Chemical characteristics of inorganic ammonium salts in PM 2 . 5 in the atmosphere of Beijing ( China )

Introduction Conclusions References


Introduction
Atmospheric particulate matter plays an important role in atmospheric visibility reduction, human health effects, acid deposition and climate (Heintzenberg, 1989;Dockery et al., 1993;Charlson and Heintzenberg, 1995;Vedal, 1997;IPCC, 2007). Fine parti- 5 cles, otherwise called PM 2.5 with an aerodynamic diameter less than 2.5 µm, contribute mainly to the above phenomena. The major inorganic particles identified in PM 2.5 are sulphate (SO 2− 4 ), nitrate (NO − 3 ), chloride (Cl − ) and ammonium (NH + 4 ), which typically comprise 25-75 % of atmospheric PM 2.5 mass (Gray et al., 1986;Heitzenberg, 1989). These species are secondary in nature and are formed in the atmosphere by physical 10 processes (nucleation, condensation and evaporation), and/or chemical reactions of precursor gases (photochemical gas phase, oxidation aqueous-phase oxidation and particulate-phase processes), such as sulphuric acid (H 2 SO 4 ), nitric acid (HNO 3 ), hydrochloric acid (HCl) and ammonia (NH 3 ). H 2 SO 4 and HNO 3 are atmospheric oxidation products of gaseous sulphur dioxide (SO 2 ) and nitrogen oxides (NO x ), respectively, 15 while NH 3 is directly emitted into the atmosphere mainly by agricultural source.
Particulate SO 2− 4 is a product of gas to particle phase reactions involving atmospheric oxidation of SO 2 by both heterogeneous and homogeneous processes. In the gas, the oxidation of SO 2 by the hydroxyl radical (OH) produces H 2 SO 4 which condenses to form SO 2− 4 . The majority of aqueous phase reactions with SO 2 occurs in cloud-water, Introduction in urban air (Stockwell et al., 2000). A second frequent path for particulate nitrate formation involves nitric acid attach on the sodium chloride (NaCl) in sea salt particles to generate sodium nitrate (NaNO 3 ) aerosol and to release hydrochloric acid to the atmosphere.
The major source of HCI in the atmosphere is probably coal combustion. Globally, release of HCI from marine aerosols by reaction with less volatile HNO 3 , H 2 SO 4 or SO 2 is of considerable importance. Reaction of NO 2 with sodium chloride under photochemical conditions, emissions from vulcanoes, and refuse incineration may constitute other sources of HCl in the atmosphere. Besides wet and dry deposition, a major sink for HCI will be neutralization by gaseous ammonia to form NH 4 Cl aerosol. 10 The rapid industrial development and urbanization, increased vehicular population and energy consumption in Beijing (39 • 55 N,116 • 23 E), the capital city of China, have led to an increased concentrations of air pollutants, especially in particulate pollution (Yao et al., 2002(Yao et al., , 2003Zhang et al., 2004). Dust-soil, industry emission, coal burning, vehicle exhaust emission and waste incineration have been identified as the major 15 sources of particulate pollution in Beijing. Traffic emissions are considered to be one of the most important sources of sub-micrometer particles in the urban area of Beijing (He et al., 2001;Zheng et al., 2005;Song et al., 2006). Zheng et al. (2005) and Song et al. (2006) indicated that, as a primary source, traffic emissions in Beijing contributed 6-7 % to PM 2.5 concentrations while the respective contribution from road dust resus- 20 pension was estimated to be 7-9 %. However, gaseous pollutants are also emitted by vehicular sources, such as NO x , which are essential for the atmospheric photochemical processes and the gas-to-particle conversions. NO x is also related to the formation of secondary particulate matter such as secondary particulate ammonium, sulphate and nitrate. These particulate species contribute over 35 % of PM 2.5 in Beijing. Coal 25 dominated energy structure is one of the major causes of air pollution in Beijing. Beijing's power plants take about one third of the total coal consumptions emitting 49 % and 27 % of the total SO 2 and NO x emissions, respectively (He et al., 2003), which contribute to formation of atmospheric inorganic fine particles (nitrates and sulphates), Introduction Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | as said above. PM 2.5 mass and chemical compositions in Beijing have been widely studied since the last decade (He et al., 2001;Yao et al., 2002Yao et al., , 2003Sun et al., 2004;Chan et al., 2005Chan et al., , 2008Wang et al., 2005;Duan et al., 2006;Guinot et al., 2006). The sources of particulate matter in Beijing include the local primary emissions, secondary formation 5 and regional transport (Yao et al., 2003;Wang et al., 2005;Chen et al., 2007;Street et al., 2007). Duan et al. (2007) reported that SO 2− 4 , NO − 3 and NH + 4 were the major water soluble ions in wintertime in Beijing, with the average concentrations of 30.8 ± 25.4, 10.8 ± 8.0 and 6.7 ± 5.4 µg m −3 , respectively. In these studies the maximum concentrations appeared on 4 January 1999, 114, 31 and 27 µg m −3 for above three ions, 10 respectively. Yao et al. (2002) found that a large part of SO 2− 4 and NO − 3 in PM 2.5 might be formed through the direct emissions of their precursor gases, such as SO 2 , NO x and NH 3 . However, data on PM 2.5 , especially regarding the semi-volatile species in PM 2.5 , e.g., NO − 3 and Cl − are very limited. In a polluted urban environment, the volatile ammonium salts (NH 4 NO 3 and NH 4 Cl) account for 10-30 % of the fine aerosol mass, 15 and the total inorganic salts account for 25-30 % of the fine aerosol mass. Different field measurements of simultaneous concentrations of NH 3 , HCl, and HNO 3 observed that the concentration products of [HCl] [NH 3 ] and [HNO 3 ] [NH 3 ] were in agreement with theoretical values predicted by thermodynamic equilibrium laws for NH 4 Cl and NH 4 NO 3 formations, respectively, (Harrison and Pio, 1983;Chang et al., 20 1986). In other cases experimental products were different from theoretical predictions mainly at relative humidity below 60 % and above 90 % (Cadle et al., 1982;Allen et al., 1989;Harrison and MacKenzie, 1990;Pio et al., 1992;Harrison and Msibi, 1994;Mehlmann and Warneck, 1995), some of them lower and others higher than theoretical values. They also found that gas-particle equilibrium conditions were not attained 25 instantaneously in the atmosphere at temperatures lower than 15 • C, requiring several minutes for the achievement of the system equilibrium. They attributed the departures of experimental data to unknown kinetic constraints on attainment of the system equilibrium. These kinetic constraints on the evaporation of ammonium containing aerosols Introduction have been restricted to chemical reaction and mass-transfer-limited particle evaporation.
Measurements of semi-volatile fine particulate species, such as inorganic ammonium salts (NH 4 NO 3 and NH 4 Cl), are complicated by the existence of the thermodynamic equilibrium between precursor gases (HCl, HNO 3 and NH 3 ) and relative partic-5 ulate ammonium salts. The concentrations of these fine particulate species will be significantly understimated in urban environments because of the loss of the semi-volatile material from particles collected on the filters during and after sampling, resulting from gas-particle and particle-particle interactions (Sickles et al., 1999a). The understimation of the semi-volatile inorganic ammonium salts will tend to over emphasize the importance of non-volatile fine particulate species such as sulfate. For this reason, there is a need for representative and reliable methods for measuring atmospheric concentrations of the semi-volatile inorganic ammonium salts without disturbing atmospheric equilibrium conditions during sampling. Denuders for absorbing gases prior to particle collection and back-up filters for absorbing HNO 3 , HCl and NH 3 evaporated from col- 15 lected particles have been widely accepted as effective tools to avoid sampling artifacts (gas-particle and particle-particle interactions). In this study, annular denuder and filter pack technique was used to accurately measure inorganic and soluble PM 2.5 without disturbing the partition equilibrium existing in the atmosphere. Following this, as part of the international collaborative research CAREBEIJING (Campaigns of Air Quality Introduction between 10:00 and 12:00 in the morning and on 19 August between 22:00 and 00:00, thus data from these events were not considered. Ambient concentrations of gases and particles were measured using the annular denuder system (Possanzini et al., 1983;Allegrini et al., 1987;Febo et al., 1989;Perrino et al., 1990Perrino et al., , 2001Perrino and Gherardi, 1999;Beine et al., 2001, Ianniello et al., 15 2002. The denuder line configuration used in this study included two sodium fluoride (1 % NaF in 9 : 1 ethanol/water solution) coated denuders for the simultaneous collection of HCl and HNO 3 , followed by two sodium carbonate plus glycerol (1 % Na 2 CO 3 + 1 % glycerol in 1 : 1 ethanol/water solution) coated denuders for the collection of HONO and SO 2 . A fifth denuder was coated with phosphorous acid (1 % H 3 PO 4 20 in 9 : 1 ethanol/water solution) for the collection of NH 3 (Perrino and Gherardi, 1999;Perrino et al., 2001;Ianniello et al., 2010). Downstream of the denuder train a cyclone collected coarse particles (>2.5 µm aerodynamic diameter cut size at flow rate of 15 l/min), while fine particles (<2.5 µm at flow rate of 15 l min −1 ) were collected on a filter pack set in series. The filter pack included a Teflon filter (Gelman Teflon, 47-mm, Introduction NH + 4 as NH 3 , which evolved from the front Teflon filter as a result of ammonium salt dissociations (NH 4 NO 3 and NH 4 Cl) (Possanzini et al., 1992;Masia et al., 1994;Perrino and Gherardi, 1999;Perrino et al., 2001;Ianniello et al., 2010). The evaporation of volatile ammonium salts from the fine particles collected on the front Teflon filter is due to fluctuations in temperature and relative humidity and/or pressure drop across this 5 filter, which perturb the gas-particle equilibrium (Appel et al., 1984;Pathak et al., 2004Pathak et al., , 2009Pathak and Chan, 2005). The dissociation products recovered on the back-up filters are indicated with term of evolved particulate species, estimating their concentrations from the nitrate, chloride and ammonium amounts measured on the back-up filters, as follows: where subscripts N and PA stand for the Nylon filter and phosphorous acid coated paper filter, respectively. The concentration of NH + 4 is the sum of its amounts measured 15 both on the Nylon filter, which absorbs a small fraction of evolved NH 3 (Masia et al., 1994), and on the coated paper filter. The remaining fine particulate species recovered on the Teflon filter are indicated with term of unevolved particulate species. Thus, adding the amounts of Cl − , NO − 3 and NH + 4 found on the Teflon filter (unevolved fine particles) to the respective amounts found on back-up filters (evolved fine particles), the 20 total concentrations of the fine particulate Cl − , NO − 3 and NH + 4 were determined without disturbing the partition equilibrium existing in the atmosphere between gaseous NH 3 and the particulate components (NH 4 Cl and NH 4 NO 3 ), as follows: The fine particulate SO 2− 4 is not volatile and, therefore, its concentration was taken as that on the Teflon filter only.
Thus, the particulate chloride, nitrate, sulphate, sodium, ammonium, potassium, magnesium and calcium in the coarse and fine fractions were measured. After sampling, the denuders and filters were extracted and samples were analyzed within 24-h 5 at the PKU laboratories by using Ion Chromatography (IC) (Dionex DX 120 connected with autosampler DX AS50 for anions and DX ICS90 connected with autosampler DX AS40 for cations).
The sampling flow rate was 15 l min −1 . The sampling volume was about 23.10 m 3 for a 24-h sampling period. In these operative conditions the collection efficiency for all 10 species was higher than 99 %. The detection limits on 24-h measurement period were 0.020 µg m −3 for NH

Temporal variation
Summary statistics for fine particulate Cl − , NO − 3 , SO 2− 4 , NH + 4 , Na + , K + , Mg 2+ , and Ca 2+ , and gaseous HNO 3 and HCl determined by means of the annular denuder technique during the winter and summer periods at Beijing are listed in Table 1. 5 The temporal patterns of these measured fine particulate and gas-phase species during the two sampling periods are reported in Fig. 1. The temporal variations are evaluated applying the paired t-test to determine the significance at the 0.05 level (p ≤ 0.05) of differences among the mean of components by examining the amount of variation between the samples. Fine particulate Cl − , NH + 4 and SO 2− 4 exhibited dis-H 2 SO 4 and HNO 3 , which is consistent with the observation that the measured particulate species seemed to be neutralized (see below). In addition, the high relative humidity conditions (daily mean 35-90 %) in summer period might dissolve a significant fraction of HNO 3 and NH 3 in humid particles, therefore enhancing fine particulate NO − 3 and NH + 4 in the atmosphere (Hesterberg et al., 1996;Krupa, 2003;Trebs et al., 20 2004Trebs et al., 20 , 2005Guinot et al., 2007;Sciare et al., 2007;Hu et al., 2008;Pathack et al., 2009Pathack et al., , 2011Ianniello et al., 2010;Sun et al., 2010). As shown in Fig. 1 (Sickles, 1999b;Walker et al., 2000Walker et al., , 2004Whitall and Paerl, 2001;Robarge 5 et al., 2002;Street et al., 2007;Ianniello et al., 2010), as shown in Table 1 and in Fig. 1. The temporal variations of fine particulate NH + 4 basically coincided with those of fine particulate SO 2− 4 , NO − 3 and Cl − , indicating that NH + 4 largely originated from the neutralization between ammonia and acidic species (Wang et al., 2006). The very similar patterns of sulfate, nitrate, chloride and ammonium suggest that they were likely 10 internally mixed and came from similar gas-to-particle processes (Huang et al., 2010).
The highest concentrations of particulate SO 2− 4 during the summer period at Beijing were attributed to the higher temperatures and solar radiation, which enhance the photochemical activities and the atmospheric oxidation (higher concentrations of ozone, hydrogen peroxide and hydroxyl radical) and, thus, the oxidation rate of SO 2 to partic-15 ulate sulphate (Puxbaum et al., 1993;Robarge et al., 2002;Bari et al., 2003;Gupta et al., 2003;Plessow et al., 2005;Wang et al., 2005;Duan et al., 2006;Street et al., 2007). Sulphate showed also a strong correlation with fine particulate K + , the most abundant cation after NH + 4 , during the winter (R 2 = 0.95, p < 0.001) and summer periods (R 2 = 0.80, p = 0.001), indicating the presence of potassium sulphate (K 2 SO 4 ). 20 Agricultural biomass burning and its impacts on the regional brown hazes in northern China, especially in Beijing, were reported by previous studies (Duan et al., 2004;Li et al., 2010). They showed that the agricultural biomass burning consisted mostly of fine potassium salts (K 2 SO 4 , KNO 3 and KCl), and they were internally mixed. and summer periods in Beijing. Thus, the evaluation of the chemical composition of fine particulate species collected on annular denuder/filter pack system can be used to calculate the NH + 4 neutralization factor, defined as the molar ratio of NH + 4 to the theoretical one, assuming complete conversion to (NH 4 ) 2 SO 4 , NH 4 NO 3 and NH 4 Cl. Thus, the ionic composition of PM 2.5 has been evaluated with regard to the balance between During the winter period the strong correlation between µmol m −3 concentrations of fine particulate NH + 4 and µmol m −3 concentrations of fine particulate SO 2− 4 (R 2 = 0.85, p = 0.001) collected on the Teflon filters and a slope of this regression line of 3.82, 10 higher than 2, suggested the complete neutralization of H 2 SO 4 by NH 3 and a predominance of particulate (NH 4 ) 2 SO 4 during the winter period (Possanzini et al., 1999;Walker et al., 2004). Thus, the excess of NH + 4 was inferred to be associated with fine particulate Cl − and NO − 3 . In addition, the ionic composition of PM 2.5 has been also evaluated with regard to the balance between the major anions (NO − 3 , SO 2− 4 , Cl − ) and 15 ammonium. Indeed, the µmol m −3 concentrations of fine particulate ammonium compared to the sum of the µmol m −3 concentrations of fine particulate nitrate, chloride and sulphate collected on Teflon filters ( Fig. 2) improved the correlation between ammonium and measured anions (R 2 = 0.98, p < 0.001). The correlation showed an ion charge balance with a slope of 1.00, confirming that sufficient NH 3 was present to com-20 pletely neutralize all acidic components (H 2 SO 4 , HNO 3 and HCl) to form (NH 4 ) 2 SO 4 , NH 4 NO 3 and NH 4 Cl. The scattering of some data could be attributed to the minor influences of other atmospheric processes, such as the formation of non volatile nitrate and chloride salts (e.g., NaNO 3 , KNO 3 , NaCl, KCl). In addition, our denuder data showed that on average 33.05 %, 23.65 % and 34.71 % of the fine particulate nitrate, chloride 25 and ammonium, respectively, evaporated from the Teflon filters during the winter period at Beijing (Fig. 3). Indeed, a comparison between µmol m −3 amount of NH + 4 with sum of µmol m −3 amounts of anions (NO good and significant correlation between these ions (R 2 = 0.90, p < 0.001). The slope of this regression line was 1.65, indicating an excellent ion balance and, thus, the presence on Nylon filters of NH 4 NO 3 and NH 4 Cl as constituent of the total fine particles. In addition, using the approach of Ansari and Pandis (1998) and of Blanchard et al. (2000) it is possible to calculate the excess NH 3 , which represents the quantity of NH 3 5 that would remain after complete neutralization of available aerosol and acid gases, is defined as: where all concentrations are in units of µmol m −3 . Our results showed that excess NH 3 existed with a mean value of 0.30 µmol m −3 (5.10 µg m −3 ) during the winter period, in-10 dicating that the atmosphere contained sufficient ammonia to form ammonium nitrate and chloride. Furthermore, there might be also the presence of other nitrate and chloride containing species, which are less volatile than NH 4 NO 3 and NH 4 Cl. Hence, a part of the fine particulate nitrate and chloride might be bound as a relatively non-volatile salts of Ca 2+ , Mg 2+ , K + or Na + . Therefore, we have calculated the free nitrate and 15 chloride concentrations (Hildemann et al., 1984;Gaidajis, 2002), defined as the fractions of nitrate and chloride in excess which are not bound with the alkali or alkaline earth metals, expressed in µmol m −3 , from data collected on the Teflon filters during the winter period:  (Duan et al., 2004;Wang et al., 2005Wang et al., , 2008. However, the observed K + concentrations were low (mean values of 1.07 µg m −3 ) and might not play a significant role in the formation of these particles in Beijing during winter period Behera and Sharma, 2010 by NH 3 and a predominance of (NH 4 ) 2 SO 4 aerosol during the summer season. In addition, the molar concentrations of fine particulate ammonium compared to the sum of the molar concentrations of fine particulate nitrate, chloride and sulphate collected 15 on Teflon filters (Fig. 2) showed a very strong correlation (R 2 = 0.90, p < 0.001) and an ion charge balance (Possanzini et al., 1999;Walker et al., 2004). The slope of this regression line was higher than unity (1.53), confirming that abundant NH 3 was present to neutralize the acidic components (H 2 SO 4 , HNO 3 and HCl) to form (NH 4 ) 2 SO 4 , NH 4 NO 3 and NH 4 Cl. Indeed, our results show also that excess NH 3 exists 20 with a mean value of 1.60 µmol m −3 (27.20 µg m −3 ) during the sampling period, indicating that ammonium salts formation occurred. In addition, our denuder data showed that on average 83.33 %, 74.64 % and 53.07 % of the fine particulate nitrate, chloride and ammonium, respectively, evaporated from the Teflon filters during the summer period at Beijing (Fig. 4) and might not play a significant role in the formation of these particles in Beijing during summer period.
These results indicate that NH 4 NO 3 , NH 4 Cl, (NH 4 ) 2 SO 4 , K 2 SO 4 and KNO 3 were formed preferentially in Beijing during winter and summer seasons.
Following similar studies on equilibria of inorganic trace gases with related ionic 10 aerosol compounds, we have tested if thermodynamic equilibrium was attained for the pure NH 3 /HNO 3 /NH 4 NO 3 (HNO 3(g) + NH 3(g) ↔ NH 4 NO 3(s or aq) ) and NH 3 /HCl/NH 4 Cl (HCl (g) + NH 3(g) ↔ NH 4 Cl (s or aq) ) systems under the meteorological conditions in Beijing during the winter and summer periods. The theoretical equilibrium dissociation constants for solid NH 4 NO 3 (K n ) and for solid NH 4 Cl (K c ) were calculated and com-15 pared to the measured concentration products of NH 3 and HNO 3 (K mn = [HNO 3 ][NH 3 ]) and of NH 3 and HCl (K mc = [HCl][NH 3 ]). The measured concentration product K mn in units of ppb 2 was determined using the measured data for 24-h sampling time while the theoretical equilibrium constant K n in units of ppb 2 was calculated according to the method in Stelson and Seinfeld (1982a), when the ambient relative humidity (RH) is 20 less than relative humidity of deliquescence (RHD) at any absolute temperature (T ) recorded in Kelvin (Hammer and Wu, 1972): ln(RHD) = 723.7/T + 1.7037 The deliquescence relative humidity (DRH) (62 % RH for pure NH 4 NO 3 , 298 K) varies 25 with air temperature and can be modified by particle size and composition under real ambient conditions (Seinfeld and Pandis, 1998 above that of deliquescence, NH 4 NO 3 is modeled in the aqueous state. During the winter the ambient RH was always lower than RHD (Fig. 3), thus, NH 4 NO 3 will exist in equilibrium with the solid phase. Instead, during the summer the ambient RH was not always lower than RHD (Fig. 4), thus, NH 4 NO 3 will be also in equilibrium with the aqueous phase and deliquescent particles. The relantionship between K mn and 5 K n , when RH was lower than RHD, is shown in Fig. 5 for winter and summer periods. The solid line indicates the theoretical equilibrium constant K n for solid NH 4 NO 3 calculated as a function of temperature (Stelson and Seinfeld, 1982a). Concentration products K mn are plotted on a logarithmic scale against a reciprocal average absolute temperature (1000/T ) and compared with the constant dissociations for both seasons.
10 1000/T values varied from 3.6 to 3.7 K −1 and from 3.3 to 3.4 K −1 in the winter and summer, respectively, during the 24-h sampling periods in both seasons. In order for solid NH 4 NO 3 to be formed in the atmosphere, the concentration products of HNO 3 and NH 3 must exceed the theoretical equilibrium dissociation constant K n (Stelson and Seinfeld, 1982a). 15 According to this, during the winter period the results of these thermodynamic calculations ( Fig. 5 and Table 2) showed that the measured concentration products were higher than the theoretical predictions at low temperatures (Walker et al., 2006), with the exception of two days, 3 and 5 in February, where the K mn are lower than K n . Thus, almost on all days NH 4 NO 3 would be expected to be formed at Beijing site under winter 20 conditions, in agreement with other results (Harrison and Pio, 1983;Allen et al., 1989;Harrison and Mackenzie, 1990;Pio et al., 1992;Possanzini et al., 1992;Mehlmann and Warneck, 1995;Danalatos and Glavas, 1999;Gupta et al., 2003). This could be due to the presence of excess NH 3 even during the winter.
During the summer period, the results of thermodynamic calculations (Fig. 5) showed 25 that almost on all days NH 4 NO 3 would not be expected to be formed at Beijing site under summer conditions. The measured concentration product values are below the thermodynamically predicted dissociation constants ( jing in the summer period (Fig. 1), as showed by other studies (Hildemann et al., 1984;Yao et al., 2002;Gupta et al., 2003;Trebs et al., 2005;Pathak et al., 2009;Guo et al., 2010). This disagreement can be understood by the fact that the theoretical predictions are achieved on the assumption that ammonium nitrate and chloride exist as pure solids or liquid particles. However, previous studies (Wu et al., 1987;Mat-5 sumoto and Tanaka, 1996;Zhang et al., 2000) confirmed that, for deliquescent particles, most of the fine particulate nitrate exists as an internal mixture with sulphate, so that HNO 3 can easily be absorbed into the droplets. This coexistence of sulphate considerably reduces the thermodynamic dissociation constant K n for NH 4 NO 3 . Thus, the fine particulate nitrate can be formed from HNO 3 and NH 3 through heterogeneous 10 reactions on fully neutralized fine particulate sulphate, which is abundantly present in an urban area. Indeed, a strong and significant correlation between sulphate and nitrate was observed (R 2 = 0.80, p < 0.001), and a high concentrations of sulphate and nitrate were found at high levels of RH (35-83 %). Indeed, good and significant correlations between sulphate and RH (R 2 = 0.40, p = 0.002), and nitrate and RH (R 2 = 0.43, 15 p < 0.001) were observed. These results can be interpreted as nitrate being produced on preexisting sulfate aerosols, which could provide sufficient surface area and aerosol water content for the heterogeneous reactions. In this case, at high RH, the amounts of the gaseous precursors, such as NH 3 and HNO 3 , have relatively less influence on the formation of the fine particulate nitrate (Zhang et al., 2010;Markovic et al., 2011). 20 Accordingly, to include the coexistence of SO 2− 4 , the NH 4 NO 3 ionic strength fraction Y = [NH 4 NO 3 ]/([NH 4 NO 3 ] + 3[(NH 3 ) 2 SO 4 ]) was calculated according to Stelson and Seinfeld (1982c) (Yao et al., 2003;Gupta et al., 2003;Trebs et al., 2005;Guo et al., 2010). Then, the new equilibrium dissociation constant K n * was derived by multiplying K n with Y , providing a K n * value significantly lower than K n at 25 • C (Fig. 6), with the exception of three days, 6, 9 and 11 in August, where K n * was higher than K mn . This, almost on all days An other possible explanation can be provided by recent studies on the solid phase changes of solid NH 4 NO 3 , leading to appreciable NH 4 NO 3 concentrations in the troposphere even at high temperatures (> 30 • C). Furthermore, potassium ions, such as 5 KNO 3 , were found to co-exist with fine particulate NH 4 NO 3 , and they changed the phase transition behaviors of solid NH4NO 3 . The presence of potassium ions into particulate NH 4 NO 3 widened its stable temperature range (32-84 • C) Wu and Chan, 2008).
These results indicate that both (NH 4 ) 2 SO 4 and NH 4 NO 3 are formed in the urban of Beijing during summer season. As far as the formation of solid NH4Cl is concerned, the same procedure was applied. It is well known that the thermodynamic equilibrium conditions for the formation of NH 4 NO 3 and NH 4 Cl aerosols are similar and depend on humidity and temperature, with NH 4 Cl showing a volatility 2-3 times higher than that of NH 4 NO 3 (Stelson and 15 Seinfeld, 1982a;Pio and Harrison, 1987;Casimiro and Nunes 1992;Matsumoto and Tanaka, 1996). At humidity lower than 75-85 % the particulate NH 4 Cl exists in the solid phase in equilibrium with the gaseous products. The theoretical equilibrium constant K c in units of pbb 2 for solid NH 4 Cl was calculated by Pio and Harrison (1987), when the ambient relative humidity is below the respective deliquescence relative humidity 20 (DRH): Thus, Fig. 7 presents the measured concentration products, K mc , for both winter and summer periods in Beijing, compared with K c . The measured concentration products 25 are plotted on a logarithmic scale against a reciprocal average absolute temperature. Since HCl is more volatile than HNO 3 , K c for NH 4 Cl was higher than the theoretical value for NH 4 NO 3 (Table 2), as seen in previous papers (Allen et al., 1989). The

Tables Figures
Back Close

Full Screen / Esc
Printer-friendly Version

Interactive Discussion
Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | behaviour of ammonium chloride is also similar to that ammonium nitrate, thus K mc exhibited a similar behaviour to K mn during both winter and summer periods. During the winter period, the results of these thermodynamic calculations (Fig. 7) showed that almost on all days NH 4 Cl will be formed in the winter because the measured [NH 3 ][HCl] products are above the predicted equilibrium constant for NH 4 Cl, as 5 seen for NH 4 NO 3 . Some of the data points in Fig. 11 of 28 and 31 January and of 3, 5, 10 and 11 February appeared to indicate an insufficient gas phase concentration product to form NH 4 Cl aerosol. Instead, our denuder data showed the presence of fine particulate chloride during these days.
During the summer period, the results of thermodynamic calculations (Fig. 7) showed that on all days NH 4 Cl would not be formed (Allen, 1989;Matsumoto and Tanaka, 1996). In fact, the measured concentration product values are below the thermodynamically predicted equilibrium constants (Table 2). Instead, our denuder data are in disagreement compared with theory because our denuder data showed that, on average, more than 70 % of the fine particulate chloride is recovered on back-up filters 15 ( Fig. 4) in the summer season indicating the presence of NH 4 Cl as constituent of atmospheric particulate chloride. Thus, ammonium chloride was also generated in the high temperature range, despite the fact that these particles should be volatized according the thermodynamic predictions. This disagreement can be understood taking into account the same interpretation for NH 4 NO 3 in the summer concerning the inter-20 nal mixture of the volatile ammonium salts (Wu et al., 1987;Matsumoto and Tanaka, 1996). Indeed, a strong and significant correlation between sulphate and chloride was observed (R 2 = 0.75, p < 0.001), and a high concentrations of sulphate and chloride were found at high levels of RH (35-83 %). Indeed, a good and significant correlation between fine particulate chloride and RH (R 2 = 0.35, p < 0.001) were observed. In 25 this case, the amount of the gaseous precursors has relatively less influence on the formation of the fine particulate chloride. This will minimise the thermodynamic constant dissociation K c for NH 4 Cl which can be generated from HCl and NH 3 through heterogeneous reactions on neutralized sulphate particles. Introduction These results indicate that NH 4 Cl is formed in the urban atmosphere of Beijing during winter and summer seasons.

Winter
The diurnal variations of fine particulate Cl − , NO − 3 , SO 2− 4 , and gaseous HNO 3 , HCl and 5 NH 3 along with meteorological parameters (temperature, T , relative humidity, RH, wind speed and direction) during the intensive winter measurements are reported in Fig. 8. The weather conditions during this study were described in detail in Ianniello et al. (2010). Both the temperature and relative humidity exhibited strong diurnal variation patterns during the entire sampling period. The mean temperature was 6.67 • C and in- The ammonium concentrations followed the chloride, nitrate and sulphate concentrations in time (Fig. 8) with higher peaks between 08:00 and 10:00, indicating the presence of NH 4 Cl, NH 4 NO 3 , and ammonium sulphate salts. In addition, the similar southeasterly and southwesterly wind directions (60 %) on 9 February, in combination with higher NO x concentrations and relative humidities, and lower temperatures during this time period, resulted in higher peaks of all four species between 08:00 and 10:00 (Ianniello et al., 2010). Obviously, NH + 4 was strongly influenced by its gas phase precursor, such as NH 3 , peaking between 08:00 and 10:00 in the morning and reveal-20 ing much higher concentrations than the anionic species. Possible evaporation of NH 3 from wet surfaces at sunrise, when relative humidities were still high might have caused a significant fraction of gaseous ammonia to dissolve in still humid particles, therefore enhancing particulate NH + 4 (Trebs et al., 2004, Ianniello et al., 2010. In addition, thermodynamic calculations obtained by 2-h samplings over the 9-10 25 February in winter showed the same results to that obtained during 24-h samplings. The only difference was that the measured concentration products and the equilibrium constants tend to be higher on the average than those derived from longer sampling periods ( and 0.60 ppb 2 , respectively, during the intensive winter measurements. As discussed above, the atmosphere is ammonia rich in Beijing during winter period and, thus, ammonium nitrate and chloride formations were permitted.
To gain an insight into the impact of regional transport on NH 3 at Beijing, 24-h

Summer
The diurnal variations of fine particulate species Cl − , NO − 3 , SO 2− 4 and NH + 4 , and gaseous HNO 3 , HCl and NH 3 , along with meteorological parameters (temperature, 20 T , relative humidity, RH, wind speed and direction, and natural radioactivity) during the intensive summer measurements are reported in Fig. 9.
The weather conditions during this time period were described in detail in Ianniello et al. (2010). Both the temperature and relative humidity exhibited strong diurnal variation patterns during the entire sampling period. The temperature ranged between 25 23.86 • C and 33.22 • C reaching maxima between 12:00 and 14:00, while the relative humidity varied from 40 % to 80.77 % peaking between 04:00 and 06:00. The wind ACPD 11,2011  possibile explanation for the detection of high concentrations of fine particulate NO − 3 in August is that there was abundant NH 3 to neutralize H 2 SO 4 and HNO 3 and HCl, which is consistent with the results obtained from our measurements on the predominance of overall neutralized particles, such as (NH 4 ) 2 SO 4 (see Sect. 3.1). In addition, the particulate matter in Beijing was likely humid due to high RH conditions, which might 15 favor the absorption of NH 3 and HNO 3 (He et al., 2001;Yao et al., 2003;Chan and Yao, 2008;Sun et al., 2010).
All four measured species showed broadly similar patterns but only fine particulate Cl − , NO − 3 and NH + 4 exhibited significant diurnal variations (p = 0.030, p = 0.021 and p = 0.025, respectively) in the summer in Beijing (Fig. 9). In addition, the similar di-20 urnal variations of these measured species are consistent with the strong correlations between NH + 4 and Cl − , NO − 3 and SO 2− 4 . Indeed, the correlation coefficients were 0.80 (p = 0.001) for NH + 4 -Cl − , 0.85 (p < 0.001) for NH + 4 -NO − 3 , and 0.90 (p < 0.001) for NH + 4 -SO 2− 4 . The variation of NH + 4 coincided with those of SO 2− 4 , NO − 3 and Cl − , indicating that NH + 4 largely originated from the neutralization between ammonia and acidic species. 25 NH + 4 concentration peaked at around 08:00 due to its accumulation under humid conditions within the stable nocturnal boundary layer between 06:00 and 08:00. This can be explained by a significant fraction of NH 3 dissolved in humid particles under high  (Stelson and Seinfeld, 1982b,c;Baek and Aneja, 2004). The observed diurnal pattern of fine particulate NO − 3 also peaked at around 08:00 within a thermally stable nocturnal boundary layer. After 08:00, NO − 3 concentrations dropped and remained at low levels between 12:00 and 18:00 due to the gas to particle partitioning of NH 4 NO 3 precursors (HNO 3 and NH 3 ), which is favored by 5 the lower temperature and high RH during night and early morning. In addtion, the low concentrations of fine particulate NO − 3 were also due to when the convective mixing of the atmosphere occurred between the late morning (12:00-13:00) of previous day and the early morning (03:00-04:00) of the subsequent day (Fig. 9). The NO − 3 concentration peaked at around 08:00 as a result of the accumulation under high RH conditions. 10 As for NH + 4 and NO − 3 , after 08:00, the observed diurnal pattern of fine particulate Cl − showed a decrease trend during daytime due to increase of the boundary layer height and dissociation of NH 4 Cl at high temperatures (Fig. 9).
Fine particulate SO 2− 4 concentrations did not show a pronounced diurnal variation as that of other fine particulate species (NH throughout day and night. This might be be explained by a more regional formation originating from distant sources. Indeed, high sulphate concentrations occurred between 08:00 and 14:00 in combination with high RH conditions and high peaks of NO x and PM 2.5 , supporting the hypothesis that the traffic was also an important mobile source of fine particulate matter. These results are consistent with the good correla-20 tions between SO 2− 4 and RH, NO x and PM 2.5 . Indeed, the correlation coefficients were 0.52 (p = 0.001) for SO 2− 4 -RH, 0.35 (p = 0.001) for SO 2− 4 -NO x , and 0.50 (p < 0.001) for SO 2− 4 -PM 2.5 . In addition, thermodynamic calculations obtained by 2-h samplings over the 17-21 August in summer showed the same results to that obtained during 24-h samplings. 25 The only difference was that the equilibrium constants tend to be higher on the average than those derived from longer sampling periods ( Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | and the mean values of K mc and K c for NH 4 Cl were 18.72 ppb 2 and 267.01 ppb 2 , respectively, during the intensive summer measurements. As discussed above, even if the atmosphere is ammonia rich in Beijing during summer period, ammonium nitrate and chloride would not be expected to be formed but the meteorological conditions favoured their formations at Beijing site in summer period. 5 As in winter, to identify the impact on the regional transport of air masses on the sampled pollutants, 24-h backward trajectories were calculated for summer intensive measurements in Ianniello et al. (2010). The dominant transport of air masses in the summer period originated in the south (53 %) and southeast (15 %) of Beijing, coinciding with the direction of Hebei province and Tianjing municipality, which are highly 10 industrialized and polluted areas. These polluted air masses from southern directions arrived slower (1.63-3.85 m s −1 ) and had much time to accumulate air pollutants in the Beijing area. This suggests that the high concentrations of fine particulate species, such as NH + 4 , NO − 3 and SO 2− 4 , were due to the impact of regional sources in Beijing during the summer period. Thus, the presence of high concentration of primary precursors in the southern regions, such as NH 3 , SO 2 and NO x , led to the high particulate ammonium, sulfate and nitrate concentrations in the atmosphere of Beijing. The indication that the areas to the south of Beijing are major sources of particulate matter for Beijing have been reported in other studies (Xia et al., 2007;Streets et al., 2007;Wehner et al., 2008;Zhao et al., 2009;Wu et al., 2009). In addition, the urban area it-20 self is a major source for traffic emission. Indeed, the local wind speeds, coming mainly from south and northwest of Beijing, arrived also slowly (0.02-3.83 m s −1 ), especially between 06:00 and 10:00 (0.02-1.30 m s −1 ) during all days, in combination with northwesterly wind direction, boundary layer variations and daytime traffic emissions in rush hours such as with highest concentrations of NO x and PM 2.5 between 06:00 and 10:00 25 (morming rush hour) for all days. Indeed, in Beijing 74 % of ground NO x originates from vehicular emissions Xu et al., 2011), while PM 2.5 accounted for 90 % of total PM emissions from vehicle exaust emissions (Zheng et al., 2005). The observed morning peaks of NO x and PM 2.5 were due to enhanced anthropogenic activity during rush hours and this suggests that the atmosphere of Beijing received regional polluted air from southern urban regions on locally produced particulate species.
These results indicate that reducing the concentrations of precursor gases, such as NH 3 and NO x , could be an effective method for alleviating secondary inorganic PM 2.5 pollution in the urban atmospheres as in Beijing.