively polluted air from continental areas were more prevalent during the uncontrolled period (Fig. 1a). Thus, changes in wind direction added to the effects of emissions control to lower PM2.5 by 54% at LG and 73% for PU, on average, during the controlled period. Lower PM2.5 concentrations corresponded to doubling of visibility (Fig. 1b). The average PM2.5 concentrations during the uncontrolled period are comparable to those reported in summer time in Shenzhen (Dai et al., 2013; Niu et al., 2006) and are approximately half of PM2.5 levels reported in winter (99.0 ± 17.6) (Niu et al., 2006). Thus, the uncontrolled period is considered to be representative of typical summertime concentrations in the absence of emission controls. On all of the study days, the measured PM2.5 concentrations in Shenzhen were below the 24 h average PM2.5 Class-II standard for urban and industrial cities of 75 mg m3 (Chinese Ambient Air Quality Standard, GB 3095e2012) (GB3095, 2012).

3.2. Elemental and organic carbon EC and OC were significant contributors to PM2.5 mass, with average contributions of 24.2 ± 4.6% and 9.5 ± 4.2% at LG and 31.4 ± 6.7% and 15.5 ± 9.2% at PU, respectively. OC concentrations at both sites were significantly increased after the controlled period (Table 1; Fig. 1). Meanwhile, EC increased significantly at the PU site, but only slightly (not significantly) at the LG site (Table 1; Fig. 1). Further discussion of OC and EC levels are provided elsewhere (Wang et al., 2014). OC:EC ratios across both sites increased from an average of 1.7 during the controlled period to 3.6 during the uncontrolled period (Fig. 1), indicating a shift in the sources of carbonaceous aerosol between the controlled and uncontrolled periods. OC and EC sources are discussed in section 3.3. The OC and EC concentrations during the uncontrolled period were comparable to prior studies in the PRD during late summer. Prior studies have reported OC in Shenzhen in the summer of 2004 ranging 4.0e20.6 mg m3 and EC ranging 1.7e3.7 mg m3, with a mean OC:EC ratio of 3.4 (Niu et al., 2006). Slightly lower concentrations were reported for Shenzhen in 2002, with mean OC levels of 7.6 ± 4.9 mg m3 and EC levels of 4.2 ± 3.1 mg m3 (Cao et al., 2004). In nearby Guangzhou in the summer of 2008, OC ranged from 1.92 to 13.7 mg m3 and EC ranged 0.69e5.07 mg m3 with a mean OC:EC of 3.41 (Ding et al., 2012). These comparisons indicate that the uncontrolled OC and EC levels in Shenzhen found in this study are typical for this region.

I.M. Al-Naiema et al. / Environmental Pollution 240 (2018) 34e43

37

Table 1 Summary of the mean concentrations of PM2.5, OC, and EC (mg m3) in Shenzhen during the controlled and uncontrolled periods at two sampling sites, along with the concentrations of SOA for isoprene, a-pinene, b-caryophyllene and toluene (ng m3). P-values correspond to t-test comparisons of the controlled and uncontrolled periods at each site, with p < 0.05 considered to be statistically significant. PM components

Longgang (LG)

Peking University (PU)

Controlled

Uncontrolled

p-value

Controlled

Uncontrolled

p-value

PM2.5 mass (mg m3) EC (mg m3) OC (mg m3) Isoprene SOA tracers (ng m3) 2-Methylglyceric acid 2-Methylthreitol 2-Methylerythritol Sum of isoprene SOA tracers a-Pinene SOA tracers (ng m3) 3-Hydroxyglutaric acid Pinic acid 2-Hydroxy-4,4-dimethylglutaric acid 3-Acetyl hexanedioic acid Sum of a-pinene SOA tracers b-Caryophyllene SOA tracer (ng m3) b- Caryophyllinic acid Toluene SOA tracer (ng m3) 2,3-Dihydroxy-4-oxopentanoic acid

24.9 ± 1.5 3.3 ± 0.4 5.9 ± 0.4

53.8 ± 1.7 3.6 ± 0.2 13.1 ± 0.5

<0.001 0.570 <0.001

12.8 ± 1.0 2.8 ± 0.4 4.3 ± 0.5

48.0 ± 2.3 4.3 ± 0.5 15.3 ± 1.7

<0.001 0.031 <0.001

1.1 ± 1.0 1.6 ± 2.5 2.8 ± 3.6 5.6 ± 6.1

4.4 ± 4.5 4.6 ± 5.8 21.2 ± 28.7 30.3 ± 38.4

<0.001

0.9 ± 0.9 0.5 ± 0.8 1.3 ± 1.6 2.7 ± 3.1

8.8 ± 7.4 9.0 ± 8.8 28.9 ± 27.6 46.6 ± 41.6

<0.001

2.8 ± 3.7 1.0 ± 1.9 2.2 ± 3.0 0.8 ± 1.7 6.8 ± 8.7

21.2 ± 25.6 4.6 ± 4.8 7.0 ± 9.8 5.7 ± 11.8 38.5 ± 40.2

<0.001

0.8 ± 1.8 0.8 ± 1.0 0.6 ± 0.6 0.3 ± 0.5 2.5 ± 2.5

17.9 ± 14.6 4.4 ± 6.1 57.6 ± 53.4 23.1 ± 21.4 103.0 ± 76.7

<0.001

0.8 ± 0.8

2.8 ± 2.4

0.034

1.5 ± 1.2

6.6 ± 3.4

0.002

5.1 ± 4.9

12.3 ± 9.5

0.06

2.0 ± 1.4

26.8 ± 9.2

0.002

3.3. Source apportionment by CMB modeling The CMB model apportioned PM2.5 OC to five primary sources (vegetative detritus, gasoline vehicles, diesel engines, coal combustion, and biomass burning) and four secondary sources (SOC from isoprene, a-pinene, b-caryophyllene, and aromatic precursors). The observed concentrations of molecular markers used in source apportionment are summarized in Table 1 and Table S1. CMB results are summarized in Fig. 2, Fig. S1, and Table S2 for controlled and uncontrolled conditions at each site. CMB results were not reported for several days that had an unacceptable model fit (August 12, 22, 26 and September 1 for LG, and August 24, 29, and 31 for PU), as indicated by R2 < 0.8 and/or c2 values > 7 (section 2.3), indicating that the selected profiles poorly fit the ambient data. The difference between the concentrations of apportioned sources and total OC mass is represented as “other OC”. On average, during controlled and uncontrolled periods, respectively the model apportioned 90% and 67% of OC in LG, and 88% and 75% of OC in PU. During the controlled period at LG, the average OC mass of 5.9 mg m3 was apportioned 80% to primary sources and 10% to secondary sources; 10% was not apportioned. The major primary OC sources at LG, on average, were gasoline vehicles (38%), diesel engines (20%), and coal combustion (12%). At PU during the controlled period, the average OC mass of 4.3 mg m3 was apportioned 80% to primary sources and 8% to secondary sources; 12% was not apportioned. The major primary OC sources were gasoline vehicles (30%), diesel engines (26%), and biomass burning (21%) (Table S2). With regard to the modelresolved secondary OC, aromatic precursors were more significant contributors than biogenic precursors by a factor of 4.5 at LG and 2.5 at PU. After the controls were lifted, OC at LG was apportioned 48% to primary sources and 19% to secondary sources; 33% was not apportioned. At PU, OC was apportioned 44% to primary sources and 31% to secondary sources; 25% was not apportioned. The major primary OC sources at both sites after the controls were lifted were gasoline vehicles, diesel engines, and biomass burning (Table S2). In regard to secondary OC, precursors were determined to be aromatic, a-pinene, and isoprene, and SOC was dominated by aromatics for both sites. The OC fraction not attributed to primary and secondary sources

is expected to derive from sources that were not included in the model. Based on prior studies in the PRD (Li et al., 2012; Zheng et al., 2011), these are expected to include cigarette smoke, cooking emissions, and road dust. In addition, SOA is likely to be underestimated (Zheng et al., 2002), as SOC formed from VOC emitted by biomass burning, semi-VOCs like long-chain alkanes, and other precursors were not included in this model (Huang et al., 2011). 3.3.1. Gasoline and diesel engines The combined vehicular emissions from gasoline and diesel engines are the largest source of OC in Shenzhen. Together, gasoline and diesel engines contributed 3.33 mgC m3 and 3.03 mgC m3 during the controlled and uncontrolled periods at LG, while they contributed 2.41 mgC m3 and 3.34 mgC m3 at PU, respectively. These results indicate a slight decrease in vehicle-derived OC at LG after the controls were lifted, which may be attributed to the drop in post-Universiade transportation demands near the main stadium that was closer to this site. Alternatively, the control on motor vehicles (alternating odd-even license plate vehicle operation) did not likely influence the change in OC at LG. Meanwhile, there was a substantially larger increase in vehicle-derived OC at PU after the controls were lifted. The stability of the CMB model, with respect to its estimate of vehicular contributions to OC, was evaluated in a sensitivity test in which the non-catalyzed gasoline profile used in the “base-case” model results was replaced with a catalyzed profile (Lough et al., 2007), as described in section 2.3 (Lough and Schauer, 2007). The summed gasoline and diesel engine contributions to OC for the base-case scenario, relative to the sensitivity test, are shown in Fig. S2 for each site. The results show a good agreement (R2  0.994) between the base case and sensitivity test. The slopes of these regressions indicate a minor underestimation (11e13% of the vehicle contribution) of the base case relative to the sensitivity test, which is within the standard error of the estimate. Thus, the selection of the non-gasoline engine profile has only a minor influence on the estimated vehicular contribution to OC, indicating that this is a robust estimate. 3.3.2. Coal combustion Coal combustion contributions were lower at both sites during the controlled period, though more significantly reduced at PU.

38

I.M. Al-Naiema et al. / Environmental Pollution 240 (2018) 34e43

Fig. 1. Wind rose plots (a) showing the frequency of wind directions in Shenzhen during the controlled period and uncontrolled periods, along with time series of the daily ambient concentrations of PM2.5 at Longgang and Peking University and visibility (b); time series of carbonaceous aerosol (OC, EC, and OC:EC) for Longgang (c) and Peking University (d).

During the controlled period, the contribution from coal combustion was slightly reduced to 0.65 ± 0.08 mgC m3 in LG and significantly reduced to 0.13 ± 0.03 mgC m3 in PU (Table S2). During the uncontrolled period, coal contributed 0.93 ± 0.13 mgC m3 at LG and 0.88 ± 0.14 mgC m3 at PU, which are comparable to annual average contributions of coal combustion sources in Shenzhen of 1.10 ± 0.12 mgC m3, reported elsewhere (Zheng et al., 2011). The more significant change observed at PU is likely due to the fact that there were more coal-operated power plants near PU compared to LG (Dewan et al., 2016).

3.3.3. Biomass burning OC from biomass burning was lower at both sites during the controlled period. The average biomass burning contribution to PM2.5 OC during the controlled period was 0.47 ± 0.12 mgC m3 at LG and 0.77 ± 0.06 mgC m3 at PU. After the controls were lifted, the average contribution of biomass burning to OC (in mgC m3) increased by a factor of 4.9 at LG and by 2.3 at PU. The increase shows the influence of the restriction on emissions from woodfired industrial boilers and biomass burning. These results indicate that reducing biomass combustion in the PRD can improve air quality in Shenzhen through reductions in the associated PM2.5 OC.

I.M. Al-Naiema et al. / Environmental Pollution 240 (2018) 34e43

Fig. 2. Source apportionment of PM2.5 OC in Shenzhen at Longgang (LG) and Peking University (PU), reported as the average relative source contributions to OC (%) during controlled and uncontrolled periods.

It is potentially an important policy measure, as previous studies have found that biomass burning contributed 20.8% to PM2.5 OC in Shenzhen during October, when households may have used biofuels for heating (Zheng et al., 2011). 3.3.4. Vegetative detritus Vegetative detritus made a minor contribution to OC in Shenzhen, with its average contribution less than 2.5% of OC at both sites (Table S2). Apportioning vegetative detritus using the CMB model is largely dependent on the concentrations of n-alkanes, which can have either biogenic or anthropogenic origins. The carbon preference index (CPI) is a diagnostic parameter that provides a means of identifying the sources of n-alkanes, where a CPI greater than three is an indication of biological origins, while a CPI of approximately one suggests anthropogenic origin (Simoneit, 1989). Average CPI value for n-alkane (C25-C34) in this study was 1.5 in LG and 1.7 in PU, with no significant change in CPI values for controlled versus uncontrolled periods. The CPI result indicates very little odd-carbon preference, as n-alkanes are primarily from evaporation and combustion of fossil fuels. Consistent with this is the small contribution from vegetative detritus to OC in Shenzhen. 3.3.5. Secondary organic aerosol SOC in Shenzhen was largely dominated by aromatic precursors, which contributed more than 70% of the apportioned SOC. The high level of SOC from aromatic relative to biogenic precursors had been previously observed in the PRD region (Ding et al., 2012), and attributed to the elevated levels of aromatic VOCs such as toluene in this industrial region (Barletta et al., 2008), compared to those quantified in other cities (Mohamed et al., 2002; von Schneidemesser et al., 2010). The contribution of SOC from aromatic, isoprene, and a-pinene precursors to OC is discussed in detail in section 3.4. 3.3.6. Source apportionment of EC Elemental carbon (EC) was simultaneously apportioned with OC by the CMB model. During the controlled period at PU, EC was primarily attributed to diesel engines (96 ± 2%; ±standard deviation), with minor contributions from non-catalyzed gasoline vehicles (1 ± 1%), coal combustion (1 ± 1%), and biomass burning (2 ± 1%). During the uncontrolled period at PU absolute

39

concentrations of EC were higher, as were the absolute contributions from diesel engines, coal combustion, and biomass burning, with 97% of EC attributed to fossil sources. At LG during the controlled period, EC was attributed largely to diesel engines (92 ± 6%), with minor contributions from coal combustion (6 ± 4%), non-catalyzed gasoline engines (1 ± 1%), and biomass burning (1 ± 1%). During the uncontrolled period, absolute concentrations of EC at LG increased, as did absolute contributions to EC from diesel engines, coal combustion, and biomass burning; the EC contribution from fossil sources was estimated to be 96%. CMB results indicated that diesel engines were largely responsible for the observed increase in EC from the controlled to uncontrolled periods at LG (from 3.3 to 2.8 mgC m3) and PU (3.8e2.8 mgC m3) and that other fossil fuel and biomass emissions were minor contributors to EC throughout this study. However, previous studies in the PRD that conducted radiocarbon source apportionment of EC reported a greater biomass burning influence than the current study does. In Guangzhou, PRD, Zhang et al. (2015) reported EC was 57 ± 5% fossil during a less-polluted event and 80 ± 2% fossil during a heavily polluted event in 2013, while Liu et al. (2014) reported that EC was 60e91% fossil EC for eight samples spanning 2012e13. Although these studies represent a different location in the PRD, these studies suggest a relatively larger contribution from contemporary carbon to EC than was found in this study. Potential reasons for this are discussed in section 3.6. 3.4. Secondary organic aerosol 3.4.1. Isoprene-derived SOC Three isoprene tracers, 2-methylglyceric acid (MGA) and 2methyltetrols (2-methylthreitol and 2-methylerythritol; MTLs) were observed in Shenzhen (Table 1, Fig. 3a). During the controlled period, the estimated contribution of isoprene SOC to OC was significantly lower (p < 0.05), by a factor of 5 in LG and 4.5 in PU. The overall isoprene SOC contributed less than 2% to the OC mass in both sampling sites, indicating that isoprene SOC is not a major source of OC in Shenzhen. On average, the concentrations of the sum of the three tracers increased significantly during the uncontrolled period from 5.6 ± 4.5 ng m3 to 30.9 ± 27.9 ng m3 in LG, and from 2.7 ± 2.0 ng m3 to 46.6 ± 29.9 ng m3 at PU, respectively (Table 1). The MGA/MTL ratio remained steady throughout: the average (±standard deviation) MGA/MTL ratio was 0.6 (±0.6) in LG, and 0.5 (±0.4) in PU. Since MGA forms through the high-NOx isoprene SOA formation pathway (Surratt et al., 2010), this result suggests that there was no substantive shift in the effect of NOx on forming isoprene SOA. It is likely that controlled period reductions in emissions of anthropogenic pollutants (e.g., sulfur dioxide that contribute to aerosol acidity when oxidized to sulfuric acid), as indicated by the lower sulfate levels (Dewan et al., 2016), decreased the extent of SOA formation at both sites. In the Southeastern United States, sulfate and isoprene SOA positively correlate (Xu et al., 2015), indicating acid-enhanced isoprene SOA formation. Similarly, in this dataset, there is a statistically significant positive correlation between sulfate and isoprene tracer concentrations (r > 0.58, p < 0.004), with trends shown in Fig. S3. Thus, reductions in biogenic SOA may be accessible by decreasing anthropogenic sulfate levels. 3.4.2. a-Pinene-derived SOC a-Pinene SOC contributed up to 3.5% of OC in Shenzhen (Table S2), and that contribution was significantly increased (p < 0.05) by an average factor of 2.6 in LG and 8.8 in PU during the

40

I.M. Al-Naiema et al. / Environmental Pollution 240 (2018) 34e43

Fig. 3. Ambient concentrations of secondary organic tracers: A) sum of three isoprene SOA tracers, B) the sum of four a-pinene SOA tracers, and C) one toluene SOA tracer at LG and PU during the controlled and uncontrolled periods.

uncontrolled period. Four a-pinene tracers (3-hydroxyglutaric acid, pinic acid, 2-hydroxy-4,4-dimethylglutaric acid, and 3-acetyl hexanedioic acid) were consistently detected in the samples collected from both sampling sites (Fig. 3b, Table 1). In general, the concentrations of these tracers in PU are higher than those in LG, likely because the PU site is surrounded by forest, unlike LG which is more urbanized with fewer green spaces. Like isoprene tracers, the sum of a-pinene tracers was significantly higher during the uncontrolled period (38.5. ± 40.2 ng m3) than in the controlled period (6.8 ± 8.7 ng m3) at LG and PU (103.0 ± 76.7 ng m3 and 2.5 ± 2.5 ng m3, respectively). a-Pinene SOA tracer levels showed consistent temporal trends to isoprene SOA tracers (Fig. 3). It has been previously reported that a-pinene SOA tracer concentrations correlate with gas phase concentrations of NOx and SO2 (Xu et al., 2015). Monoterpene SOA reduction during the controlled period, then, is likely due to emission controls’ reduction of the ambient concentrations of these species. Although there was no significant correlation between apinene SOA tracers and nitrate ions, there is a significant moderate correlation between the sum of a-pinene SOA tracers and sulfate at PU (r ¼ 0.56, p ¼ 0.005) and a significant strong correlation at LG (r ¼ 0.79, p < 0.001, Fig. S3). The ratios of biogenic SOA tracers-tosulfate were consistently greater during the uncontrolled period, suggesting that SOA formation was enhanced during the uncontrolled period. Thus, a-pinene SOA may also be reduced by controlling primary emissions.

3.4.3. b-Caryophyllene-derived SOC b-Caryophyllene-derived SOC is a minor source of OC in

Shenzhen, with an average contribution of less than 1.8% (Table S2). Upon emission control, the concentrations of b-caryophyllene SOC decreased significantly at both sites (Table 1).

3.4.4. Aromatic VOC-derived SOC The SOC from aromatic VOCs was estimated based on the ambient concentrations of 2,3-dihydroxy-4-oxopentanoic acid (DHOPA). The average ambient concentrations of DHOPA rose from 5.1 ± 4.7 ng m3 in LG and 2.0 ± 1.4 ng m3 in PU to 12.3 ± 9.5 ng m3 and 26.8 ± 9.2 ng m3, respectively, after the pollution controls were lifted. The increase in DHOPA is statistically significant at both sites: LG (p ¼ 0.06) and PU (p ¼ 0.002). SOC from aromatic VOCs was the most abundant contributor to OC among the quantified sources and its contribution to OC during the uncontrolled period was as high as 23% at the PU site. Aromatic SOC contributed much more to PM2.5 OC than did biogenic SOC from isoprene and a-pinene. These observations can be explained by prior observations that the rates of aromatic VOC emissions, such as toluene and xylenes, are higher than emission rates of biogenic VOC in industrial megacities in the PRD region (Wang et al., 2013). Several studies of megacities such as Mexico City (Stone et al., 2010) and the PRD (Ding et al., 2012; Wang et al., 2013), have also reported higher contributions from anthropogenic precursors to SOC than biogenic precursors. This result illustrates that anthropogenic SOC contributes more than biogenic sources to the OC fraction of PM2.5 in megacities, despite global SOA budgets largely dominated by biogenic sources (Henze et al., 2008).

I.M. Al-Naiema et al. / Environmental Pollution 240 (2018) 34e43

3.5. Isotope analysis 3.5.1. Radiocarbon (14C) For the radiocarbon results, the total carbon was determined to be predominantly fossil for the entire period; these total carbon results would include both OC and EC contributions. By following ambient levels and percent contributions of both fossil and contemporary carbon, it is possible to determine whether any relative or absolute change in source contributions occurred over the study period. Ambient concentrations indicated that fossil and contemporary carbon increased at both sites during the uncontrolled period, but the concentration of contemporary carbon increased to a greater degree. PU had more fossil signature than LG, in both controlled and uncontrolled periods. 3.5.2. Stable isotopes: carbon (13C) Stable carbon and nitrogen (d 13C and d 15N) for both sampling locations are shown in Fig. S4. At LG, the average d 13C during the controlled period was 27.2 ± 0.5‰ and increased significantly to 26.6 ± 0.2‰ after the emission control (p ¼ 0.003). At PU the average d 13C was constant in both periods (26.5). The stable carbon fraction of total carbon can be affected by the end members, or isotope signature of the primary emission sources; however, it can also be affected by kinetic isotope effects during reaction of gas or particle phase species. More investigation would be needed to determine what is driving the small change in the d 13C at LG. However, since the difference is only apparent at LG during the controlled period, this does further stress the difference between the sites during the controlled period. The sites were significantly different in d 13C during the controlled period (p < 0.001); during the uncontrolled period, the 13C data from the two sites were not significantly different (p ¼ 0.440 using student two population (two-tailed) t-tests). There was no difference in the 15N by site or by the presence or absence of controls. 3.6. Comparison of CMB and radiocarbon source apportionment results Prior to comparing CMB and 14C results, the CMB-estimated source contributions to EC were summed with those for OC, so that fossil and contemporary contributions to total carbon could be compared across the two methods, as seen in Fig. 4. CMB results were excluded from source apportionment results for days with

41

unacceptable model fit, as described in section 2.3, whereas 14C measurements were performed as a composite and included all sampling days. Because CMB and 14C are distinct ways of quantifying the average difference between controlled and uncontrolled periods, 14C can indicate the overall bias in the CMB results and potential bias introduced by excluding non-fit days. For the uncontrolled period, the radiocarbon measurements and CMB estimates of fossil carbon agree within 2% of their total carbon contribution (Table S3) for both sampling locations. Thus, the unapportioned OC unresolved by the CMB model (25.6% at LG and 20.1% at PU) is contemporary in origin. For the controlled period, the radiocarbon measurements indicated a smaller fossil carbon fraction than was estimated by the CMB model, by an average of 19% of total C at LG and 8% at PU (Table S3). However, the radiocarbon measurements of fossil and contemporary carbon are either within or near to one standard deviation of the mean CMB estimates. While the radiocarbon was composited by control status, the tracer-based CMB apportionment was daily and demonstrated large day-to-day variability in source contributions. The overestimate of fossil C in the CMB model may result from an underestimation of biomass burning contributions to EC, because the cereal straw burning profile (Zhang et al., 2007) has a low EC-to-OC ratio. As discussed in section 3.3, EC was attributed almost entirely to fossil sources during the controlled period, which does not match previous measurements of EC radiocarbon in the PRD. In addition, if levoglucosan the primary biomass burning tracer degrades during transport, CMB would under-estimate the contribution of biomass burning to OC (Lai et al., 2014). Similar to the uncontrolled period, we conclude that the OC unapportioned by the CMB model is derived from contemporary sources (Table S3). The underestimate of contemporary OC by CMB modeling relative to 14C measurements was also observed in Bakersfield, California (Sheesley et al., 2017). In sum, while the CMB model provides source specificity in the apportionment of PM2.5 C to its sources, the 14C measurements provide constraint in interpreting the unapportioned OC fraction. Here, the unapportioned OC is shown to be contemporary in origin; biomass burning and/or biogenic SOC are the likely origins. This suggests that improved CMB model representation is needed for a more complete apportionment of OC by this approach. This could include improved source characterization and improved handling of degradation or atmospheric lifetime of tracers in the model.

Fig. 4. Comparison of the fossil and contemporary sources of total carbon (equivalent to the sum of organic and elemental carbon) estimated by a) radiocarbon (14C) analysis, and b) chemical mass balance (CMB) modeling.

42

I.M. Al-Naiema et al. / Environmental Pollution 240 (2018) 34e43

4. Conclusions Key findings from this study include:  PM2.5 concentrations in Shenzhen were significantly lower during the period of emission controls by 54% at LG and 73% at PU. However, because the wind blew from different directions during these two periods, it is not evident what extent of these reductions were due to stricter emission controls versus changes in wind direction.  OC was significantly lower during the controlled period, and CMB source apportionment modeling indicated significant reductions in OC contributions from coal combustion, biomass burning, and SOC from isoprene, a-pinene, b-caryophyllene, and aromatic VOCs.  The correlation of biogenic SOA tracers with sulfate suggested that anthropogenic emissions via acidic PM enhanced SOA formation during the uncontrolled period.  Aromatic SOC contributed up to 8% of OC during the controlled period and up to 23% during the uncontrolled period, indicating that anthropogenic VOC strongly influence SOA formation.  Measurements of 14C content indicated the importance of fossil and contemporary sources of OC and EC, with both decreasing in their ambient concentrations during the controlled period.  Radiocarbon estimates for the fossil contribution to carbon agree with CMB source apportionment within the uncertainty of the CMB estimates. Together, these data indicate that the unapportioned OC fraction in CMB is mainly from contemporary sources (i.e. biomass burning and biogenic SOA). Acknowledgments I.M.A. and E.A.S. were supported by the University of Iowa. R.J.S. and S.Y. were supported by Baylor University. We thank James J. Schauer from the University of Wisconsin-Madison for leadership in coordinating this research study. Appendix A. Supplementary data Supplementary data related to this article can be found at https://doi.org/10.1016/j.envpol.2018.04.071. References Al-Naiema, I., Estillore, A.D., Mudunkotuwa, I.A., Grassian, V.H., Stone, E.A., 2015. Impacts of co-firing biomass on emissions of particulate matter to the atmosphere. Fuel 162, 111e120. https://doi.org/10.1016/j.fuel.2015.08.054. Barletta, B., Meinardi, S., Simpson, I.J., Zou, S., Sherwood Rowland, F., Blake, D.R., 2008. Ambient mixing ratios of nonmethane hydrocarbons (NMHCs) in two major urban centers of the Pearl River Delta (PRD) region: Guangzhou and Dongguan. Atmos. Environ. 42, 4393e4408. https://doi.org/10.1016/ j.atmosenv.2008.01.028. Cao, J.J., Lee, S.C., Ho, K.F., Zou, S.C., Fung, K., Li, Y., Watson, J.G., Chow, J.C., 2004. Spatial and seasonal variations of atmospheric organic carbon and elemental carbon in Pearl River Delta Region, China. Atmos. Environ. 38, 4447e4456. https://doi.org/10.1016/j.atmosenv.2004.05.016. Dai, W., Gao, J., Cao, G., Ouyang, F., 2013. Chemical composition and source identification of PM2.5 in the suburb of Shenzhen, China. Atmos. Res. 122, 391e400. https://doi.org/10.1016/j.atmosres.2012.12.004. Dewan, N., Wang, Y.Q., Zhang, Y.X., Zhang, Y., He, L.Y., Huang, X.F., Majestic, B.J., 2016. Effect of pollution controls on atmospheric PM2.5 composition during Universiade in Shenzhen, China. Atmosphere 7. https://doi.org/10.3390/ atmos7040057. Ding, X., Wang, X.-M., Gao, B., Fu, X.-X., He, Q.-F., Zhao, X.-Y., Yu, J.-Z., Zheng, M., 2012. Tracer-based estimation of secondary organic carbon in the Pearl River Delta, south China. Journal of Geophysical Research-Atmospheres 117. https:// doi.org/10.1029/2011jd016596. EPA, 2004. United States Environmental Protection Agency-Chemical Mass Balance Model -CMBv8.2. Gao, Y., Liu, X., Zhao, C., Zhang, M., 2011. Emission controls versus meteorological conditions in determining aerosol concentrations in Beijing during the 2008

Olympic Games. Atmos. Chem. Phys. 11, 12437e12451. https://doi.org/10.5194/ acp-11-12437-2011. GB3095, 2012. Ambient air quality standards in China. In: Ministry of Environmental Protection and General Administration of Quality Supervision, Inspection and Quarantine of the People's Republic of China. Gelencser, A., May, B., Simpson, D., Sanchez-Ochoa, A., Kasper-Giebl, A., Puxbaum, H., Caseiro, A., Pio, C., Legrand, M., 2007. Source apportionment of PM2.5 organic aerosol over Europe: primary/secondary, natural/anthropogenic, and fossil/biogenic origin. Journal of Geophysical Research-Atmospheres 112. https://doi.org/10.1029/2006jd008094. Guo, S., Hu, M., Guo, Q., Zhang, X., Schauer, J., Zhang, R., 2013. Quantitative evaluation of emission controls on primary and secondary organic aerosol sources during Beijing 2008 Olympics. Atmos. Chem. Phys. 13, 8303e8314. https:// doi.org/10.5194/acp-13-8303-2013. € Kruså, M., Zencak, Z., Sheesley, R.J., Granat, L., Engstro € m, E., Gustafsson, O., Praveen, P.S., Rao, P.S.P., Leck, C., Rodhe, H., 2009. Brown clouds over South Asia: biomass or fossil fuel combustion? Science 323, 495e498. https://doi.org/ 10.1126/science.1164857. He, L.-Y., Huang, X.-F., Xue, L., Hu, M., Lin, Y., Zheng, J., Zhang, R., Zhang, Y.-H., 2011. Submicron aerosol analysis and organic source apportionment in an urban atmosphere in Pearl River Delta of China using high-resolution aerosol mass spectrometry. J. Geophys. Res.: Atmosphere 116. https://doi.org/10.1029/ 2010JD014566 n/a-n/a. Henze, D.K., Seinfeld, J.H., Ng, N.L., Kroll, J.H., Fu, T.M., Jacob, D.J., Heald, C.L., 2008. Global modeling of secondary organic aerosol formation from aromatic hydrocarbons: high- vs. low-yield pathways. Atmos. Chem. Phys. 8, 2405e2420. https://doi.org/10.5194/acp-8-2405-2008. Huang, D., Xu, J., Zhang, S., 2012. Valuing the health risks of particulate air pollution in the Pearl River Delta, China. Environ. Sci. Pol. 15, 38e47. https://doi.org/ 10.1016/j.envsci.2011.09.007. Huang, X.-F., He, L.-Y., Hu, M., Canagaratna, M., Kroll, J., Ng, N., Zhang, Y.-H., Lin, Y., Xue, L., Sun, T.-L., 2011. Characterization of submicron aerosols at a rural site in Pearl River Delta of China using an aerodyne high-resolution aerosol mass spectrometer. Atmos. Chem. Phys. 11, 1865e1877. https://doi.org/10.5194/acp11-1865-2011. Huang, X.F., Yun, H., Gong, Z.H., Li, X., He, L., Zhang, Y.H., Hu, M., 2014. Source apportionment and secondary organic aerosol estimation of PM2.5 in an urban atmosphere in China. Sci. China Earth Sci. 57, 1352e1362. https://doi.org/ 10.1007/s11430-013-4686-2. Kleindienst, T.E., Jaoui, M., Lewandowski, M., Offenberg, J.H., Lewis, C.W., Bhave, P.V., Edney, E.O., 2007. Estimates of the contributions of biogenic and anthropogenic hydrocarbons to secondary organic aerosol at a southeastern US location. Atmos. Environ. 41, 8288e8300. https://doi.org/10.1016/ j.atmosenv.2007.06.045. Kong, S., Han, B., Bai, Z., Chen, L., Shi, J., Xu, Z., 2010. Receptor modeling of PM2.5, PM10 and TSP in different seasons and long-range transport analysis at a coastal site of Tianjin, China. Sci. Total Environ. 408, 4681e4694. https:// doi.org/10.1016/j.scitotenv.2010.06.005. Kroll, J.H., Seinfeld, J.H., 2008. Chemistry of secondary organic aerosol: formation and evolution of low-volatility organics in the atmosphere. Atmos. Environ. 42, 3593e3624. https://doi.org/10.1016/j.atmosenv.2008.01.003. Kuang, B.Y., Lin, P., Huang, X.H.H., Yu, J.Z., 2015. Sources of humic-like substances in the Pearl River Delta, China: positive matrix factorization analysis of PM 2.5 major components and source markers. Atmos. Chem. Phys. 15, 1995e2008. https://doi.org/10.5194/acp-15-1995-2015. Lai, C., Liu, Y., Ma, J., Ma, Q., He, H., 2014. Degradation kinetics of levoglucosan initiated by hydroxyl radical under different environmental conditions. Atmos. Environ. 91, 32e39. https://doi.org/10.1016/j.atmosenv.2014.03.054. Li, Y.-C., Yu, J.Z., Ho, S.S.H., Yuan, Z., Lau, A.K.H., Huang, X.-F., 2012. Chemical characteristics of PM2.5 and organic aerosol source analysis during cold front episodes in Hong Kong, China. Atmos. Res. 118, 41e51. https://doi.org/10.1016/ j.atmosres.2012.05.026. Liu, H., Wang, X., Zhang, J., He, K., Wu, Y., Xu, J., 2013. Emission controls and changes in air quality in Guangzhou during the Asian Games. Atmos. Environ. 76, 81e93. https://doi.org/10.1016/j.atmosenv.2012.08.004. Liu, J., Li, J., Zhang, Y., Liu, D., Ding, P., Shen, C., Shen, K., He, Q., Ding, X., Wang, X., 2014. Source apportionment using radiocarbon and organic tracers for PM2. 5 carbonaceous aerosols in Guangzhou, South China: contrasting local-and regional-scale haze events. ES T (Environ. Sci. Technol.) 48, 12002e12011. https://doi.org/10.1021/es503102w. Liu, J.F., Mu, Y.J., Zhang, Y.J., Zhang, Z.M., Wang, X.K., Liu, Y.J., Sun, Z.Q., 2009. Atmospheric levels of BTEX compounds during the 2008 Olympic Games in the urban area of Beijing. Sci. Total Environ. 408, 109e116. https://doi.org/10.1016/ j.scitotenv.2009.09.026. Lough, G.C., Christensen, C.G., Schauer, J.J., Tortorelli, J., Mani, E., Lawson, D.R., Clark, N.N., Gabele, P.A., 2007. Development of molecular marker source profiles for emissions from on-road gasoline and diesel vehicle fleets. J. Air Waste Manag. Assoc. 57, 1190e1199. https://doi.org/10.3155/1047-3289.57.10.1190. Lough, G.C., Schauer, J.J., 2007. Sensitivity of source apportionment of urban particulate matter to uncertainty in motor vehicle emissions profiles. J. Air Waste Manag. Assoc. 57, 1200e1213. https://doi.org/10.3155/1047-3289.57.10.1200. Mohamed, M.F., Kang, D., Aneja, V.P., 2002. Volatile organic compounds in some urban locations in United States. Chemosphere 47, 863e882. https://doi.org/ 10.1016/S0045-6535(02)00107-8. Niu, Y., He, L., Hu, M., Zhang, J., Zhao, Y., 2006. Pollution characteristics of

I.M. Al-Naiema et al. / Environmental Pollution 240 (2018) 34e43 atmospheric fine particles and their secondary components in the atmosphere of Shenzhen in summer and in winter. Science in China Series B-Chemistry 49, 466e474. https://doi.org/10.1007/s11426-006-2010-0. Olsson, I., 1970. The use of oxalic acid as a standard. In: Wiley, Proceedings of the 12th Nobel Symposium. Uppsala University, New York. Rogge, W.F., Hildemann, L.M., Mazurek, M.A., Cass, G.R., Simoneit, B.R., 1993. Sources of fine organic aerosol. IV: particulate abrasion products from leaf surfaces of urban plants. ES T (Environ. Sci. Technol.) 27, 2700e2711. https://doi.org/ 10.1021/es00049a008. Schauer, J.J., Kleeman, M.J., Cass, G.R., Simoneit, B.R., 2002. Measurement of emissions from air pollution sources. 5. C1 C32 organic compounds from gasolinepowered motor vehicles. ES T (Environ. Sci. Technol.) 36, 1169e1180. https:// doi.org/10.1021/es0108077. Schauer, J.J., Rogge, W.F., Hildemann, L.M., Mazurek, M.A., Cass, G.R., Simoneit, B.R.T., 1996. Source apportionment of airborne particulate matter using organic compounds as tracers. Atmos. Environ. 30, 3837e3855. https://doi.org/10.1016/ 1352-2310(96)00085-4. Sheesley, R.J., Nallathamby, P.D., Surratt, J.D., Lee, A., Lewandowski, M., Offenberg, J.H., Jaoui, M., Kleindienst, T.E., 2017. Constraints on primary and secondary particulate carbon sources using chemical tracer and 14C methods during CalNex-Bakersfield. Atmos. Environ. 166, 204e214. https://doi.org/ 10.1016/j.atmosenv.2017.07.025. Simoneit, B.R., 1989. Organic matter of the tropospheredV: application of molecular marker analysis to biogenic emissions into the troposphere for source reconciliations. J. Atmos. Chem. 8, 251e275. https://doi.org/10.1007/ BF00051497. Simoneit, B.R.T., Schauer, J.J., Nolte, C.G., Oros, D.R., Elias, V.O., Fraser, M.P., Rogge, W.F., Cass, G.R., 1999. Levoglucosan, a tracer for cellulose in biomass burning and atmospheric particles. Atmos. Environ. 33, 173e182. https:// doi.org/10.1016/s1352-2310(98)00145-9. Stone, E.A., Hedman, C.J., Zhou, J., Mieritz, M., Schauer, J.J., 2010. Insights into the nature of secondary organic aerosol in Mexico City during the MILAGRO experiment 2006. Atmos. Environ. 44, 312e319. https://doi.org/10.1016/ j.atmosenv.2009.10.036. Stone, E.A., Nguyen, T.T., Pradhan, B.B., Dangol, P.M., 2012. Assessment of biogenic secondary organic aerosol in the Himalayas. Environ. Chem. 9, 263e272. https://doi.org/10.1071/en12002. Surratt, J.D., Chan, A.W.H., Eddingsaas, N.C., Chan, M.N., Loza, C.L., Kwan, A.J., Hersey, S.P., Flagan, R.C., Wennberg, P.O., Seinfeld, J.H., 2010. Reactive intermediates revealed in secondary organic aerosol formation from isoprene. Proc. Natl. Acad. Sci. U.S.A. 107, 6640e6645. https://doi.org/10.1073/ pnas.0911114107. Szidat, S., Jenk, T.M., G€ aggeler, H.W., Synal, H.A., Fisseha, R., Baltensperger, U., Kalberer, M., Samburova, V., Reimann, S., Kasper-Giebl, A., Hajdas, I., 2004. Radiocarbon (14C)-deduced biogenic and anthropogenic contributions to organic carbon (OC) of urban aerosols from Zürich. Switzerland. Atmospheric Environment 38, 4035e4044. https://doi.org/10.1016/j.atmosenv.2004.03.066. Tan, J.-H., Duan, J.-C., Chen, D.-H., Wang, X.-H., Guo, S.-J., Bi, X.-H., Sheng, G.-Y., He, K.-B., Fu, J.-M., 2009. Chemical characteristics of haze during summer and winter in Guangzhou. Atmos. Res. 94, 238e245. https://doi.org/10.1016/ j.atmosres.2009.05.016. von Schneidemesser, E., Monks, P.S., Plass-Duelmer, C., 2010. Global comparison of VOC and CO observations in urban areas. Atmos. Environ. 44, 5053e5064. https://doi.org/10.1016/j.atmosenv.2010.09.010. Wang, J.Z., Ho, S.S.H., Ma, S.X., Cao, J.J., Dai, W.T., Liu, S.X., Shen, Z.X., Huang, R.J., Wang, G.H., Han, Y.M., 2016. Characterization of PM2.5 in Guangzhou, China: uses of organic markers for supporting source apportionment. Sci. Total Environ. 550, 961e971. https://doi.org/10.1016/j.scitotenv.2016.01.138. Wang, S., Wu, D., Wang, X.-M., Fung, J.C.-H., Yu, J.Z., 2013. Relative contributions of secondary organic aerosol formation from toluene, xylenes, isoprene, and monoterpenes in Hong Kong and Guangzhou in the Pearl River Delta, China: an emission-based box modeling study. Journal of Geophysical Research-

43

Atmospheres 118, 507e519. https://doi.org/10.1029/2012jd017985. Wang, S.X., Zhao, M., Xing, J., Wu, Y., Zhou, Y., Lei, Y., He, K.B., Fu, L.X., Hao, J.M., 2010. Quantifying the air pollutants emission reduction during the 2008 Olympic Games in Beijing. ES T (Environ. Sci. Technol.) 44, 2490e2496. https://doi.org/ 10.1021/es9028167. Wang, W.T., Jariyasopit, N., Schrlau, J., Jia, Y.L., Tao, S., Yu, T.W., Dashwood, R.H., Zhang, W., Wang, X.J., Simonich, S.L.M., 2011. Concentration and photochemistry of PAHs, NPAHs, and OPAHs and toxicity of PM2.5 during the Beijing Olympic games. ES T (Environ. Sci. Technol.) 45, 6887e6895. https://doi.org/ 10.1021/es201443z. Wang, Y.Q., Zhang, Y.X., Zhang, Y., Li, Z.Q., He, L.Y., Huang, X.F., 2014. Characterization of carbonaceous aerosols during and post-Shenzhen UNIVERSIADE period. China Environ. Sci. 34, 1972e1978. Watson, J.G., Cooper, J.A., Huntzicker, J.J., 1984. The effective variance weighting for least squares calculations applied to the mass balance receptor model. Atmos. Environ. 18, 1347e1355. https://doi.org/10.1016/0004-6981(84)90043-X. Widory, D., 2006. Combustibles, fuels and their combustion products: a view through carbon isotopes. Combust. Theor. Model. 10, 831e841. https://doi.org/ 10.1080/13647830600720264. Xu, L., Guo, H.Y., Boyd, C.M., Klein, M., Bougiatioti, A., Cerully, K.M., Hite, J.R., Isaacman-VanWertz, G., Kreisberg, N.M., Knote, C., Olson, K., Koss, A., Goldstein, A.H., Hering, S.V., de Gouw, J., Baumann, K., Lee, S.H., Nenes, A., Weber, R.J., Ng, N.L., 2015. Effects of anthropogenic emissions on aerosol formation from isoprene and monoterpenes in the southeastern United States. Proc. Natl. Acad. Sci. U.S.A. 112, 37e42. https://doi.org/10.1073/pnas.1417609112. Yuan, Z.B., Yu, J.Z., Lau, A.K.H., Louie, P.K.K., Fung, J.C.H., 2006. Application of positive matrix factorization in estimating aerosol secondary organic carbon in Hong Kong and its relationship with secondary sulfate. Atmos. Chem. Phys. 6, 25e34. https://doi.org/10.5194/acp-6-25-2006. Zhang, Y.-L., Huang, R.-J., El Haddad, I., Ho, K.-F., Cao, J.-J., Han, Y., Zotter, P., Bozzetti, C., Daellenbach, K., Canonaco, F., 2015. Fossil vs. non-fossil sources of fine carbonaceous aerosols in four Chinese cities during the extreme winter haze episode of 2013. Atmos. Chem. Phys. 15, 1299e1312. https://doi.org/ 10.5194/acp-15-1299-2015. Zhang, Y.-L., Li, J., Zhang, G., Zotter, P., Huang, R.-J., Tang, J.-H., Wacker, L., Prevot, A.S., Szidat, S.n, 2014. Radiocarbon-based source apportionment of carbonaceous aerosols at a regional background site on Hainan Island, South China. ES T (Environ. Sci. Technol.) 48, 2651e2659. https://doi.org/10.1021/ es4050852. Zhang, Y.-X., Min, S., Zhang, Y.-H., Zeng, L.-M., He, L.-Y., Bin, Z., Wei, Y.-J., Zhu, X.-L., 2007. Source profiles of particulate organic matters emitted from cereal straw burnings. J. Environ. Sci. 19, 167e175. https://doi.org/10.1016/S1001-0742(07) 60027-8. Zhang, Y., Schauer, J.J., Zhang, Y., Zeng, L., Wei, Y., Liu, Y., Shao, M., 2008. Characteristics of particulate carbon emissions from real-world Chinese coal combustion. ES T (Environ. Sci. Technol.) 42, 5068e5073. https://doi.org/10.1021/ es7022576. Zheng, M., Cass, G.R., Schauer, J.J., Edgerton, E.S., 2002. Source apportionment of PM2.5 in the southeastern United States using solvent-extractable organic compounds as tracers. ES T (Environ. Sci. Technol.) 36, 2361e2371. https:// doi.org/10.1021/es011275x. Zheng, M., Salmon, L.G., Schauer, J.J., Zeng, L., Kiang, C.S., Zhang, Y., Cass, G.R., 2005. Seasonal trends in PM2.5 source contributions in Beijing, China. Atmos. Environ. 39, 3967e3976. https://doi.org/10.1016/j.atmosenv.2005.03.036. Zheng, M., Wang, F., Hagler, G.S.W., Hou, X.M., Bergin, M., Cheng, Y.A., Salmon, L.G., Schauer, J.J., Louie, P.K.K., Zeng, L.M., Zhang, Y.H., 2011. Sources of excess urban carbonaceous aerosol in the Pearl River Delta region, China. Atmos. Environ. 45, 1175e1182. https://doi.org/10.1016/j.atmosenv.2010.09.041. Zotter, P., El-Haddad, I., Zhang, Y., Hayes, P.L., Zhang, X., Lin, Y.H., Wacker, L., Schnelle-Kreis, J., Abbaszade, G., Zimmermann, R., 2014. Diurnal cycle of fossil and nonfossil carbon using radiocarbon analyses during CalNex. J. Geophys. Res.: Atmosphere 119, 6818e6835. https://doi.org/10.1002/2013JD021114.