Spatial and seasonal variability of PM 2 . 5 acidity at two Chinese megacities : insights into the formation of secondary inorganic aerosols

Aerosol acidity is one of the most important parameters influencing atmospheric chemistry and physics. Based on continuous field observations from January 2005 to May 2006 and thermodynamic modeling, we investigated the spatial and seasonal variations in PM 2.5 acidity in two megacities in China, Beijing and Chongqing. Spatially, PM2.5 was generally more acidic in Chongqing than in Beijing, but a reverse spatial pattern was found within the two cities, with more acidic PM2.5 at the urban site in Beijing whereas the rural site in Chongqing. Ionic compositions of PM2.5 revealed that it was the higher concentrations of NO−3 at the urban site in Beijing and the lower concentrations of Ca2+ within the rural site in Chongqing that made their PM2.5 more acidic. Temporally, PM2.5 was more acidic in summer and fall than in winter, while in the spring of 2006, the acidity of PM2.5 was higher in Beijing but lower in Chongqing than that in 2005. These were attributed to the more efficient formation of nitrate relative to sulfate as a result of the influence of Asian desert dust in 2006 in Beijing and the greater wet deposition of ammonium compared to sulfate and nitrate in 2005 in Chongqing. Furthermore, simultaneous increase of PM 2.5 acidity was observed from spring to early summer of 2005 in both cities. This synoptic-scale evolution of PM 2.5 acidity was accompanied by the changes in air masses origins, which were influenced by the movements of a subtropical high over the northwestern Pacific in early summer. Finally, the correlations between [NO−3 ]/[SO 2− 4 ] and [NH + 4 ]/[SO 2− 4 ] suggests that under conditions of high aerosol acidity, heterogeneous reactions became one of the major pathways for the formation of nitrate at both cities. These findings provided new insights in our understanding of the spatial and temporal variations in aerosol acidity in Beijing and Chongqing, as well as those reported in other cities in China.


Introduction
Acidic aerosols can increase the risks to human health by direct inhalation and indirectly by activating hazardous particulate materials (Amdur and Chen, 1989;Health Effects Institute, 2002).Wet/dry deposition of acidic aerosols also lead to severe degradation of ecosystems (Larssen et al., 2006).Most acidic aerosols are hygroscopic, and as such act to reduce atmospheric visibility (Watson, 2002) as well as disturbing the radiative balance of the atmosphere (Boucher and Anderson, 1995;Crumeyrolle et al., 2008).They are also of great importance to atmospheric chemistry through their influence on many heterogeneous reactions and the behaviors of reactants and oxidants (Seinfeld and Pandis, 1998;Jang et al., 2002).Aerosol acidity can also affect the solubility of iron and phosphorus in the atmospheric aerosols (Meskhidze et al., 2005;Shi et al., 2011;Nenes et al., 2011), which has Published by Copernicus Publications on behalf of the European Geosciences Union.
important implications for ocean biogeochemistry and global climate change (Jickells et al., 2005).
Acidic aerosol species in cities are usually dominated by sulfate (SO 2− 4 ) and nitrate (NO − 3 ), mostly converted from the precursors SO 2 and NO x , respectively, and are partly or fully neutralized by ammonium (NH + 4 ) and basic cations such as Ca 2+ and Mg 2+ .Na + and Cl − may also be important species influencing aerosol acidity in coastal area where sea salt plays a role.Aerosol acidity cannot be directly measured due to its low water content (Meng et al., 1995;Nenes et al., 1998), and is generally assessed using three different kinds of parameters, namely, strong acidity, ion-balanced acidity and in situ acidity.
Strong acidity, measured from the aqueous extracts of aerosol samples, represents the absolute acidity of the aerosols, but it cannot show any in situ characteristics due to the large excesses of water (Pathak et al., 2004).Ionbalanced acidity refers to the estimation of H + concentration by subtracting the equivalent cations, other than H + , from anions (Zhang et al., 2007a).It is more widely used in a relative way to indicate the neutralizing level with the equivalent ratio of cations/anions (Adams et al., 1999;Zhang et al., 2002Zhang et al., , 2007a;;Chu et al., 2004;Sun et al., 2010;Johansen et al., 1999;Takami et al., 2007;Chou et al., 2008).In situ aerosol acidity, in the form of the concentration of free H + or pH in the deliquesced particles at the ambient condition, is most likely to influence the chemical behavior of aerosols.It can be estimated from a variety of thermodynamic models, such as E-AIM, SCAPE and GFEMN (Pathak et al., 2004(Pathak et al., , 2009;;Yao et al., 2006;Takahama et al., 2006;Zhang et al., 2007a).However, it should be noted that ion-balanced and in situ aerosol acidity are empirical approaches that both depend on the choice of ion species.For example, due to their low abundance relative to ammonium in fine particles, basic Ca 2+ and Mg 2+ are usually ignored in the estimation of aerosol acidity, which might be less appropriate during dust events (Ziemba et al., 2007).
The characteristics of aerosol acidity may vary from region to region due to the spatiotemporal variability in the emission of primary aerosols and gaseous precursors, as well as regional differences in the climatic driving forces.The earliest observations on aerosol acidity in China were initiated in the 1980s in regions in the south and southwest (Huang et al., 1988;Shen et al., 1992;Zhao et al., 1994), but they generally focused on the acidification of fog and cloud in respect to severe acid rain, with most of the sites located in rural and remote areas.In recent years in China, there have been many field observations on aerosol acidity in the megacities of different regions, such as Beijing (Yao et al., 2002;Dillner et al., 2006;Sun et al., 2010), Shanghai (Yao et al., 2002;Xiu et al., 2005;Wang et al., 2006), Hong Kong (Pathak et al., 2003(Pathak et al., , 2004a, b) , b) and Chongqing (Quan and Zhang, 2008;Aas et al., 2007).
While these studies suggest a general pattern of higher acidity in southern China than in the north, only a few of them presented parallel inter-region comparisons, with little information on seasonal variation.For example, Wang et al. (2006) and Pathak et al. (2009) investigated aerosol acidity at different regions in China, in spring and summer, respectively.Moreover, even for a specific region, there are large discrepancies between the studies.For Shanghai, Xiu et al. (2005) found that aerosols were almost completely neutralized, a finding that is contrary to the results reported by Yao et al. (2002) and Wang et al. (2006) (Chan and Yao, 2008).The discrepancies might be attributed to a variety of factors, including the procedures of sampling analysis and changes in emission strength and meteorological condition.Increasingly there is a need to understand how these factors have influenced the variability of aerosol acidity.
Along with the increase in domestic NO x emissions in recent years (Zhang et al., 2007b), the concentration and proportion of NO − 3 in aerosols have been found to have increased significantly in most Chinese megacities (Richter et al., 2005;Chan and Yao, 2008;Shen et al., 2008), and become a major concern in the acidity of aerosols as well as their wet/dry deposition (Han et al., 2006;Larssen et al., 2006;Song et al., 2008).The formation pathway of NO − 3 depends on not only the availability of NH + 4 and meteorological condition (such as temperature), but also the characteristics of the preexisting particles, such as aerosol acidity, water content and alkaline mineral composition (Pakkanen et al., 1996;Hu and Abbatt, 1997;John et al., 1990;Zhuang et al., 1999).In particular, Pathak et al. (2009) recently reported high concentration of NO − 3 with high acidity and low NH + 4 at Beijing and Shanghai, with their formation being largely attributed to the hydrolysis of N 2 O 5 on the preexisting particles.This differs from the findings of most previous field observations, which indicated that high concentration of NO − 3 are found in association with high NH + 4 (Pathak et al., 2009).However, the observations of Pathak et al. (2009) were limited to summertime.
In this study, the aerosol acidity at Beijing and Chongqing, two megacities in northern and southwestern China, respectively, was examined in parallel during a 15-month period of field observation, and the characteristics of fine particles (PM 2.5 , particles of aerodynamic diameter <2.5 µm) were investigated in detail.Based on the measurements of ionbalanced and in situ acidity, we investigated the spatial and seasonal patterns of PM 2.5 acidity at these two cities.We also discussed the factors that determined these spatial and temporal variations.
2 Experimental method and model description

Sampling and analysis
Weekly PM 2.5 samples were collected at both urban and rural sites of Beijing and Chongqing using a three-channel lowflow sampler (Aerosol Dynamics Inc., Berkeley, CA).Details of the sampling sites have been provided previously (He et al., 2001;Zhao et al. 2010).In brief, the urban and rural sites at Beijing were inside Tsinghua University (TH, 40 The sampling procedure was also given by He et al. (2001).Operating at a flow rate of 0.4 l min −1 , one of the three channels collected PM 2.5 on a Teflon filter with a Teflon impactor followed by a glass denuder.The glass denuder is coated with a 2 % carbonate solution prepared in 50:50 water:methanol to remove the acidic gases.HNO 3 volatilized from the Teflon filter is collected on a nylon filter.Hence water soluble ions are determined from this Teflon filter but the reported particulate NO − 3 is the sum of NO − 3 on both the Teflon and nylon filters.The other two channels collected PM 2.5 with a single Teflon filter and quartz filter for measuring elements and carbonaceous components, respectively, which were not used in this study.Each sample was collected continuously for a week.From 28 January 2005 to 5 May 2006, 106 and 180 PM 2.5 samples were collected at Beijing and Chongqing, respectively.
Hourly meteorological data for both Beijing and Chongqing were obtained from the website http://www.wunderground.com,including temperature, dewpoint, wind speed, visibility and precipitation.The spatial distribution of geopotential height was derived from the archived meteorological data of NOAA's Air Resources Laboratory (ARL, http://ready.arl.noaa.gov/).

Ratio of cation/anion
In this study, the equivalent charge ratio (eq/eq) of the major cations NH + 4 and Ca 2+ to anions SO 2− 4 and NO − 3 was used to indicate the neutralizing level of PM 2.5 , as the other ions generally had little influence on the acidity at Beijing and Chongqing (to be discussed in Sect.3.1).The equivalent charge ratio was defined as following (Adams et al., 1999;Zhang et al., 2002): where all the species denote the concentrations of their equivalent charges (likewise for all the ratios of species without brackets in the following text).In this equation, R C/A ≥ 1 indicates that most of the acids can be neutralized, while R < 1 indicates the aerosol is acidic.

In situ aerosol acidity
Both free H + concentration ([H + ] Ins , the square brackets indicate the molar concentration of the species inside, used here and henceforth) and in situ pH in the deliquesced particles were used as indicators of aerosol acidity, which can be estimated from a chemical thermodynamic model (E-AIM2, http://www.aim.env.uea.ac.uk/aim/).E-AIM2 is a state-of-the-art model that can accurately simulate the liquid and solid phase of ionic compositions in the mixing system H + -NH + 4 -SO 2− 4 -NO − 3 -H 2 O at a given temperature and relative humidity (Clegg et al., 1998).The model input of E-AIM2 includes weekly averaged temperature, relative humidity, [SO 2− 4 ], [NO − 3 ], [NH + 4 ] and total H + ([H + ] Total ), which is estimated from the ionic balance of the relevant species (Yao et al., 2006;Zhang et al., 2007;Pathak et al., 2009): The aerosol pH was calculated as: where γ is the activity coefficient on mole fraction basis and [H + ] Frac is the molar fractions of aqueous phase H + (Zhang et al., 2007a).In addition to these two parameters, [H + ] Ins and the concentration of water content ([H 2 O]) were derived from E-AIM2.The lack of information about the organic acids generally has little influence on the estimation of aerosol acidity due to their low abundance in aerosols (Zhang et al., 2007a;Pathak et al., 2009).However, larger bias may exist because of a lack of information about basic Ca 2+ and Mg 2+ , especially in samples containing high concentrations of mineral dust (Ziemba et al., 2007).Although there are models, such as SCAPE, that take into account a system with these basic ions included, they cannot be used in the current study because the required gaseous HNO 3 and NH 3 , the input parameters for the models, were not measured here.It should also be noted that the E-AIM2 model only simulates the average results over the whole week without considering the influence from temporal variations in aerosol composition, temperature and relative humidity (Yao et al., 2006).

Trajectory computation and clustering
Backward trajectories of air masses arriving at Beijing and Chongqing were calculated using the HYSPLIT model (Version 4.8) to investigate the influence of different air masses on aerosol composition and acidity.The meteorological data fields used to run the model are 6-hourly FNL archived data, which are available at NOAA's ARL archives.For single trajectory calculation, the model was run 4 times per day (UTC 00:00, 06:00, 12:00 and 18:00) with the arrival level at 500 m (below the boundary layer) or 2000 m (above the boundary layer).
As the typical errors of individual trajectories were estimated to be 20 % of the traveled distance (Stohl, 1998), the trajectories over the whole sampling period were classified into seasonal transport patterns using the HYSPLIT model.A detailed procedure of the clustering analysis is provided in the supplementary material according to the model description (Draxler et al., 2006).The percentage change in total spatial variance (TSV) was used to determine what is the reasonable number of clusters in each season: a large increase in TSV indicates that different clusters are being paired and therefore that the cluster process should stop.

Abundance of ionic species in PM 2.5
The annual concentrations of PM 2.5 and ionic species were averaged from March 2005 to February 2006 for Beijing and Chongqing, as shown in Table 1.PM 2.5 mass concentration was similar at all the three sites in Chongqing (∼130 µg m −3 ), all of which were higher than those at Beijing (118 µg m −3 for TH and 68 µg m −3 for MY).This indicates high regional background levels of PM 2.5 in the surrounding area in Chongqing.As with PM 2.5 mass, higher concentrations of total ionic species were found in Chongqing (41.3-45.0µg m −3 ) than in Beijing (28.3-39.1 µg m −3 ).The proportion of ionic species in PM 2.5 at the urban site in Beijing (33.0 %) was close to those at Chongqing (31.8-34.9%), but the higher fraction in MY (41.4 %) suggests a more important role of ionic species in rural areas of Beijing.Components other than ionic species contributed similar amounts to PM 2.5 at Beijing and Chongqing, i.e. carbonaceous species and crustal dust accounted for 36-40 % and 6-8 % of PM 2.5 mass, respectively (Zhao et al., 2010;He et al., 2011).SO 2− 4 , NO − 3 and NH + 4 dominated the ionic species, with a contribution of up to 85-90 % at both Beijing and Chongqing.As with PM 2.5 at Chongqing, the three species when considered in combination had a small spatial variation, while at Beijing their relatively small difference between MY and TH (MY/TH: 76.8 %) compared to PM 2.5 (MY/TH: 57.7 %) indicates that they are of greater regional significance than the other aerosol species.Concentrations of SO 2− 4 for all the sites were higher at Chongqing than at Beijing, whereas NO − 3 showed the opposite spatial pattern with higher concentrations at Beijing than at Chongqing.This is mainly attributed to regional differences in energy structure and meteorology (see Sect. 3.2).At the mean time, it should be noted that a higher proportion of SO 2− 4 in PM 2.5 was found for MY (0.19) than TH (0.13), but similar as that for the sites in Chongqing (0.18-0.20).This was probably due to the more homogeneous spatial distribution of sulfate in Beijing than the other aerosol species, as has been found by other studies (Zhao et al., 2009;Guo et al., 2010).It also indicates that the regional influence of sulfate in Beijing was as important as in Chongqing.
Compared to the above major ionic species, Ca 2+ , Mg 2+ , Na + , K + and Cl − constituted a minor fraction (10-15 %) of ionic species at Beijing and Chongqing.The annual concentrations of basic Ca 2+ and Mg 2+ were higher at Chongqing (totaling 0.9-1.5 µg m −3 ) than at Beijing (totaling 0.7-1.1 µg m −3 ).Their low abundance relative to NH + 4 suggests that they have only a weak influence on neutralizing the acidic species, as found in most studies on aerosol acidity (Yao et al., 2006;Zhang et al., 2007a;Pathak et al., 2009).However, as an indicator of mineral dust, which was found in high concentrations during the spring and winter at Beijing and Chongqing (Zhao et al., 2010), Ca 2+ was included in the acidity analysis to get an idea of the regional influence of alkaline dust.

Spatial distribution
The ratio of cation/anion (R C/A ) in PM 2.5 was calculated for all the sites according to Eq. ( 1).As shown in Table 1, annual R C/A from March 2005 to February 2006 were 0.97 and 1.04 at TH and MY, respectively, whereas those in Chongqing were substantially lower, ranging from 0.76 to 0.86, respectively.This indicates that the aerosols were much more acidic at Chongqing than at Beijing.This pattern is consistent with the findings of Wang et al. (2006), who reported that aerosols over southern China were less neutralized during springtime than aerosols over northern China.
The urban-rural distributions of R C/A for PM 2.5 in Beijing and Chongqing were opposite to each other, which can be explained by their compositional difference in NO − 3 /SO 2− 4 and Ca 2+ /NH + 4 .Both of these two ratios were higher at urban sites, possibly due to more vehicle sources and construction activities, however they showed different spatial gradients within the two cities.Compared to the difference of NO − 3 /SO 2− 4 at Chongqing (<10 %; urban: 0.16-0.17,rural: 0.15), the ratio was found ∼30 % higher at urban TH (0.60) than rural MY (0.47) at Beijing.At the meantime, Ca 2+ /NH + 4 showed a larger urban-rural difference at Chongqing (∼25 %) than at Beijing (∼10 %).Therefore, it was the higher concentrations of NO − 3 within the urban site in Beijing and the lower concentrations of Ca 2+ within the rural site in Chongqing that made their PM 2.5 more acidic.This differs from findings for Pittsburgh where NH + 4 levels determined the spatial distribution of aerosol acidity at urban and semi-rural sites (Liu et al., 1996).

Seasonal variation
The seasonal averages of R C/A of PM 2.5 in Beijing and Chongqing are shown in Table 2.In Beijing, PM 2.5 at TH was more acidic in the summer of 2005 and spring of 2006 than other seasons (R C/A ≤ 1 for all seasons).Similar seasonal variation in R C/A was also observed at rural MY, but its higher R C/A indicates that PM 2.5 was almost neutral in all seasons (R C/A ≥ 1), except for the spring of 2006 (R C/A = 0.80).These results are similar to previous findings that aerosols were more acidic in warm seasons than in cold seasons at Beijing (Wang et al., 2000;Pathak et al., 2009), as indicated from R C/A reproduced from their reported datasets shown in Table 2.However, no consistent pattern in interannual trends can be discerned for each season.During the summer, R C/A was >1 in 2001-2003 (Wang et al., 2005) and 2006 (Sun et al., 2010), and <1 in 1999-2000(Wang et al., 2000;He et al., 2001) and 2005 (Pathak et al., 2009; this study); for spring, R C/A was >1 in 1994-1995 (Wang et al., 2000) and 2001-2003(Wang et al., 2005), and <1 in 1999-2000 (He et al., 2001;Dillner et al., 2006Dillner et al., ) and 2005Dillner et al., -2006 (this study) (this study); for winter, R C/A was >1 in most years except 1999-2000 when it was only 0.63 (He et al., 2001), indicating the aerosols were much more acidic during spring than during other seasons.Similar characteristics in seasonal variation of aerosols are also observed in other northern cities.As shown in Table 2, while they were neutralized in most cases (R C/A > 1), very acidic aerosols were also observed at Yungang (summer of 1988;Wang et al., 2000), Xi'an (winter of 1996Xi'an (winter of -1997;;Zhang et al., 2002) and Jinan (spring of 2004-2005;Yang et al., 2007).
As in Beijing, R C/A was high in winter and low in summer and fall of 2005 at both urban and rural sites of Chongqing.However, the ratios for the two springs were not the same, with much more acidic aerosols being observed in 2005 (R C/A = 0.58-0.62)than in 2006 (R C/A = 0.94-1.00).No consistent inter-annual trend in seasonal acidity could be found at Chongqing, either.For example, Liu et al. (1988) found R C/A of PM 2 to be 1.77 in the fall of 1980s, much higher than the ratios from our observation (0.79-1.02).The large difference between the two studies cannot be simply explained by the increased acidity of aerosol over the past 20 yr because similarly low R C/A (0.78-1.01) were also measured in TSP by Zhao et al. (1994Zhao et al. ( ) during 1987Zhao et al. ( -1988 in Chongqing (the ratio was even lower for fine particles).This phenomenon shifts the likely explanation to factors other than the changes in the emission of acidic aerosols and their precursors.As at Chongqing, aerosol R C/A at other southern cities have also exhibited an inconsistent inter-annual trend in recent years.The R C/A of PM 2.5 in Shanghai was significantly higher in the summer and winter of 2004 (1.38 and 0.78, respectively; Wang et al., 2006) than in summer of 2005 and winter of 2001 (0.4-0.5; Xiu et al., 2004;Pathak et al., 2009), while for Hong Kong the ratio in winter was also distinctively higher in 2002 (1.06;Cheung et al., 2005) than 2001 (0.73; Louie et al., 2005).
In spite of the uncertainties in using R C/A to compare the aerosol acidity between different studies, such as the variable sampling methods, the representativeness of sampling periods, and analytical procedures, the above findings collectively suggest that the seasonal variation of aerosol acidity in northern and southern China may be influenced by a variety of factors (emission strength, meteorological condition and characteristics of preexisting particles and precursors, etc.).For either long-term or short-term field observations, it is consequently risky to attribute the inter-annual changes of aerosol acidity to any of these factors alone.For example, based on a comparison with the acidic aerosols reported by Dillner et al. (2006) for spring 2001, Sun et al. (2010) simply attributed their fully neutralized aerosols in Beijing during the summer of 2006 to the reduced SO 2 emissions or increased NH 3 emissions in this region.
It is interesting to observe that the inter-annual variation in spring PM 2.5 acidity for Chongqing was opposite to that for Beijing.For both cities, their covariation at urban and rural sites indicates that the inter-annual trend was of regional scale.Thus, the weekly R C/A for each city was averaged to investigate the short-period variation in aerosol acidity within each season.
As shown in Fig. 1a, R C/A showed extensive weekly fluctuation for both cities, with larger variation at Beijing (0.39-1.60) than at Chongqing (0.51-1.13).However, R C/A at Beijing was higher in the spring of 2005 (>1) than in 2006 (<1, from the week of 24-31 March), while at Chongqing continuously higher R C/A was observed in the spring of 2006 (0.8-1.1) than in 2005 (0.6-0.8).This pattern is better presented as normalized R C/A (R C/A minus the averaged ratio during the whole sampling period, as shown in Fig. 1b), indicating a weak intra-seasonal variation of aerosol acidity for both Beijing and Chongqing during the two springs.Also of note is that R C/A at both Beijing and Chongqing showed a similar decreasing trend during February-June 2005 with a sharp in- crease at the end, which implies that the aerosol acidity of both cities had been influenced by large-scale driving forces, as discussed in Sect. 4.

Seasonal variation of in situ aerosol acidity
In situ aerosol pH, [H + ] Ins and [H 2 O] of PM 2.5 at Beijing and Chongqing were shown in Fig. 2

. [NH +
4 ], [SO 2− 4 ] and [NO − 3 ] were averaged for urban and rural sites and used as the input data to simplify the comparison between the two cities.
The in situ PM 2.5 acidity showed similar seasonal variation as previously indicated by R C/A , but gave additional insight into the hygroscopic properties of aerosols.As shown in Fig. 2a, it was only in summer and fall of 2005 and spring of 2006 that deliquescent aerosols were found to be abundant at Beijing with free H + , while most of them remained in solid phase during the spring and winter of 2005.The spring of 2006 at Beijing had the most acidic aerosols, with an in situ pH of only −0.618 to 0.404, while there were only two weeks in the spring of 2005 when PM 2.5 was found acidic.Moreover, although high [H + ] Ins existed in PM 2.5 during the summer of 2005 (average: 0.030 µmol m −3 , range: 0.002-0.126µmol m −3 ) and spring of 2006 (average: 0.034 µmol m −3 , range: 0.006-0.122µmol m −3 ), the former was less acidic because of its much higher [H 2 O] (average: 3.074 µmol m −3 , range: 0.100-8.466µmol m −3 ) than the latter (average: 0.186 µ mol m −3 , range: 0.034-0.468µmol m −3 ).
www.atmos-chem-phys.net/12/1377/2012/Contrasting with Beijing, PM 2.5 at Chongqing was deliquescent throughout the year with high [H + ] Ins and [H 2 O], while a significant variation of in situ acidity between the two springs was also clearly evident, as shown in Fig. 2b.The most acidic aerosols were found during February-June 2005, when in situ pH remained at its lowest level (0.52-1.38) due to a faster increase in [H + ] Ins (20 times) than in [H 2 O] (5 times).Interestingly, both parameters decreased simultaneously decreased to their lowest level of the whole observation period in the week of 25 June to 1 July 2005, resulting in a significant increase of in situ pH and thus much less acidic PM 2.5 .This was also coincident with the week when R C/A showed a remarkable increase (Fig. 1).
In contrast to results revealed by R C/A , a noteworthy finding for the variation of in situ pH is that PM 2.5 was more acidic at Beijing than at Chongqing during the springs.This was mainly due to the drier climatology and lower water content in aerosols at Beijing that favored high in situ acidity, even though there might be less free H + .Similar finding was also reported in Hong Kong where variation of in situ PM 2.5 acidity was a function of relative humidity (RH) and even the more neutralized particles could have a high acidity under the influence of dry air masses from the Chinese mainland (Pathak et al., 2004a).This result highlighted the importance of the in situ acidity relative to other parameters (Pathak et al., 2004b).
In addition, as one of the most important parameters determining the in situ acidity, RH clearly exhibited opposite trends from winter into spring at Chongqing during 2005 and 2006, which can partly explain the inter-annual variation of PM 2.5 acidity.As shown in Fig. 2b, for the period from February to May, it increased from ∼60 to ∼80 % in 2005, but decreased from ∼80 to ∼50 % in 2006.These long-playing reverse seasonal trends were likely to have been influenced by large-scale synoptic system anomalies.

Factors influencing the spring-summer variation of PM 2.5 acidity
As a case study, we examined in the following the covariation of PM 2.5 acidity from spring into summer 2005 for Beijing and Chongqing, as well as the opposite inter-annual variation in PM 2.5 acidity in these cities during the springs of 2005 and 2006.

Asian summer monsoon
The covariation of R C/A for PM 2.5 at Beijing and Chongqing from February to June 2005, with a sharp increase at the end of June, indicates a synoptic-scale influence (Roger and Andrew, 2002).
The anomaly of the Asian summer monsoon in June 2005 was the abnormal behavior of the subtropical high over the Northwestern Pacific (Pacific High, in short) and the trough/ridge systems over mid-and high latitudes (Lu et al., 2007;Mu et al., 2008).The northward movement of the Pacific High, which is one of the most important parameters indicating the evolution of spring into summer in East Asia, was delayed until the end of June 2005.As shown in Fig. 3a (Lu et al., 2007), the ridge of the Pacific High remained around 13-16 • N before 26 June 2005, 3-5 • to the south of the normal position.During 26-28 June, however, it suddenly moved from 17 to 28 • N at a speed of 3-4 • per day, and its representative positions before and after the movement are indicated by the geopotential heights of 500 hPa in Fig. 3b   increased from 13 % in March-April (cluster 5 in Fig. 4a) to 41 % in May (clusters 4 and 5 in Fig. 4c) and 59 % in June (clusters 3 and 4 in Fig. 4d).At the same time, Chongqing was dominated by air masses from the east of the city, which increased from 50 % in March-April (cluster 4 in Fig. 5a) to 65 % in May (clusters 1 and 3 in Fig. 5c) and 73 % in June (clusters 1, 4 and 5 in Fig. 5c).As indicated by the aerosol optical depth (AOD) shown in the Supplement Fig. S1a-c, these source regions were found to have high aerosol loading from March to June in 2005, which clearly suggested that the aerosol acidity was increasing over a broad region of mainland China, with a stronger influence in the south than in the north.
Along with the northward movement of the Pacific High at the end of June 2005, the acidic aerosols over Beijing and Chongqing were replaced by the cleaner air from the northwest and southeast, respectively, which coincided with the simultaneous decrease of PM 2.5 acidity at both cities.This effect was also evident from significantly weakened AOD during 29-30 June over a wide region that used to be covered by highly acidic aerosols (Supplement Fig. S1d).In July, Beijing was again dominated by air masses from the south with a monthly contribution (65 % of all air masses) and [H + ] Ins level comparable to those in June.However, the Asian summer monsoon was found to have a much greater significance in Chongqing, where the air masses in July were dominated by those having been transported over long distances from the south of China and from southeastern Asia, with high monthly contribution (74 % of all air masses), but low aerosol acidity.
These lines of evidence collectively suggest the major role of the Asian summer monsoon in determining the regional evolution of PM 2.5 acidity from the spring to the summer of 2005 for Beijing and Chongqing.However, it can not explain the inter-annual variation of PM 2.5 acidity during the springs of 2005 and 2006, since no obvious difference was found between the transport patterns of their air mass trajectories.As shown in Figs.4b and 5b, respectively, the dominant air masses for Beijing and Chongqing from 3 March to 5 May 2006 were also from the northwest and east of China, a situation that was similar to that from 4 March to 6 May 2005 (Figs.4a and 5a, respectively).Therefore, there must be other factors that caused the inter-annual variation of PM 2.5 acidity in the springs of Beijing and Chongqing.

Asian desert dust
Mineral dust can affect aerosol acidity by either directly neutralizing the acidic aerosol or increasing the surface area of heterogeneous reaction for the acids.As an indicator of mineral dust, higher Ca 2+ as well as higher ratios of Ca 2+ /NH + 4 were observed in spring of 2005 and 2006 for Chongqing and Beijing, respectively.This is consistent with the findings of Wu et al. (2009) who reported that emission of Asian desert dust was more active in the spring of 2006 than in the spring of 2005 for Beijing, and our related study (Zhao et al., 2010) found Asian dust to be more active in the spring of 2005 than of 2006 for Chongqing.3 on the surface of mineral particles.It is well known that the reactions with alkaline mineral components are of several magnitudes faster for gaseous HNO 3 than for NO 2 and SO 2 (Ooki and Uematsu, 2005;Vlasenko et al., 2006), which were all abundant in the atmosphere of Beijing (Bergin et al., 2001).When Asian dust was transported to Beijing, CaCO 3 could react with HNO 3 to form Ca(NO 3 ) 2 , providing more hygroscopic surfaces for the heterogeneous reaction with SO 2 and NO 2 , as has been directly observed by single particle analysis during dust storms at Beijing (Li and Shao, 2009).However, NO − 3 /SO 2− 4 showed little increase (∼1.5 %) at MY in the spring of 2006 compared to 2005, as seen in Fig. 6a.This is perhaps due to the lack of precursors of NO − 3 , which was more concentrated in urban area of Beijing, and the unstable nature of NH 4 NO 3 , which could easily be decomposed into gaseous NH 3 and HNO 3 during transport from the urban area to MY.
On the other hand, compared to that in the spring of 2005 a higher increase in SO 2− 4 was observed at MY (23.3 %) than at TH (11.3 %) in the spring of 2006.The elevated SO 2− 4 concentration at the rural MY than at the urban TH (a pattern not observed for NO − 3 or NO − 3 /SO 2− 4 ) can be explained by coupling the SO 2− 4 formation with the inter-annual variation in transport pathways of air masses during spring.Compared to the transport pattern in the spring of 2005 (Fig. 4a), polluted air masses were more frequently transported from the west and south of Beijing in the spring of 2006 (Fig. 4b), which favored a higher regional contribution of SO 2− 4 at MY than during periods when other transport pathways were in play (Jia et al., 2008;Zhao et al., 2009).Moreover, faster transformation of local SO 2 to SO 2− 4 at Beijing could also lead to higher increase of SO 2− 4 in MY due to the more acidic and hygroscopic aerosols in the southern and southwestern air masses.However, it should be noted that the difference in NH + 4 concentrations between the two springs was nearly the same for the two sites at Beijing, perhaps because of the recapture of decomposed NH 3 from NH 4 NO 3 by the unneutralized SO 2− 4 or HSO − 4 during transport.Although the influence of Asian dust at Beijing may partly explain the inter-annual variation of PM 2.5 acidity for the springs of 2005 and 2006, this does not seem to be an explanatory factor at Chongqing.Firstly, it was a significant decrease in NH + 4 concentration in the spring of 2005, which was 28.0 to 30.2 % lower than that for the spring of 2006 (Fig. 6b), that essentially led to the elevation of aerosol acidity.The increased concentration of mineral dust might have changed the gas-particle equilibrium of NH 3 /NH + 4 by limiting the transfer of NH 3 to NH + 4 in fully neutralized aerosols (Luo et al., 2007), but it could hardly influence the highly acidic aerosols at Chongqing during the spring of 2005.Secondly, significant monthly variation in the transport of northwestern air masses was found from March through June 2005 at either the boundary layer (Fig. 5) or the higher atmosphere (Supplement Fig. S2), but the PM 2.5 acidity at Chongqing (as indicated by R C/A in Fig. 1 and [H + ] Ins in Fig. 2) remained at a consistently high level, and indeed showed a slight increase.

Wet deposition of NH + 4 in Chongqing
Particulate NH + 4 mainly comes from the gaseous NH 3 and has a residence time of 4-6 days compared to only 1 day for gaseous NH 3 (Adams et al., 1999).The variation in NH + 4 concentration at Chongqing for the two springs can be influenced by many factors, including the emission strength of precursor NH 3 , the gas-particle equilibrium of NH 3 /NH + 4 , and patterns of atmospheric transport, diffusion and deposition (Asman et al., 1998).
NH 3 emissions from natural sources, including animal waste, natural and fertilized soils, and vegetation, usually depend on temperature, which showed little difference between the two springs in Chongqing, as shown in Fig. 6b.Anthropogenic sources, such as industrial process, are considered to be stable during all seasons.The gas-particle equilibrium of NH 3 /NH + 4 is usually related to NH 4 NO 3 and NH 4 Cl, which are unstable; however, these were not the major form of NH + 4 in Chongqing due to the high PM 2.5 acidity dominated by SO 2− 4 .
The influence of atmospheric transport, diffusion and deposition on NH + 4 concentration at Chongqing can be assessed from the variation in air mass trajectories and meteorological factors.For the two springs at Chongqing, little difference was observed in the patterns of air mass backward trajectories, as previously discussed.Meanwhile, as shown in Fig. 6b, surface temperature, wind speed and relative humidity also showed weak variations.Together they suggest that atmospheric transport and diffusion at Chongqing play a minor role in explaining the significant inter-annual variation of NH + 4 during the springs of 2005 and 2006.In contrast to all the above factors, the amount of precipitation was 35.5 % higher in the spring of 2005 than of 2006, which is comparable to the differences for NH + 4 (28.0-30.2%) and R C/A (26.3-30.7 %).As shown in Fig. 7a, the precipitation was negatively correlated with NH + 4 in PM 2.5 at the JB site from February 2005 to April 2006 (R = −0.63,p = 0.01), indicating that the wet removal of NH + 4 was favored by the increase in precipitation.In fact, southwestern China experienced a long drought from the fall of 2005 to the spring of 2006, and the number of days on which rain fell during spring 2005 was 20 to 50 % more than in the spring of 2006 for most cities in the Sichuan Basin (Supplement Fig. S3; meteorological data from http://www.wunderground.com).Although the precipitation in spring at Beijing also showed large inter-annual variation (Fig. 6a), a similar effect of increased precipitation was not evidence due to the much smaller rain volumes during both springs (40 mm and 11 mm in 2005 and 2006, respectively).
Along with these lines of evidence, the chemistry of wet deposition at Chongqing, derived from the Acid Deposition Monitoring Network in East Asia (http://www.eanet.cc/product/index.html), also suggests the significant influence of precipitation on the variation of NH + 4 in PM 2.5 .Since NH + 4 and SO 2− 4 were the major species determining PM 2.5 acidity at Chongqing, the ratio of NH + 4 /SO 2− 4 (eq/eq) was used to better present the chemical behavior of NH + 4 in determining the acidity of PM 2.5 and precipitation.As shown in Fig. 7b, during February-November 2005 when precipitation was relatively abundant, the ratios of NH + 4 /SO 2− 4 in the precipitation and PM 2.5 were significantly negatively correlated with each other at Chongqing (R = −0.88,p < 0.001).However, only a weak correlation (R = 0.44, p = 0.38) was found for the dry seasons from November 2005 to April

Fig. 1 .
Fig. 1.Seasonal variations of (a) R C/A (uncertainty = standard deviation) and (b) normalized R C/A of PM 2.5 at Beijing and Chongqing.
(23 June) and Fig. 3c (29 June), respectively.The evolution of air mass sources at Beijing and Chongqing was investigated before and after the northward movement of the Pacific High.The period from 4 March to 27 June 2005 was divided into three phases, including 4 March-6 May (to be compared with the same period in the spring of 2006), 7-31 May and 1-27 June.For each phase, backward trajectories of air masses at 500 m above ground level of Beijing and Chongqing were classified according to the procedures in Sect.2.3.Trajectories of longer duration were calculated for Chongqing (120 h) than for Beijing (72 h) because of their different spread of travel.The increasing aerosol acidity from March to June 2005 in the two cities was closely associated with the contribution of air masses from areas between the Northern China Plain to the south of Beijing and from central China to the east of Chongqing.As shown in Fig. 4, Beijing was gradually dominated by air masses originating from south of the city, which
In order to compare the R C/A and ionic species of the two springs in parallel, data of weekly samples from 4 March to 6 May 2005 and 3 March to 5 May 2006 were averaged to represent the spring of 2005 and 2006 for each site, respectively.As shown in Fig. 6a, the Ca 2+ concentration at Beijing was 23.8 to 30.6 % higher in the spring of 2006 than of 2005, while the concentrations of SO 2− 4 and NO − 3 also increased by 11.3 to 23.3 % and 1.9 to 8.8 %, respectively, with little variation for NH + 4 .Due to the small contribution of neutralization from mineral components, the increased SO 2− 4 and NO − 3 in the spring of 2006 remained acidic.Meanwhile, a significant increase (45.5 %) was also found for NO − 3 /SO 2− 4 at urban TH in the spring of 2006 compared to 2005, which strongly suggests the influence from enhanced production of NO −

Table 1 .
Annual mass concentrations of PM 2.5 , its major ionic species and their equivalent ratios at Beijing and Chongqing from March 2005 to February 2006.
the acidity of the two cities.As shown in Table 1, NO −