Polycyclic Aromatic Hydrocarbons in PM 2 . 5 and PM 2 . 5 – 10 in Urumqi , China : Temporal Variations , Health Risk , and Sources

PM2.5 and PM2.5–10 samples were simultaneously collected in Urumqi from January to December 2011, and 14 priority polycyclic aromatic hydrocarbons (PAHs) were determined. The mean concentrations of total PAHs in PM2.5 and PM2.5–10 were 20.90~844.22 ng m−3 and 19.65~176.5 ng m−3 respectively, with the highest in winter and the lowest in summer. Above 80% of PAHs were enriched in PM2.5, which showed remarkable seasonal variations compared to coarse particles. High molecular weight (HMW) PAHs were predominant in PM2.5 (46.61~85.13%), whereas the proportions of lower molecular weight (LMW) and HMW PAHs in PM2.5–10 showed a decreasing and an increasing trend, respectively, from spring to winter. The estimated concentrations of benzo[a]pyrene equivalent carcinogenic potency (BaPeq) in PM2.5 (10.49~84.52 ng m−3) were higher than that of in PM2.5–10 (1.15~13.33 ng m−3) except in summer. The estimated value of inhalation cancer risk in PM2.5 and PM2.5–10 were 1.63 × 10−4~7.35 × 10−3 and 9.94 × 10−5~1.16 × 10−3, respectively, far exceeding the health-based guideline level of 10−4. Diagnostic ratios and positive matrix factorization results demonstrated that PAHs in PM2.5 and PM2.5–10 were from similar sources, such as coal combustion, biomass burning, coking, and petroleum combustion, respectively. Coal combustion was the most important source for PAHs both in PM2.5 and PM2.5–10, accounting for 54.20% and 50.29%, respectively.


Introduction
Polycyclic aromatic hydrocarbons (PAHs) are ubiquitous constituents of particulate matter in the atmosphere and well-known to be carcinogenic and/or mutagenic [1,2], and are emitted into the atmosphere from natural as well as anthropogenic sources.Natural sources include forest fires and volcanic eruptions, while anthropogenic sources include meat cooking, motor vehicles, road dust (including particles generated by tire wear and brake lining erosion), natural gas home appliances, tobacco smoke, asphalt, boilers, coal combustion, wood burning [3,4], solid waste incineration, and petroleum spills and discharge [5].
Urumqi, the capital of Xinjiang Uygur Autonomous Region of China, is in central Xinjiang, which is on the north foot of Tianshan Mountain and the south edge of Jungger Basin.It is located in the center of Asia, north of the Taklimakan desert and south of the Gurbantunggut desert (Figure 1).Three sides of it are surrounded by the Tianshan Mountain, of which the highest elevation is 5000 m.The particles are brought into the urban area from the Quasigeer basin through the north-south route [6,7].Rapid economic development has caused more and more serious air pollution problems in Urumqi, which is one of the most polluted cities during winter in China [8].During the six months of heating, the city was covered by the haze for at least one third of the time.Li et al. (2008) reported that the mean concentrations of PM 10 in the winter of 2006 in Urumqi was 305.6 µg m −3 with a maximum of 1048.5 µg m −3 , and the mean concentrations of PM 2.5 (particle diameter ≤2.5 µm) and TSP (Total Suspended Particulate, particle diameter ≤100 µm) in the winter of 2007 were 187 and 385 µg m −3 with a maximum of 487 and 1313 µg m −3 , respectively, far higher than the National Ambient Air Quality Standard of China [6].
Urumqi, the capital of Xinjiang Uygur Autonomous Region of China, is in central Xinjiang, which is on the north foot of Tianshan Mountain and the south edge of Jungger Basin.It is located in the center of Asia, north of the Taklimakan desert and south of the Gurbantunggut desert (Figure 1).Three sides of it are surrounded by the Tianshan Mountain, of which the highest elevation is 5000 m.The particles are brought into the urban area from the Quasigeer basin through the north-south route [6,7].Rapid economic development has caused more and more serious air pollution problems in Urumqi, which is one of the most polluted cities during winter in China [8].During the six months of heating, the city was covered by the haze for at least one third of the time.Li et al. (2008) reported that the mean concentrations of PM10 in the winter of 2006 in Urumqi was 305.6 µg m −3 with a maximum of 1048.5 µg m −3 , and the mean concentrations of PM2.5 (particle diameter ≤2.5 µm) and TSP (Total Suspended Particulate, particle diameter ≤100 µm) in the winter of 2007 were 187 and 385 µg m −3 with a maximum of 487 and 1313 µg m −3 , respectively, far higher than the National Ambient Air Quality Standard of China [6].Research on atmospheric PAHs in China is mainly focused on the Yangtze River Delta and the Pearl River Delta region [9][10][11][12][13][14].In contrast, much fewer studies (most published in Chinese) have been carried out in Urumqi [7,8,[15][16][17][18][19][20].Early, Peng et al. (2006) analyzed the sources of PAHs in TSP and in PM10 in Urumqi with the composition of carbon isotope [20].Diao et al. (2005) analyzed PAHs using a GC/MS technique in Urumqi Xinshi district from December 2004 to January 2005 (winter) and reported that the major source of PAHs in Urumqi Xinshi district was coal combustion and vehicle exhaust [15].Lately, Limu et al. (2013) analyzed the PAHs in PM2.5 and PM2.5-10 in Urumqi's southern urban area from 26 September 2010 to 3 March 2011 (autumn and wintertime) [7].The results showed that the ΣPAHs (sum of 15 PAHs) ranged from 0.11 to 1058.08 ng m −3 in PM2.5 and 0.01 to 90.89 ng m −3 in PM2.5-10, respectively, and 90% of the ΣPAHs existed in PM2.5.The above studies were based on the winter or autumn seasons, and the source analysis of PAHs in Urumqi was limited to the ratio and the principal component analysis methods.Although Ren et al. (2017) reported particle size distribution of PAHs in the atmosphere of Urumqi and analyzed their source using the Positive Matrix Factorization (PMF) model, it was also limited to the heating season [8].Urumqi belongs to typical temperate semiarid continental climate, with a large temperature difference between winter and summer.Winter is long and cold, and the temperature is stable.It is windy in spring and autumn, and the precipitation is concentrated in summer, which has the effect of washing and purifying pollutants in the atmosphere.The pollution characteristics of PAHs in different seasons may vary greatly.However, to date, there are few detailed studies about the pollution levels, particle size distributions, temporal variations, health risk, identification, and apportionment of the specific pollution sources of PAHs Research on atmospheric PAHs in China is mainly focused on the Yangtze River Delta and the Pearl River Delta region [9][10][11][12][13][14].In contrast, much fewer studies (most published in Chinese) have been carried out in Urumqi [7,8,[15][16][17][18][19][20].Early, Peng et al. (2006) analyzed the sources of PAHs in TSP and in PM 10 in Urumqi with the composition of carbon isotope [20].Diao et al. (2005) analyzed PAHs using a GC/MS technique in Urumqi Xinshi district from December 2004 to January 2005 (winter) and reported that the major source of PAHs in Urumqi Xinshi district was coal combustion and vehicle exhaust [15].Lately, Limu et al. (2013) analyzed the PAHs in PM 2.5 and PM 2.5-10 in Urumqi's southern urban area from 26 September 2010 to 3 March 2011 (autumn and wintertime) [7].The results showed that the ΣPAHs (sum of 15 PAHs) ranged from 0.11 to 1058.08 ng m −3 in PM 2.5 and 0.01 to 90.89 ng m −3 in PM 2.5-10 , respectively, and 90% of the ΣPAHs existed in PM 2.5 .The above studies were based on the winter or autumn seasons, and the source analysis of PAHs in Urumqi was limited to the ratio and the principal component analysis methods.Although Ren et al. (2017) reported particle size distribution of PAHs in the atmosphere of Urumqi and analyzed their source using the Positive Matrix Factorization (PMF) model, it was also limited to the heating season [8].Urumqi belongs to typical temperate semiarid continental climate, with a large temperature difference between winter and summer.Winter is long and cold, and the temperature is stable.It is windy in spring and autumn, and the precipitation is concentrated in summer, which has the effect of washing and purifying pollutants in the atmosphere.The pollution characteristics of PAHs in different seasons may vary greatly.However, to date, there are few detailed studies about the pollution levels, particle size distributions, temporal variations, health risk, identification, and apportionment of the specific pollution sources of PAHs using PMF model in PM 2.5 and PM 2.5-10 in Urumqi throughout the year.Our study sought to fill this knowledge gap.

Sampling
Samples were collected at the rooftop (9 m above the ground) of the Xinjiang branch of Chinese academy of sciences (N43 • 51 51", E87 • 34 9") in Xinshi district in Urumqi.This point is 500 m away from the main road (Beijing road).Xinshi district, located in the northwest of Urumqi city with a total area of 143 km 2 and a total population of 730,307 (2010), is the National Technology Industry Development Zone.It is also the transportation center to the north Xinjiang, representing an area of mixed residential, commercial, and industrial activities.
On five rainless days each month from January to December 2011, a middle-volume sampler (NL20; Tokyo Dylec Co, Tokyo, Japan) equipped with quartz fiber filter (Whatman, Mainstone, UK) was used to collect 24-h PM 2.5 and PM 2.5-10 particles simultaneously at a flow rate of 20 L/min.The filters were wrapped within aluminum foil and baked for 4 h at 450 • C before sampling.After sampling, samples were also wrapped in prebaked aluminum foil, sealed with clean Teflon bags and stored at −20 • C until analysis [7].

Pretreatment and Analysis of Samples
Extraction and analysis of PAHs was performed according to a previously published protocol [10].Before solvent extraction, deuterium PAH standards (Naphthalene-d 8 , Acenaphthene-d 10 , Phenanthrene-d 10 , Chrysene-d 12 , and Perylene-d 12 ), were added to the samples as surrogates.Each sample was ultrasonically extracted three times with 30 mL dichloromethane for 30 min each time.Then, the extracts were combined and concentrated to 1~2 mL by a rotary evaporator.Solvent was exchanged to n-hexane purified using a 1:2 alumina/silica column chromatography.Two fractions were eluted.The first fraction containing nonpolar compounds was eluted by 30 mL of hexane.The second fraction was eluted by 70 mL of DCM/hexane (3:7 v/v).The two fractions were combined, concentrated to ~1.0 mL by a rotary evaporator, and blown down to 0.3 mL under a gentle nitrogen stream.

GC/MS Analysis
All samples were analyzed by a gas chromatography-mass spectrometer (GC-MS) (Agilent 6890-5973N, Santa Clara, CA, USA) equipped with a HP-5MS elastic quartz capillary column (30 m × 0.25 mm × 0.25 mm).Injection of 1 µL samples was conducted with an automatic sampler in the splitless mode.
The chromatographic condition was as follows: carrier gas was helium with a purity of 99.9%; Inlet and transfer line temperatures were 290 • C; the oven temperature program was initiated at 65 • C (held for 2 min) and increased to 290 • C at 5 • C/min (held for 20 min).Mass spectrometric conditions were: electron impact (EI) selective ion monitoring mode, electron impact energy 70 ev [21]; multiplier voltage, 1624 V; scanning range (M/Z), 50~500 amu (unit of molecular mass).

Quality Control and Quality Assurance (QA/QC)
All samples were subjected to strict quality control and assurance, according to U.S. EPA-610 method [7].Limits of detection (LOD) for each compound were calculated as three times the signal-to-noise ratio, which was taken as the standard deviation of the lowest level standard [22].In this work, the LOD of PAHs was in the range of 0.02~0.12ng m −3 .The recovery experiment was done by spiking the standard solution onto blank filters (n = 3).Amounts of target compounds in the standard solution are similar to those in real samples.After being evaporated to dryness, the spiked filters were analyzed in a manner the same as the real samples [22].The recoveries of deuterium polycyclic aromatic hydrocarbons in the samples were Naphthalene-d 8 32% ± 9%, Phenanthrene-d 10 38% ± 10%, Acenaphthene-d 10 42% ± 8%, Chrysene-d 12 88% ± 11%, and Perylene-d 12 69% ± 9%.Field and laboratory blanks were also extracted and analyzed in the same way as the field samples.Target compounds were very low (0~0.05ng m −3 ) and in most cases not detectable in the blanks.Data reported here are all corrected for the blanks, but not corrected for the recoveries.

Source Apportionment with Positive Matrix Factorization (PMF)
Positive Matrix Factorization (PMF), a receptor-based source apportionment model, is based on a least squares method.In this study, the USEPA PMF 5.0 model was applied to identify the sources of observed ambient PAHs.The principles, detailed concepts, and applications of the PMF model for source apportionment can be found in previous studies [23,24].In our study, we have tested 3~7 factors and discovered that three-factor solution produced a good fit to the data and was the most stable and interpretable.Additional factors made negligible contributions to PAHs and caused factor splitting for tracers.Therefore, more than three factors did not produce meaningful results compared with that of three factors.After the initial run, the Q values were stable, and the Q values in the robust mode were approximately equal to the degrees of freedom [25].All of the scaled residuals were between −3 and 3.There was a good correlation between measured and predicted PAH concentrations, indicating that most of the species were simulated well.

Seasonal Variations of PAHs in Aerosols
Table 1 presents seasonal variation of mean concentrations of Σ 14 PAHs in PM 2.5 and PM 2.5-10 , whose values were 134.05 ng m −3 and 67.59 ng m −3 in spring, 20.90 ng m −3 and 19.65 ng m −3 in summer, 214.30ng m −3 and 24.48 ng m −3 in autumn, and 844.2 ng m −3 and 176.5 ng m −3 in winter, respectively.The result suggested that the distribution of PAHs in particulate matter was significantly associated with seasonal variation.Namely, whether in PM 2.5 or in PM 2.5-10 , the mean concentration of total PAHs was the lowest in summer and the highest in winter.This could be explained by adverse weather conditions for pollutant dispersion in cold season [31].In PM 2.5 , the concentrations of total PAHs in autumn were higher than in spring, while an opposite trend was found in PM 2.5-10 .In addition, the 2.5 /( 2.5 + 2.5-10 ) value of total PAHs decreased from 80% over in autumn and winter with colder temperature to 51.57% in summer.This behavior may be explained by higher incidence of resuspension and abrasion processes in summer [32].The result was in agreement with Manoli et al. (2002), showing that the 2.5 /( 2.5 + 2.5-10 ) value of total PAHs during cold periods was 96.1~98.4%, and in relatively warm periods slightly decreased to 92.2~97.8% in Thessaloniki [33].However, Guo et al. (2003) observed that the proportion of PM 2.5 -bound total PAHs in PM 10 was 72% in winter, and it was increased to 79% in summer in Hong Kong [34].
According to different molecular weight, 14 kinds of PAHs were classified into lower molecular weight PAHs (LMW, 2-and 3-rings PAHs) including Phe and Ant, middle molecular weight PAHs (MMW, 4-rings PAHs) including Flu, Pyr, BaA, Chr, and higher molecular weight PAHs (HMW, ≥5 rings of PAHs) including BbF, BkF, BeP, BaP, IcdP, BghiP, DahA, and Cor [32]. Figure 2 shows that in PM 2.5 , the proportion of HMW PAHs was dominant in all season, accounting for 46.61~85.13%.Especially in the autumn, the proportion reached 85.13%, while the MMW PAHs and LMW PAHs proportions reached the minimum value, 9.82% and 5.04%, respectively.The highest proportion of MMW PAHs and LMW PAHs were found in winter and summer, 30.69% and 38.02%, respectively.In PM 2.5-10 , the proportions of middle molecular weight PAHs did not exhibit distinct seasonality, whereas the proportions of lower molecular weight and HMW PAHs showed decreasing and increased trends, respectively, from spring to winter.

Seasonal Variations of PAHs in Aerosols
Table 1 presents seasonal variation of mean concentrations of Σ14PAHs in PM2.5 and PM2.5-10, whose values were 134.05 ng m −3 and 67.59 ng m −3 in spring, 20.90 ng m −3 and 19.65 ng m −3 in summer, 214.30ng m −3 and 24.48 ng m −3 in autumn, and 844.2 ng m −3 and 176.5 ng m −3 in winter, respectively.The result suggested that the distribution of PAHs in particulate matter was significantly associated with seasonal variation.Namely, whether in PM2.5 or in PM2.5-10, the mean concentration of total PAHs was the lowest in summer and the highest in winter.This could be explained by adverse weather conditions for pollutant dispersion in cold season [31].In PM2.5, the concentrations of total PAHs in autumn were higher than in spring, while an opposite trend was found in PM2.5-10.In addition, the ρ2.5/(ρ2.5 + ρ2.5-10) value of total PAHs decreased from 80% over in autumn and winter with colder temperature to 51.57% in summer.This behavior may be explained by higher incidence of resuspension and abrasion processes in summer [32].The result was in agreement with Manoli et al. (2002), showing that the ρ2.5/(ρ2.5 + ρ2.5-10) value of total PAHs during cold periods was 96.1~98.4%, and in relatively warm periods slightly decreased to 92.2~97.8% in Thessaloniki [33].However, Guo et al. (2003) observed that the proportion of PM2.5-bound total PAHs in PM10 was 72% in winter, and it was increased to 79% in summer in Hong Kong [34].
According to different molecular weight, 14 kinds of PAHs were classified into lower molecular weight PAHs (LMW, 2-and 3-rings PAHs) including Phe and Ant, middle molecular weight PAHs (MMW, 4-rings PAHs) including Flu, Pyr, BaA, Chr, and higher molecular weight PAHs (HMW, ≥5 rings of PAHs) including BbF, BkF, BeP, BaP, IcdP, BghiP, DahA, and Cor [32]. Figure 2 shows that in PM2.5, the proportion of HMW PAHs was dominant in all season, accounting for 46.61~85.13%.Especially in the autumn, the proportion reached 85.13%, while the MMW PAHs and LMW PAHs proportions reached the minimum value, 9.82% and 5.04%, respectively.The highest proportion of MMW PAHs and LMW PAHs were found in winter and summer, 30.69% and 38.02%, respectively.In PM2.5-10, the proportions of middle molecular weight PAHs did not exhibit distinct seasonality, whereas the proportions of lower molecular weight and HMW PAHs showed decreasing and increased trends, respectively, from spring to winter.Figure 3 illustrates that the ratios of individual PAH concentrations in winter to summer (w/s) in fine particles had a wider range (from 6.61 (Phe) to 89.48 (BaA)) than that in the coarse particles (from 4.45 (Phe) to 18.75 (BkF)).This result was similar to that in Liaoning (6.5~125.8 and 1.7~37.6,respectively) and Turkish Zonguldak (7.0~70.7 and 1.5~12.0,respectively) [26].In PM2.5 and PM2.5-10, the w/s values of BbF and BkF, which mainly existed in the particle phase, were higher than those of Phe and Ant with higher volatility, indicating different source contributions to PAHs between summer Figure 3 illustrates that the ratios of individual PAH concentrations in winter to summer (w/s) in fine particles had a wider range (from 6.61 (Phe) to 89.48 (BaA)) than that in the coarse particles (from 4.45 (Phe) to 18.75 (BkF)).This result was similar to that in Liaoning (6.5~125.8 and 1.7~37.6,respectively) and Turkish Zonguldak (7.0~70.7 and 1.5~12.0,respectively) [26].In PM 2.5 and PM 2.5-10 , the w/s values of BbF and BkF, which mainly existed in the particle phase, were higher than those of Phe and Ant with higher volatility, indicating different source contributions to PAHs between summer and winter.Except Ant, the w/s values of PAHs in fine particles were higher than in coarse particles, suggesting that PAHs might be transported from coarse particles to fine particles in winter [26].In addition, the 2.5 /( 2.5 + 2.5-10 ) values of Flu~Cor in summer were lower than in other season, and the 2.5 /( 2.5 + 2.5-10 ) value of 4-ring PAHs in winter were greater than in other season (Table 1).
Atmosphere 2018, 9, x 7 of 13 and winter.Except Ant, the w/s values of PAHs in fine particles were higher than in coarse particles, suggesting that PAHs might be transported from coarse particles to fine particles in winter [26].In addition, the ρ2.5/(ρ2.5 + ρ2.5-10) values of Flu~Cor in summer were lower than in other season, and the ρ2.5/(ρ2.5 + ρ2.5-10) value of 4-ring PAHs in winter were greater than in other season (Table 1).

Risk Assessment of PAHs Exposure
BaP has been widely used as an indicator of total PAHs in the inhalation risk assessment due to its high carcinogenicity.The mean concentration of BaP in PM2.5 was 6.82 ng m −3 , far exceeding the standard (NAAQSS GB 3095-2012, 1.0 ng m −3 ), and its concentration in PM2.5-10 was 0.78 ng m −3 .In order to further explain the effects of PAHs exposure on human during sampling, the toxicity of PAHs was evaluated by BaP toxicity equivalent concentration (BaPeq) [23,35]: The toxic equivalent factor (TEFi) values were taken from Nisbet and LaGoy (1992) [36] and are shown in Table 2.

Risk Assessment of PAHs Exposure
BaP has been widely used as an indicator of total PAHs in the inhalation risk assessment due to its high carcinogenicity.The mean concentration of BaP in PM 2.5 was 6.82 ng m −3 , far exceeding the standard (NAAQSS GB 3095-2012, 1.0 ng m −3 ), and its concentration in PM 2.5-10 was 0.78 ng m −3 .In order to further explain the effects of PAHs exposure on human during sampling, the toxicity of PAHs was evaluated by BaP toxicity equivalent concentration (BaP eq ) [23,35]: The toxic equivalent factor (TEF i ) values were taken from Nisbet and LaGoy (1992) [36] and are shown in Table 2.

Diagnostic Ratios (DR) Analysis
Some of the identified components and ratios in PAHs can reflect part of the characteristics of the emission sources in the study area.The DR method was commonly utilized for identifying PAH sources by comparing ratio of pairs of frequently found PAH emissions [44].The Flu/Pyr and IcdP/BghiP were isomer pairs.Therefore, they were photolytically degraded at comparable rates [45,46].Thus, their ratios preserved the original compositional information during atmospheric transport.Yunker et al. (2002) summarized the results of different sources and proposed that for Flu/(Pyr + Flu), the petroleum boundary ratio appears close to 0.40, and the ratio between 0.40 and 0.50 was characteristic of petroleum combustion (liquid fossil fuel combustion), whereas a ratio >0.50 was characteristic of grass, wood, or coal combustion; for IcdP/(BghiP + IcdP), a ratio <0.20 indicates unburned petroleum (petrogenic source), 0.20~0.50implies petroleum combustion (liquid fossil fuel combustion), and >0.50 was implies grass, wood, or coal combustion [47].
Figure 4 shows that the variation ranges of Flu/(Flu + Pyr) in PM 2.5 and PM 2.5-10 were 0.49~0.76and 0.43~0.81,respectively.The variation ranges of IcdP/(IcdP + BghiP) in PM 2.5 and PM 2.5-10 were 0.51~0.68 and 0.46~0.66,respectively.A large number of data were mainly distributed in the grass/wood/coal combustion area, and only some of the data were distributed in petroleum combustion, which accounted for a total of 12.77%.The result indicated that the main source of PAHs in Urumqi was grass/wood/coal combustion, and the minor source was petroleum combustion.The information about the specific source was obtained by the ratio method, but it was unable to estimate the contribution from the specific source categories.However, such information can be obtained by PMF easily.Therefore, the combination of these two methods could make the analysis results more reliable.
Atmosphere 2018, 9, x 9 of 13 accounted for a total of 12.77%.The result indicated that the main source of PAHs in Urumqi was grass/wood/coal combustion, and the minor source was petroleum combustion.The information about the specific source was obtained by the ratio method, but it was unable to estimate the contribution from the specific source categories.However, such information can be obtained by PMF easily.Therefore, the combination of these two methods could make the analysis results more reliable.

Sources of PAHs with Positive Matrix Factorization (PMF)
To further identify and evaluate the major sources of PAH contaminants in PM2.5 and PM2.5-10 in Urumqi, the PMF model (EPA PMF 5.0) was used.We selected 3 factors which were most stable and interpretable, as shown in Figure 5.During the process of analyzing the generated results, there were some differences in the source contribution rates of PAHs in PM2.5 and PM2.5-10, but two profiles are identified as the same sources, which have similar composition and present the same tracer species.

Sources of PAHs with Positive Matrix Factorization (PMF)
To further identify and evaluate the major sources of PAH contaminants in PM 2.5 and PM 2.5-10 in Urumqi, the PMF model (EPA PMF 5.0) was used.We selected 3 factors which were most stable and interpretable, as shown in Figure 5.During the process of analyzing the generated results, there were some differences in the source contribution rates of PAHs in PM 2.5 and PM 2.5-10 , but two profiles are identified as the same sources, which have similar composition and present the same tracer species.
Atmosphere 2018, 9, x 9 of 13 accounted for a total of 12.77%.The result indicated that the main source of PAHs in Urumqi was grass/wood/coal combustion, and the minor source was petroleum combustion.The information about the specific source was obtained by the ratio method, but it was unable to estimate the contribution from the specific source categories.However, such information can be obtained by PMF easily.Therefore, the combination of these two methods could make the analysis results more reliable.
In the sampling period, emissions from coal combustion (54.20%, 50.29%), biomass burning (33.43%, 21.25%), and coking and petroleum combustion (12.37%, 28.47%) contributed to PAHs bound PM in Urumqi, with coal combustion being the most important source.This result was consistent with the previous results obtained before changing coal to natural gas in Urumqi [8], and we found that the contribution of coal combustion was apparently smaller than that in 2002 (84%) [20].

Figure 1 .
Figure 1.Location of sampling site.

Figure 1 .
Figure 1.Location of sampling site.

Figure 2 .
Figure 2. Distributions of different PAHs in fine and coarse particles.

Figure 2 .
Figure 2. Distributions of different PAHs in fine and coarse particles.

Figure 3 .
Figure 3. Winter to summer ratios of atmospheric PAHs for fine and coarse particles.The w/s was calculated by the concentration of PAHs in winter divided by concentration of PAHs in summer.

Figure 3 .
Figure 3. Winter to summer ratios of atmospheric PAHs for fine and coarse particles.The w/s was calculated by the concentration of PAHs in winter divided by concentration of PAHs in summer.
a Range (mean); Note: Spring includes the samples of March, April, and May; Summer includes the samples of June, July, and August; Autumn includes the samples of September, October, and November; Winter includes the samples of January, February, and December.

Table 2 .
BaP eq concentrations (ng m −3 ) for fine and coarse particles in different seasons.BaP eq (%) refers to the proportion of several carcinogenic polycyclic aromatic hydrocarbons (CANPAHs, such as BaA, BbF, BkF, DahA, and IcdP) in total BaP eq .