Gaseous and aerosol pollutants during fog and clear episodes in South Asian urban atmosphere

Atmospheric Environment 42 (2008) 7775–7785

Contents lists available at ScienceDirect

Atmospheric Environment journal homepage: www.elsevier.com/locate/atmosenv

Gaseous and aerosol pollutants during fog and clear episodes in South Asian urban atmosphere K.F. Biswas a,1, Badar M. Ghauri b, Liaquat Husain a, c, * a

Wadsworth Center, New York State Department of Health, Empire State Plaza, Albany, NY 12201, USA Pakistan Upper Atmosphere Research Commission (SUPARCO), Division of Space and Environment, P.O. Box 8402, University Road, Karachi, Pakistan c Department of Environmental Health Sciences, School of Public Health, State University of NY at Albany, Albany, NY 12201, USA b

a r t i c l e i n f o

a b s t r a c t

Article history: Received 21 September 2007 Received in revised form 24 April 2008 Accepted 26 April 2008

We report the first measurements of acidic gases and ammonia (NH3) during fog and clear episodes in Lahore, a highly polluted mega-city of South Asia, along with concentrations of PM2.5 (particles of <2.5 mm aerodynamic diameter) and ions. An annular denuder system was used to measure acidic gases, NH3, and PM2.5 from December 2005 to February 2006 in Lahore, a mega-city in Pakistan. The denuders yielded average concentrations (mg m3) as follows: ammonia, 50; nitrous acid, 19.6; sulfur dioxide, 19.4; hydrochloric acid, 1.16; nitric acid, 1.00; and oxalic acid, 0.6. The filters yielded average concentrations (mg m3): PM2.5, 209; sulfate, 19.2; nitrate, 18.9; chloride, 7.43; oxalate, 0.97; ammonium, 16.1; potassium, 3.49; calcium, 0.89; sodium, 0.76; and magnesium, 0.08. Emissions from local sources, e.g., fossil fuel consumption by motorized transport and power plants, farming, burning of agricultural residues, industrial and construction activities contributed the 2 major proportion of pollutants in Lahore. Concentrations of ionic species, e.g., NO 3 , SO4 , 2þ 2þ Naþ, NHþ 4 , Mg , and Ca , and gaseous species, e.g., HCl, HNO3, SO2 and (COOH)2 showed a distinct diurnal variation. Mixing heights and photochemistry played major roles in defining the diurnal pattern. Fog appeared to profoundly enhance the oxidation of sulfur dioxide. High moisture content of fog resulted in uptake of the gases in fog droplets. Ó 2008 Elsevier Ltd. All rights reserved.

Keywords: PM2.5 Pakistan Acidic gases India Denuder

1. Introduction Although pollution controls have significantly decreased atmospheric concentrations of pollutants in the industrialized nations in the Western Hemisphere, the concentrations are expected to continue to grow in Asia. Rapid urbanization, and lack of efficient monitoring and control of pollution, along with particular natural and meteorological phenomena like Asian brown haze or prolonged episodes of winter fog, make the South Asian atmospheric chemistry * Corresponding author. Wadsworth Center, New York State Department of Health, Empire State Plaza, Albany, NY 12201, USA. Tel.: þ1 518 473 4854; fax: þ1 518 473 2895. E-mail address: [email protected] (L. Husain). 1 Present address: Center for Marine Science and Department of Chemistry and Biochemistry, University of North Carolina Wilmington, NC 28403, USA. 1352-2310/$ – see front matter Ó 2008 Elsevier Ltd. All rights reserved. doi:10.1016/j.atmosenv.2008.04.056

a very complex one. Information on the airborne concentrations of toxic pollutants in developed countries is relatively abundant. However, for less-developed nations and for countries with a predominantly arid climate, the availability of data is far lower. For instance, air pollution in South Asian countries is rapidly growing as a result of increasing populations, expanding economies, and urbanization. Since late 1990s, some cities of eastern India and northeastern Pakistan have been experiencing intense fog over a period of a few months in winter (Hameed et al., 2000). The fogs thoroughly disrupt rail- and air traffic and other human activities, and even caused loss of lives in vehicular accidents due to poor visibility. Hameed et al. (2000) and Rattigan et al. (2002) measured  sulfate (SO2 4 ), nitrate (NO3 ), sulfur dioxide (SO2), and selected trace elements in air collected during a highpollution episode in Lahore. Lahore (31350 N, 74 210 E;

7776

K.F. Biswas et al. / Atmospheric Environment 42 (2008) 7775–7785

Fig. 1. Location of Lahore and its neighboring cities in Pakistan and India.

Fig. 1), with a population of about 10 million, is the second largest city in Pakistan (Husain et al., 2007). During a global monitoring program, the World Health Organization (WHO) identified concentrations of total suspended particulates in Lahore as being among the highest in the world (Smith et al., 1996). Out of 63,000 annual deaths in Lahore, about 1250 are caused by air pollution which is responsible for various chronic diseases, especially asthma (www.cleanairnet.org/ caiasia/1412/article-70340.html). The main causes are the high density of pollution sources (for example, from numerous small industrial activities) and the lack of effective pollution controls. The city has recently been covered by fog for a period of 1–2 months during winter. The measurements of Hameed et al. (2000) and Rattigan et al. (2002) were based on a 2-week sampling largely during fog episodes, and only SO2 was determined among the gaseous species. To more fully understand the sources of the pollutants, and the chemical processes occurring, we conducted an extensive study from late December 2005 to February 2006. The primary objectives of this study were as follows: (1) to chemically characterize aerosols both during fog episodes and during periods without fog, (2) to study the impact of fog on the chemical composition of the ambient air, and (3) to identify the major sources of pollutants. 2. Experimental 2.1. Sampling Aerosols and gaseous species were simultaneously collected using a Thermo Electron Corporation Reference

Ambient Air Sampler (RAAS) PM2.5 sampler coupled with an annular denuder system (ADS) from December 23, 2005 to February 14, 2006; the sampler was placed at a height of 10 m above the ground at the campus of Punjab University (31.57 N, 74.31 E). In addition to vehicular emissions from motorways, emissions from coal combustion and biomass burning are the main local sources of air pollution in this industrial city (Rattigan et al., 2002; Husain et al., 2007). The RAAS sampler as operated had two cyclones and four collection channels followed by a 47-mm filter module. For maintenance of a cut-point of 2.5 mm the flow through each cyclone was set to 24 L min1. Only channel 1 (16.67 L min1) and channel 2 (7.33 L min1) from cyclone A were used in this study. Initially pre-weighed 47-mm Teflon filters ran on channel 1, and pre-fired 47-mm quartz filters ran simultaneously on channel 2. The air samples were collected using annular denuder/ filter pack systems (URG-2000-01L) manufactured by University Research Glassware, Inc. This is an Eight Channel Sequential Fine Particle Sampler, each with two 120-mm long glass-walled annular denuders connected in series, followed by a 47-mm filter module. The annular denuder tubes were coated with appropriate coating solutions (citric acid: 1% w/v in methanol; sodium carbonate: 1% w/v, 1% w/v glycerol in a 1:1 methanol/water solution), and the coated tubes were dried with ‘zero’ air at a rate of 3 L min1. The first annular denuder was coated with Na2CO3 to remove SO2, nitrous acid (HONO) and nitric acid (HNO3), and the second was coated with citric acid to remove NH3. The denuder trains were assembled and were leak-checked under clean laboratory conditions. With each batch of

K.F. Biswas et al. / Atmospheric Environment 42 (2008) 7775–7785

7777

seven denuder assemblies (ADS plus filter) sent out in the field, a blank denuder assembly was included. It was left for 7 days inside the sampler but was not connected to the air flow. Twenty-four samples each of aerosols and gas were collected during the sampling period. The flush-end of the citric acid coated annular denuder was attached directly to the filter module. The filters (47 mm, 2.0 mm pore size Zefluor, and 47 mm, 1.0 mm pore size Nylasorb; Pall Corporation) were positioned on the Teflon-coated stainless steel screen so that the air stream particulates (diameter <2.5 mm) were trapped on the Teflon-coated side of the filter. At the downstream end of the filter pack, a Nylasorb filter was used to capture HNO3 resulting from the dissociation of NH4NO3 particles on the Teflon filter. The NO 3 mass measured on the nylon back-up filter was used to correct the values of NO 3 measured on the Teflon filter. Measured NHþ 4 concentration reported here may, however, be underestimated due to NHþ 4 loss by volatilization of NH4NO3. The degree of ammonium loss from denuded nylon and Teflon filters were found to vary between campaigns made by Yu et al. (2006), ranging from an average of 3.5–17.8%. Losses of ammonium tended to increase with increasing diurnal temperature, relative humidity swings, and the degree of gas-to-particle equilibrium perturbation (Yu et al., 2006). Since this study was conducted during winter with average temperatures around 15  C, the NHþ 4 loss should be less than 17%. A cyclone with a cut-point of 2.5 mm preceded the ADS. Air samples were collected for 12 h, generally from 8:00 a.m. to 20:00 p.m. and from 20:00 p.m. to 8:00 a.m. (local time). During the sampling period, the daytime temperatures varied from 15 to 25  C and nighttime temperatures dropped to the single digits. Rain or drizzle occurred on 1, 2, 15, 16, and 17 January. Fog was intermittently present between 23 December and 4 January, although the number and density of fog events in December and January were lower than had been seen over the previous years. Filters and denuder tubes were refrigerated and transported cold (4  C) to our Wadsworth Center laboratory soon after the end of sampling.

detection limits of the analytical method, calculated as three times the standard deviation of the blank signals, were respectively, 0.1, 0.34, 0.22, 0.07, 0.44, 0.87, 0.22, 0.29,    0.15, 0.42, 0.05 and 0.25 mg m3, for BrO 3 , Cl , NO2 , Br ,  2 2 þ þ þ 2þ 2þ and Ca . Teflon NO3 , SO4 , C2O4 , Na , NH4 , K , Mg filters were acid-digested in a microwave oven (Model CEM Mars 5) and analyzed for 25 trace elements with a Hewlett Packard inductively coupled plasma mass spectrometer (Model HP 4500). Selenium was among the trace elements of our interest. Se was found to be below the detection limit (7.53 ng m3) in two third of Lahore aerosol samples. However, we calculated SO2 4 /Se ratios using measured concentrations of SO2 4 and Se (above the detection limit) to facilitate the discussion on emission sources in Section 3.4.

2.2. Chemical analysis

2.4. Data handling

Teflon filters were equilibrated for 24 h and weighed before and after aerosol collection in a room meeting EPA requirements for temperature (20–23  C) and relative humidity (30–40%), for measurement of PM2.5 mass. Filters were ultrasonicated in purified (18 MU cm1) water for 90 min at 60–70  C, and the pH of the extracts was determined (Bari et al., 2003). Coating solutions impregnated with gaseous species of the denuder tubes were rinsed thoroughly into purified water (Bari et al., 2003). Aqueous extracts of filters and adsorbed coating solutions were stored at 4  C. Analysis of extracts for ions of interest, i.e.,   bromate (BrO 3 ), chloride (Cl ), nitrite (NO2 ), bromide 2 2 ), sulfate (SO ), oxalate (C O (Br), nitrate (NO 3 4 2 4 ), sodium þ þ þ (Na ), ammonium (NH4 ), potassium (K ), magnesium (Mg2þ), and calcium (Ca2þ), was done by DIONEX ion chromatograph (Model DX 500) using an AS9 column for anions and a CS14 column for cations. The extract of the citric acid adsorbent was analyzed only for NHþ 4 . The

In order to investigate the source of aerosols and gases collected during the sampling, we calculated backward air parcel trajectories with the HYSPLIT4 (HYbrid SingleParticle Lagrangian Integrated Trajectory) model, accessed via the NOAA (National Oceanic and Atmospheric Administration) Air Resources Laboratory READY (Real-time Environmental Applications and Display sYstem) Website (Draxler and Rolph, 2003). The model was used with archived meteorological data to predict the path of an air parcel prior to arrival at a given location. Three-day HYSPLIT4 air mass backward trajectories were calculated for the 12-h sampling periods (December 23, 2005–February 14, 2006). Representative backward trajectories during the sampling period are shown in Fig. 2 and are discussed in Section 3.4. The air trajectories were computed at 500 m and the height was adjusted isobarically. Trajectories were plotted for the beginning, middle and end of each of the 48-h periods. Each of the points represents 1-h air

2.3. Use of meteorological information The meteorological data, including temperature, atmospheric pressure, wind speed, relative humidity, sun index, fog cover, and visibility, were recorded every 30 min at nearby Lahore International Airport and were obtained from the Weather Underground Website (www.wunderground. com/history/airport/OPLA/2006). Our sampling site is 8.9 km west-southwest of Lahore International Airport. In addition, during January, instruments located at our sampling site recorded 5-min meteorological data. Over the study period, the daytime temperatures varied from 15 to 25  C and nighttime temperatures were in the single digits. The average wind speed during the night, 1.7 m s1, was lower than the mean 3.5 m s1 during the daytime (Husain et al., 2007). A combination of operators’ observations and meteorological records of Lahore International Airport and instrument at the sampling site was used to calculate the duration of fog during collection of each sample. Samples collected under completely fog-free condition have been classified as clear-episode samples (clear day or clear night; Table 1). Samples collected during periods when fog was present more than >80% of the time are considered as fog-episode samples (Table 1).

7778

Table 1 Composite gaseous and ionic concentration in the PM2.5 of Lahore, Pakistan from December 2005 through February 2006 Type of samples

PM2.5

HCl

HONO

HNO3

SO2

(COOH)2

NH3

Cl

NO3

SO2 4

C2O2 4

Naþ

NH4þ

Kþ

Mg2þ

Ca2þ

Atmospheric concentration (mg m3) 1.16 0.85 0.08 2.52

19.6 10.8 0.05 48.9

1.00 0.57 0.18 2.63

19.4 12.9 2.11 43.1

0.60 0.52 0.07 2.19

50.1 16.9 21.1 81.3

7.43 4.98 1.76 25.7

18.9 9.31 3.85 35.9

19.2 9.90 6.48 39.2

0.97 0.40 0.41 1.72

0.76 0.49 0.26 2.27

16.1 8.66 5.73 39.7

3.49 1.64 0.87 7.27

0.08 0.05 0.03 0.18

0.89 0.63 0.14 2.75

Fog episode (n ¼ 3) Mean SD Min Max

183 90.8 82.7 260

1.30 1.72 0.08 2.52

11.3 12.01 0.05 24.0

0.38 0.28 0.18 0.57

3.5 1.7 2.11 5.4

0.27 0.15 0.16 0.38

44.7 29.9 21.2 78.4

7.66 5.51 1.76 12.7

12.6 6.19 6.37 18.7

16.0 8.21 9.18 25.1

0.84 0.29 0.64 1.05

0.52 0.03 0.50 0.54

14.8 7.5 9.00 23.3

3.24 1.87 1.18 4.82

0.07

0.48 0.41 0.20 0.95

Clear-day (n ¼ 8) Mean SD Min Max

214 98.5 102 356

1.44 0.83 0.19 2.44

19.7 10.8 4.90 32.4

1.10 0.45 0.51 1.68

27.9 11.1 14.2 43.1

0.70 0.51 0.07 1.33

54.9 16.7 25.9 81.3

5.54 2.43 3.07 9.71

16.2 7.21 9.49 28.5

15.5 8.72 6.48 29.5

0.94 0.40 0.59 1.48

0.62 0.38 0.26 1.18

11.9 6.09 5.73 21.1

3.36 1.76 1.57 5.95

0.11 0.06 0.04 0.18

1.37 0.83 0.63 2.75

Clear night (n ¼ 5) Mean SD Min Max

211 78.0 151 330

1.08 0.94 0.25 2.13

21.8 4.30 17.3 27.6

1.34 0.90 0.58 2.63

19.1 5.45 11.4 24.3

0.95 0.85 0.29 2.19

51.9 5.23 45.7 58.9

6.49 3.77 3.73 12.8

23.7 8.94 14.2 35.9

23.4 10.6 11.9 39.2

1.05 0.55 0.41 1.72

1.25 0.73 0.30 2.27

18.5 7.78 10.9 28.4

3.79 1.33 2.85 6.02

0.04 0.01 0.03 0.05

0.58 0.20 0.29 0.75

Clear episode Clear-day/night Clear/Fog episode

213 1.01 1.16

1.26 1.34 0.97

20.7 0.90 1.83

1.22 0.82 3.23

23.5 1.46 6.72

0.82 0.73 3.10

53.4 1.06 1.19

6.01 0.85 0.79

20.0 0.68 1.58

19.4 0.66 1.22

0.99 0.89 1.18

0.93 0.50 1.80

15.17 0.64 1.02

3.6 0.88 1.10

0.07 2.74 1.12

0.97 2.37 2.02

K.F. Biswas et al. / Atmospheric Environment 42 (2008) 7775–7785

All sampling days (n ¼ 24) Mean 209 SD 92.1 Min 82.7 Max 450

K.F. Biswas et al. / Atmospheric Environment 42 (2008) 7775–7785

7779

Fig. 2. Representative air mass back trajectories during the 12-h sampling periods over which we measured the (a) highest, (b) lowest, and (c) moderate to high concentrations of certain species, and (d) a period of atmospheric stagnation.

movement. Close spacing of the data points indicates low wind speed and stagnant air. In the cause of characterizing carbonaceous aerosols in Lahore, Husain et al. (2007) calculated the mixing heights from the HYSPLIT4 air trajectory model at 3-h interval in the month of December. Between midnight and 5:00 a.m. the mixing height was consistently 250 m, i.e., at the model minimum. At 6:00 a.m., the height began to increase, reaching the maximum around midday, when it averaged 1000 m and it then decreased to the model minimum by around 6:00 p.m. We use intercomponent correlation analysis as a tool for possible source identification of aerosol and gaseous species of interest. Correlations (expressed as correlation coefficients, r2) that are significant at the 0.001 level (twotailed) are considered and reported for that purpose (Clarke and Cooke, 1983). Concentrations of gaseous species and concentrations of PM2.5 and aerosol components are each summarized as average along with one standard deviation of the measured values. Throughout this paper, the term ‘average’ and ‘mean’ are used to indicate arithmetic mean. 3. Results and discussions 3.1. PM2.5 mass The average concentration of PM2.5 was 209  92.1 mg m3 (Table 1) during the total sampling period which included

daily and nightly samples collected in the absence or presence of fog or sun light. Concentrations of total suspended particulate reported by Smith et al. (1996), Hameed et al. (2000) and Rattigan et al. (2002) were also in the same order of magnitude. The mean PM2.5 concentration in Lahore was several-fold higher than those measured in polluted urban locations of Seoul and Hong Kong and an order of magnitude higher than the annual average concentration reported for New York City (Table 2). The ratio of daily to nightly average PM2.5 concentrations in Lahore (Table 1) in the absence of fog is unity. Around 67% of PM2.5 mass at Lahore was attributed to carbonaceous aerosols, which are originated from fossil fuel combustion and biomass burning (Husain et al., 2007). Although vehicular traffic, the major localized source of PM2.5, decreases during the night, the atmospheric concentration of PM2.5 does not decrease correspondingly. Apparently, due to decreased mixing heights of the nighttime atmosphere compared to daytime, the atmospheric concentration of particulates tends to build up during the night. Hence, any significant diurnal difference in PM2.5 concentration is less likely to be observed. The average PM2.5 concentration ratio between fog and clear episodes is also close to unity (Table 1). During Lahore sampling campaign, relative humidity was found to range from 20 to 60% during clear periods, while 87–93% during fog episodes. We can assume a super saturation (RH > 100%)

1.70 1.13–2.33 2.09 1.28–3.20 1.12 0.44–1.70 2.43 1.36–3.24

4.27 2.09–7.11

1.10 0.04–14.6

4.32 3.00–5.71

The average concentrations for the measured gaseous species are reported in Table 1, along with the values for one standard deviation, and minimum and maximum concentrations. In our earlier studies at Lahore (Hameed et al., 2000), gaseous species were not included. The scale of variation in concentrations of HONO, SO2, and NH3 during the sampling period is remarkable. During clear episodes, day/night ratios of hydrochloric acid (HCl) (1.34), HNO3 (0.82), SO2 (1.46) and (COOH)2 (0.73) represent some diurnal difference. The ratios for NH3 (1.06), and HONO (0.9) are not indicative of any significant diurnal pattern. The concentration of HONO in Lahore is remarkably higher than concentrations measured in some metropolitan cities of the world, e.g., Seoul, Nara and New York City (Table 2). The secondary pollutants HONO and HNO3 are produced in the atmosphere from natural and anthropogenic emissions of primary gas-phase pollutant such as nitrogen oxides (NOx). Anthropogenic emissions of NOx are primarily from power generation, motorized transportation, and other industrial and domestic combustion processes (Finlayson-Pitts and Pitts, 2000). Nitrogen oxides react with hydroxyl radical and produce HNO3 and HONO via complex photochemical reactions. The key reactions of HNO3 and HONO formation are the homogeneous reactions (Finlayson-Pitts and Pitts, 2000):

1.46 0.81–2.65 Nara, Japan (Matsumoto and Okita, 1998) Mean NA Range

1.61 0.26–3.99

0.26 0.14–1.18

0.15 0.15–0.16

2.99 0.2–41.8

0.548 0.15–3.47 0.32 0.14–2.55 10.5 0.34–106

3.25 0.33–48.5

NA NA NA 1.30 3.22

Eastern North Carolina, USA (Mcculloch et al., 1998) Mean 0.74 Range 0.32–3.51

26.0

2.78

3.89

25.9 8.90 1.5 0.43 NA NA NA

NA

NA

6.00 NA

3.2. Gaseous species

)

1.09 4.51 NAa

17.3

19.4 2.11–43.1 19.6 0.05–48.9 1.16 0.08–2.52

1.00 0.18–2.63

4.34

18.9 3.85–35.9 7.43 1.76–25.7 50.1 21.2–81.3

8.70

19.2 6.48–39.2

NHþ 4 SO2 4 NO 3 CL NH3 SO2 HNO3 HONO

during the fog episode. Fog and associated higher RH seemed to have caused 14% reduction, on average, in PM2.5 concentrations compared to that during the clear episode. A further drop in PM2.5 concentration due to scavenging of particle upon fog formation would be expected. Nevertheless, carbonaceous aerosol, which is hydrophobic by itself, constituted around two-third of the total PM2.5 mass. While fresh soot is hydrophobic, it can mix with ammonium sulfate and become activated with aging (Husain et al., 2007). But, it revealed from the SO2 4 to Se ratio that Lahore aerosols were not aged (discussed in Section 3.4). Moreover, liquid water contents of Lahore fog droplets were estimated to be 0.05, 0.1 and 0.15 g m3, respectively, during light, medium, and heavy fog. Dutkiewicz and Husain (1998) observed the bulk scavenging efficiency for SO2 4 , which is hydrophilic, to vary only 15% (i.e. from 0.55 to 0.65) with a 74% increase of liquid water content (L) (i.e. from 0.24 to 0.91 g m3). Therefore, 50–67% increase in liquid water content is less likely to cause as much scavenging of predominantly hydrophobic aerosols in Lahore as to observe a discrete difference between PM2.5 concentrations during clear- and fog episodes.

a

Not available.

1.66 1.19–2.27

0.45 New York City, USA (Bari et al., 2003) Mean 15.7

Hong Kong (Pathak et al., 2003) Mean NA

Seoul, Korea (Lee et al., 1999) Mean 57.0

209 82.7–450 Lahore(this study) Mean Range

Atmospheric concentration (mg m

3

HCl PM2.5 Type of samples

Table 2 Gaseous and ionic concentrations in the PM2.5 of Lahore relative to other locations of the world

4.19

K.F. Biswas et al. / Atmospheric Environment 42 (2008) 7775–7785 16.1 5.73–39.7

7780

OH þ NO2 / HNO3

(1)

NO þ OH þ M / HONO þ M

(2)

HONO þ hn / OH þ NO

(3)

However, the important reactions to produce HNO3 and HONO during nighttime are heterogeneous hydrolysis of

K.F. Biswas et al. / Atmospheric Environment 42 (2008) 7775–7785

N2O5 and NO2 on soot, SO2 4 and sea salt aerosols via the following reactions: N2 O5 þ H2 O / 2HNO3

(4)

2NO2 þ H2 O / HONO þ HNO3

(5)

Nitrous acid (HONO) formed in reaction (2) is rapidly photolyzed in reaction (3) at the wavelengths 400 nm during the day (Calvert et al., 1994). Thus, the HONO is mainly accumulated at night and is subsequently photolyzed by OH after sunrise (Staffelbach et al., 1997; Acker et al., 2005). However, only 10% drop in daytime HONO concentrations was evident relative to nighttime concentrations in Lahore aerosols. It is to be noted that several sampling days were fog-covered and/or the sky was overcast. Hence, it was less likely to observe the intuitive effect of photolysis on daytime HONO concentration. Reactions (4) and (5) could be responsible for the higher HNO3 concentration during the night in Lahore. It is to be noted that N-containing species such as (NOx) and peroxyacetylnitrate (PAN) are potential interfering  agents and can produce NO 2 and NO3 inside the Na2CO3 coated denuder resulting in the overestimation of HONO and HNO3 concentrations (Bari et al., 2003). PAN thermally decomposes to peroxyacetyl radical and NO2, hence the artifact from it will be similar to NO2. In urban areas, the artifact correction was observed to be less than that observed in rural, semirural and suburban areas, and in general ranged from 2 to 15% for HONO and 3.4 to 13.2% for HNO3 (Koutrakis et al., 1988; Perrino et al., 1990). The atmospheric concentration of HCl was generally higher during the day. In general, HCl is mainly produced by coal combustion and refuse incineration, and by municipal and industrial waste incineration, as a primary pollutant, and by the reaction of acidic gases (e.g. HNO3, H2SO4, NO2) with NaCl (s, aq) in salt particles as a secondary pollutant (Eldering et al., 1991). Sea salt aerosols were not supposed to be significant precursors of atmospheric HCl in Lahore, because, first, Cl/Naþ in Lahore aerosol (9.76) was higher than the corresponding ratio (1.16) in seawater (Stumm and Morgan, 1996) by nearly a factor of 9, and second, while Lahore is located 1000 km north of the Arabian sea coast, the prevailing wind flow was from the farther north of Lahore during our sampling period. In Lahore, HCl is likely to be a primary pollutant emitted from anthropogenic operations, e.g. coal burning, brick manufacturing and incineration of wastes (Harrison et al., 1997). Incineration primarily takes place during the day, and hence, daytime HCl concentration was generally found to be elevated. Daytime concentration of SO2 was also higher compared to nighttime. The main sources of SO2 emission in an urban area like Lahore were likely to be fossil fuel usages in industry and transport (Harrison et al., 1997; Husain et al., 2007). The average concentration of oxalic acid, (COOH)2, was elevated during the night in Lahore. Vehicular traffic, emissions from soil and vegetation, burning of coal, waste and biomass, and secondary formation from anthropogenic or natural gas-phase precursors are the major sources of oxalic acid (Wang et al., 2007). Although the vehicular

7781

traffic slows down in the night in Lahore, the concentration of oxalic acid tend to increase, because emission sources, e.g., wood or coal burning for cooking or heating, and soil and vegetation, still continue to produce oxalic acid. Unlike formic or acetic acid (monocarboxylic acids), oxalic acid is likely to be associated with particles (Andreae et al., 1988; Khwaja, 1995). More discussion on water-soluble oxalic acid follows in Section 3.3. In the atmosphere, NH3 and its protonated form, NHþ 4, are ubiquitous. Ammonia is the primary basic gas in the atmosphere, and plays a critical role in the formation of fine particulates through its reactions with HNO3 and H2SO4. The concentration of NH3 was exceedingly high in the Lahore samples relative to values reported for locations of the world (Table 2). In general, the sources of NH3 emission include direct volatilization from livestock farming, mineral fertilizers (particularly urea), agricultural crops and a wide range of non-agricultural facilities and/or activities, including sewage, catalytic converters, and industrial processes (Sutton et al.,1995, 2000). The residence times of NH3 can be a few hours to a few days (Asman and van Jaarsveld, 1992). Lahore and its surrounding areas have very large farming and livestock. Hence, the exceedingly high NH3 concentrations must be derived from decomposition of agricultural and animal wastes. Because this source of NH3-emission is fairly continuous, concentration of NH3 in Lahore did not show any significant diurnal variation (Table 1). 3.3. Aerosol species Concentrations (average  SD and range) of water2 þ þ þ 2þ and soluble ions (i.e., Cl, NO 3 , SO4 , Na , NH4 , K , Mg Ca2þ) in Lahore PM2.5 are summarized for fog and for clear episodes averaged with daily and nightly data in Table 1.   Concentrations of BrO 3 , NO2 and Br were, with a few exceptions, below the detection limits and are not discussed in this paper. A strong diurnal difference in aerosol ion concentrations is evident from the ratios between daily and nightly values. For instance, during clear episodes, day2 þ þ to-night concentration ratios of NO 3 , SO4 , Na , NH4 , Mg2þ, and Ca2þ were, respectively, 0.68, 0.66, 0.5, 0.64, 2.74 and 2.37, while ratios for other ions varied from 0.85 to 2 unity. The chemistry of NO 3 and SO4 is usually complicated by photochemical reactions. The relatively low atmospheric mixing height and the slow vertical mixing process during the night can be identified as major reasons for the elevated nocturnal concentrations of PM2.5 and most of the aerosol ions. Building construction and demolition activities and traffic-induced soil erosion are likely to have contributed to the higher average Mg2þ and Ca2þ concentrations during daytime than at night, during the clear episodes. Oxalic acid has been detected as the major fraction of water-soluble organic compounds in urban, rural, and even remote backgrounds (Wang et al., 2007). The concentration of particulate oxalic acid (C2O2 4 ) in Lahore was comparable to the levels in Shanghai, Tokyo, Hong Kong, Los Angeles, Chiba, New York City and Vienna, suggesting similar conditions occurring among those big cities (Wang et al., 2007 and references therein). Oxalate showed strong 2 (r2 ¼ 0.84) and NO correlation with SO2 4 3 (r ¼ 0.80).

K.F. Biswas et al. / Atmospheric Environment 42 (2008) 7775–7785

Previous studies showed that in-cloud formation and heterogeneous formation can yield a good correlation and C2O2 (Yao et al., 2002, 2003). The between SO2 4 4  robust correlation that we found between C2O2 4 and NO3 suggested that motor vehicle exhaust hydrocarbon emissions that are perhaps one of the major sources of oxalate. Oxalate also showed significant correlations with Kþ 2 þ (r2 ¼ 0.69) and NHþ 4 (r ¼ 0.72). Because fine-mode K and are accepted tracers of biomass burning (Andreae and NHþ 4 Merlet, 2001), the strong correlation is an indication of biomass burning as source of these species. Karthikeyan and Balasubramanian (2006) reported a significant increase in the concentrations of oxalic acid fine particles during biomass burning in Singapore. The average NHþ 4 concentration in Lahore was a factor of 16 higher than the concentration measured in an NH3-rich environment at a hog farm in eastern North Carolina, USA 2 (Table 2). The strong correlation of NHþ 4 with SO4 2 2 2 (r2 ¼ 0.87), NO (r ¼ 0.77) and with C O (r ¼ 0.72) 3 2 4 indicate that the ions coexist in PM2.5 (Fig. 3). A molar ratio 2  2 of 1.24  0.18 for NHþ 4 /(2SO4 þ NO3 þ 2C2O4 ), and the strong correlation (m ¼ 1.27, r2 ¼ 0.89) between [NHþ 4 ] and  2 2[SO2 4 ] þ [NO3 ] þ 2[C2O4 ], also suggest that Lahore 2 aerosols are neutralized. The molar ratio of NHþ 4 /SO4 equaled 4.6  0.97, indicating that ammonium sulfates in Lahore air mostly exist in the form of (NH4)2SO4 rather than NH4HSO4. 3.4. Emission sources for Lahore aerosols

0.012

SO4 NO3 C2O4 0.25

0.008 0.15

[C2O4]/µmol m-3

Given that the average wind speed was 3.5 m s1 during the sampling period, emissions from point sources such as factory stacks in Lahore, and its neighboring regions in Pakistan and northern India, are likely to be the major sources of atmospheric pollutants in Lahore aerosols and gases. There are textile industries in Faisalabad and Lahore, steel mills in Lahore, factories of light engineering and electrical goods in Gujarat and Gujranwala, petroleum refineries, thermal power plants and fertilizer manufacturers in Multan (Qadir and Zaidi, 2006). Brick kilns, in which the fuel used is often high-sulfur coal, are recognized to be a major source of air pollutants in peri-urban areas of Lahore as well as other cities of Pakistan. Emissions from vehicles and vegetation burnings are also considered to be major sources of pollutants like PM2.5, SO2, and NOx (www. rrcap.unep.org; Husain et al., 2007), as well as Kþ, NHþ 4 and C2O2 4 . An agriculture-based economy like that of Pakistan, is likely to have a major contribution of NH3 to the atmosphere from livestock farming, although biological degradation of agricultural wastes and sewage, and biomass burning, also are important sources in Lahore. Husain (1989) and Husain and Dutkiewicz (1992) made extensive use of SO2 4 to Se (selenium) ratios to infer the age of SO2 4 aerosols and hence the distance of emission sources. Both Se and SO2 are emitted in fossil fuel combustion, but Se condenses within w10 km and is largely (>90%) present in aerosols. Sulfur dioxide, in contrast, is slowly oxidized in the gas phase, at w1% h1 (Newman, 1981); it adds to aerosol SO2 resulting in 4 increase in the SO2 4 /Se ratio with time (Husain and Dutkiewicz, 1992). The SO2 4 /Se ratio in fresh aerosols is

typically w1000–1500. This ratio increases with time (aging) of air parcel from the source, as SO2 becomes oxidized to SO2 4 via gas and aqueous-phase mechanisms. Sulfate to Se ratios in Lahore aerosols were found to vary from 910 up to 2534, with a median of 1658 (and average ratio 1660  482), which may suggest that sources of emissions were not too distant. However, as there are several other parameters (e.g., fog, precipitation, wind speed), which can alter the rate of SO2 oxidation, only SO2 4 / Se ratio cannot be stand-alone way to determine the distance of emission sources. Air trajectories and chemical signatures can identify the source region(s) of emissions (Husain and Samson, 1979; Rahn and Lowenthal, 1984; Dutkiewicz et al., 1987). Threeday air mass backward trajectories showed that the sampled air masses predominantly originated from north and northwest, and occasionally southwest, of Lahore. According to the 3-day air parcel back trajectories of the sampling days the air masses with SO2 4 /Se ratio <1500 were found to originate from northern, northeastern, northwestern and western Pakistan, and from northern and northeastern India (Fig. 2a, c, d). Emission sources located upwind of Lahore in, e.g., Rawalpindi and Sialkot (northwest of Lahore), Faisalabad (west) and Multan (southwest), Amritsar (east), and Jalandhar and Ludhiana (southeast) were presumed to be major contributors in fresh or moderately aged aerosols. The backward trajectories in Fig. 2a correspond to the sample with the highest 2 þ concentrations of PM2.5, SO2 4 , C2O4 , K , HONO (respectively, 450, 37.6, 1.46, 7.27, 48.9 mg m3), and very high concentrations of NHþ 4 and NH3 (respectively, 30.2 and 69.9 mg m3). Given the low wind speed, those species were most likely to be attributable to local pollution sources in Lahore, Amritsar, Faisalabad, and Multan. Trajectories in Fig. 2b correspond to the ‘cleanest’ among all samples, with low concentration of PM2.5 (82.7 mg m3) and other species

[SO4], [NO3]/µmol m-3

7782

m = 0.2, r 2 = 0.87 (SO4) m = 0.27, r 2 = 0.77 (NO3) m = 0.01, r 2 = 0.72 (C2O4)

0.05 0.0

0.5

1.0

1.5

0.004

NH4/µmol m-3 Fig. 3. Relationship of ammonium with sulfate, nitrate and oxalate.

K.F. Biswas et al. / Atmospheric Environment 42 (2008) 7775–7785

3.5. Effect of fog on SO2 oxidation and other gases

2.5

2.0

30

1.5

20

1.0

10

0.5

0

3

6

9

[HNO3] and [HCl]/µg m-3

[HONO]/µg m-3

HONO HNO3 HCl

40

0

60

40

20

Fogs typically form under conditions that promote pollutant build-up in the fog layer (stable conditions, low wind speeds). Under the conditions favoring fog formation, water condenses on pre-existing aerosol particles to form droplets. This condensed water offers the medium for aqueous-phase reactions in which atmospheric species, like SO2, are rapidly oxidized. The simultaneous gas dissolution, aqueous-phase reactions, and deposition of fog drops alter both the amounts and distribution of atmospheric species (Fahey et al., 2005), as evidenced by the lower concentrations of HONO, HNO3, HCl, SO2, (COOH)2, and NH3 during fog (Table 1). In Fig. 4, we have plotted the concentrations of HONO, HNO3 and HCl against the number of hours over which the fog persisted in the 12-h sampling period. A general trend of declining concentration of the gases is noticeable for the gases. Because of relatively higher relative humidity (up to >100%) during fog, these

50

80

% of S as aerosol sulfate

(except for NH3 at 78.4 mg m3) and the lowest SO2 4 /Se ratio, 910. Although the air mass originated in and passed over industrialized northern India, scattered showers on the previous day could have scavenged the particulate and gaseous species, producing the relatively low concentration. Fig. 2c presents a northwesterly air mass arriving at a moderate speed, giving rise to a PM2.5 of 270 mg m3, and moderate to high concentrations of other species. Pollution sources near the Afghanistan-Pakistan border and across Pakistan, along with local and/or regional sources such as those of Faisalabad and Lahore were likely to have been the origin of the measured species. Local emission sources were clearly the most important ones during the events of atmospheric stagnation, as seen in Fig. 2d.

7783

0.0

Hours of fog during the 12-h sampling period Fig. 4. Concentrations of acidic gases, HONO, HNO3, and HCl, as a function of duration of fog.

0

0

2

4

6

8

10

Hours of fog during the 12-h sampling period Fig. 5. Percentage of sulfur present as aerosol as a function of duration of fog.

gaseous species readily dissolve into fog droplets, and are eventually removed from the smaller size fractions such as PM2.5. Heterogeneous oxidation of SO2 in cloud and fog drops has been extensively studied in our laboratory and elsewhere (Husain, 1989; Husain and Dutkiewicz, 1990, 1992; Rattigan et al., 2002; Reilly et al., 2001; Fahey et al., 2005), as a function of both drop size and available oxidants. Because exceedingly high concentrations of S(IV) and S(VI) were measured at Lahore (Hameed et al., 2000 and this study), assessment of the impact of fog on gas-to-particle conversion of sulfur is important. Since much of the sulfur is emitted as SO2 in the combustion of fossil fuel, initially little sulfur is present as aerosol SO2 4 . However, gas- and and the aqueous-phase oxidation converts SO2 to SO2 4 proportion of the latter increases. Hence, the impact of fog can be examined by determining the fraction of total S present as SO2 4 , while total S is the sum of S present as SO2(gas) and SO2 4 (particle). In Fig. 5, we have plotted the fraction of S present as SO2 4 against the number of hours over which the fog persisted in the 12-h sampling period. An increase in the fraction of S as SO2 4 with greater fog duration is clearly evident. Samples representing at least 2 h of fog contained 50% of the S in oxidized form; samples representing larger fog durations contained as high as w80% of the S in oxidized form. 4. Conclusion The extent of pollution of Lahore is manifested in the average concentrations of PM2.5 mass (209 mg m3), NO 3 3 3 þ (18.9 mg m3), SO2 4 (19.2 mg m ), and NH4 (16.1 mg m ) that we have measured. High concentrations of HONO (average, 19.6 mg m3) and NH3 (average, 50 mg m3) are also indicative of a heavily polluted atmosphere. Local and regional emission sources, e.g., vehicular traffic, power

7784

K.F. Biswas et al. / Atmospheric Environment 42 (2008) 7775–7785

plants, manufacturing industries, livestock farms, agricultural biomass burning, and building activities, contribute the major proportion of pollutants in Lahore. Diurnal variations in the concentrations of measured species illustrated the significance of variability in the mixing height of the atmospheric boundary layer and other meteorological factors, e.g., wind speed, sun index, and relative humidity. The occurrence of fog was found to complicate the atmospheric chemistry and distribution of aerosols and gases. Fog clearly influenced the process of formation of particulate SO2 4 from SO2. Higher relative humidity during fog conditions enhanced uptake of gaseous species in fog droplets. Given the continuing increases in SO2 emissions in South Asia, it is reasonable to anticipate that fog will contribute more SO2 4 , and hence more PM2.5 mass. We believe that a comprehensive long-term study needs to be undertaken, to assess the interaction between fog and emitted pollutants in Lahore and other South Asian cities. This, of course, is a characteristic problem of all urbanized regions of the world.

Acknowledgements The financial support for this work was partially provided by the US National Science Foundation through grant number ATM 0503850. We thank A. Bari for preparation of the denuders; and Saleem Khan, Zulfiqar Ali, M. Mannan, and M. Khalid for their help in sampling; and Adil R. Khan for analysis of cations. One of us (LH) acknowledges the United Nations Development Program, and the National Talent Pool, Ministry of Labor, Government of Pakistan, for a travel grant to conduct the field campaign in Lahore, and to Raza Hussain for hospitality during his stay at SUPARCO, Karachi.

References Acker, K., Mo¨ller, D., Auel, R., Wieprecht, W., Kala b, D., 2005. Concentrations of nitrous acid, nitric acid, nitrite and nitrate in the gas and aerosol phase at a site in the emission zone during ESCOMPTE 2001 experiment. Atmospheric Research 74, 507–524. Andreae, M.O., Merlet, P., 2001. Emission of trace gases and aerosols from biomass burning. Global Biogeochemical Cycles 15, 955–966. Andreae, M.O., Talbot, R.W., Andreae, T.W., Harris, R.C., 1988. Formic and acetic acids over the central Amazon region, Brazil, dry season. Journal of Geophysical Research 93, 1616–1624. Asman, W.A.H., van Jaarsveld, H.A., 1992. A variable-resolution transport model applied for NHx in Europe. Atmospheric Environment 26A, 445–464. Bari, A., Ferraro, V., Wilson, L.R., Luttinger, D., Husain, L., 2003. Measurements of gaseous HONO, HNO3, SO2, HCl, NH3, particulate sulfate and PM2.5 in New York, NY. Atmospheric Environment 37, 2825–2835. Calvert, J.G., Yarwood, G., Dunker, A., 1994. An evaluation of the mechanism of nitrous acid formation in the urban atmosphere. Research on Chemical Intermediates 20, 463–502. Clarke, G.M., Cooke, D., 1983. A Basic Course Course in Statistics, second ed. Edward Arnold, London. Draxler, R.R., Rolph, G.D., 2003. HYSPLIT (HYbrid Single-Particle Lagrangian Integrated Trajectory) Model access via NOAA ARL READY Website (http://www.arl.noaa.gov/ready/hysplit4.html). NOAA Air Resources Laboratory, Silver Spring, MD. Dutkiewicz, V.A., Husain, L., 1998. Inefficient scavenging of aerosol sulfate by non-precipitating clouds in polluted air. Atmospheric Environment 32 (16), 2793–2801.

Dutkiewicz, V.A., Parekh, P.P., Husain, L., 1987. An evaluation of regional elemental signatures relevant to the northeastern Unites States. Atmospheric Environment 21, 1033–1044. Eldering, A., Solomon, P.A., Salmon, L.G., Fall, T., Cass, G.R., 1991. Hydrochloric acid: a regional perspective on concentration and formation in the atmosphere of Southern California. Atmospheric Environment 25A, 2091–2102. Fahey, K.M., Pandis, S.N., Collett Jr., J.L., Herckes, P., 2005. The influence of size-dependent droplet composition on pollutant processing by fogs. Atmospheric Environment 39, 4561–4574. Finlayson-Pitts, B.J., Pitts Jr., J.N., 2000. Chemistry of the Upper and Lower Atmosphere: Theory, Experiments, and Applications. Academic Press, San Diego. Hameed, S., Ishaq Mirza, M., Ghauri, B., Siddiqui, Z.R., Javed, R., Khan, A.R., Rattigan, O.V., Qureshi, S., Husain, L., 2000. On the widespread winter fog in northeastern Pakistan and India. Geophysical Research Letters 27 (13), 1891–1894. Harrison, R.M., Smith, D.J.T., Piou, C.A., Castro, L.M., 1997. Comparative receptor modelling study of airborne particulate pollutants in Birmingham (United Kingdom), Coimbra (Portugal) and Lahore (Pakistan). Atmospheric Environment 31 (20), 3309–3321. Husain, L., Samson, P.J., 1979. Long-range transport of trace elements. Journal of Geophysical Research 84 (C3), 1237–1240. Husain, L., 1989. A technique for determining in-cloud formation of SO4. Geophysical Research Letters 16 (1), 57–60. Husain, L., Dutkiewicz, V.A., 1990. A long-term (1975–1988) study of – regional contributions and concentration atmospheric SO2 4 trends. Atmospheric Environment Part A – General Topics 24 (5), 1175–1187. Husain, L., Dutkiewicz, V.A., 1992. Elemental tracers for the study of homogeneous gas phase oxidation of SO2 in the atmosphere. Journal of Geophysical Research 97, 14635–14643. Husain, L., Dutkiewicz, V.A., Khan, A.J., Ghauri, B.M., 2007. Characterization of carbonaceous aerosols in urban air. Atmospheric Environment 41, 6872–6883. Karthikeyan, S., Balasubramanian, R., 2006. Determination of watersoluble inorganic and organic species in atmospheric fine particulate matter. Microchemical Journal 82, 49–55. Khwaja, H.A., 1995. Atmospheric concentrations of carboxylic acids and related compounds at a semiurban site. Atmospheric Environment 29, 127–139. Koutrakis, P., Wolfson, J.M., Slater, J.L., Brauer, M., Spengler, J.D., Stevens, R. K., Stone, C.L., 1988. Evaluation of an annular denuder/filter pack system to collect acidic aerosols and gases. Environmental Science & Technology 22, 1463–1468. Lee, H.S., Kang, C.-M., Kang, B.-K., Kim, H.-K., 1999. Seasonal variations of acidic air pollutants in Seoul, South Korea. Atmospheric Environment 33, 3142–3152. Matsumoto, M., Okita, T., 1998. Long term measurements of atmospheric gaseous and aerosol species using an annular denuder system in Nara, Japan. Atmospheric Environment 32 (8), 1419–1425. McCulloch, R.B., Few, G.S., Murray Jr., G.C., Aneja, V.P., 1998. Analysis of ammonia, ammonium aerosols and acid gases in the atmosphere at a commercial hog farm in eastern North Carolina, USA. Environmental Pollution 102 (S1), 263–268. Newman, L., 1981. Atmospheric oxidation of sulfur oxide: a review as viewed from power plant studies. Atmospheric Environment 15, 2231–2239. Pathak, R.K., Yao, X., Chan, C.K., 2003. Sampling artifacts of acidity and ionic species in PM2.5. Environmental Science & Technology 38, 254– 259. Perrino, C., Santis, F.D., Febo, A., 1990. Criteria for the choice of a denuder sampling technique devoted to the measurement of atmospheric nitrous and nitric acids. Atmospheric Environment 24A, 617–626. Qadir, M.A., Zaidi, J.H., 2006. Characteristics of the aerosol particulates in the atmosphere in an urban environment at Faisalabad, Pakistan. Journal of Radioanalytical and Nuclear Chemistry 267, 545–550. Rahn, K.A., Lowenthal, D.H., 1984. Elemental tracers of distant regional pollution aerosols. Science 223, 132–139. Rattigan, O.V., Ishaq Mirza, M., Ghauri, B.M., Khan, A.R., Swami, K., Yang, K., Husain, L., 2002. Aerosol sulfate and trace elements in urban fog. Energy & Fuels 16 (3), 640–646. Reilly, J.E., Rattigan, O.V., Moore, K.F., Judd, C., Sherman, D.E., Dutkiewicz, V.A., Kreidenweis, S.M., Husain, L., Collett Jr., J.L., 2001. Drop size-dependent S(IV) oxidation in chemically heterogeneous radiation fogs. Atmospheric Environment 35, 5717–5728. Smith, D.J.T., Harrison, R.M., Luhana, L., Pio, C.A., Castro, L.M., Tariq, M.N., Hayat, S., Quraishi, T., 1996. Concentrations of particulate airborne

K.F. Biswas et al. / Atmospheric Environment 42 (2008) 7775–7785 polycyclic aromatic hydrocarbons and metals collected in Lahore, Pakistan. Atmospheric Environment 30, 4031–4040. Staffelbach, T., Neftel, A., Horowitz, L.W., 1997. Photochemical oxidant formation over Southern Switzerland: 2. model results. Journal of Geophysical Research 102, 23363–23373. Stumm, W., Morgan, J.J., 1996. Aquatic Chemistry, Chemical Equilibria and Rates in Natural Waters, third ed. John Wiley & Sons, Inc., New York. Sutton, M.A., Place, C.J., Eager, M., Fowler, D., Smith, R.I., 1995. Assessment of the magnitude of ammonia emissions in the United Kingdom. Atmospheric Environment 29, 1393–1411. Sutton, M.A., Dragosits, U., Tang, Y.S., Fowler, D., 2000. Ammonia emissions from non-agricultural sources in the UK. Atmospheric Environment 34, 855–869.

7785

Wang, Y., Zhuang, G., Chen, S., An, Z., Zheng, A., 2007. Characteristics and sources of formic, acetic and oxalic acids in PM2.5 and PM10 aerosols in Beijing, China. Atmospheric Research 84, 169–181. Yao, X., Fang, M., Chan, C.K., 2002. Size distributions and formation of dicarboxylic acids in atmospheric particles. Atmospheric Environment 36, 2099–2107. Yao, X., Fang, M., Chan, C.K., Hu, M., 2003. Size distributions and formation of ionic species in atmospheric particulate pollutants in Beijing, China: 2 – carboxylic acids. Atmospheric Environment 37, 3001–3007. Yu, X.-Y., Lee, T., Ayers, B., Kreidenweis, S.M., Malm, W., Collett Jr., J.L., 2006. Loss of fine particle ammonium from denuded nylon filters. Atmospheric Environment 40, 4797–4807.