Characterization of Chemical Composition in PM 2 . 5 in Beijing before , during , and after a Large-Scale International Event

To commemorate the 70th anniversary of the victory of the Chinese people's Anti-Japanese War and the World AntiFascist War, an international parade was held in Beijing in September 2015. In order to ensure satisfactory air quality during this event, a phased emission control measures were taken in Beijing and its surrounding provinces. The 24-h PM2.5 samples were collected in Beijing from August 1 to September 15, 2015 covering the period before, during and after this large-scale event. The observed PM2.5 data, meteorological data, emission reduction measures, and air mass trajectory simulation results were systematically analyzed to understand the pollution characteristics and chemical compositions of PM2.5 in Beijing. The results indicated that PM2.5 concentration during the two emission control phases was reduced by 61.7% comparing to the non-control period, but the regional transport of pollutants and meteorological conditions had a more prominent impact on PM2.5 than emission reduction during phase 2. The secondary water-soluble ions including SO4, NO3, and NH4 were found as the main ions present in PM2.5. During the entire emission control period, organic carbon (OC) and elemental carbon (EC) mass concentrations were decreased by 53.1% and 57.9%. A PM2.5 mass balance was analyzed, and it was found that the organic matter accounted for 29.3, 37.6 and 28.5% of the PM2.5 mass before, during and after the emission control, while the contribution of mobile sources to PM2.5 was relatively outstanding after a series of emission control measures.


INTRODUCTION
Northern China has experienced significant economic growth over the past 30 years as a result of rapid industrialization and urbanization.However, such economic prosperity has also put tremendous pressure on the local and regional environment (Ji et al., 2014).Especially, the frequently occurred haze pollution problems in this region have attracted widespread attention in recent years (Li and Han, 2015;Huang et al., 2015).For example, northern China experienced severe haze episodes with long duration which affected unusually sizeable areas in January 2013, October 2014 and December 2015, respectively (Wang et al., 2014a;Tang et al., 2015).Many studies have shown that PM 2.5 is a main cause of haze problem, which can have adverse effects on visibility, human health, and even global climate (Song et al., 2012;Fann and Risley, 2013;Wang et al., 2014b).The investigation of PM 2.5 characteristics and its formation mechanism is thus of critical importance.
Beijing, the capital city of China, is one of the largest megacities in the world.It has more than 21.5 million residents and 5.6 million vehicles in 2014 (BMBS, 2015).Located in northern China, its main topography is plain, but it is surrounded by the Taihang and Yanshan mountains at approximately 1000-1500 m above sea level in the west, north, and northeast, respectively (Chen et al., 2015).This special terrain environment is beneficial to trap air pollutants.Previous studies indicated that air pollution in Beijing was affected not only by its local emission, but by pollutant transport from its surrounding provinces (Tao et al., 2013).Under the stable weather condition, high density air pollution can be easily formed.As the center of politics, culture, international exchange and technological innovation of China, Beijing hosted various large-scale international events, such as the 2008 Olympic Games and the 2014 Asia-Pacific Economic Cooperation (APEC) Summit.In order to ensure satisfactory air quality during such events, the Chinese government has implemented stringent air pollution control measures (e.g., shutting down factories, restriction of vehicles on road) in Beijing and its surrounding areas (Meng et al., 2015).These measures have provided a valuable opportunity to investigate the characteristics of regional air pollution under the effect of emission reduction.Some studies have been reported to examine the properties and formation mechanism of PM 2.5 in Beijing.For example, Okuda et al. (2011) investigated the difference in the concentrations of air pollutants between the period of the 2008 Olympic Games and the same periods during the prior three years, and they observed that many ions in PM 2.5 decreased during the Game period, but NO 3 -increased significantly.Schleicher et al. (2012) analyzed the elemental characteristics of PM 2.5 before, during, and after the period of the 2008 Olympic Games, and they found that elements from anthropogenic sources were reduced more efficiently during the Olympic Games than those from geological sources.Gao et al. (2013) reported that the concentrations of SO 4 2-, NO 3 -, and NH 4 + in PM 2.5 were closely related to meteorological conditions, primary emissions, and regional transport, and the high pollution observed after the full-scale emission control was attributed to regional transport from the south of Beijing.Xu et al. (2015) examined the impacts of emission controls on aerosol composition, size distributions, and oxidation properties, and their results indicated that both highly oxidized secondary organic aerosol (SOA) and secondary inorganic aerosol decreased during the 2014 APEC Summit.
Based on observation data, analysis of emission reduction measures, and air quality simulations, Wen et al. (2016) investigated the PM 2.5 chemical composition change and the relationship between emissions and air pollutant concentrations during the 2014 APEC Summit, and they found that total PM 2.5 concentrations were reduced by 54% when compared to non-emission control period.
To commemorate the 70th anniversary of the victory of the Chinese People's Anti-Japanese War and the World Anti-fascist War, a large-scale international parade was held in Beijing on September 3, 2015.In order to improve the air quality during the parade, the government of Beijing cooperated with the governments of its surrounding regions including Hebei province, Tianjin city, Shanxi province, Shandong province, Inner Mongolia Autonomous Region, and Henan province (Fig. 1).A series of emission reduction measures were gradually implemented before and during this event, such as reducing the number of vehicles in operation by about 50%, shutting down factories, closing construction activities, and enhancing the cleanliness of urban roads.Different from previous emission reduction activities for large-scale events, a phased emission reduction program was adopted, including phase 1 from August 20 to 27, 2015 with emission reduction only in Beijing, and phase 2 from August 28 to September 4, 2015 with emission reduction both in Beijing and its surrounding provinces.Such phased emission control measures may result in different pollution characteristics of PM 2.5 .As a result, it is of importance to study the chemical components of PM 2.5 during this special event for better understanding the pollution sources and formation mechanism in Beijing.The objective of this study is then to reveal the impact of the phased emission reduction on PM 2.5 pollution characteristics.The PM 2.5 samples were collected in Beijing from August 1 to September 15, 2015.The observed PM 2.5 and meteorological data as well as the emission reduction measures were systematically analyzed.The obtained results would help understand the formation and development of PM 2.5 pollution and thus provide scientific basis for air pollution control in Beijing.

Emission Reduction Measures
Beijing and its surrounding provinces had taken a series of emission reduction measures (Fig. 1).More than 111 businesses with high emissions in Beijing were required to stop or limit their production from August 20 to September 4, 2015.Meantime, the odd-even rule was implemented in Beijing to restrict traffic emissions.Vehicles with license plates ending in odd number were banned from roads on even-numbered days, and those with plates ending in even number were banned from roads on odd-numbered days.In addition, 80% of government vehicles were prohibited on roads.During this emission control period, all construction activities were stopped.Trucks transporting mud and stone as well as heavy-emission vehicles were prohibited from roads in Beijing.Moreover, the road sweeping in Beijing was enhanced.Through a series of these measures, the emission reductions were 36.5% for SO 2 , 49.9% for NO x , 50.3% for PM 10 , 49.0% for PM 2.5 and 32.4% for VOCs in Beijing (BMEPB, 2015).Furthermore, major provinces around Beijing had also taken a series of measures to reduce emissions from August 28 to September 4, 2015, and their emission reductions were over 30%.More than ten thousand enterprises were closed or ordered to limit production, and about 9000 construction sites were shut down during the emission control period in Beijing and its surrounding provinces (BMEPB, 2015).Table 1 summarizes the emission reductions and reduction ratio in Beijing.

Sample Collection
PM 2.5 was sampled during the period from August 1 to September 15, 2015 (Fig. 1).A total of 41 daily samples were collected, and there were no samples for 5 days taken due to rain and equipment failure.The samples were collected on the roof (35 m above ground) of a building in Beijing Normal University (39°57′N, 116°21′E), a site having mixed residential and traffic sources, but without major industrial sources.The sampling site was about 6.5 km away from the northwest of the city center (Tiananmen Square), and it could fairly represent the air quality in Beijing.The sampling time was from 9:00 a.m. to 9:00 a.m. of the next day.When faced with exceptional circumstances (rainy day and equipment failure), the temporary adjustment of sampling time was made.For the collection of PM 2.5 samples, a medium volume sampler (URG-3000ABC, USA) with a flow-rate of 8.35 L min -1 was used.Teflon filters (47 mm, Whatman, UK) and quartz filters (47 mm, Whatman, UK) were used as impaction substrates on alternate days.The Teflon samples were used for the analysis of PM mass, elements and inorganic ions, while the quartz samples were used for the analysis of carbonaceous species (Wen et al., 2016).All of the samples were stored in a fridge at < 4°C before analysis.

Chemical Analysis
For analyzing PM mass concentrations, the Teflon filter were weighed before and after each sampling event using a microbalance (Sartorius-Denver TB-215D, accuracy, 0.01 mg) after stabilizing under constant temperature (20 ± 5°C) and humidity (40 ± 2%) for over 48h.For analyzing the concentrations of water-soluble ions, half of each Teflon filter sample was extracted ultrasonically by 10 mL of deionized water (with resistivity of 18.25 MΩ cm -1 ).After passing through microporous membranes (pore size, 0.45 µm; diameter, 25 mm), the filtrates were determined for pH with a pH meter.The concentrations of six anions (F -, Cl -, NO 2 ) and five cations (Na + , K + , Ca 2+ , Mg 2+ , NH 4 + ) were analyzed by Ion Chromatography (861 Advanced Compact IC, Metrohm) (Wang et al., 2005).
For analyzing the concentrations of elements, half of each Teflon filter sample was digested at 170°C for 4 h in a high-pressure Teflon digestion vessel with 3 mL of concentrated HNO 3 , 1 mL of concentrated HClO 4 , and 1 mL of concentrated HF.After cooling, the solutions were dried and treated with 1 mL of concentrated HNO 3 , and then diluted to 10 mL with deionized water.A total of 23 elements (Na, Mg, Al, S, Ca, Sc, Ti, V, Cr, Mn, Fe, Co, Ni, Cu, Zn, As, Se, Sr, Cd, Sb, Ce, Eu, and Pb) were determined for their concentrations by inductively coupled plasma mass spectrometry (ICP-MS, Agilent 7500A) (Fu et al., 2008).
The thermal optical reflectance carbon analysis method for element carbon (EC) and organic carbon (OC) content was performed following the IMPROVE protocol (Chow et al., 2004).The quartz filter sample was heated stepwise in pure He gas (no oxygen) at temperatures of 120°C (OC1), 250°C (OC2), 450°C (OC3), and 550°C (OC4), respectively, transforming the particulate carbon into CO 2 .The filter was then heated in an environment of He with 2% oxygen, at temperatures of 550°C (EC1), 700°C (EC2), and 800°C (EC3), by which temperature the EC of the samples was released.OC in the carbonization process would form the pyrolyzed organic carbon (OP).The OC was defined by OC1 + OC2 + OC3 + OC4 + OP, while the EC was defined by EC1 + EC2 + EC3 -OP (Zhang et al., 2015).
Methodological blank values were determined and subtracted from analytical results of the samples.All the analysis procedures were subject to strict quality control to avoid any possible contamination.

Meteorological Data
Meteorological parameters including wind speed, wind direction, ground temperature, relative humidity, visibility, and precipitation amount were obtained from the "Weather Underground" website (http://www.wunderground.com).Data obtained from this website were supplied by Beijing weather station which is located in Beijing Capital International Airport (40°04′N, 116°35′E), about 25 km apart from the northeast of the city center.All of the meteorological data were arranged to half-hour interval according to the PM 2.5 sampling time (9:00 am to 9:00 am next day).In addition, morning temperature profile was  (Cheng et al., 2002).

Air Mass Trajectory Analysis
In this study, air mass trajectory analysis was used to examine the impact of pollutant regional transport on pollution characteristics in Beijing.The 24-h back trajectories were calculated for each hour at 100 m height using the Hybrid Single-Particle Lagrangian Integrated Trajectory (HYSPLIT, NOAA) 4 model (Draxler and Hess, 1997).The trajectories were grouped into three clusters during phase 1 of emission reduction and five clusters during phase 2 using the algorithm of cluster analysis.The clustering of trajectories could minimize the inter-cluster differences among trajectories while maximizing the inter-cluster difference, and it has been widely used in many studies (Chen et al., 2015;Li et al., 2015).

General Characteristics of PM 2.5
Fig. 2 shows the time series of PM 2.5 mass concentrations and major meteorological parameters during the entire study period.The daily PM 2.5 concentrations varied significantly from 10.3 to 122.6 µg m -3 and averaged at 49.6 ± 31.0 µg m -3 .As indicated in Fig. 2, the concentration of PM 2.5 in Beijing was affected by wind direction and wind speed.The concentration of PM 2.5 was much higher when the predominant wind direction was south.However, when the wind direction changed from the south to the north, the concentration of PM 2.5 rapidly decreased.Haze episodes with such cycle driven by meteorological conditions have also been observed many times in Beijing (Sun et al., 2014;Chen et al., 2015).A highly negative correlation was found between PM 2.5 concentration and visibility, with a correlation coefficient of -0.77.It was also found that the mean concentration of PM 2.5 was as low as 18.6 µg m -3 during the 20%-best-visibility days, while it reached as high as 90.8 µg m -3 during the 20%-worst-visibility days.This prominent contrast highlighted the importance of PM 2.5 on visibility impairment (Li et al., 2013).In addition, mixing layer height which determines the diffusion of air pollution in the vertical direction also had a great influence on PM 2.5 , and a higher negative correlation coefficient of -0.66 was observed with PM 2.5 except for rainy days.Although a few days in phase 2 were rainy, the rainfall was relatively low.Some previous studies showed that the removal effect of light rain on PM 2.5 was not obvious, but the resulted low mixing layer height and high relative humidity during rainy days could promote the accumulation and generation of PM 2.5 (Wei et al., 2015), and thus a relatively high PM 2.5 concentration was observed in phase 2 as shown in Fig. 2(d).
Table 2 presents the average concentration of PM 2.5 before, during and after the emission control measures (August 1 to 19; August 20 to 27; August 28 to September 4; September 5 to 15).During the emission control period, the concentration of PM 2.5 was reduced by 61.7%, comparing with no emission control period.This indicates that the stringent emission control played an important role on air quality improvement.It was noted that the average concentration of PM 2.5 increased by 53.8% in phase 2 as compared to phase 1 of the emission control period although a more rigorous emission reduction was implemented in phase 2. This unexpected PM 2.5 increase could be explained by the unfavorable meteorological condition, which was characterized by the occurrence of southerly wind (Fig. 2).According to cluster analysis results of 24 h air mass backward trajectories shown in Fig. 3 and the predominant wind direction shown in Fig. 2, nearly all of the air mass was from the north of Beijing in phase 1, but about 43% of the air mass in phase 2 was from southern Beijing.Previous study indicated that the anthropogenic emission was small in northern Beijing, while it was much higher in southern Beijing where many factories were located (Li et al., 2014).This could lead to higher PM 2.5 concentration in phase 2 when southerly wind occurred.Moreover, during phase 2, the observed average PM 2.5 concentration in Baoding and Tianjin (e.g., south of Beijing) where trajectory 2, 3, and 4 (Fig. 3(b)) passing through was 43.2 and 33.8 µg m -3 , respectively, while the average PM 2.5 concentration in Chengde (e.g., north of Beijing) where trajectory 1 and trajectory 5 passing through was 13.5 µg m -3 (CNEMC, 2015).Such relatively high PM 2.5 concentration in the regions south of Beijing may also contribute to the increased PM 2.5 concentration in phase 2. Furthermore, as shown in Fig. 3(b), the air mass from the south of Beijing passed near the ground surface, making more polluted air mass being carried to Beijing and thus leading to increased PM 2.5 .The above analysis illustrated that, to a certain extent, the transport of pollutants and meteorological conditions had a more prominent impact on PM 2.5 than emission reduction.

Ionic Components of PM 2.5
Table 3 presents the monitoring results of water-soluble ions in PM 2.5 during the study period.In this study, the secondary water-soluble ions (SWSI) including SO 4 2-, NO 3 -, and NH 4 + were found as the main ions present in PM 2.5 , accounting for 73.3% of the total water-soluble ions during the entire study period although this proportion varied before, during and after the emission control period.Fig. 4 shows the time series of the mass concentrations of SO 4 2-, NO 3 -, NH 4 + , Ca 2+ , Mg 2+ , K + , Na + , and Cl -during the study period.It was observed that SO 4 2-, NO 3 -and NH 4 + correlated well with PM 2.5 , and the correlation coefficient was 0.95, 0.94 and 0.98, respectively.The concentrations of these three ions were high on August 2, August 6, August 11 to 13, and September 14 to 15 but low during the emission reduction period, illustrating that these three ions had a great contribution to PM 2.5 .Some studies have shown that Cl -might be derived from coal combustion when the mass ratio of Cl -/Na + was larger than 1.17 for sea water (Wang et al., 2016).The ratios of Cl -/Na + were 1.99 ± 0.98, 1.45 ± 0.65, and 3.16 ± 2.45 before, during, and after the emission control period, respectively.It indicated that the effect of coal combustion emission control was better.
The concentration of Ca 2+ and Mg 2+ was dramatically  decreased during the emission control period and rapidly increased after emission reduction.This indicated that the measures of stopping construction and strengthening of road sweeping had a significant impact on the concentration of PM 2.5 because the main sources of Ca 2+ and Mg 2+ were soil dust, road dust and construction dust.K + mainly came from biomass burning (Zhang et al., 2013).Because of summertime and the emission control during the study period, the concentration of K + was at a low level.The total water-soluble ions (TWSI) of PM 2.5 accounted for 41.5 ± 4.3%, 35.5 ± 10.0%, and 41.0 ± 6.3% of the total mass of PM 2.5 before, during, and after the emission control period, respectively.The proportion of the TWSI in PM 2.5 decreased significantly during the emission control period.During the entire emission control period, the average concentration of SWSI (SO 4 2-, NO 3 -, and NH 4 + ) was decreased by 75.3%, 61.3%, and 67.3%, respectively, comparing with the period before emission control.The drop of SO 4 2-concentration was the largest because SO 4 2was mainly converted from its precursor SO 2 which was a major gaseous pollutant.The regional cooperative emission control has ensured a significantly decreased SO 2 emission.However, a least decrease in the concentration of NO 3 -was observed.This was because the NO 3 -was mostly converted from its precursor NO x which was mainly from vehicle emissions.Although the NO x emission was reduced because of the traffic control measures, the absolute amount of vehicles running in Beijing was still remained at a high level.The drop of NH 4 + concentration was slightly less than that of SO 4 2-.The NH 4 + was mainly resulted from the reaction between NH 3 and the acid components of NO 3 and SO 4 2- (He et al., 2011).The NH 3 emission was mainly from human activities and natural sources, and the cooperative emission control measures led to reduced NH 3 emission.Consequently, a decreased NH 4 + concentration could be resulted due to the reduced NH 3 emission and the decreased NO 3 -and SO 4 2-in the atmosphere (Rumsey et al., 2011).Comparing the concentrations of the main ions in PM 2.5 during phase 1 with those during phase 2, SO 4 2-, NO 3 -, and NH 4 + increased by 1.02, 1.67 and 1.33 µg m -3 , respectively.According to the above analysis, a significant portion of air mass during phase 2 was from the south of Beijing, which had large emissions of SO 2 and NO x , the precursors of secondary inorganic particles.Because of the regional transport of pollutants and the higher humidity, the concentrations of SO 4 2-, NO 3 -, and NH 4 + during phase 2 were much higher than those during phase 1.
The equivalent charge ratio of the major cations Na + , NH 4 + , K + , Mg 2+ , and Ca 2+ to major anions SO 4 2-, NO 3 -, and Cl -was used to indicate the neutralizing level of PM 2.5 .The cation equivalent (CE) = Na + + NH 4 + + K + + Mg 2+ + Ca 2+ , and the anion equivalent (AE) = SO 4 2-+ NO 3 -+ Cl -.In this equation, CE/AE ≥ 1 indicated that most of the acids could be neutralized, while CE/AE < 1 indicated the aerosol was acidic (He et al., 2011).Fig. 5 shows the correlations between CE and AE before and during the emission control period.The slope of the linear regression was 0.96 with a correlation coefficient R 2 = 0.98 before emission control period, indicating that the aerosol was acidic.And the slope of the linear regression was 1.03 with a correlation coefficient R 2 = 0.98 during emission control period, indicating that the anions were fully neutralized by cations.The comparisons between the two periods showed that the proportion of acidity in the atmosphere reduced during the emission control period, which indicated the effect of SO 2 and NO x emission control was significant.
The mass ratio of NO 3 -/SO 4 2-had been used as an indicator of the relative importance of stationary versus mobile sources of sulfur and nitrogen in the atmosphere (Yao et al., 2002).Typical mass ratio of NO x and SO x was reported as 8:1 to 13:1 from gasoline and diesel combustion emissions and 1:2 from coal combustion emissions, respectively (Chen et al., 2015).Therefore, the higher ratio of NO 3 -/SO 4 2-, the more contribution of mobile sources than stationary sources to PM 2.5 .The emission control of stationary sources was usually better in developed countries, so their mobile sources had a greater contribution to PM 2.5 with a higher ratio of NO 3 -/SO 4 2-.For example, the ratio of NO 3 -/SO 4 2-in PM 2.5 in Los Angeles was reported as 2 (Kim et al., 2000).However, in most areas of China, stationary source of emission from coal burning was still the main source of atmospheric pollutants, and the ratio of NO 3 -/SO 4 2-in PM 2.5 was normally about 0.3 to 0.5 (Xu et al., 2012;Tao et al., 2013).The ratio of NO 3 -/SO 4 2-was 1.7 ± 0.4 during the emission control period, which was higher than that before (1.2 ± 0.4) and after (1.2 ± 0.5) emission control.Meanwhile, as shown in Fig. 4, the concentration of NO 3 was higher than that of SO 4 2-during the entire emission control period.This implies that the emission reductions during the study period from stationary sources were more than those from mobile sources.As a result, after taking a series of emission control measures, the PM 2.5 pollution from mobile source was more outstanding.

Inorganic Element in PM 2.5
The mean concentrations of 23 elements during entire periods were given in Table 3.The percentages of total detected elements accounted for 12.2 ± 2.5%, 10.7 ± 3.4%, and 13.5 ± 3.0% of PM 2.5 before, during, and after the emission control period, respectively.S was one of the major elemental components in PM 2.5 , its origin is mainly anthropogenic being produced by burning of fossil fuels, in vehicle, and power production (Cohen et al., 2004).The concentration of S was dramatically decreased during the emission control period, which indicated that the measures of strict supervision of anthropogenic sources had a significant impact on PM 2.5 .The crustal elements (i.e., Na, Mg, Al, Ca, and Fe) were mainly from soil or construction dust.The concentration of this five elements was decreased by 39.9% to 76.6%.It indicated that the measures of stopping construction and strengthening of road sweeping had a significant impact on the concentration of PM 2.5 .
Cu was used as an indicator for wear debris of vehicle (Lin et al., 2014).Pb is coal combustion's tracer.The variation of Cu/Pb during entire study periods can be indicated that the relative importance of mobile versus stationary sources.The ratio of Cu/Pb was 0.42 ± 0.38, 0.54 ± 0.20, and 0.41 ± 0.27 before, during, and after the emission control period, respectively.The ratio variation agreed with NO 3 -/SO 4 2-, which further explained that the PM 2.5 pollution from mobile sources was relatively outstanding during the emission control period.

Carbonaceous Species of PM 2.5
Carbonaceous species represent a significant fraction of PM 2.5 in Beijing.In this study, carbonaceous species were divided into two major subtypes: organic carbon (OC) and elemental carbon (EC).It was found that OC and EC accounted for 16.8 ± 4.6% and 5.8 ± 1.6% of PM 2.5 during the entire study period.The mean concentrations of OC and EC during four periods were given in Table 3.During the entire emission control period, OC and EC mass concentrations were decreased by 53.1% and 57.9%, respectively, as compared with those before the emission control.
The relationship between OC and EC was very useful in assessing the origin of carbonaceous particles.A strong correlation between OC and EC indicates that they were predominantly emitted from a single source (Park et al., 2001).Fig. 6 presents the correlation between OC and EC before and during the emission control period.A relatively strong correlation was found between OC and EC both before (R 2 = 0.73) and during (R 2 = 0.82) the emission control.This indicates that the OC and EC constituents may come from one single emission source, probably vehicle emissions, as both OC and EC were emitted significantly from internal combustion engines, especially diesel vehicles.Several investigators used OC/EC ratios to identify the sources of carbonaceous species (Ram et al., 2008).Some studies reported OC/EC ratios of 1.0 to 4.2 for traffic emissions and 2.5 to 10.5 for coal combustion emissions, while OC/EC ratio greater than 2 had been used for the identification and evaluation of secondary organic carbon (SOC) (Chen et al., 2006).In this study, the mass ratio of OC/EC before and during emission control was 3.0 ± 0.6 and 3.2 ± 0.4, respectively.Using EC tracer method (Castro et al., 1999), the SOC concentrations before and during emission control were determined as 2.6 ± 2.3 and 1.6 ± 0.9 µg m -3 , accounting for 24.0 ± 15.7% and 30.7 ± 10.0% of the total mass of OC, respectively.Compared the SOC/OC ratios from this work and other research, the value were 56.5% (Xu et al., 2015) and 50.2% (Yao et al., 2016), which was higher than this study period.It indicated that SOC for the carbon component contribution was not obvious, and the main source of carbonaceous aerosols was vehicular emissions and coal combustion sources.

PM 2.5 Mass Balance
In this study, PM 2.5 mass balance was calculated (Fig. 7).The mass balance was classified into primary components and secondary components.The primary components include primary organic aerosol (POA), EC, soil dust, and pollution elements (S, V, Cr, Mn, Ni, Cu, Zn, As, Sr, Cd, Sb, and Pb).The secondary components include secondary organic aerosol (SOA), SO 4 2-, NO 3 -, and NH 4 + .The mass of SOA and POA could be converted from SOC and primary organic carbon (POC) by multiplying a coefficient of 1.6 (Chow et al., 1996).Soil dust concentrations were calculated by the concentrations of the elements of Al, Ca, Fe, and Ti according to Eq. ( 1).In this study, Si was not measured and its concentration was estimated to be four times of Al (Wen et al., 2016).
As shown in Fig. 7, the decrease of SO 4 2-percentage was the largest among all the SWSI ions, with SO 4 2-accounting for 15.2%, 9.6%, 10.3%, and 15.1% of the PM 2.5 , respectively, during the four study periods.This indicates that the strict emission control in industrial enterprises had a significant impact on PM 2.5 concentration.The percentage of NO 3 -during the four periods was 16.1%, 16.3%, 16.3%, and 14.8%, respectively, with a little difference observed after emission control, and this is consistent with the results of ion analysis.In addition, Fig. 7 shows a clear increase in SOA on August 20, when emission control was firstly implemented in Beijing.During the entire emission control period, the organic matter (POA + SOA) accounted for 37.6% of the PM 2.5 mass, while this proportion was only 29.3% and 28.5% before and after the emission control.The soil dust accounted for 12.1%, 10.2%, 10.9%, and 12.6% of the total mass of PM 2.5 , respectively, during four study periods.The percentage of soil dust mass in PM 2.5 during emission control period decreased considerably, indicating that the measures of stopping construction and strengthening of road sweeping were effective.The variation trend of pollution elements was the same as that of soil dust.In addition, the percentage of soil dust, pollution elements, and SWSI in PM 2.5 during phase 2 was slightly higher than that during phase 1.

CONCLUSIONS
A large-scale parade was held in Beijing on September 3, 2015 to commemorate the 70th anniversary of the victory of the Chinese People's Anti-Japanese War and the World Anti-fascist War.A phased emission control measures were taken in Beijing and its surrounding provinces in order to improve the air quality during the parade.Such large-scale open experiments provided an excellent opportunity of investigating the pollution characteristics in Beijing.In this study, PM 2.5 samples were collected in Beijing before, during, and after the emission control for this international parade event.The observed PM 2.5 and meteorological data as well as the emission reduction measures were systematically analyzed.Air mass trajectory analysis based on HYSPLIT model was also used to examine the impact of pollutant regional transport on the pollution characteristics in Beijing.It was found that the concentration of PM 2.5 decreased significantly during the emission control period, but the concentration during phase 2 of emission control increased by 32.6% as compared to that during phase 1 due to the impact of regional pollutant transport and meteorological conditions.The secondary water-soluble ions including SO 4 2-, NO 3 -, and NH 4 + were found as the main ions present in PM 2.5 in Beijing.During the emission control period, the drop of the concentration of Ca 2+ , Mg 2+ , and SO 4 2-was the largest comparing with other ions.The results also indicated that the measures of stopping construction, strengthening of road sweeping, and strict supervision of industrial enterprises had a significant impact on PM 2.5 .The correlation between OC and EC concentrations and the OC/EC ratio showed that the main source of carbonaceous aerosols was vehicular emissions and coal combustion sources.Moreover, the ratio of NO 3 -/SO 4 2-and Cu/Pb during the emission control period was higher than before and after the emission control.It was shown that after taking a series of emission control measures, the contribution of mobile sources to PM 2.5 in Beijing was relatively outstanding.

Fig. 2 .
Fig. 2. Time series of (a) predominant wind direction and wind speed; (b) temperature (T) and relative humidity (RH); (c) visibility and mixing layer height (MLH) of daily maximum; and (d) PM 2.5 concentrations and precipitation.

Fig. 3 .
Fig. 3. Cluster analysis results of 24 h air mass backward trajectories during phase 1 (a) and phase 2 of emission reduction (b).

Fig. 5 .
Fig. 5. Correlations between cations and anions before (a) and during (b) emission control period.

Fig. 6 .
Fig. 6.Correlation between OC and EC concentrations before (a) and during (b) emission control period.

Fig. 7 .
Fig. 7.Chemical compositions in mass percentage for the PM 2.5.

Table 1 .
Emission reduction and reduction ratio during the emission control period in Beijing.

Table 2 .
Summary of mass concentrations of PM 2.5 and average meteorological parameters at different periods.

Table 3 .
Mass concentration of major components in PM 2.5 at different periods (µg m -3 ).