Pollution Properties of Water-Soluble Secondary Inorganic Ions in Atmospheric PM 2 . 5 in the Pearl River Delta Region

Based on the online observation of PM2.5 mass concentration, its water-soluble inorganic ions, and their gaseous precursors during August of 2013 to March of 2014 at the atmospheric supersite in the Pearl River Delta (PRD) region, the inter-action of the secondary compositions and their precursors was discussed, and the pollution properties of the secondary inorganic ions were revealed. During the whole measurement period, the average concentrations of SO4, NO3 and NH4 were 16.6 μg m, 9.0 μg m and 10.2 μg m, respectively, with total contribution to PM2.5 of 55.8%, indicating the significant role of secondary transformation in PM2.5 pollution. The seasonal average total contributions of SO4, NO3 and NH4 to PM2.5 varied from 46.0% to 64.3%, lowest in summer and highest in winter. The contributions of SO4 and NH4 to PM2.5 were relatively stable; while those of NO3 in different seasons were distinct, even dominating PM2.5 in some pollution cases in winter. NH3 was abundant with an annual average concentration of 15.2 μg m, facilitating the neutralization of H2SO4 and HNO3 with the average [NH4]/(2[SO4] + [NO3]) equivalent charge ratio of 1.1. The maximum daily peak concentration of HNO3 was as high as 18.6 μg m, providing an evidence for the strong oxidizing property of the atmosphere in the PRD region. The theoretical equilibrium constant (Ke) of NH4NO3 is always lower than the observed concentration product (Km = [NH3] × [HNO3]) in spring and winter with higher HNO3 concentrations; while in over 60% of the time during summer and autumn, mainly during daytime, Ke was higher. In general, the strong oxidizing property and NH3 played important roles in the fine particle pollution in the PRD region.


INTRODUCTION
Atmospheric aerosols influence our life in many aspects.They impose obvious effects on the global climate change and human health (Dockery et al., 1994;Nel, 2005), and visibility degradation (Sokolik and Toon, 1996;Jung and Kim, 2006).Consequently, more and more attention has been paid to them in recent years.It is well known that aerosols with different compositions and from diverse sources have distinct effects.In order to understand their effects and provide scientifically sound evidence for effective control policies making, accurate knowledge on physical and chemical properties of aerosol is urgently required.Previous studies showed that concentrations and compositions of PM 2.5 vary with seasons, presenting generally high values in winter and low ones in summer in general; and carbonaceous matter and water-soluble ions such as sulfate (SO 4 2-), nitrate (NO 3 -) and ammonium (NH 4 + ) are usually the main contributors.HNO 3 and H 2 SO 4 can be transformed into aerosol by neutralization reactions (Seinfeld and Pandis, 1998).Semi-volatile NH 4 NO 3 is formed via reversible phase equilibrium with NH 3 and HNO 3 (Pio and Harrison, 1987).Such equilibrium between gas-and particle-phase is strongly influenced by ambient temperature (T) and relative humidity (RH) (Mozurkewich, 1993).
The Pearl River Delta (PRD) region is one of the biggest city clusters with extremely invigorating economy and dense population in the world.Rapid urbanization and economic development have been accompanied by the serious deterioration of regional air quality as well as significant changes of air pollution properties: emissions of primary pollutants such as SO 2 and inhalable particulate matter (PM 10 ) have been greatly reduced by certain abatement measures, while formations of secondary products such as ozone and fine particles in high concentrations have become the main issues, especially after the National Ambient Air Quality Standards (GB 3095-2012) was promulgated.The chemical compositions of PM 2.5 and PM 10 in the PRD region have been reported before (e.g., Cao et al., 2004;Hagler et al., 2006;Liu et al., 2008;Yue et al., 2010a;Huang et al., 2014).The gaseous precursors are crucial to investigate their formation mechanisms and behaviors in the atmosphere.However, simultaneous on-line measurements of water-soluble ions of PM 2.5 and related precursor gases (HNO 3 , SO 2 and NH 3 ) are still very rare (Hu et al., 2008), and seasonal variations of them has not been reported in this region.In this study secondary water-soluble inorganic ions, including SO 4 2-, NO 3 -and NH 4 + in PM 2.5 , and the related gaseous pollutants including HNO 3 , NH 3 , SO 2 and NO 2 were measured simultaneously at a regional atmospheric supersite in the PRD region of Guangdong Province (Zhong et al., 2013) during four different seasons.The goal of this study was to understand the inter-action of the secondary water-soluble ions and their gaseous precursors and thus to reveal the pollution properties of the secondary inorganic ions in the PRD region, providing scientific support for designing effective fine particle pollution control strategies.

Observation Site and Periods
The measurements were conducted at the Guangdong Atmospheric Supersite located in Taoyuan Town, Heshan county, Jiangmen city, Guangdong Province (112.929°E,22.728°N; altitude 60 m), about 80 km southwest from Guangzhou urban area, 50 km southwest and 30 km northeast from Foshan and Jiangmen urban area, respectively.The site was heavily influenced by the intense air pollutant emissions from Guangzhou and Foshan area in spring, autumn and winter (Zhong et al., 2013;Peng et al., 2014).
Measurements were conducted in four different periods, i.e., 5 to 24 August (summer), 30 September to 19 October (autumn), and 29 November to 17 December (winter) in the year of 2013, and 7 to 26 March (spring) in 2014, to represent the four seasons, respectively.Correspondingly, the average concentrations of the four periods were taken as annual average values.

Instrumentation
Mass concentrations of PM 2.5 , its water-soluble ions (SO 4 2-, NO 3 -and NH 4 + ) and the corresponding gaseous precursors including HNO 3 , NH 3 , SO 2 and NO 2 were measured on-line at the supersite.
The water-soluble inorganic ions in PM 2.5 and gaseous HNO 3 and NH 3 were detected by a Gas and Aerosol Collector-Ion Chromatography (GAC-IC) system made in Peking University, and the principle and configuration of the GAC-IC system was similar to that introduced in detail by Dong et al. (2012).In the system, a set of dull-polished wet annular denuder with a steam jet aerosol collector was employed for the continuous sampling of both gas-and aerosol-phase pollutants.The inorganic gaseous and particulate species were detected by two separate ion chromatography (IC; ICS-90, Dionex, USA).The IC to measure anions was equipped with a 4 × 25 mm guard column (IonPac AG 14) followed by a 4 × 250 mm analytical column (Ion Pac AS 14) with a mixture of 3.5 mM CO 3 2and 1.0mM HCO 3 -as eluent solution.The other IC for cations was installed with a 4 × 25 mm guard column (Ion Pac CG 12) followed by a 4 × 250 mm analytical column (IonPac CS 12) with 20 mM methane sulfonic acid as eluent solution.Both ICs were electronically suppressed to reduce the background signal.The inlet for the system was set up with a URG PM 2.5 cut-off with a flow rate of about 16.7 L min -1 .The time resolution was 30 minutes, and the measured data was hourly averaged for analysis.The detection limits of gaseous HNO 3 and NH 3 were 0.183 µg m -3 and 0.023 µg m -3 , respectively, and those for SO 4 2-, NO 3 and NH 4 + were 0.159 µg m -3 , 0.034 µg m -3 , and 0.030 µg m -3 .The amounts of the collected liquefied gaseous and particulate pollutants of each sample were recorded to judge the stability of this system, and standard samples were detected before and after each separate measurement period in the four different seasons to ensure the consistence of the data.
PM 2.5 mass concentration was measured by a tapered element oscillating microbalance (Model TEOM 1405), SO 2 by a trace level SO 2 analyzer (Model 43iTLE), NO 2 by a trace level NO/NO 2 /NO x analyzer (Model 42iTL), and O 3 by a UV spectrophotometry O 3 analyzer (Model 49i); these instruments were produced by the Thermo Fisher Scientific, USA.
In addition, particle number size distributions and meteorological parameters including temperature, RH, precipitation, wind speed (WS), wind direction (WD), UVA and visibility were also measured in this research.Particle number size distributions from 3 to 900 nm were monitored by a couple of scanning mobility particle sizers (TSI model 3936, TSI Inc., St. Paul, MN, USA).UVA was detected by a multiband ultraviolet radiation meter (Model UV-S-A-T/UV-S-B-T, Kipp&Zonen B.V., Holland); visibility was measured by a Belfort Model 6000 visibility detection instrument from USA, and the other meteorological parameters were reported by a portable weather station (Model WXT520,Vaisala,Finland).

Meteorological Conditions
Meteorological conditions of the four seasons in the PRD region are distinct.Table 1 shows selected meteorological parameters for four seasons at the supersite.Overall, mean temperature and RH exhibited obvious seasonality with the highest values in summer and the lowest values in winter.Total precipitation was virtually insignificant in autumn while abundant in summer.UVA and atmospheric visibility also peaked in summer, but showed troughs during spring instead of in winter, resulted from a special continuous wet weather called Huinan Weather with very high RH occurred in spring.As expected, the predominant wind direction was northeast except south in summer.In general, higher wind speeds were frequently observed in summer.

General Observations
The annual average mass concentration of PM 2.5 was 64.2 µg m -3 , higher than the secondary grade mass concentration limit of the National Ambient Air Quality Standards (GB 3095-2012) in China (35 µg m -3 ) by 83.4%.The annual average concentrations of SO 4 2-, NO 3 -and NH 4 + were 16.6 µg m -3 , 9.0 µg m -3 and 10.2 µg m -3 , respectively, with total contribution to PM 2.5 (SNA/PM 2.5 ) of 55.8%.It was significantly higher than those measured at other places of China, such as Shangdianzi regional site in North China (Meng et al., 2013), Xi'an (Han et al., 2009), Beijing, Chongqing, and Shanghai (Yang et al., 2011), indicating the important role of secondary transformation played in the PM 2.5 pollution in the PRD region.Among the major gaseous precursors of the water-soluble ions, the annual concentration of SO 2 was 31.6 µg m -3 , meeting the National Ambient Air Quality Standards (GB 3095-2012).However, the annual average concentration of NO 2 was 47.9 µg m -3 in violation of the national standards, indicating the significant influence from vehicle emissions.Due to high chemical activity and its sticky nature, the atmospheric lifetime of HNO 3 was relatively short, leading to low concentration level with annual average of 3.3 µg m -3 , compared with other species.However, the HNO 3 concentrations were higher than those at other places such as in North Carolina, USA of 0.9 µg m -3 (Walker et al., 2006), in Oberbärenburg, Germany of 0.8 µg m -3 (Plessow et al., 2005) and at Hong Kong urban area of 1.3 µg m -3 (Yao et al., 2006).In addition, the daily peak concentrations of HNO 3 were 1.8-18.6µg m -3 , significantly higher than those in Mexico City of 1.4-8.4µg m -3 (Zheng et al., 2008) and at Beijing urban area of < 8 µg m -3 (Yue et al., 2013).Both provided an evidence for the strong oxidizing property of the atmosphere in the PRD region.The annual average NH 3 concentration was 15.2 µg m -3 , obviously higher than that at Shangdianzi regional site in North China of 8.7 µg m -3 (Meng et al., 2013).It suggested that active agricultural production (e.g., composting and N-fertilizer application), livestock feeding, and biological activity imposed important effect on the atmosphere in the PRD region and higher temperature was also probably one of the key factors leading to higher NH 3 concentration at the supersite than that in North China, as industrial process, vehicle emission and biomass combustion contributed a total of less than 10% to NH 3 sources in Guangdong Province (Pan et al., 2014).
As diagrammed in Fig. 1, concentrations of PM 2.5 , SO 4 2and NH 4 + had the same seasonal variation pattern: winter > autumn > spring > summer, and the rank of NO 3 -concentrations was the same as that in Hong Kong (Louie et al., 2005): winter > spring > autumn > summer; all showed highest values in winter and lowest ones during summer.The weather over PRD region during wintertime is influenced by Siberian cold current as East Asian winter monsoon is mainly influenced by Siberian High during this time (Wu and Wang, 2002).The weak to moderate northeasterly winds with increased air parcel residence time over potential source regions in winter in the PRD region (Louie et al., 2005) favored the accumulation, condensation, and transformation of atmospheric pollutants and the strong northerly winds can bring superregional transport pollutants to the PRD region, both leading to high concentrations of PM 2.5 and its chemical components.During summer, a low-pressure trough draws moist warm air inland from the ocean with an increase in precipitation.Frequent precipitation and good dispersion conditions for air pollutants brought good air quality with low concentrations of particles in summer.Two possible reseasons for the higher NO 3 -concentration in spring than autumn which was different from that of PM 2.5 , SO 4 2-and NH 4 + were as follows: (1) Although the intense solar radiation in autumn was favorable for NO 3 -formation through oxidation reaction, the high temperature and low RH pushed the equilibrium of NH 3 (g) + HNO 3 (g) < = > NH 4 NO 3 (s,aq) to produce gaseous NH 3 and HNO 3 , which was confirmed in the subsequent discussions on gas to particle conversion.(2) The lower gaseous precursor concentrations of NO 2 and HNO 3 than those in spring would also lead to lower NO 3 -concentration.HNO 3 showed a consistent seasonal variation pattern (winter > spring > autumn > summer) with NO 2 and NO 3 -, but the average concentrations of HNO 3 in spring (3.7 µg m -3 ) and autumn (3.3 µg m -3 ) were close with a difference of around 10%.The dominant daytime source of HNO 3 is through the oxidation of NO 2 by the hydroxyl radical.HNO 3 deposition is normally considered to be a terminal sink of NO x , but recent studies have indicated that HNO 3 deposited on the surface can be released back into the atmosphere in the form of NO 2 after heterogeneous reactions with NO (Saliba et al. 2001).In addition, HNO 3 •H 2 O complex could be photolyzed into HONO or NO (Ramazan et al., 2006).Solar radiation was a double-edged sword for HNO 3 .In summertime, clean south wind from South China Sea usually dilutes the pollutants over PRD region and improves the air quality and visibility to large extent.The precursors of HNO 3 , NO 2 and OH radical should also be low to produce HNO 3 during the time when southerly wind is prevalent, even though there is existence of strong solar radiation.Similarly, NO 2 was recorded to be lower during the autumn than in spring and winter, unlikely to cause high HNO 3 in the autumn either.
The seasonal variation of NH 3 concentrations was unique in this research with the highest average concentration in spring and the lowest one in winter.It was also different from NH 3 variation mode at the Shangdianzi regional site in North China with highest concentrations occurred in summer, as over there agricultural activity was a major source of NH 3 and the main compositions of the fertilizer used were diammonium phosphate and urea, which are very easy to volatile and produce NH 3 in summer, and the high temperature makes against the reaction of NH 3 and HNO 3 to form particulate NH 4 NO 3 (Meng et al., 2013).NH 3 observation of this work was also different from the one observed by Zheng et al. (2015) in an industry zone of Nanjing, which was heavily influenced by industry emissions.However, the atmospheric temperature in the PRD region is significantly higher than that in North China, and meteorological conditions favoring agricultural production appeared earlier in a year, so the agricultural activity such as composting and fertilization will be more active during spring than summer, resulting in more NH 3 emission.Moreover, the far higher relative humidity in summer is in favor of the formation and existence of NH 4 NO 3 .Affected by such complex factors, the highest seasonal average NH 3 concentrations in PRD appeared in spring followed by that in summer.NH 3 concentrations at the supersite were obviously higher than those in Beijing and at Xinken during the same seasons (Yue et al., 2013;Hu et al., 2008).

Diurnal Variations of the Pollutants
The diurnal variations of the average temperature, RH, wind speed, UVA, PM 2.5 , water-soluble ions, O 3 , HNO 3 , SO 2 and NH 3 during the measuring period are shown in Fig. 2. The lowest temperature occurred in the morning at about 6 o'clock (18.8°C), and the maxima in the afternoon at about 15 o'clock of 24.3°C.The diurnal variation pattern of RH was the inverse of temperature with maxima in the early morning and minima around 14 o'clock.The highest average wind speed was observed at noon of about 2 m s -1 .In contrast, the wind speeds in the early morning were lower.This trend of wind speed diurnal variation was different from that observed at a coastal regional site in the PRD region with higher wind speed in the morning (Hu et al., 2008).The UVA increased quickly after sun rise and showed peak values at noon with obvious decrease in the afternoon followed, and come to nearly zero at night.Overall, the meteorological conditions in the afternoon were characterized by high wind speed, high temperature, and low RH.In addition, the boundary layer height was expected to be higher in the afternoon.Early in the morning, by contrast, it was characterized with low wind speed, low temperature, high RH, and low boundary layer height.That is to say, it was favorable for the dispersion and dilution of the pollutants in the afternoon while beneficial for pollutant accumulation and gas to particle transformation early in the morning.It was the main causes for the fact that PM 2.5 and its inorganic aerosol ions showed lower concentrations in the afternoon and higher ones early in the morning.SO 2 and NH 3 , mainly from primary emission sources, also showed similar diurnal variation trends.In contrast, the diurnal variations of O 3 and HNO 3 were similar to that of UVA, but with later peak time, as they are products of atmospheric photochemistry reactions.

Contributions of Secondary Ions to PM 2.5
The contributions of SO 4 2-, NO 3 -and NH 4 + to PM 2.5 are shown in Fig. 3. Overall, they contributed 25.9%, 14.0%, and 15.9% to PM 2.5 , respectively.Among them, SO 4 2-was the biggest contributor, with a small fluctuation range of 23.9-27.4% in different seasons.The contributions of NH 4 + were 12.7-17.7%,but the fractions of NO 3 -varied significantly, from 6.3% in summer to 19.2% in winter.The total contribution of SNA/PM 2.5 changed from 45.6% to 64.3%, showing high values during spring and winter but low ones in summer and autumn.The reasons for the low fraction during summer and autumn are as follows: (1) Biological activity during these seasons is vigorous and will emit a large amount of volatile and semi-volatile organic compounds, supplying abundant precursors for organic aerosols.(2) The intense solar radiation provides a good condition for the photochemical reactions to produce secondary organic aerosols.As reported that the secondary organic matters accounted for 56% in the organic fine aerosols in Shenzhen in PRD during summer, while in winter the fraction was only 6% (Niu et al., 2006).In short, the increase in organic matter contribution was probably the proximate cause for the obvious decrease in water-soluble inorganic ion contribution to PM 2.5 .The seasonal fraction of SNA/PM 2.5 at the supersite was significantly higher than those in Guangzhou of 36.7% in summer and of 41.3% in winter (Tao et al., 2010;Huang et al., 2014).

Charge Balance of SO 4 2-, NO 3 -and NH 4 +
The mean molar ratio of NH 4 + to SO 4 2-was 3.3, far more than 2, indicating that abundant NH 3 was present to neutralize H 2 SO 4 and NH 4 HSO 4 was negligible.The excess of NH 4 + was inferred to be associated with NO 3 -and Cl -.The annual average equivalence charge ratio of [NH 4 + ]/ (2[SO 4 2-] + [NO 3 -]) at the supersite was 1.2, implying that NH 4 + in PM 2.5 can totally balance SO 4 2-and NO 3 -, and mainly existing in the form of (NH 4 ) 2 SO 4 and NH 4 NO 3 .The average equivalence charge ratios in different seasons were comparable, being 1.1 in summer and winter, and 1.2 in spring and autumn.They were larger than those in Guangzhou, Xinken, Beijing, Shanghai and Xi'an in the range of 0.5-0.9,where NH 4 + cannot balance the entire SO 4 2-and NO 3 -, and a part of NO 3 -may be in the form of KNO 3 (Wang et al., 2006;Zhang et al., 2007;Hu et al., 2008;Tao et al., 2010;Yue et al., 2013).The abundant NH 3 at the supersite was the major cause for the high equivalence charge ratio of

Gas to Particle Conversion
The linear correlation between SO 4 2-and NH 4 + (R 2 = 0.55 or 0.68) was significantly better than that between NO 3 and NH 4 + (R 2 = 0.27 or 0.36) during summer and autumn, respectively; while such linear correlation was comparable during winter and spring, with R 2 of 0.79 and 0.85 for the former and R 2 of 0.75 and 0.84 for the latter.It indicated that the heterogeneous reaction between HNO 3 and NH 3 in the particle-phase to form NH 4 NO 3 probably played a more important role in winter and spring than in summer and autumn.In order to determine if the meteorological conditions, mainly atmosphere temperature and RH, at the supersite favor the formation of NH 4 NO 3 or not, the measured concentration product (K m = [NH 3 ] × [HNO 3 ]) of HNO 3 and NH 3 was calculated using the measured data and then compared with the theoretical equilibrium constant (K e ) calculated according to the method mentioned in (Mozurkewich, 1993).The data were divided into two 2-in the deliquescent aerosol particles reduces K e compared to that of pure NH 4 NO 3 solution (Stelson and Seinfeld, 1982).The ionic strength fraction (γ) of NH 4 NO 3 in NH 4 + /NO 3 -/SO 4 2system can be calculated according to (Stelson and Seinfeld, 1982): And K e * for this system was derived by multiplying K e with γ and was compared with K m .K e (Group 1) or K e * (Group 2) larger than K m suggests that NH 4 NO 3 tends to dissociate and conversely the formation of NH 4 NO 3 is possible.The ratios of K m /K e (or K m /K e *, Ke* was used when RH is larger than the DRH through the paper) by hour at the supersite in four seasons were derived, and categorized into two types according to whether it was below 1.As shown in Table 2, K m was mostly larger than K e (or K e *) during spring and winter; while in over 60% of the time in summer and autumn, K m was less than K e (or K e *).One possible cause was that HNO 3 concentrations were much higher during spring and winter than in summer and autumn.The lower temperature in spring and winter (< 20°C on average) than those in summer and autumn (> 25°C on average) was probably also an important factor.It has been also reported that K e was larger during daytime and smaller at night than K m in autumn at Xinken site (Hu et al., 2008), presenting less NH 3 and NH 4 + but more HNO 3 and NO 3 than those at supersite.

Annual average
Spring Summer Autumn Winter

Case Study
Autumn and winter are the period with heavy air pollution and frequent regional haze in the PRD region; while the comparison between K m and K e implies that the characteristics of particle pollution in the two different seasons will be distinct.Consequently, a complex air pollution episode with both daily PM 2.5 and O 3 concentrations violating the secondary grade concentration limit of the National Ambient Air Quality Standards (75 µg m -3 and 200 µg m -3 , respectively) in autumn and a particle pollution episode with daily PM 2.5 exceeding the secondary grade concentration limit of the same standards but low O 3 concentrations in winter were discussed respectively to investigate the different mechanisms of such episodic cases.

A Complex Pollution Episode in Autumn
Due to the strong solar radiation in autumn, PM 2.5 and O 3 frequently exceeded the national standards even on the same day.The pollution episode from 2 to 6 October 2013 was a typical example (Fig. 4).The daily maximum UVA was 38-50 W m -2 ; wind speed showed a clear diurnal variation which increased during daytime to around 3 m s -1 ; temperature and RH varied in the ranges of 22-31°C and 30-74%, respectively.On the whole, the conditions were not favorable for pollutant accumulation but beneficial for secondary transformation related with atmospheric photochemical reactions.HNO 3 had a similar diurnal variation trend to O 3 , since both of them were produced from atmospheric photochemical reactions, and O 3 provided hydroxyl radical for NO 2 oxidation and thus increased the HNO 3 concentration concurrently.The trends among NO 3 -, NH 4 + and NH 3 were just in opposite.This observation can be explained by the development of the planetary boundary layer as the day progressed.Also, it has been suggested that gas-particle partitioning can play a dominant role in the fate of aerosol ammonium nitrate over a large spatial scale (DeCarlo et al., 2008).During the polluted days from 3 to 5 October, K m /K e was below 1 during daytime, indicating that NH 4 NO 3 tended to dissociate.As clearly illustrated in Fig. 4 that aerosol NO 3 -concentration and the mass ratio of [NO 3 -]/[SO 4 2-] showed increasing trends, and the theoretical equilibrium constant was lower than the product of observed concentrations at night, implying that the formation of NH 4 NO 3 was favored.SNA contributed around 60% to PM 2.5 and sulfate played a more important role in the fine particle pollution.On the days of 2 and 6 October, new particle formation events with burst and obvious growth of ultrafine particles especially significant increase of the nucleation mode particle number concentration (Wu et al., 2007, Yue et al., 2009) were observed with lower concentrations of PM 2.5 .During these days, most of the ratios of K m /K e were below 1 even at night and the mass ratios of [NO 3 -]/[SO 4 2-] kept at low levels.Therefore, it was unlikely that NO 3 -participated in new particle formation.Different fractions of SNA/PM 2.5 was probably due to the fact that the new particle formation event on 2 October was mainly dominated by H 2 SO 4 and NH 3 , while organic compounds might contribute significantly to the event on 6 October (Yue et al., 2010b).In addition, the burst and subsequent growth of ultrafine particles on every single day during this episode especially with high PM 2.5 mass concentrations should be an indicator for the strong atmospheric oxidizing property in autumn of the PRD region.

A Particle Pollution Episode in Winter
A particle pollution episode with daily PM 2.5 concentrations violated the second grade limit of the national standards but low O 3 concentrations was observed during 30 November to 5 December in 2013.Time series of selected parameters during this period were illustrated in Fig. 5.During this episode, the daily maximum UVA was 26-38 W m -2 , significantly lower than that during the complex air pollution episode in autumn, causing obviously lower O 3 concentrations.The wind speed was relative low (< 2 m s -1 ), indicating stagnant atmospheric conditions, which were favorable for the accumulation and transformation of air pollutants.The RH variation range was 21-78%, similar to that during the complex air pollution episode in autumn, while the temperature was significantly lower, in the range of 10-23°C, leading to very low K e and high ratios of K m /K e , and promoting the gas to particle conversion of HNO 3 and NH 3 to form NH 4 NO 3 .HNO 3 also showed high concentrations during the daytime and low ones at night, similar to the trend of O 3 .SNA accounted for about 60% of PM 2.5 .The SO 4 2-concentration was generally higher than that of NO 3 -.However, from the late afternoon on 2 December to early morning on 4 December with low temperature and high RH, concentrations of NO 3 -exceeded SO 4 2-most of the time, concurring with high concentrations of PM 2.5 .The ratios of K m /K e were also higher than 1, with particularly higher values on 3 and 4 December.The quick and effective gas to particle conversion of HNO 3 and NH 3 to NH 4 NO 3 should be the major cause for the higher NO 3 concentrations than SO 4 2-.No new particle formation events were observed during this episode, probably because of the low UVA values and high concentration of PM 2.5 , indicating weak source but strong sink for the newly formed particles and their precursors.

CONCLUSIONS
The concentrations of PM 2.5 , its water-soluble inorganic ions and the related gaseous precursors during four seasons were measured and discussed at the regional atmospheric supersite in the PRD region.The average concentrations of + , SO 4 2-and NO 3 -should be mainly in the form of (NH 4 ) 2 SO 4 and NH 4 NO 3 in PM 2.5 .The high HNO 3 concentrations provided an evidence for the strong oxidizing capability of the atmosphere in the PRD region.
The seasonal average SNA/PM 2.5 contributions varied from 46.0% to 64.3%, lower in summer and autumn, and higher in winter and spring.The contributions of SO 4 2-and NH 4 + to PM 2.5 were relatively stable, which fluctuated in the range of 23.9-27.4% and 12.7-17.7%,respectively; while the contributions of NO 3 -in different seasons were diverse with a large fluctuation range of 6.3-19.2%.During winter and spring, the comparable linear correlations of NO 3 -versus NH 4 + (R 2 = 0.75 or 0.84) and SO 4 2-versus NH 4 + (R 2 = 0.79 or 0.85), the higher ratios of K m /K e , and larger contributions of NO 3 -to PM 2.5 reflected that the heterogeneous reaction between HNO 3 and NH 3 to form NH 4 NO 3 in the particle-phase played a more important role in fine particle pollution.And the fast and effective gas to particle convention of HNO 3 and NH 3 to NH 4 NO 3 can cause even higher concentrations of NO 3 -than SO 4 2-during some pollution cases in winter.
Overall, the strong atmospheric oxidizing property and NH 3 played significant roles in the fine particle pollution in the PRD region, and effective NH 3 control measures will lead to substantial reduction in PM 2.5 mass concentration and the frequency of haze events.

Fig. 1 .
Fig. 1.Concentrations of O 3 , PM 2.5 and the water-soluble ions and their major gaseous precursors during different seasons.

Fig. 2 .
Fig. 2. Average diurnal variations of selected parameters during the whole measurement.

Table 1 .
Selected meteorological parameters in the four seasons.

Table 2 .
Percentages of the K m /K e (or K m /K e *) ratios during different seasons.