Chemical Characteristics of PM 2 . 5 during 2015 Spring Festival in Beijing , China

Air pollution especially of PM2.5 pollution is a serious problem in Beijing. In order to quantify the effect of a festival in which pollution was expected to be reduced, we collected and analyzed PM2.5 samples in urban Beijing during the 2015 Spring Festival (from February 9 to March 6 2015). We divided the sampling period into three phases: non-haze, haze and firework days. The average concentration of PM2.5 was highest on firework days at 248.9 μg m, followed by haze days (199.9 μg m) and non-haze days (90.8 μg m). The air quality of non-haze days during the holiday was better than that during non-holiday periods. Secondary inorganic ions (SO4, NO3 and NH4) were enriched on the haze days, while those on firework days contained large amounts of Cl and K, but small amounts of NO3 and NH4. Ratios of NO3/SO4, SO4/K and Cl/K effectively distinguished the characteristics of PM2.5 between firework events and haze days. Ion balance calculations indicated that the acidity of PM2.5 from firework days was higher than that from haze and non-haze days. A method using enrichment factors (EF) found that crustal elements (EF < 10 on all three types of days) included Ca, Al, Fe, Na, Co, Ni, P, Ti, and V; firework elements (EF > 10 on firework days, significantly higher than haze days) were made up of Ba, Cr, Cu, Mg, Pb, S, Si, and Zn; common anthropogenic pollution elements (EF > 10 in all three types of days) were As, Cd, Cu, Pb, S, Sb, Zn. Differences in chemical characteristics indicated that holidays such as the Spring Festival can affect air pollution patterns in two ways: a decrease in the population and vehicles but an increase in activities such as firework displays.


INTRODUCTION
Air pollution caused by fine particles with a diameter of 2.5 µm or less (PM 2.5 ) has aroused worldwide attention during the past decades because of its significant negative effects on visibility, human health and global climate (Okada et al., 2001;Kang et al., 2004;Sun et al., 2006;Sun et al., 2014).In China, the average annual number of haze days has fluctuated but generally increased (Gao, 2008).In Beijing, hazy episodes happened more frequently than in others cities (Che et al., 2009;Tang et al., 2013;Li et al., 2016).The annual average number of haze days in Beijing increased from about 50 days yr -1 in the 1980s to 72 days yr -1 in 2008 (Hu and Zhou, 2009;Zhao et al., 2012;Chen et al., 2014).Pollution in Beijing is generally attributed to industry, road dust, vehicle emissions, coal and biomass burning (Okuda et al., 2004;Wang et al., 2006;Duan et al., 2012;Zhang et al., 2013;Yu et al., 2015).Thus, strategies to improve air quality should take all these factors into account.
Nevertheless, there are still some special anthropogenic factors that lead to dramatic pollution episodes, one of which is the use of fireworks.Burning fireworks releases huge amounts of gaseous pollutants, such as sulfur dioxide (SO 2 ) (Huang et al., 2012), nitrogen oxides (NO x ), carbon dioxide (CO 2 ), and carbon monoxide (CO) (Wang et al., 2007;Vecchi et al., 2008;Huang et al., 2012), as well as suspended particles containing water-soluble ions, organic/elemental carbon and trace metals (Drewnick et al., 2006;Sarkar et al., 2010;Jiang et al., 2015;Kong et al., 2015).It has been noted before that firework displays could lead to a sharp increase in PM 2.5 concentrations (Wang et al., 2007;Feng et al., 2012;Tsai et al., 2012;Han et al., 2014) and generates haze pollution within a short time (Sarkar et al., 2010;Tian et al., 2014b).In China, high-intensity firework displays mainly occur during festivals or special events, especially the Spring Festival.As a mega city, a "Banned Fireworks" policy has been in place in Beijing since 1993, but was replaced by a "Limited Fireworks" policy in 2006 (Jing et al., 2014).In preparation for the Spring Festival, almost half of the population in Beijing (about 9 million in 2013, http://news.xinhuanet.com/local/2013-02/18/c_124355887.htm) leave for their hometown.Consequently, factories closed down and the population density and vehicle numbers sharply decrease, reducing the emission intensity of the main pollutants.Meanwhile, fireworks burning becomes an important pollution source during the Spring Festival.There have been some studies on the characteristics of fireworks pollution, including its impacts on fine and coarse particles (Chang et al., 2011;Huang et al., 2012;Li et al., 2013;Jing et al., 2014;Zhao et al., 2014;Jiang et al., 2015;Lin et al., 2016), source apportionment (Feng et al., 2012;Tian et al., 2014b) and human health risk assessment (Zhao et al., 2012;Shi et al., 2014;Yang et al., 2014).
We analyzed the chemical characteristics of PM 2.5 collected during the 2015 Spring Festival period (February 9 th -March 6 th 2015).The sampling period was separated into pollution events and non-haze days.The pollution events periods involved firework displays and haze episodes.This allowed us to further study the short-term holiday effects.We hypothesized that the results of this study would provide support for controlling short-term and particular forms of air pollution (such as firework displays) in urban areas.

Sampling
PM 2.5 samples were collected on the rooftop of a residential bungalow (about 4 m above ground level), in Guangning village (39.94°N, 116.15°E),Shijingshan District, Beijing.It is surrounded by a viaduct (city expressway, linear distance of 250 m), a hill and two large thermal power plants (Jingxi and Datang).During the sampling period, the two thermal power plants were shut down.PM 2.5 particles were collected on Whatman filters (90 mm quartz membrane, diameter 2.2 microns) by a medium volume air sampler (model: TH-150C, flow rate: 100 L min -1 ).The sampling time was approximately 12 h, normally from 8:00 a.m. to 8:00 p.m., once every three days on non-haze days.If there was an unexpected pollution event, additional samples were collected.Based on the regularity of setting off firework, the samples were collected over three periods during February 18 th and 19 th , including before the Spring Festival (8:10 a.m.-4:10 p.m.), Spring Festival's eve (4:20 p.m.-0:20 a.m.), and after the Spring Festival (0:30 a.m.-8:30 a.m.).On March 5 th (the Lantern Festival), the sample was collected from 6:00 p.m. (the beginning of setting off the firework) to 6:00 a.m.(the next morning).Detailed sampling information is presented in Table 1.The samples were placed in sulfuric acid paper after sampling, wrapped with aluminum foil in the dark, and then stored in a refrigerator (-4°C).The filters were weighed before and after sampling on an analytical balance to an accuracy of 0.0001g (TB-215D, Denver) at a constant temperature (20°C) and relative humidity (40%), then divided by the sampling volumes to obtain PM 2.5 mass concentrations.Another three blank filters were dealt with simultaneously using the same method as above.All the procedures were strictly quality-controlled to avoid any possible contamination of the samples.

Water-Soluble Ions
One-fourth of each sample was extracted with 10 mL deionized water (resistivity18.2MΩ cm -1 ).After a 30 min in a microwave to dissolve the particles, the solution was drawn into a 1 mL syringe, filtered and injected into a 10 mL centrifuge tube with 0.22 mm filter cap.Water-soluble ions (Na + , NH4 + , K + , Mg 2+ , Ca 2+ , Cl -, NO 3 -, and SO 4 2-) were analyzed by ion chromatography (IC, Dionex).Three blank samples were analyzed at the same time to ensure the accuracy of analysis.Anions were analyzed on an ICS-2100 ion chromatograph, which consisted of a separation column (Dionex Ionpac AS11) and guard column (Dionex Ionpac AG11).Cations were analyzed on a DX-600 ion chromatograph, which consisted of a separation column (Dionex CS12A), a guard column (Dionex AG12A), a selfregenerating suppressed conductivity detector (Dionex Ionpac ED50), a gradient pump (Dionex Ionpac GP50), and a chromatographic box (LC20).The details were given elsewhere Yuan et al. (2003).

Meteorological Data and Air Quality Data Collection
During the sampling periods, meteorological factors (hourly average) such as relative humidity (RH), visibility, temperature, wind speed and direction were obtained from Weather Underground (http://www.wunderground.com), which archive meteorological data was generally used by previous researches about air quality (Wang et al., 2009;Guo et al., 2010;Verma et al., 2013) Air quality data (average of 24 hours, monitored at the end of the PM 2.5 sampling) were collected from the National Urban Air Quality Real-time Publishing Platform (http://113.108.142.147:20035/emcpublish/).To accurately represent local air quality of the sampling site, we selected the Gucheng site (4.4 km away from Shijingshan PM 2.5 sampling site).The Data data include NO 2 , SO 2 concentrations.

Variation of PM 2.5 Mass Concentrations
Meteorological data during the sampling period are shown in Table 1.Daily temperatures ranged from 3.7°C to 13.2°C.Wind speed was low (5.1-17.2km h -1 ).Relative humidity (RH) was from 19% to 67%.Concentrations of PM 2.5 , SO 2 and NO 2 during the sampling period in Shijingshan are presented in Fig. 1.The PM 2.5 concentration range was 67-294 µg m -3 , which was almost always higher than the limit given by National Ambient Air Quality Standard (NAAQS, GB3095-2012) Grade II (75 µg m -3 ), exceeding it by times except on Feb.12, Feb.18 (8:10 a.m.-4:10 p.m.) and Feb. 27.In contrast, SO 2 and NO 2 were below the limits (150 µg m -3 for SO 2 , 80 µg m -3 for NO 2 ).
According to the "Observation and forecasting levels of haze" (Sarkar et al., 2010), when relative humidity is < 80%, haze is an atmospheric phenomenon during which dry particles obscure the clarity of the sky, giving a visibility is less than 10.0 km.Using this definition, the sampling days can be separated into the non-haze days and haze days.There were four samples (Feb. 19,(0)(1)(2)(3)(4)(5)(6)(7)(8)Feb. 19,Mar. 5,6:00 a.m.,Feb. 18,, 0:20 a.m.) collected during intensive firework displays, a typical custom of the Spring Festival.These samples were therefore defined as 'fireworks days'.Thus, the sampling period was divided into nonhaze, haze and fireworks phases (see Table 1).The PM 2.5 average concentration on non-haze days was 90.8 µg m -3 .Compared with results from other studies in Beijing, the mass concentration of non-haze day in the 2015 Spring Festival was lower than that on typical clear days (Wang et al., 2006;Shen et al., 2009;Sun et al., 2013c).This suggests that PM 2.5 concentrations on non-haze days during the Spring Festival were strongly affected by the sharp decrease in population and human activity.The PM 2.5 average concentrations were 199.9 and 248.9 µg m -3 on haze and fireworks days respectively, approximately 2.2 and 2.7 times higher than this on non-haze days.The PM 2.5 concentration on firework days was approximate 1.2 times higher than on haze days.Compared with other studies, the average PM 2.5 concentration on firework days was similar to that of 255.3 µg m -3 measured in Shanghai (Feng et al., 2012), but higher than the 184.0 µg m -3 and 116.5 µg m -3 measured in 2006 and 2012 in Beijing (Wang et al., 2007;Cheng et al., 2014), which may be explained by the sampling location of the above two studies being part of the area covered by the"Limited Fireworks" policy, while the sampling site in this study was not.However, the PM 2.5 concentration was much lower than that measured in other firework display studies in Lucknow (352 µg m -3 during Diwali festival) (Barman et al., 2009), Delhi (588 and 389 µg m -3 during Deepawali festivals in 2007 and 2008, respectively) (Tiwari et al., 2012), and the YanShuei Area of Southern Taiwan (437 µg m -3 during the beehive fireworks display) (Lin et al., 2014).For the former two studies, high relative humidity, low wind speeds in November (the Diwali and Deepawali festivals month) contributed to the high PM 2.5 concentrations.Furthermore, there is uncertainty in the scale and intensity of the firework displays, which could strongly effect air pollution levels, such as that during "the beehive fireworks display" at one site in Taiwan which seemed to seriously aggravate air quality much more than civil firework burning.It is acknowledged that firework displays release large amounts of gaseous pollutants (Ravindra et al., 2003) and particulate matter (Vecchi et al., 2008), which results in elevated levels of PM 2.5 .A study of in Chengdu in 2009-2010 also observed that the average PM 2.5 mass concentration during a firework display (212.7 µg m -3 ) was higher than that on haze days (187.3 µg m -3 ) (Wang et al., 2013).Thus, much attention has been paid to this serious short-term air pollution.The average concentrations of SO 2 on firework days was 57.8 µg m -3 , higher than on haze days and non-haze days.However, the average concentration of SO 2 on firework days was less than that measured in a study in 2006 in Beijing (Wang et al., 2007).This was probably because that 2015 was the first year in which a sulphur-free propellant technology was used in civil fireworks, the same type as those used in the APEC (2014) fireworks display (http://report.qianlong.com/33378/2014/11/11/8785@9991966.htm).In addition, (Hooper and Johnson, 1999) fireworks sales in 2015 decreased by 41% compared with 2014 since people are gradually aware of the seriousness of pollution (http://www.askci.com/chanye/2015/02/24/123461ywk.shtml).
Previous studies have shown that the relatively stagnant meteorological factors and low mixed boundary layer can prevent pollutants diffusion and result in the accumulation of pollution concentrations and species in atmosphere (Sun et al., 2013a, b;Sun et al., 2013c).Low wind speed and high relative humidity as external meteorological factors promote high PM 2.5 concentration, especially in winter (Sun et al., 2006;Sun et al., 2013c;Tian et al., 2014a).In our study, low wind speed and low temperatures (shown in Table 1) covered all three periods (i.e., haze, non-haze and firework days).Although the emission density of common anthropogenic pollutants declined during the sampling period, meteorological factors might play a primary role in the emergence of haze during the Spring Festival.The average ambient visibilities during the firework days averaged 7.8 km.On Feb. 19, 0:30 a.m.-8:30 a.m., when the average PM 2.5 concentrations reached the maximum (293.7 µg m -3 ), the mean visibility reached its lowest value of 5 km.This is likely due to the firework displays generally happening at midnight (12:00 a.m.) and dawn (about 5:00 a.m.) as is the traditional Chinese custom.Qin et al. (2013) reported similar results, noticing that the visibility was only 1.2 km when the PM 10 , PM 2.5 and PM 1 hourly concentrations reached a maximum value on the eve of the Spring Festival.Aerosol loadings and visibility degradation are negatively correlated (Watson, 2002;Lee et al., 2005;Shen et al., 2009;Shen et al., 2014).

Water-Soluble Inorganic Ion in PM 2.5
Water-soluble inorganic ions (WSIIs) are important components of atmospheric particles and play a significant role in the ambient air quality and visibility.Fig. 2 illustrated the variations of WSIIs during the sampling periods.A high proportion of WSIIs in particulates occurred on haze and firework days, at 44.7% (76.0 µg m -3 ) and 50.8% (126.3 µg m -3 ) respectively, whereas during non-haze days it was 21.6% (19.6 µg m -3 ).
On non-haze days, the dominant ions were the secondary inorganic ions (SNA: SO 4 2-, NO 3 -and NH 4 + ), which accounted for approximately 73.6% of the total WSIIs.The proportion of WSIIs on typical winter days approaching 90% (Gao et al., 2015;Chen et al., 2017), was higher than the proportion on non-haze days during the 2015 Spring Festival.The short-term holiday effect was very clear.On haze days, ion concentrations in PM 2.5 were in order NO 3 -> SO 4 2-> NH 4 + > Cl -> K + > Ca 2+ > Na + > F ->Mg 2+ .Approximately 83.9% of the total measured WSIIs were attributed to SNA, and the concentrations of SNA on haze days was 63.7 µg m -3 , 5.5 times higher than that on nonhaze days.Thus, secondary aerosol species were relatively abundant during haze events.In particularly, the concentration of NH 4 + on haze days was 6.2 times that on non-haze days.SO 4 2-and NO 3 -are formed by the conversion of gaseous precursors (SO 2 and NO 2 ), which on haze days in our study were higher than on non-haze days (shown in Fig. 1).Higher SO 2 emissions and adverse weather conditions (e.g., low temperature, low wind speed and high RH) during haze days could accelerate aqueous reactions and increase conversion efficiency of the accumulating secondary particles (He et al., 2001(He et al., , 2002;;Sun et al., 2013bSun et al., , 2016;;Zhang et al., 2016).However, compared with the data during the haze episode, the order of the ion concentrations on fireworks days was very different, with SO 4 2-> K + > Cl -> NO 3 -> NH 4 + > Mg 2+ > Ca 2+ > F -> Na + .The concentration of SNA was 62 µg m -3 accounting for 49.6% of the total WSIIs on firework days.The concentrations of NH 4 + and NO 3 -were both clearly lower than on haze days.NO 3 -is generally produced by fossil fuel combustion and vehicle emissions, and traffic may also be an important source of ammonia in Beijing (Ianniello et al., 2010;Wang et al., 2014).Therefore, the decreases of NH 4 + and NO 3 -concentrations were largely related to people leaving Beijing for their hometowns for the Spring Festival, leading to a decline in the population and vehicle.These results are similar to those of other studies of the impacts of firework displays conducted in Jinan by Yang et al. (2014).The chemical compositions of particulates during firework days was strongly enriched in Cl -and K + , which made up 21.5% and 25.2% of total WSIIs.The concentration of K + in particulates during firework days was approximately 13 and 49 times that of haze days and non-haze days, respectively.Prior studies also have shown that K + can be enriched in firework aerosols and may be a tracer of PM from fireworks (Drewnick et al., 2006;Vecchi et al., 2008;Cheng et al., 2014).Wang et al. (2007) reported that firecrackers contain 75% KNO 3 .Notably, KNO 3 , KClO 3 and KClO 4 are usually used as oxidizer, and converted into KNO 2 and KCl if there is sustained burning of fireworks (Yang et al., 2014).
To further distinguish the characteristics of PM 2.5 between fireworks days and haze days, several ion ratios were calculated and compared (Fig. 3).Previous research has recognized the NO 3 -/SO 4 2-ratio as an indicator of the influence of mobile versus stationary pollution sources (Arimoto et al., 1996;Hu et al., 2002;Ye et al., 2003).However, the ratio of NO 3 -/SO 4 2-not only indicates the intensity of primary emissions, but also depends on chemical reactions.In our data, the ratio of NO 3 -/SO 4 2-decreased from 1.5 on haze days to 0.9 on non-haze days and 0.6 on firework days.In Xi'an, the ratio was 0.4 on non-haze and 0.7 on haze days (Shen et al., 2009).During a heavy haze pollution episode in 2013, the ratio was 0.73 in Beijing (Wang et al., 2014).High mass ratio in this study suggests more contribution of mobile sources in the observation episode.A low NO 3 -/SO 4 2-ratio on firework days was attributed to increased SO 4 2− caused by larger precursor SO 2 emissions from fireworks.Compared with firework days, the high ratio of NO 3 -/SO 4 2-on haze days was mainly due to the increase of nitrate above that of sulfate.K + and Cl -are mainly produced by primary emissions.The ratio of Cl -/K + (0.9) was lowest on firework days because the concentrations of K + were enriched.The abundance of K + on firework days also affected the ratio of SO 4 2-/K + , which varied considerably when compared with that on non-haze days and haze days.Examination of these three ratios would be useful to help distinguish firework days from other events.
The mean equivalent ratios of A/C during non-haze, haze, and firework episodes were 0.84, 0.95, and 1.10 for PM 2.5 respectively.Correlations between cation and anion equivalents for the three types of PM 2.5 samples were observed (Fig. 4).Most of which were close to unity, indicating that the eight ion species in these samples were the major ions comprising PM 2.5 and they were basically neutral for this study even though carbonate and bicarbonate were not measured in the samples.The mean equivalent ratios of A/C suggested that particles on fireworks days were more acidic than those on haze and non-haze days.The relatively higher A/C ratios during fireworks days could be due to the higher concentrations of SO 4 2-.

Element Species in PM 2.5
Total mass concentrations of 24 elements accounted for about 12.3%, 8.0% and 12.7% of PM 2.5 on non-haze, haze and fireworks days, respectively.Al, Ca, Fe, and S were the dominant elements on non-haze days, accounting for 82.3% of all measured elements (Fig. 5).The concentrations of As, Ba, Cd, Cr, Cu, Pb, S, Se, and Zn during haze days were 2.1-10.4times higher than on non-haze days.Concentration of the other measured elements on haze days were all < 1 µg m -3 , together accounting for 87.9% of all measured elements.
The concentrations of typical fireworks elements Al (8.0 µg m -3 ), Mg (5.7 µg m -3 ), S (8.8 µg m -3 ), and Ba (2.4 µg m -3 ) were higher during firework days than non-haze and haze days, and the sum of these accounted for 79.0% of all measured elements.The levels of some pollution elements (e.g., Al, Ba, Cu, Mg, Pb, S, Sr, and Zn) on firework days were 1.6-18.6times higher than those on haze days, which indicated that the high concentrations of these elements was at least practically due to the fireworks.A similar result was found in India, where metal concentrations during Diwali fireworks were 5-80 times greater than those on non-haze days (Chatterjee et al., 2013).However, Do et al. (2012) found that high levels of Na, K, Si, Al, Mg, Ba, Ca, and Fe were present in the atmosphere prior to the fireworks display.Enhanced concentrations of Pb, Sr, Cu, and Ba during fireworks days had also been observed in Beijing (Wang et al., 2007) and Chengdu (Wang et al., 2013).
Enrichment factors of elements in PM 2.5 during the nonhaze, haze, and fireworks are shown in Table 2.According to Lantzy and Mackenzie (1979), crust (surface soil) was primary source of elements in the atmosphere when that element's enrichment factor (EF) is less than 10, which means that the element from soil or rock weathering.If the EF is between 10 and 10 4 , it means that the element enrichment was the result of crust and human activities (Tang et al., 2006).Following this approach, we classified 24 elements into three groups: a. Crustal elements (EF < 10 on all three types of days), including Ca, Al, Fe, Na, Co, Ni, P, Ti, and V, which could be mainly produced by high-intensity of crustal dust.The EF of crustal elements remained stable during the sampling period, which is due to these elements being largely from natural sources (e.g., soil and construction materials) and rarely transported into the atmosphere, hence, their contribution to PM on haze days is limited.
Stable weather conditions and high RH are also important factors (Tian et al., 2014b).b.Common anthropogenic pollution elements (EF > 10 in all three types of days), such as As, Cd, Cu, Pb, S, Sb and Zn, which are mainly attributed to anthropogenic sources.Studies have indicated that the elements (e.g., As, Cd, Cr, Cu, Ba, Pb, Se, and Zn) would mainly arise from anthropogenic sources (e.g., industrial activities, fossil fuel combustion and vehicle emissions) (Monaci et al., 2000;Marcazzan et al., 2001;Gao et al., 2015).c.Fireworks elements (EF > 10 on firework days, and significantly higher than on haze days), made up of Ba, Cr, Cu, Mg, Pb, S, Si, and Zn, since they are generally used as fuels, oxidizers and coloring agents in fireworks.Tsai et al. (2012) found firework powders burn and release a large amount of metal-related aerosol particles into the atmosphere.Cu, Sr, and Ba are used to create fireworks coloring effects (blue, red, and green, respectively) (Do et al., 2012).Mg is used as a fuel and a producer of sparks (Lancaster and Shimizu, 1998).CaCl 2 and CaSO 4 are used to produce orange sparks (Moreno et al., 2007).

CONCLUSION
This study provides insights into the chemical compositions of PM 2.5 during the 2015 Spring Festival Period and explores the differences in chemical characteristics of PM 2.5 between non-haze days, haze days and firework days as well as the overall effects of short-term holidays such as the Spring Festival.
In most days during Spring Festival, the PM 2.5 concentration exceeded the NAAQS (GB3095-2012) Grade II standard (75 µg m -3 ) and ranged from 67 to 294 µg m -3 .High PM 2.5 average concentrations were observed during firework days, approximately 2.7 and 1.2 times higher than that on non-haze days and haze days, respectively.Meanwhile, the concentration of SO 2 and NO 2 were both below the national allowable concentrations (150 µg m -3 for SO 2 , 80 µg m -3 for NO 2 ) during the sampling period.The SO 2 concentration during firework days was less than that measured in previous research into the effects of fireworks because of the introduction of sulphur-free propellant technology in civil fireworks.Adapted from: The composition of the continental crust* (Wedepohl, 1995).
SO 4 2-, NO 3 -and NH 4 + were the dominant ions during haze days, while the concentrations of NH 4 + and NO 3 significantly decreased during firework days.The PM 2.5 chemical compositions on firework days were dominated by Cl -and K + , largely exceeding those on non-haze days and haze days.Particulates on firework days were more acidic than those on haze and non-haze days, mainly due to the high concentrations of SO 4 2-.According to enrichment factors of elements in PM 2.5 during the sampling period, we classified 24 elements into crustal elements (Ca, Al, Fe, Na, Co, Ni, P, Ti, and V), common anthropogenic pollution elements (As, Cd, Cu, Pb, S, Sb and Zn) and fireworks elements (Ba, Cr, Cu, Mg, Pb, S, Si, and Zn).This allowed for further investigation of the differences in chemical characteristics between common air pollution and short-term special air pollution (e.g., from fireworks).
Our results demonstrate that the air quality of normal days in holidays was better than that in non-holidays.But burning of fireworks contributes to sudden and dramatic air pollution, which is much more severe than haze pollution in this study.Differences in chemical characteristics indicated that holidays such as the Spring Festival can affect air pollution patterns in two ways: a decrease in the population and vehicles but an increase in activities such as firework displays.Thus, special human activities during holiday periods should be seriously considered when we take measures to improve air quality.

Feb 12 PMFig. 1 .
Fig. 1.Time series of PM 2.5 , SO 2 and NO 2 mass concentrations during the sampling period in Shijingshan.The left, middle and right bars in the Figure denote measured values of PM 2.5 concentration, SO 2 and NO 2 concentrations, respectively, and the dashed line is the national secondary ambient air quality standard for daily PM 2.5 concentration (75 µg m -3 ).

Table 2 .
Enrichment factors of elements in PM 2.5 during the sampling period.