Characterization , Meteorological Influences and Source Identification of Carbonaceous Aerosols during the Autumn-winter Period in Tianjin , China

Organic and elemental carbon (OC and EC) were measured in PM2.5 and PM10 at urban and suburban sites in Tianjin, China, from September 2009 to February 2010. Average PM2.5 and PM10 mass levels were 124 and 243 μg/m, respectively, with 53% of the PM10 mass present as PM2.5. The average OC and EC concentrations in PM2.5 were 14.5 and 4.3 μg/m, respectively, while those in PM10 were 33 and 6.1 μg/m, respectively. Relation of OC and EC to the trace elements As, Zn, K and Pb and principal component analysis indicated that multiple sources contributed to the OC and EC. Secondary organic carbon was found to constitute 46% and 66% of the total organic carbon in PM2.5 and PM10, respectively, in autumn and 41% and 56%, respectively in winter.


INTRODUCTION
Ambient particulate matter (PM), especially in urban and highly industrialized areas, contains an important fraction of carbonaceous material (Nunes and Pio, 1993;Castro et al., 1999;Ho et al., 2003;Dan et al., 2004;Cao et al., 2006;Meng et al., 2007).Atmospheric particulate carbon is comprised of a complex mixture of substances containing carbon atoms, in two main fractions: organic carbon (OC) and elemental carbon (EC).EC is mainly emitted from anthropogenic combustion sources and does not undergo chemical transformations.It is used as a good indicator of primary anthropogenic air pollutants (Larson and Cass, 1989;Kim et al., 2000;Lin and Tai, 2001).OC, containing PAHs, PCBs and other components with potential mutagenic and carcinogenic effects, has both a primary and secondary origin (Cao et al., 2003;Cao et al., 2005;Ram et al., 2008).Primary organic carbon (POC) is formed during combustion and emitted in particulate form, while secondary organic carbon (SOC) is formed in the atmosphere through gas-toparticle conversion processes of volatile organic compounds, either as result of the condensation of low vapor pressure volatile organics or chemical adsorption of gaseous species on aerosol particle surfaces (Pandis et al., 1992;Pankow, 1994;Turpin and Huntzicker, 1995).Formation of SOC has frequently been observed during laboratory investigations in smog chamber experiments from light irradiation of gaseous mixtures containing organic compounds (McMurry and Grosjean, 1985;Stern et al., 1987;Pankow, 1994).
Rapid industrialization and economic development has occurred in China during recent years and may increase the emission of various pollutants.Lack of extensive and effective pollution-control measurement means that the amount of particulate matter emitted into atmosphere will increase and impact the health of the public.Many studies have been performed in the world focusing on carbonaceous species of PM, however, limited studies have been done in China (Cao et al., 2003;Dan et al., 2004).To date, research on carbonaceous particles in PM 2.5 and/or PM 10 in China were reported only in few cities, such as Taiwan (Lin and Tai, 2001), Xian (Cao et al., 2005), Taiyuan (Meng et al., 2007), the Pear River Delta Region (Cao et al., 2003) and Beijing (Dan et al., 2004).As far as Tianjin is concerned, rare studies have been reported.Tianjin is considered as one of the most developed region of China, and high PM levels and poor visibility have become a serious problem (Cao et al., 2003).Coal and biomass burning, vehicle exhaust and industrial emissions all contribute to the ambient PM in Tianjin.Therefore, observations for OC and EC analysis are needed for Tianjin.
In this paper the concentrations and distributions of particulate elemental and organic carbon in Tianjin are presented.The paper is aimed to quantify the contribution of carbonaceous species to PM 2.5 and PM 10 , identify potential sources and meteorological influences, and estimate SOC production during the sampling period.Information obtained in this study will also allow evaluation of the changes in air quality in Tianjin and help to develop further pollution control strategies for particulate matter.

Sampling Sites and Description
Tianjin is a typical metropolis in northern China, with a population of approximately 10 million, covering a land area of about 11,919 km 2 .It is located to the southeast of Beijing, on the lower reaches of the Haihe River and adjacent to the Bohai Sea.As one of the most rapidly developing regions in China, Tianjin has been suffering from severe particulate matter pollution problems during recent years.Eight sampling sites were chosen in this study as shown in Fig. 1.DongLi monitoring station site (DL) and BeiChen monitoring station site (BC) were chosen to represent suburban sites in Tianjin.The other six sites were located in the urban area.Detailed information about sampling sites is given in Table 1.

PM Sampling
PM samples were collected at 8 sites from September, 2009 to February, 2010.PM 2.5 samples were collected  The site was situated at about 12 m above ground level in the environmental monitoring station of DongLi District BeiChen monitoring station site (BC) N39°13'33.60"E117°11'6.10" The site was situated at about 10 m above ground level in the Tianjin BeiChen hi-tech industrial park.

Traditional urban sites
HeXi monitoring station site (HX) N39°5'33.39"E117° 12'6.23 The site was situated at about 10 m above ground level in the environmental monitoring station of HeXi District.
HeBei-environmental protection bureau site (HB) The site was situated at about 15 m above ground level in the meteorological observatory tower of Tianjin simultaneously at 3 sites (TT, SZ and DL).The sampling span was six days and sampling was for 24 h.Middlevolume air samplers were deployed to collect PM samples at a flow rate of 100 L/min (TH-150C, Tianhong Intelligent Instrument Plant of Wuhan, China).PM samples were collected using Pall quartz fiber filters (90 mm in diameter, type 2500QAT-UP, Pall Life Sciences, U.S.A.) over 24 h.After sampling, the exposed filters were stored in a refrigerator at about 4°C until chemical analysis to limit losses of volatile components.To reduce the carbonaceous species background level in the filter, quartz filters were pretreated at 900°C for 180 min and then placed in clean polyethylene Petri dishes.The Petri dishes were then wrapped with Teflon tape and aluminum foil, and stored in a freezer until field measurement.The filters were maintained under conditions of 40% relative humidity (RH) and 25°C for over 48 h, and were weighed using an electronic microbalance with ± 1 mg sensitivity before and after sampling.In this study, a blank filter of each site was treated with the same procedure as the sampling filter during the sampling and analytical processes.The blank filter was exposed to the same environments as the sampling filter except for being used as a particle sampling filter.Then the OC content of the blank filter was subtracted from the sampling filter OC content to correct the positive artifacts due to adsorption of gas-phase organics onto the filter during and/or after sampling collection.However, negative artifacts due to volatilization of particle-phase organics from the particle sample were not quantified in this study and the OC values obtained can be considered as conservative results.In all, 231 PM 10 samples and 81 PM 2.5 samples were collected in the observation period.

Experimental Procedure for Analysis of Carbonaceous Species
The samples were analyzed for OC/EC using DRI Model 2001 (Thermal/Optical Carbon Analyzer).The IMPROVE A thermal/optical reflectance (TOR) protocol was used for the carbon analysis.The protocol heats a 0.526 cm 2 punch aliquot of a sample quartz filter stepwise at temperatures of 140°C (OC1), 280°C (OC2), 480°C (OC3), and 580°C (OC4) in a non-oxidizing helium (He) atmosphere, and 580°C (EC1), 740°C (EC2), and 840°C (EC3) in an oxidizing atmosphere of 2% oxygen in a balance of helium.The carbon that evolves at each temperature is oxidized to carbon dioxide (CO 2 ), and then reduced to methane (CH 4 ) for quantification with a flame ionization detector.As temperature increases in the inert helium, some of the organic carbon pyrolyzes to elemental carbon, resulting in darkening of the filter deposit.This darkening is monitored by reflectance of 633 nm light of a He-Ne laser.When oxygen is added, the original and pyrolyzed elemental carbon combusts and the reflectance increases.The amount of carbon measured after oxygen is added until the reflectance achieves its original value is reported as optically detected pyrolyzed carbon (OPC).The eight fractions OC1, OC2, OC3, OC4, EC1, EC2, EC3, and OPC are reported separately in the data sheet.The IMPROVE protocol defines OC as OC1 + OC2 + OC3 + OC4 + OPC and EC as EC1 + EC2 + EC3 -OPC.The analyzer was calibrated with known quantities of CH 4 every day.Replicate analyses were performed at the rate of one per group of 10 samples.Blank filters were also processed, similar to the protocol used for samples, in order to assess the analytical detection limits.Concentrations of carbonaceous species in the blanks were below the detection limits in all analysis.
While the TOR method employed in this study is commonly used for PM measurements, it should be recognized that other methods of determining the OC and EC fractions can give different results (Chow et al., 2001).
The elements of PM 10 and PM 2.5 collected simultaneously by polypropylene filters (90 mm in diameter, Beijing Synthetic Fiber Research Institute, China) using the same model air sampler were analyzed by ICP-AES.For the detailed procedures see Kong et al. (2011).Meteorological data of the sampling period were obtained from the Tianjin Meteorological Bureau.

PM 2.5 and PM 10 Concentrations and Ratios
Descriptive statistics for all valid observations of PM 2.5 and PM 10 concentrations from 8 sites in Tianjin operating from September 2009 to February 2010 are summarized in Table 2.The average daily concentration of PM 10 at the 8 sites ranged from 53 to 1024 g/m 3 , with an average and standard deviation of 243 ± 151 g/m 3 .Among the 8 sites, the HD site where PM 10 samples were collected only in the winter period had the highest mass concentration of 285 ± 218 g/m 3 .For the other 7 sites sampled both in autumn and winter, average PM 10 mass at the HB site were the highest (283 ± 185 g/m 3 ).In contrast, the average PM 10 mass concentration analyzed at the SZ site was the lowest, with a mean concentration of 201 ± 144 g/m 3 .DL and BC sites, two suburban sites undergoing urbanization, affected by increasing emission sources distributed around the regions, have PM 10 mean concentrations of 221 ± 126 and 246 ± 139 g/m 3 , respectively.The average concentrations of PM 10 at HX, TT and HQ sites were 227 ± 114, 215 ± 104 and 256 ± 127 g/m 3 , respectively.In the present study, the daily average PM 10 concentrations of the 8 sites all exceeded the Class 3 of Chinese PM 10 standard of 150 g/m 3 , indicating serious particulate emissions in this period in Tianjin.
PM 2.5 samples were simultaneously collected at three sites: two traditional urban sites (TT and SZ) and one suburban site (DL).The average 24-h PM 2.5 concentration was 124 g/m 3 , with a large standard deviation of 83 g/m 3 , indicating a serious fine particulate pollution and multisource contribution to airborne PM 2.5 in this period in Tianjin.Among the 3 sites, unlike the distribution of PM 10 concentrations, the highest levels of PM 2.5 appeared at the SZ site, at which the site average concentration of PM 10 was the lowest in the present study.DL had a similar level of PM 2.5 at 129 ± 74 g/m 3 .The TT site had lower PM 2.5 concentrations, with a mean concentration of 117 ± 73 g/m 3 .The PM 2.5 mass concentrations analyzed in Tianjin were higher than those detected in Europe (Van Dingenen et al., 2003;Saarnio et al., 2008;Pindado et al., 2009;Putaud et al., 2010) and Brazil (12 g/m 3 ) (Bourotte et al., 2005); and higher than Cao's study during the same period in the Pearl River Delta Region (73 g/m 3 ) of China (Cao et al., 2003), but were much lower than that found in Taiyuan (193 g/m 3 ), China (Meng et al., 2007).Although there is no specific legislation for PM 2.5 levels in China, the average daily concentration of PM 2.5 (124 g/m 3 ) in Tianjin during the measurement period exceeded the 24-h National Ambient Air Quality Standard (NAAQS) of the USA (35 g/m 3 ).In order to build an ecological city in Tianjin, further control measures should be conducted to reduce particulate emissions.However, the anthropogenic emissions contributions were considered to be highest during the year of the observation period, and lower PM 2.5 and PM 10 mass concentrations may be found on warmer days in Tianjin and further analysis will be done.

Concentrations of Carbonaceous Species
A summary of the measurement results for 24-h average concentrations of OC, EC and TC from 8 sites of Tianjin are given in Table 2 and their seasonal variations are summarized in Fig. 2. The average OC concentrations in PM 2.5 and PM 10 were 14.5 ± 11.2 and 32.6 ± 20.6 g/m 3 , respectively and the average EC concentrations in PM 2.5 and PM 10 were 4.3 ± 2.9 and 6.1 ± 2.9 g/m 3 , respectively.Particulate carbon species (TC, calculated by the sum of OC and EC) accounted for an averaged 15.2% of the fine particulate (PM 2.5 ) mass loading and 15.9% of PM 10 .PM 2.5 consisted of 11.7% organic carbon and 3.5% elemental carbon, while PM 10 consisted of 13.4% organic carbon and 2.5% elemental carbon.EC is confined to sub-micron size particles and has a longer lifetime in the atmosphere.In previous studies (Venkataraman and Friedlander, 1994;Offenberg and Baker, 2000), the ambient distribution of EC was found to be associated with fine particles and contained almost 75% of the total EC in polluted urban areas.In this study, 71% of EC was associated with fine particle mode and 29% of EC was associated with coarse particle mode on average.Our study is in agreement with prior research.However, it is different from some previous research (Pickle et al., 1990;Offenberg and Baker, 2000), which showed that the accumulation of organic compounds was mainly in the sub-micrometer (< 1 m diameter) particle size range, 44% of OC was found to associate with the fine particle (PM 2.5 ) mode and 56% of OC was associated with the coarse particle mode in Tianjin.
Seasonal variability of carbonaceous concentrations has been analyzed in many studies.Higher concentrations are always found in the colder period, because of enhanced consumption of coal, biomass burning and gasoline fuel and typical meteorological conditions (lower intensity of sunshine, lower inversion height and less efficient photochemical decomposition).
The concentration of TC (calculated by the sum of OC and EC) measured in PM 10 samples in this study was higher in winter (from December to February), with a mean concentration of 46.1 g/m 3 , and was lower in autumn (from September to November), with a mean concentration of 31.5 g/m 3 , showing an evident seasonal variation, which was consistent with the normal seasonal trend of prior research (Castro et al., 1999;Dan et al., 2004).OC had higher concentrations in winter compared to autumn, with average concentrations of 40.0 and 25.5 g/m 3 , respectively.The relatively high abundance of OC during the wintertime was attributed to the enhanced contribution from local emissions and meteorological conditions (lower mixing height and trapping of pollutants due to temperature inversion).
For PM 2.5 samples collected in 3 sites, higher concentration levels of OC were also found in the winter period, however, higher EC concentrations were found in the autumn season rather than that in winter.
OC and EC concentrations varied among the observation sites, indicating that carbonaceous aerosols in Tianjin were influenced more or less by local factors at various types of sampling sites.Except for the HD site where samples were collected in winter only, the highest concentrations of OC and EC in PM 10 samples (37.0 ± 20.8 and 7.1 ± 3.8 g/m 3 , respectively) were observed at the BC site: a suburban site undergoing urbanization, with some factory emissions.Another suburban site (DL) also had higher TC concentrations of PM 10 samples compared to sites including TT, SZ and HX.For PM 2.5 samples, the highest OC and EC concentrations were observed at the DL site, higher than that collected at TT and SZ.The results showed that serious carbonaceous pollution was not only in traditional urban sites, but also in suburban sites undergoing urbanization.
Compared to other cities in the world, the cities of China always had much higher OC and EC concentrations in fine particles, therefore a high health risk could exist in China.High OC and EC concentrations (14.5 and 4.3 g/m 3 ) of PM 2.5 samples analyzed in the present study indicated that severity of the pollution of carbonaceous particles in Tianjin atmospheric environment.

Correlations and Sources of Carbonaceous Species
EC is known as a primary anthropogenic particulate constituent, mainly emitted from combustion processes of biomass or fossil fuels, while OC can be directly emitted to the atmosphere in particulate form along with EC or can be produced by gas to particle conversion processes as secondary sources (Lin and Tai, 2001;Dan et al., 2004;Ram et al., 2008).Since primary OC and EC are mostly generated from the same sources, emission and transformation characteristics of carbonaceous aerosol can be estimated based on the relationship between organic and elemental carbon (Kim et al., 2000;Lin and Tai, 2001).
Linear regression analysis between OC and EC is useful to assess the primary nature of carbonaceous aerosols.As shown in Fig. 3, good OC-EC correlations, differentiated according to seasons, were obtained for PM 2.5 samples (r = 0.85 for autumn, p < 0.0001; r = 0.93 for winter, p < 0.0001) and PM 10 samples (r = 0.81 for autumn, p < 0.0001; r = 0.69 for winter, p < 0.0001) collected in Tianjin, indicating the primary nature of the OC in this region.The different linear regression equations in two seasons could be related to the seasonal fluctuation of source emissions and SOC formation.
Elevated OC/EC ratios were found in winter compared to autumn, and this could be attributed to several reasons.First, coal consumption for winter heating contributes more to OC than EC, and also increases the emission of volatile organic precursors.Second, low temperature leads to the adsorption and condensation of semi-volatile organic compounds onto existing solid particles.Third, the low mixing layer height in winter would enhance the SOC Fig. 3. Relationship between organic carbon and elemental carbon concentrations in PM 2.5 and PM 10 samples in Tianjin (n: the number of samples).
formation (Strader et al., 1999).This seasonal pattern of higher wintertime OC/EC ratio was also observed in studies in Beijing (Dan et al., 2004), Guangzhou and Hong Kong (Duan et al., 2007).
The HD site, where samples were only collected in winter, had the highest OC/EC ratio for PM 10 samples (8.9).Among the other sites, the OC/EC ratios of PM 10 had similar levels between 4.4 and 6.0.HQ had the higher OC/EC ratio (6.0) and SZ had the lowest OC/EC ratio (4.4) for PM 10 samples.For PM 2.5 samples, the DL site had a higher OC/EC ratio (4.3) compared to SZ (3.8) and TT (3.1) sites.
The ratios of OC to EC concentrations in PM 2.5 for this study are compared with other measurements in the world in Table 3. Together with the present study in Tianjin, Chongju (Korea) (Lee and Kang, 2001), Shanghai (China) (Ye et al., 2003), PRDR (China) (Cao et al., 2003), Los Angeles (USA), Anaheim (USA) (Chow et al., 1994) and Seoul (Korea) (Kim et al., 2007) had a OC/EC ratio that ranged from 1.2 to 3.7, consistent with the results of PM 2.5 observation of western interagency monitoring of protected visual environment (IMPROVE) sites in USA (Richard et al., 2002).And the ratios of OC/ EC at Milan (Italy) (Lonati et al., 2007), Taiyuan (China) (Meng et al., 2007) and Xian (China) (Cao et al., 2005) were larger than 4.0.
Several authors have used EC/TC ratio as an indicator to differentiate the dominant contributor among fossil fuel burning and biomass burning.The relatively lower ratios (0.1-0.2) are used as typical of biomass burning sources and higher ratios (~0.5) indicate contribution from fossil fuel combustion sources (Gillies et al., 2000;Gillies et al., 2001;Ram et al., 2008;Sudheer and Sarin, 2008).The EC/TC ratios of PM 10 samples were always less than 0.2 at the sites throughout the sampling period.And for PM 2.5 samples, EC/TC ratios were a little higher than 0.2, which could also represent biomass burning contribution.Biomass burning in residential consumption and open fire was suggested as one of the most important carbonaceous emission sources in China (Cao et al., 2006;Liu and Shao, 2007).In the analysis of Beijing, biomass burning was the largest source of EC, accounting for 40% of the total (Liu and Shao, 2007).
However, it is not sure if these ratios reflect emission characteristics accurately as emission patterns in China may be different from that in other countries.Source apportionment analysis conducted in Tianjin previously indicated that coal combustion fly ash was an important contributor to ambient particles (Bi et al., 2007).In order to further characterize the carbonaceous emission sources in Tianjin, OC and EC were compared with pollution elements from ICP-AES analysis.The relationship of elements including As, Zn, K and Pb, with OC and EC at sampling sites is shown in Table 4.As has long been recognized as coal combustion makers, Zn and Pb are known as tracers of motor vehicle emission, and K is usually considered to relate to biomass combustion (Meng et al., 2007;Dan et al., 2004).OC and EC of PM 10 and PM 2.5 samples collected at different sampling sites all have significant positive correlations with Zn and most have positive correlations with Pb, suggesting that vehicle  emission was a main contributor of TC burden.Good positive correlations were also usually found between carbonaceous compounds and K and As, indicating that biomass burning and coal combustion were the main sources of OC and EC in Tianjin.Particular matter for the PM 2.5 analysis at the TT site and PM 10 analysis at the HQ site, OC and EC only have good correlations with Zn, showing enhanced vehicle emission contributions.
Principal component analysis (PCA) had been applied for source analysis in many studies (Cao et al., 2005;Moon et al., 2006;Li et al., 2010;Wang et al., 2010), eight carbon fraction concentrations were identified by PCA to quantify source contributions (Table 5, Table 6).For PM 2.5 samples, three factors accounted for 93.80% of the total variance in the data.Factor 1, which explained 68.37% of the variance, showed high loading for OC1, OC2, OC3, EC1 and OPC.OC1 was enriched in biomass burning; OC2 could point to coal combustion; EC1, OC3 and OPC were considered markers of gasoline motor-vehicle exhaust (Cao et al., 2005).Thus, factor 1 represented multi-source emissions including coal combustion, biomass burning and gasoline motor-vehicle exhaust.Factor 2, which explained 17.07% of the variance, had high loadings for EC2 and EC3, reflected the contribution of diesel-vehicle exhaust (Watson et al., 1994).Factor 3, which explained 8.36% of the variance, had high loadings for OC4.OC4 are enriched in the road dust profile (Chow et al., 2004).For PM 10 samples, three factors accounted for 90.26% of the total variance in the data.Factor 1, explaining 60.76% of the variance, showed high loading for OC1, OC2, OC3 and In the present study, multi-source contribution to OC and EC was found in this period in Tianjin, however, much work needs to be done to thoroughly apportion OC and EC sources.

Correlations of OC, EC and TC Levels with Meteorological Parameters
Meteorological conditions, including air pressure, wind speed (WS), temperature (Temp.), and relative humidity (RH), may affect concentrations of OC and EC.Correlation coefficients for relationships between OC and EC concentrations (average concentrations of sampling sites) and meteorological parameters collected during the observation period were evaluated and are listed in Table 7.In some studies, wind speed was found to influence pollution concentrations due to its cleanup function (Pindado et al., 2009).Good negative correlations were found between wind speed and OC and EC of PM 10 and PM 2.5 fractions in the present study, supporting the cleanup function of wind.TC and OC of PM 10 samples had good negative correlations with temperature; this could be caused by elevated emissions in colder days.Pressure varied little during the sampling period and did not show evident correlations with OC and EC of particles.No significant correlation was observed between OC and EC concentrations and RH in this research.

Estimation of Secondary Organic Carbon Concentrations
While elemental carbon is essentially a primary pollutant, OC consists of a complicated mixture of species from both primary sources and secondary sources.SOCs are produced in the atmosphere via various chemical and physical transformation processes involving the oxidation of volatile organic reactive gases (VORGs) with reactive species such as ozone, hydroxyl and NO x radicals followed by coagulation/condensation onto the pre-existing aerosol particles becoming a part of aerosols already present in the atmosphere (Ram et al., 2008).The importance of SOC related to haze, visibility, climate and heath has been known for decades, but not enough research has been done (Cao et al., 2003).Although several constraints and uncertainty exist, the OC/EC ratios were adopted in many studies as an indicator to identify SOC formation.For control strategies for particulate matter, it is necessary to quantify the contributions of the primary and secondary organic carbon to carbonaceous aerosol.
The direct measurement of secondary organic carbon (SOC) is not possible as they are derived from the various physical and chemical transformation processes.Rather, they are estimated either by counting the primary organic carbon, using EC as a tracer, and then subtracting it from the total organic carbon measured (Turpin and Huntzicker, 1995;Castro et al., 1999;Ram et al., 2008) or by summing up the concentrations of all such oxidation products found in the aerosols (Schauer et al., 1996).Since the abundances of VORGs and reactive radicals vary on temporal and spatial scales, SOCs are also considered to vary in the same manner (Ram et al., 2008).
In this research, SOC was estimated by subtracting the primary organic carbon from total organic carbons, using the following equation: where OCsec is the secondary OC (SOC), OCtot is the total OC (TOC), and (OC/EC)pri is the primary ratio estimate.
In this study, a total of 231 PM 10 samples and 81 PM 2.5 samples were collected during the sampling period.Average SOC concentrations of PM 2.5 and PM 10 are listed in Table 8.The estimated SOC concentrations for PM 2.5 were 5.8 and 7.4 g/m 3 , accounting for 46% and 41% of the total OC, was 4.9% and 5.6% compared to the particle mass in the autumn and winter, respectively.For PM 10 , the average concentrations of estimated SOC were 16.9 and 22.3 g/m 3 , accounting for 66% and 56% of OC concentrations, was 7.4% and 9.0% compared to the particle mass in the autumn and winter, respectively.The results suggested that SOC was an important component of organic carbon and particle mass.Due to emission and meteorological characters, SOC concentrations were higher in winter than in autumn, specifically 1.32 times higher for PM 10 and 1.28 times higher for PM 2.5 .However, SOC contributed more to OC in autumn than winter, was 66% in contrast to 56% and 46% in contrast to 41% for PM 10 and PM 2.5 , respectively.In Castro et al.'s (1999) study, SOC was reported to contribute to a minimum of about 17% of TOC during winter contrast to the maximum 78% in summer in Portugal.Also Na et al. (2004) reported that the contribution of SOC to the TOC burden tends to be lower during the season with lower photochemical activity than that during the time of enhanced photochemical activity.The present research supported this view to some degree.
The relative amount of secondary organic carbon ranged from 52 to 75%, 44 to 67% with an average of 66% and 56% of the TOC for PM 10 in autumn and winter season, respectively.And it ranged from 40 to 52%, 33 to 53%, with an average of 46% and 41% of TOC for PM 2.5 in autumn and winter, respectively.The concentration ratio of secondary organic carbon in PM 2.5 to that in PM 10 sampled simultaneously indicated that 34% and 33% of the secondary organic carbon was distributed in PM 2.5 in autumn and winter, respectively.This showed that the secondary organic particles observed in this research were associated more with the coarse particle mode.
In some research in other urban areas, Cao et al. (2003) analyzed SOC accounting for 43% and 38% of organic carbon concentrations in winter for PM 2.5 and PM 10 in PRDR, respectively.Castro et al. (1999) obtained SOC accounting for 17%, 39%, and 37% of OC at Birmingham, Oporto, and Coimbra in winter for PM 10 .Lin and Tai (2001) found SOC contributed to 40% and 32% of OC for PM 2.5 and PM 10 in Kaohsiung, respectively.The estimations of SOC in this research were in agreement with those in urban areas.
Formation of SOA involves two different processes: production of condensable organic compounds through oxidation reactions (chemical) and subsequent nucleation/ condensation on the pre-existing aerosols (physical) (Ram et al., 2008).However, their formation mechanism needs to be further analyzed in order to assess their contribution to aerosols and their atmospheric chemistry and environment effects on a regional scale.

CONCLUSIONS
PM 2.5 , PM 10 and carbonaceous aerosol were investigated at eight sites in Tianjin from September 2009 to February 2010.Average PM 2.5 and PM 10 concentrations were 124 and 243 g/m 3 , respectively, with PM 2.5 constituting 53% of the PM 10 mass.The average OC concentrations in PM 2.5 and PM 10 were 14.5 and 33 g/m 3 , respectively and the average EC concentration in PM 2.5 and PM 10 were 4.3 and 6.1 g/m 3 , respectively.The average OC/EC ratios for PM 2.5 and PM 10 in the Tianjin were 3.7 and 5.4, respectively.OC and EC in Tianjin were found to be strongly correlated, which indicated that the dominant source of OC and EC was primary emission.Comparisons of OC and EC with trace elements including As, Zn, K, and Pb and PCA analysis demonstrated multi-sources contributions to carbonaceous species in Tianjin.Good negative correlations were found between wind speed and OC and EC of PM 10 and PM 2.5 fractions, indicating the cleanup function of wind.Estimating by primary OC/EC ratio, secondary organic carbon was found to constitute 46% and 66% of the total organic carbon in PM 2.5 and PM 10 , respectively in autumn and 41% and 56% respectively in winter period.To formulate effective control strategies for air pollutant and build an ecological city in Tianjin, further studies of the origin and composition of the primary and secondary particles are needed.

Fig. 2 .
Fig. 2. Seasonal variation of EC, OC abundances ( g/m 3 ) of PM 10 (a) and PM 2.5 (b) at different sampling sites in Tianjin during September 2009-February 2010.The seasons have been presented in terms of two different time periods: autumn (Sep.-Nov.)and winter (Dec.-Feb.).

Table 1 .
Descriptions of sampling sites

Table 3 .
Comparison of OC and EC of PM 2.5 at Tianjin with other cities.

Table 4 .
Pearson product-moment correlation coefficients between OC, EC concentrations in PM 2.5 and PM 10 samples and As, Zn, K and Pb at sampling sites in Tianjin.

Table 7 .
Pearson product-moment correlation coefficients between OC, EC and TC concentrations in PM 2.5 , PM 10 samples and meteorological parameters.Carbonaceous concentration used average concentration of sampling sites.

Table 8 .
Levels of secondary organic carbon (SOC) at 8 sites in Tianjin estimated from primary OC/EC ratios.