Pollution Characteristics and Source Apportionment of PM 2 . 5-Bound n-Alkanes in the Yangtze River Delta , China

PM2.5-bound n-alkanes in Shanghai (SH), Nanjing (NJ) and Ningbo (NB) cities from November 2014 to August 2015 were investigated. Averaged concentrations of the total 25 n-alkanes (∑n-alkanes, C16–C40) in SH, NJ and NB were 97.4 ± 73.9, 83.8 ± 57.1 and 187.1 ± 87.1 ng m, respectively. Obvious spatial and seasonal variations were attributed to the differences of emission sources and meteorological conditions. Analysis of the diagnostic ratios and specific molecular markers of n-alkanes suggested that high plant wax and vehicle emissions were the major sources of n-alkanes in the YRD region. Strong inputs of microbial components in summer were found and attributed to the emission from plankton in the ocean. The annual average contributions of higher plant wax to n-alkanes (%wax) in SH, NJ and NB were estimated to be 47.5%, 50.1% and 34.5%, respectively. Anthropogenic sources were responsible for the n-alkanes in NB, while biogenic sources contributed much more n-alkanes in NJ and SH. Based on the result of backward analysis, the emissions of n-alkanes in NB and NJ were mainly from local sources when the air masses came from the sea and south China with low n-alkanes concentrations. When the air masses originated from north China, the transport of contaminant aggravated the pollution of n-alkanes in SH.


INTRODUCTION
With the acceleration of urbanization and rapid development of economy, severe air pollution in China has received an unprecedented attention in recent years (Huang et al., 2014;Jiang et al., 2015;Li et al., 2015;Lv et al., 2015;Han et al., 2016).Long-term exposure to air pollutants, including organic compounds, metals and pathogenic bacterium, could be associated with increased incidences of cardiorespiratory disease mortality and diminished life expectancy (Rusconi et al., 2011;Pope et al., 2013).Fine organic aerosols contributed a significant fraction of PM 2.5 and are composed of elemental carbon or black carbon and a mixture of large variety of organic species define as organic carbon (OC).OC directly affected air quality and solar radiation (Gustafsson et al., 2009).By acting as a cloud condensation nuclei, they also affect microphysical properties (Haywood and Boucher et al., 2000;Singh et al., 2016).OC is a large and significant fraction of atmospheric particular matter (Chow et al., 1994;Zheng et al., 1997).OC accounted for an average 32.4% of PM 2.5 concentration in Munbai city (Joseph et al., 2012).At two non-urban sites from a polluted region in Europe (Southern Poland), OC contributed 34.4% of PM 2.5 (Blaszczak et al., 2016).Recent study in the Pearl River Delta, China showed that the average contribution of OC to PM 2.5 was estimated to be 23.4% (Wang et al., 2016).
As a group of nonpolar organic compounds, n-alkanes are the primary pollutants and show relatively high concentrations, usually range from 0.1 to 2000 µg m −3 in the atmosphere (Chen et al., 2014;Ren et al., 2016).They mainly originate from both anthropogenic sources (e.g., fossil fuel combustion, petroleum residues and biomass burning) and biogenic sources (including higher plants waxes, bacteria, pollen and insects) (Simoneit et al., 1989;Rogge et al., 1993a).Due to their stability characteristics, n-alkanes could provide insight into the origins and longrange transport of aerosols (Schauer et al., 1999;Zheng et al., 2002).Perrone et al. (2014) found that C21-C22 n-alkanes were predominantly emitted from diesel engine exhausts.Gas/particle distribution of n-alkanes indicated that light n-alkanes mostly presented in the gas phase and heavy n-alkanes were absorbed in the particle phase (Xie et al., 2014).In addition, based on diagnostic ratios and specific molecular markers, the contributions of combustion of coal and wood in residential heating and traffic emissions to PM 2.5 -bound n-alkanes in Ostrava were identified (Mikuška et al., 2015).Although distributions of n-alkanes were investigated in some coastal and inland cities in China, sources of PM 2.5 -bound n-alkanes remain unclear (Duan et al., 2010;Li et al., 2010;Cao et al., 2013;Wang et al., 2016a).Moreover, severe air pollution frequently occurred in the cluster of cities because of the effects of local and regional air pollutants.Aliphatic hydrocarbons (n-alkanes, n-fatty acids and n-alcohols) contributed an important fraction of ambient aerosols, ranging from 20 to 80% (Chrysikou et al., 2009).N-alkanes accounted for 0.7% of the OC in Sanya during the winter season (Wang et al., 2015).Study in Shanghai showed that n-alkanes contributed 0.47% of the OC (Cao et al., 2013).Although n-alkanes comprise of a minor fraction of the OC, they are significant secondary organic aerosol (SOA) precursor (Aumont et al., 2012).n-alkanes with more than 16-carbon number are estimated to have a health impact on the skin, and even engender skin cancer (Cheng et al., 1999).
The Yangtze River Delta (YRD) is located in the eastern part of China, including Shanghai city, Zhejiang and Jiangsu province.The YRD is one of the most developed regions, which accounts for 2.2% of geographical area and 18.8% of Gross Domestic Product (GDP) of China (http://www.ndrc.gov.cn/zcfb/zcfbtz/201606/t20160603_806388.html).In recent years, haze occurred more frequently in the YRD region (Zhang et al., 2015;Fu et al., 2016;Sun et al., 2016;Zhao et al., 2016).Most studies focused on the formation of new particles and sources of PM 2.5 in the YRD region.However, limited information about the pollution characteristics of n-alkanes was reported (Wang et al., 2005(Wang et al., , 2016a;;Cao et al., 2013).In this paper, observation campaigns were simultaneously conducted in three core cities of the YRD region: Shanghai (SH), Nanjing (NJ) and Ningbo (NB).Seasonal and spatial distributions of total 25 PM 2.5 -bound n-alkanes were studied.Potential sources were identified by diagnostic parameters like carbon preference index (CPI), carbon number (C max ), and the plant wax alkane (WaxCn), by which the contribution between anthropogenic and biogenic sources could be estimated (Simoneit et al., 1991;Kavouras et al., 1999).Backward trajectory analysis was also performed to evaluate the influences of local and regional sources in the YRD region.The related results could enhance the understanding of pollution characteristics and sources of PM 2.5 -bound n-alkanes at a regional scale during severe air pollution.

Sampling Site
Three core cities (SH, NJ and NB) in the YRD region were chosen for measuring PM 2.5 -bound n-alkanes (Fig. 1).Sampling sites in SH and NJ were located at the Shanghai Academy of Environmental Science (SAES) and Nanjing University, respectively, close to commercial and residential areas.In addition, a suburban site of NB was set in the Fig. 1.Location of sampling sites in the Yangtze River Delta.Hong et al., Aerosol and Air Quality Research, 17: 1985-1998, 20171987 University of Nottingham (Ningbo China).All the samplers were installed on the rooftop of the building, with the height of 18-21 m above the ground.

N-Alkanes Analysis
Quartz filter was cut into pieces, and extracted by accelerated solvent extraction (ASE, Dionex 350) with dichloromethane (DCM)/methanol (2:1, V/V).Prior to the extraction, n-Tetradecane-d30, n-Tricosane-d48 and n-Hexatriaconta-d74 were spiked into the filters as surrogate.The samples were extracted three times under 1500 psi pressure at 120°C.The extracts of each sample filters were combined, filtered, and concentrated to 3 mL by rotary evaporator (RE52-AA, Shanghai yarong biochemistry instrument factory), and then condensed to 1 mL under a stream of high-purity nitrogen.The concentrated solutions were stored in 2-6°C until analysis.
Total 28 n-alkanes were analyzed by a gas chromatography/mass spectrometer detector (GC/MS, Agilent 7890A/5975C) in a selected ion mode (SIM) with a 30 m HP-5 MS capillary column (i.d.0.25 mm, 0.25 mm thickness).Prior to the analysis, n-Tetracosane-d50 was spiked into the solution as internal standard.The sample was injected 1 µL in splitless mode at 290°C and high purity helium (99.999%) was used as the carrier gas at a stable flow of 1.2 mL min -1 .The GC temperature was initiated at 60°C (held for 10 min) and increased to 300°C at 5 °C min -1 , then held for 40 min.The electron impact energy was set at 70eV with the ion source temperature of 290°C (Cao et al., 2013).

Quality Control and Quality Assurance
Quality control and quality assurance were performed according to China's national standards (HJ618-2011 Determination of atmospheric articles PM 10 and PM 2.5 in ambient air by gravimetric method; HJ194-2005 Manual methods for ambient air quality monitoring).Briefly, the aerosol sampler was checked and calibrated regularly during the sampling campaign.Flow rate was calibrated by intelligent flow calibrator.Ensure the integrity of quartz filters and only the samples collecting over 20 h were available.Due to the lack of samplers, there was no collocation sampling performed to collect the duplicate samples.
The detection limits (LOD) for individual n-alkanes ranged from 0.02 to 0.09 ng m -3 , concentrations of analytes below the detection limits were assumed equal to 1/2 of LOD.Field blanks were used to determine background contamination during the transportation and storage of samples.Method blanks (solvent) were also measured.None of the target compounds in method blanks was found.The concentrations of n-alkanes detected in the field blank samples were < 1.5% of those determined in the PM 2.5 samples.Based on six-point calibration curves for individual compounds, internal calibration was used to quantified the n-alkanes, and the correlation coefficient (R 2 ) was > 0.999.The recoveries of the surrogate standards ranged from 63.7% to 116%.The instrument response error and the operational error during the experiment could result in the recovery upper 100%.The reported concentrations were not surrogate-recovery corrected.

Seasonal and Spatial Variations of n-Alkanes Concentrations
The annual average concentrations of ∑n-alkanes in NB, NJ and SH were 184.2, 83.5 and 93.7 ng m -3 , with the ranges of 9. 45-807.2, 16.8-218.3 and 15.1-287.5 ng m -3 , respectively (Fig. 2).Seasonal patterns of PM 2.5 -bound nalkanes in NJ and SH were quite similar during the sampling periods.However, the ∑n-alkanes concentrations of NB were much higher than those of NJ and SH.Because the sampling site of NB was located in a suburban area, which might contribute more n-alkanes emitted from plant waxes.Meanwhile, the large petroleum chemical industry close to the monitoring site in NB had a great influence on the increase of n-alkanes.
The concentrations of PM 2.5 -bound n-alkanes in the YRD region were compared with domestic and foreign cities as shown in Table 1.In this study, the annual concentrations of n-alkanes were much lower than what have been reported from two megacities Beijing and Tianjin in China (Ren et al., 2016;Li et al., 2010), but more than two orders of magnitude above Tibet, a remote background area in southwest China (Chen et al., 2014).The concentrations of n-alkanes in Guangzhou (Wang et al., 2016) were lower than that in the YRD region.In this study, the concentrations of ∑n-alkanes in SH were higher than that measured in September 2009 (Cao et al., 2013), while the concentrations of ∑n-alkanes in NJ were much lower than those of previous studies (Wang and Kawamura, 2005).Compared to those foreign cities, pollution level of n-alkanes in the YRD region were lighter than that of Kabul in Afghanistan (Wingfors et al., 2011) and Augsburg in Germany (Schnelle-Kreis et al., 2005), but much heavier than that of Atlanta in USA (Yan et al., 2009).In winter, the concentrations of n-alkanes in Ostrava (Mikuška et al., 2015) were similar to those in the YRD region (Table 1).
Seasonal pattern of ∑n-alkanes in NB and SH was listed in the following order: winter > spring > autumn > summer, and those in NJ followed the ranking of winter > autumn > spring > summer (Fig. 2).For the three cities, the lowest concentrations of ∑n-alkanes appeared in summer (38.52 ng m -3 in NB, 25.65 ng m -3 in NJ and 26.3 ng m -3 in SH).The pattern was similar to that of fourteen cities in China investigated by Wang et al. (2006).The variations between them were probably ascribed to different sources and meteorological conditions.The concentrations of n-alkanes in the YRD region showed a negative correlation with temperature and relative humidity (Table 2), more precipitation in favored of the removal of particular matter in summer.Most of the n-alkanes were in the vapor phase during summer while most of them presented in the particulate phase in cold time because of the temperature change.In contrary, stable atmospheric structure, high inversion frequency and low mixing heights in winter would be disadvantageous to the diffusion of n-alkanes.The distributions of n-alkanes in New Delhi suggested that the Planetary Boundary Layer (PBL) was the most important factor among all the meteorological parameters considered (Yadav et al., 2013).

Carbon Number Distribution
The specific molecular markers of n-alkanes could provide some insight into the potential sources that contribute to aerosol pollution varying from city to city (Li et al., 2010).Higher plant is an important contributor to the n-alkanes with the long chain length (> C24) and predominance of odd carbon number, while anthropogenic sources, including fossil fuel burning and vehicular emission, usually consisted of low and medium chain length (< C24) (Rogge et al., 1993;Simoet, 1999).C max , the carbon number exhibiting the highest concentration, isusually used to distinguish biogenic from anthropogenic sources.Anthropogenic sources such as fossil fuels majorly emitted low C max alkanes, while the plants were generally with high C max alkanes.For example, Perrone et al. (2014) found that C21-C22 n-alkanes were predominantly emitted from diesel engine exhausts.
As shown in Fig. 3, the high concentrations presented in the range of C20 to C34 with a peak at C29 or C27 in all the three cities were obtained.They were characterized by the predominance of odd carbon number in the range of C27 to C33.In this study, n-alkanes from C25 to C34 remained high concentrations (Figs. 2, 3, 4), accounting for 67.1-80.6% of the ∑n-alkanes.The result for C27 to C33 suggested that biogenic sources do make some contributions to organic aerosol loadings in the YRD region (Simoneit, 2002).C29 was the C max for all the three sites in spring, fall and winter, while the C max of C27 was found in NB and SH in summer.The C max could be partly corroborated by the fact that season variations of biomass burning activities in the YRD region were observed, according to the NASA fire map.It was difficult to distinguish the n-alkanes from biomass burning or plant wax emissions.Considering agricultural straw burning in the fields was a common practice in the YRD region during harvest seasons (Wang et al., 2009), further research is needed to determine if biomass burning is an important contributor to PM 2.5 -bound n-alkanes in the YRD region.
The PM 2.5 -bound n-alkanes in three cities exhibited similar distributions of carbon number in all the seasons except for summer (Fig. 3).The distribution of n-alkanes in summer reflected the contributions of various emission sources from different cities.The concentrations of ∑n-alkanes in SH were higher than that in NJ during the spring season.However, in winter, NJ and SH had the same pollution level.NJ and SH showed no obvious difference of the C27-C40 n-alkanes concentration in fall, suggesting their similar intensity from higher plant emission.High concentrations of the C20-C26 n-alkanes in NJ were found in fall (Figs.3, 4), indicating the contributions of anthropogenic activities to the PM 2.5 -bound n-alkanes during this period.It was interesting that the opposite findings occurred in spring, the n-alkanes concentrations from C26 to C33 in NJ were lower than that in SH.The results could be also explained by the seasonal variation of meteorological conditions, which could not change the relative abundances of n-alkanes in the atmosphere (Wang and Kawamura, 2005).
The concentrations of n-alkanes from C20 to C25 in NB were much higher than that in SH and NJ during the winter period (Fig. 3), which was attributed to the contribution of fossil combustion to PM 2.5 -bound n-alkanes (Rogge et al., 1993b;Simoneit, 1999).The monitoring site in NB located at the suburban area, emission from surrounding industrial resulted in high n-alkanes concentrations.
In summer, percentages of the n-alkanes less than C20 in NB, NJ and SH accounted for 6.1, 7.6 and 7.7% of the PM 2.5bound n-alkanes, respectively (Fig. 4).They were estimated to be twice of that in other seasons, indicating a stronger input of microbial components to PM 2.5 .No obvious difference of the n-alkanes distributions among all the cities was found, suggesting the influence of regional sources.According to the previous study, the n-alkanes less than C20 mainly came from oceanic low plankton, including bacteria and algae (Li et al., 2010).Wind rose (Fig. 5) showed that southeast was the dominant wind direction during the summer time.The southeast wind might bring much n-alkane less than C20 emitted from plankton in the oceanto the monitoring site, which caused the high concentrations of low carbon n-alkanes in summer.
N-alkanes from C23 to C25 exhibited high concentrations in all the seasons indicated that gasoline was the dominant vehicle emissions sources in the YRD region (Mazureket al., 1989;Rogge et al., 1993b;Schnelle-Kreis et al., 2005).
(3) CPI in NB, NJ and SH ranged from 0.96 to 2.12, 1.22 to 2.12 and 1.20 to 2.68 with the means of 1.34, 1.62 and 1.58, respectively (Table 3).CPI value was adopted to discriminate between anthropogenic and biogenic influences.The CPI value of NB was lower than SH and NJ, which indicated a stronger contribution of fossil fuels burning in NB, while higher plants had more significant impacts on n-alkanes of NJ and SH.For all the three sites, the highest CPI value in spring was found, suggesting the impacts from biogenic activities, such as the changing of leaves, fungi and pollen spreading, and blooming during growing   1(31.8-48.3) 53.3(37.3-66.8) 45(31.1-57.3) 50.4(37-68.4) 47.5(31.1-68.4)season (Chen et al., 2014).The CPI values in this study were higher than other cities, such as Beijing, Tianjin, Guangzhou and Sanya (Li et al., 2010;Wang et al., 2015;Ren et al., 2016;Wang et al., 2016), and were close to the reported CPI value (1.59) in Tibet (Chen et al., 2014).CPI1 ranged from 0.72 to 2.43 in NB, 0.96 to 2.74 in NJ and 0.84 to 3.16 in SH with the means of 1.46, 1.59 and 1.43, respectively (Table 3).All the cities exhibited relative low CPI1 values in fall and winter, indicating the significant impacts of human activities.In contrary, the high CPI1 values in summer and spring showed the emission from plant waxes with some extent biomass burning.The annual CPI1 value in Tibet (1.32) was little lower than that in the YRD region (Chen et al., 2014).A study of n-alkanes in Tianjin showed that the CPI1 were close to 1 (Li et al., 2010).The average of CPI2 in NB, NJ and SH were 1.32, 1.65 and 1.50, respectively (Table 3).Consistent with the highest CPI in spring noted above, the highest CPI2 also occurred in spring, further supporting the conclusion that the largest impact from biogenic sources occurred during the growing season.The CPI2 value in the YRD region were close to the reported values (1.5 in winter and 1.1 in summer) in Guangzhou China (Wang et al., 2016b).
Therefore, seasonal variation of CPI (including CPI1, CPI2) values in three cities might provide the evidence for regional and specific sources of PM 2.5 -bound n-alkanes.Similar ranges of CPI values in SH and NJ indicated the similar sources of n-alkanes, originating from both anthropogenic and biogenic emissions, especially the influence from higher plants.Low CPI values in NB were attributed to the combustion of fossil fuel at a large scale.

Contribution of Plant Wax
The concentrations of wax n-alkanes (%waxCn) have been used to estimate the relative contributions of biogenic versus anthropogenic sources (Simoneit, 1985).The %waxCn (Simoneit et al., 1991) was calculated by the following equation: where Cn was the concentration of plant wax alkanes, negative values of Cn were taken as zero (Simoneit et al., 1991).Higher %waxCn implied a greater contribution from plants.
The annual average %waxCn in NB, NJ and SH were 34.5%, 50.1% and 47.5%, with the range of 19.4% to 57.1%, 39.3% to 68.7% and 31.1% to 68.4%, respectively (Table 3).Biogenic sources varied from different cities: the %waxCn of NB was much lower than that of NJ and SH, where biogenic sources contributed almost 50% of the n-alkanes.The results further support the analysis results of carbon distribution and CPI.Biogenic sources such as high plant waxes and anthropogenic sources (including vehicle emission, petroleum, combustion of coal and biomass) in NJ and SH had common impacts on the distribution of n-alkanes in the atmosphere.However, anthropogenic activities were the dominant sources contributing to the PM 2.5 -bound n-alkanes in NB.Ren et al. (2016) found the n-alkanes in Beijing were mainly from incomplete combustion of fossil fuels and petroleum residue and the %waxCn was about 17% in winter.The average %waxCn of Guangzhou in winter and summer were 17.3% and 12.0%, respectively (Wang et al., 2016b).They suggested that vehicular emission was the major source, and epicuticular wax was an important contributor of biogenic sources to the n-alkanes.In addition, in a remote area of Tibet, high concentration of n-alkanes occurred in winter when home heating was prevailed (Chen et al., 2014).Recent study in Bologna (northern Italy) showed that the average %waxCn in the surroundings of a municipal waste incinerator were 25.8% (Sarti et al., 2017).In this study, due to the sloughing of epicuticular waxes, biogenic source might take an ever more prominent role in large abundance of n-alkanes in spring (%waxCn up to 32.3%).The %waxCn of three cities in the YRD region were higher than most of the cities reported in Table 2.

Source Identification by PCA
As discussed above, diagnostic ratios and specific molecular markers could provide some information about the source identification of n-alkanes.However, further estimation is needed to determine the weights of each source.Principal component analysis (PCA, SPSS version 19.0 IBM corp. 2010) is widely used for data reduction and interpretation (Wang et al., 2015).Relationship between principal component and the chemical compounds is indicated by the factor loadings and then relates to the source composition.The main steps are as follows: (1) normalization of all n-alkanes concentrations; (2) analyzation of variables' factor scores from PCA; (3) regression.The PCA result was acceptable when the Kaiser-Meyer-Olkin (KMO) value was above 0.6 (Wang et al., 2015).And the KMO values of n-alkanes in three cities were all above 0.7, therefore the results of PCA in this study were reliable.
The results of PCA were listed in Table 4. Three components were yielded which account for 93.3, 90.5 and 89.6% of the total variance in the subset of variables.Factor 1 (PC1) indicated biogenic sources because of its significant correlation with long-chain n-alkanes (> C25).Factor 1 accounted for 51.5, 50.2 and 43.3% in NB, NJ and SH.The PCA results demonstrated the significant influence from biogenic source, which was consistent with the results of diagnostic ratios and specific molecular markers discussed above.Though it was difficult to distinguish the biomassburning n-alkanes from the nature plant wax derived nalkanes.Considering agricultural straw burning in the fields was a common practice in the YRD region during harvest seasons (Wang et al., 2009), biomass combustion might be an important source of n-alkanes in the YRD region.
Factor 2 (PC 2) contributed 26.7, 30.3 and 30.4% of total variance in NB, NJ and SH were interpreted as the sources of vehicular and coal burning due to its association with medium molecular weight n-alkanes (C17-C24).Along with the carbon distribution analysis, vehicular emissions played an important role in PM 2.5 -bound alkanes in the YRD region, and this was not consistent with the result of Beijing, which coal burning was considered as a major source (Duan et al., 2010).It should be noticed that only using PCA method could not distinguish the vehicle emission from coal burning derived alkanes.
Factor 3 (PC3) in this study was described as contribution of microbial component due to its strong correlation with low molecular weight n-alkanes (C13-C17).Factor 3, accounting for 15.1, 10 and 15.9% of the total variance in NB, NJ and SH, were associated with the influence from microbial component, especial the emission from plankton in the ocean.

Backward Trajectory Analysis
Based on their stable characteristics, n-alkanes in atmospheric particles could provide insight into the origin and long-term transmission of aerosols.In this study, the clustering analysis of backward trajectory was performed in the monitoring cities (NB, SH and NJ), starting at every 0:00 (UTC) in 72 h, 500 m altitude during the whole sampling period (Figs. 6(a), 6(b) and 6(c)).Back-trajectories in NB originated from Siberia (Cluster 1, 35%), west Yellow Sea (Cluster 2, 38%) and south China (Cluster 3, 28%).The air masses of NJ came from north (Cluster 1&2, 78%) and south China (Cluster 3, 22%).The air masses of SH originated mainly from north (Cluster 1&2, 44%), Yellow Sea (Cluster 3, 26%) and central China (Cluster 4&5, 31%).The trajectories were automatically clustered and then categorized manually according to various origins and different seasons.Cluster consistent of the trajectories from the same direction and they were related to the sampling time.The concentrations of n-alkanes under different air masses cluster were calculated as the average value of the corresponding sample.The seasonal distributions under different backward trajectory conditions (%) were also calculated according to the corresponding sampling time of the trajectories.Fig. 6(d) showed the average concentrations of n-alkanes for three cities under different air mass clusters.The concentrations of n-alkanes of NB showed the peak value (264.56 ng m -3 ) when the air masses originated from Cluster 2 during the winter and fall (Fig. 7).The air masses of Cluster 2 were mainly generated from west Yellow sea, representing clear marine masses with low n-alkanes concentrations.Thus, the n-alkanes of NB were mainly from local sources and diluted by the clear mass.Similar with NB, the n-alkanes of NJ appeared maximum value under the air masses from cluster 1 (south China).The highest concentrations of nalkanes in SH were observed when the air masses originated from Siberia and north China (cluster 1) during winter and fall (Fig. 7).The air mass with heavy contaminants transported from north China to SH, and caused the elevated concentrations of n-alkanes in the atmosphere.

CONCLUSIONS
PM 2.5 -bound n-alkanes were determined since November 2014 to August 2015 in three core cities in the YRD region.Obvious spatial and seasonal variations were discovered.The concentrations of n-alkanes in NB were much higher than that in NJ and SH, while all the cities exhibited the minimum concentration in summer.NJ and SH tended to have similar pollution intensity in all the seasons except for spring.
The carbon distribution of n-alkanes in the YRD region presented high concentration in the range of C20 to C34 with a peak at C29 or C27.Three distributions of n-alkanes varied from seasons and indicated different sources.The high concentrations of n-alkanes from C23 to C25 were observed and suggested that vehicle emission were major anthropogenic source of PM 2.5 -bound n-alkanes in the YRD region.The strong input of microbial components to organic aerosols in summer was found.
Diagnostic ratios and source specific molecular markers  of n-alkanes implied that high plant wax and anthropogenic activities (vehicle emission, combustion of coal and biomass) were major sources of n-alkanes in the YRD region.Biogenic sources contributed more n-alkanes in SH and NJ, while anthropogenic sources were responsible for n-alkanes in NB.Biomass burning should be an important contributor of n-alkanes in harvest time.
Clustering analysis of the backward trajectories was performed through different monitoring cities (NB, SH and NJ).The air masses coming from the sea and south China brought less pollutant and diluted the concentration of nalkanes in the YRD region, the emission of n-alkanes were mainly from local sources in NB and NJ.The transport of contamination aggravated the pollution of n-alkanes in SH when the air masses originated from north China.

Fig. 3 .
Fig. 3. Mass concentrations of PM 2.5 -bound alkanes in three cities during various seasons.

Fig. 6 .
Fig. 6.Clusters analysis of backward trajectory in a) NB, b) NJ and c) SH, starting at UTC0000 in 72 h, 500 m altitude during the whole sampling period.The concentrations of n-alkanes under different air masses cluster (d) in three cities in the YRD region were analyzed.

Table 1 .
Comparison of concentrations (ng m -3) and diagnostic parameters with other cities around the world.

Table 2 .
Meteorological data and its correlation with n-alkanes in the YRD region.
a WS: wind speed.b T: temperature.c RH: relative humidity.d Significant correlation at 0.01 level.

Table 3 .
Cmax, CPI and %wax values of PM 2.5 -bound alkanes in the Yangtze River Delta.

Table 4 .
PCA results of PM 2.5 -bound alkanes in the Yangtze River Delta.