ARTICLE IN PRESS

Atmospheric Environment 39 (2005) 3771–3784

www.elsevier.com/locate/atmosenv

The ion chemistry and the source of PM2.5 aerosol in Beijing

Ying Wanga,b, Guoshun Zhuanga,b,c,, Aohan Tanga, Hui Yuana, Yele Suna,
Shuang Chena, Aihua Zhenga
a
Department of Chemistry, Center for Atmospheric Environmental Study, Beijing Normal University, Beijing 100875, China
b
Department of Environmental Science and Engineering, Center for Atmospheric Chemistry Study, Fudan University,
Shanghai 200433, China
c
The Institute of Atmospheric Physics, CAS, NZC/LAPC, Beijing 100029, China

Abstract

Daily PM2.5 aerosol samples were collected at five sites in Beijing for a 3-year period from 2001 to 2003.
Concentrations of the water-soluble ions (SO2



3




4 , NO3 , Cl , F , PO4 , NO2 , CH3COO , HCOO , MSA, C2O4 ,
2

+ 2+ + 2+ +
NH4 , Ca , K , Mg , Na ) and 23 elements were measured for a total of 334 samples. A relatively even spatial
distribution throughout Beijing and a significant seasonal variation were observed. SO2


+
4 , NO3 , Cl , NH4 , Ca
2+
, and
+
K were the major ions and existed mainly in the form of (NH4)2SO4, NH4NO3, NaCl, KCl, and CaCl2 in aerosol
particles. Most ions showed high concentrations in winter and low in summer. Secondary ions, mainly SO2
4 , NO3 ,

exhibited high concentrations in both summer and winter due to the secondary transformation accelerated under high
humidity and strong solar radiation in summer and the higher concentration of SO2 from coal burning and the lower
removal rate in winter. The formations of SO2
4 and NO
3 were determined largely by temperature and NH4 ,
+

respectively. Temperature, relative humidity, rainwater frequency, and air mass origin might be the main factors
regulating the aerosol distribution. Crustal ions exhibited sporadic but high peaks in spring due to the intrusion of dust
from west and northwest of China. The Ca2+/Al ratio was used to indicate the mixing of different dust sources. Factor
analysis showed that the secondary formation of coal/biomass burning products, crust, industrial and traffic emissions
were the major sources of the fine aerosols in Beijing. Traffic source became more significant with motorization in recent
years.
r 2005 Elsevier Ltd. All rights reserved.

Keywords: Water-soluble part; Ions; Speciation; Seasonal variation; Secondary transformation; Sources

1. Introduction quality, has been listed as one of the top ten pollution
cities in the world, as it has experienced a rapid increase
As the capital of China, Beijing (39.9N, 116.4E) with in both energy consumption and vehicle quantities for
14.6 million inhabitants distributed over 16,800 km2 the past two decades, plus the invaded dust from outside
has been facing serious particulate pollution in air Beijing throughout the entire year, especially in spring.
While many studies on Total Suspended Particulate
Corresponding author. Department of Chemistry, Center (TSP, particle size smaller than 100 mm) and particle size
for Atmospheric Environmental Study, Beijing Normal Uni- smaller than 10 mm (PM10) have been reported since the
versity, Beijing 100875, China. Tel.: +86 21 55664579; fax: 1980s (Winchester et al., 1981; Cheng et al., 2000; Kim
+86 10 82902844. et al., 2002; Xiao and Liu, 2004), there were much less
E-mail address: [email protected] (G. Zhuang). studies on particle size smaller than 2.5 mm (PM2.5),

1352-2310/$ - see front matter r 2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.atmosenv.2005.03.013
 ARTICLE IN PRESS
3772 Y. Wang et al. / Atmospheric Environment 39 (2005) 3771–3784

which is commonly defined as fine particles. The first pollutants as well as their impact on air quality, we
comprehensive study on PM2.5 in Beijing was done in present a 3-year measurement of the aerosol compo-
1989–1990 (Chen et al., 1994). The average concentra- nents, their chemical forms, the seasonal variations at
tion of PM2.5 for the four seasons was reported to be in five sites in Beijing, and also investigated the possible
the range of 70–90 mg m
3. He et al. (2001) and Ye et al. factors determining the variation and the formation
(2003) reported SO2


4 , NO3 , and NH4
+
to be the mechanisms of secondary ions to explain the observa-
dominant ionic species in PM2.5, which accounted for tion, and the relative contribution of both natural and
about one-third of the total PM2.5 mass in Shanghai and anthropogenic sources.
Beijing. Yao et al. (2002) investigated the formation of
SO2


4 and NO3 in PM2.5 in understanding the origin of
these species. They found that a large part of these 2. Experimental
species might be formed through the direct emissions of
SO2, NOX, and NH3. Hu et al. (2002) studied the 2.1. Sampling
seasonal variation of ionic species in fine particles in
Qingdao during 1997–2000. Their results showed that Aerosol samples of PM2.5 were collected in four
the concentrations of major ions, SO2

+
4 , NO3 , NH4 , seasons from 2001 to 2003 at five sampling sites, i.e. (1) a

+
Cl , and K , were higher in the heating season than in traffic site, located in the campus of the Beijing Normal
the non-heating season. Energy consumption and University (BNU) between the 2nd and 3rd Ring Roads,
vehicle emission are found to be major sources of PM2.5. (2) an industrial site near the Capital Steel Company
PM2.5 particles can more readily penetrate into lungs (CS), (3) a residential site, Yihai Garden (YH), located
and increase respiratory and mutagenic diseases near the South 4th Ring Road, and (4) the rural sites at
(Shwartz et al., 1996; Hughes et al., 1998). There is a Miyun (MY) and Pinggu (PG) in the suburban areas of
widely accepted hypothesis that chemical components of Beijing, China, using a medium-volume sampler (model
particles and their capacity to carry potentially toxic (TSP/PM10/PM2.5)-2; ow rate, 77.59 L min
1). The
substances may be the main factors for health effects, traffic site is located on the roof (40 m high) of the
but it is not yet clear as to which factors are determinant. Science and Technology Building at BNU. The indus-
It was reported that water-soluble ions such as sulfate, trial site is located on the roof (4 m high) of a building,
nitrate, and other acid-rain-related pollutants had severe which is close to the Capital Steel Company. The
effects on human health (Raizenne et al., 1996). Besides, residential site is located on the roof (40 m high) of a
some epidemiological data indicated that the possible residential building. The three sampling sites in urban
seasonal effects of particulate, for example, there were areas could basically be the representatives over
larger number of deaths and hospitalizations with the urban Beijing. The location of sampling sites was shown
same particulate exposition in summer rather than in in Fig. 1 (a). PM2.5 were collected on Whatmans 41
winter (Ostro, 1995; Michelozzi et al., 1998). Therefore, filters (Whatman Inc., Maidstone, UK) for element and
understanding the composition, variation, sources, and ion analysis. The sampling time was nominally 12 h with
the formation of secondary ionic species, with an sampling starting at 8:00 a.m. every day. The samples
emphasis on sulfate and nitrate, has been the key issue were put in polyethylene plastic bags right after
and hence the urgency in the study on air quality and its sampling and preserved in a refrigerator. All those
control policy. filters were weighed before and after sampling with an
Coal is used both industrially and domestically in analytical balance (Sartorius 2004MP, reading precision
Beijing and SO2 is mainly got from coal burning. In 10 mg) after stabilizing under constant temperature
recent years, as the clean fuels such as natural gas and (2075 1C) and humidity (4072%). A total of 334
liquefied petroleum gas have replaced coal to be a part aerosol samples were collected and used in this study.
of the domestic heating, and also because low-sulfur All the procedures were strictly quality-controlled to
coal is recommended for industrial use, SO2 emissions avoid any possible contamination of the samples.
have been reduced. On the contrary, NOx emissions and,
in turn, nitrate as well as fine particle PM2.5, have 2.2. Chemical analysis
increased rapidly, as the amount of vehicles in Beijing
has grown up at a rate of 15% per year since 1990, and 2.2.1. Ion analysis
has now exceeded 2.1 million. In addition, the industrial One-fourth of each sample and blank was extracted
emissions from 4019 factories, currently in Beijing, may ultrasonically by 10 ml water, which was deionized to a
have worsened the surrounding environment. resistivity of 18 MO cm
1. After passing through micro-
With the development of economy and the control of porous membranes (pore size, 0.45 mm; diameter,
pollutant emissions, the status of air quality in Beijing 25 mm; made by the affiliated plant of Beijing chemical
might be different from years ago. To understand the school), the filtrates were determined for pH with a pH
current air pollution and the characteristics of air meter (model, Orion 818). Each filtrate was stored at
 ARTICLE IN PRESS
Y. Wang et al. / Atmospheric Environment 39 (2005) 3771–3784 3773

Fig. 1. (a) Sampling sites, (b) representative back trajectory on 20 March 2002 ending at Beijing; the triangles represent air mass
locations at 6-h intervals and (c) the four sectors got from the classification of 2-day back trajectories ending at Beijing.

4 1C in a clean tube for analysis. Ten anions (SO2
4 , 2.2.2. Element analysis
3

NO





3 , Cl , F , PO4 , NO2 , CH3COO , HCOO ,


The sample filters were digested at 170 1C for 4 h in a
2
+ 2+ +
MSA, C2O4 ) and five cations (NH4 , Ca , K , high-pressure Teflon digestion vessel with 3 ml concen-
Mg2+, Na+) were analyzed by Ion Chromatography trated HNO3, 1 ml concentrated HCl, and 1 ml concen-
(IC, model, Dionex 600), which consists of a separation trated HF. After cooling, the solutions were dried, and
column (Dionex Ionpac AS11 for anion and CS12A for then diluted to 10 ml with deionized water (resistivity of
cation), a guard column (Dionex Ionpac AG 11 for 18 MO cm
1). Total 23 elements (Al, Fe, Mn, Mg, Ti,
anion and AG12A for cation), a self-regenerating Sc, Na, Eu, Ce, Sr, Ca, Co, Cr, Ni, Cu, Pb, Zn, Cd, V, S,
suppressed conductivity detector (Dionex Ionpac As, Se, and Sb) were determined by inductively coupled
ED50), and a gradient pump (Dionex Ionpac GP50). plasma atomic emission spectroscopy (ICP-AES, model,
The gradient weak base eluent (76.2 mM NaOH+H2O) ULTIMA, made by JOBIN-YVON Company, France).
was used for anion detection, while the weak acid eluent The detailed analytical procedures are given elsewhere
(20 mM MSA) for cation detection. The recovery of (Zhuang et al., 2001; Sun et al., 2004). Elements were
each ion was in the range of 80–120%. The relative used as the indicators of different sources in this study.
standard deviation (SD) of each ion was less than 5%
for the reproducibility test. The limits of detection 2.3. Meteorological data and the trace gases
(S=N ¼ 3) were less than 0.04 mg L
1 for anions and
0.006 mg L
1 for cations. The quality assurance was The meteorological data, including temperature,
routinely carried out by using Standard Reference atmospheric pressure, wind speed, relative humidity
Materials (GBW 08606) produced by the National (RH), cloud cover, vapor pressure etc. was downloaded
Research Center for Certified Reference Materials, from National Climate Data Center (http://cdc.cma.-
China. Blank values were subtracted from sample gov.cn). Air Pollution Index (API ¼ 100 corresponds to
determinations. The details are given elsewhere (Yuan the Chinese air quality standard II) of PM10, SO2, NO2,
et al., 2003). CO, and O3 in Beijing were collected from Beijing
 ARTICLE IN PRESS
3774 Y. Wang et al. / Atmospheric Environment 39 (2005) 3771–3784

Environmental Protection Bureau (http://www.bjepb. oxidation might be an important process in the

gov.cn) and converted to concentrations. The formulas transformation of trace gases. Besides, there was a slight
used for converting API to concentration is decrease of SO2 from 2001 to 2003, which appeared as a
result of coal that has been replaced by clean fuels step-
C ¼ C low þ ½ðI
I low Þ=ðI high
I low Þ ðC high
C low Þ, by-step in Beijing in recent years.
Based on the calculated data, in Beijing summer was
where C is the concentration and I is the API value. Ihigh relatively wet with monthly average amount of pre-
and Ilow, the two values most approaching to value I in cipitation of 66.1, 57.7, 34.2 mm for June, July, and
the API grading limited value table, stand for the value August, respectively, while winter was very dry with
larger and lower one than I, respectively; Chigh and Clow precipitation of 0.0, 9.6, 2.9 mm for December, January,
represent the concentrations corresponding to Ihigh and and February, respectively. The ambient temperature in
Ilow, respectively. Beijing started to increase in April and reached a
Daily and seasonal variations of SO2 and NO2 from maximum in July–August (approximately 27 1C), then
2001 to 2003 are presented in Fig. 2. Based on their decreased to a minimum of
5 1C in January. In
variations, four seasons were defined: spring (March, summer, the removal of aerosol particles by wet
April, May), summer (June, July, August), autumn deposition is maximal and the possibility for photo-
(September, October, November), and winter (Decem- chemical formation of aerosol particles from their
ber, January, February). The concentration of SO2 in gaseous precursors is also at a maximum (high O3
winter (average 163.65 mg m
3) was 11 times higher than concentration and high temperature), whereas in winter
that in summer (average 14.84 mg m
3). As SO2 is mainly the primary emission is most efficient and may be
from coal burning in Beijing, the enhanced coal burning accumulated in the atmosphere caused by the low
for heating might lead to the high concentration of SO2 inverse layer. The wind speed was higher in spring
in winter. The slight higher NO2 in winter might be due (average 3.5 m s
1) and autumn (average 3.8 m s
1) and
to burning activities, traffic, and industrial emissions. O3 lower in summer (average 2.0 m s
1) and winter (average
showed high values in summer (average 118.67 mg m
3) 1.8 m s
1), indicating that the long-range transport of
and low in winter (average 30.47 mg m
3). As O3 aerosols from outside Beijing is efficient in spring and
indicates the oxidation ability of atmosphere, its high autumn, especially in the spring dust periods (with the
concentration in summer suggests that photochemical average wind speed of 5.5 m s
1).

500
SO2
400

300

200

100

0
250
NO2
200
150

100
50
0
00/01
00/03
00/05
00/07
00/09
00/11
01/01
01/03
01/05
01/07
01/09
01/11
02/01
02/03
02/05
02/07
02/09
02/11
03/01
03/03
03/05
03/07
03/09
03/11

Fig. 2. Daily and seasonal variations of SO2 and NO2 from 2001 to 2003 in Beijing. (Unit: mg m
3. The lines represent the Chinese
standard level II: 150 for SO2, and 120 for NO2, respectively).
 ARTICLE IN PRESS
Y. Wang et al. / Atmospheric Environment 39 (2005) 3771–3784 3775

2.4. Air mass trajectories 3. Results and discussions

To identify the potential importance of different 3.1. PM2.5 mass concentrations

source regions on aerosol composition at the sampling
sites, the air mass trajectories were calculated using the Mass concentrations of PM2.5 in Beijing were in the
HYSPLIT 4 Model of the Air Resources Laboratory of range of 11.1–1393.0 mg m
3, with a mean and SD of
NOAA (Draxler and Rolph, 2003). On every sampling 154.37145.7 mg m
3 as shown in Table 1, which were
day, a 2-day back trajectory was computed at 16:00 or higher than the values of 77.5 mg m
3 in 1989 (Chen et
4:00 h UTC, and at 1000 m above the starting point, al., 1994), 129.0 mg m
3 in 1999 and 2000 (Yao et al,
located at ground level. FNL Meteorological data have 2002), and 109.6 mg m
3 in 2001 (Wang et al., 2004)
been used as the input. The model output is a set of measured in Beijing by other studies. Compared with the
latitude–longitude coordinates of the air parcel esti- daily average of the standard value of 65 mg m
3 for
mated position. A representative trajectory is shown in PM2.5 (US EPA, 1997), almost 79% of the daily PM2.5
Fig. 1(b). With the statistical analysis of all these mass concentrations in 2001–2003 were higher than the
calculated trajectory directions, it has been found that standard in this work, showing that fine particle
the whole area could be classiinto five sectors with pollution is extremely terrible in Beijing.
different air . These sectors were shown in Fig. 1(c): (1) Strong temporal variations in PM2.5 mass concentra-
north–northeast (N–NE) sector represents the Northeast tions were observed in Beijing. As illustrated in Table 1,
Plain, (2) south–southeast–east (S–SE–E) sector repre- PM2.5 mass concentrations were highest in winter,
sents the eastern China as well as the western Pacific, (3) decreased through spring, autumn and tended to be
southwest (SW) sector represents the southwest China, lowest in summer. The PM2.5 mass averaged
(4) west–northwest (W–NW) sector represents the long 214.2 mg m
3 in winter, exceeding the 154.3 mg m
3
fetch of dust air masses, and (5) an additional sector annual average by 39%. The averages in spring, autumn,
(local) represents the short trajectories that come from and summer were 162.1, 105.2, and 93.3 mg m
3,
the local areas over Beijing, which are not shown in Fig. respectively.
1(c). These five sectors also represented long- or short- The daily and seasonal variations of PM2.5, trace
range transport. The first four sectors represent the long- gases, and meteorological parameters at BNU and their
range transport, while the last sector represents the correlations were shown in Fig. 3. Generally, PM2.5
short-range transport. appeared weakly correlated with these parameters, as

Table 1
Average concentrations (mg m
3) and the standard deviations (SD) of PM2.5 and ions in PM2.5 in four seasons in Beijing

PM2.5 Spring Summer Autumn Winter Total

Mean SD Mean SD Mean SD Mean SD Mean SD

No. 101 — 86 — 40 — 107 — 334 —

Mass 162.06 179.94 93.29 56.26 105.22 39.00 214.23 159.34 154.26 145.65
pH 5.99 0.87 5.48 0.49 5.92 0.54 5.10 0.89 5.57 0.84
Na+ 0.61 0.64 0.24 0.17 0.21 0.15 0.88 0.52 0.55 0.54
NH+ 4 6.47 6.75 10.10 6.97 6.33 5.80 10.64 8.83 8.72 7.66
K+ 1.09 0.97 1.29 1.25 0.76 0.74 2.48 2.16 1.55 1.63
Mg2+ 0.24 0.20 0.10 0.07 0.06 0.06 0.20 0.17 0.17 0.16
Ca2+ 2.54 2.46 0.73 0.60 1.16 1.70 1.68 1.67 1.63 1.90
F
0.25 0.22 0.10 0.09 0.09 0.09 0.55 0.36 0.29 0.31
CH3COO
0.21 0.57 0.15 0.26 0.06 0.10 0.22 0.39 0.18 0.41
HCOO
0.04 0.05 0.08 0.08 0.07 0.07 0.13 0.12 0.08 0.10
MSA 0.07 0.32 0.06 0.07 0.02 0.03 0.01 0.03 0.04 0.18
Cl
2.92 2.19 1.41 1.31 1.09 0.99 5.28 3.99 3.07 3.13
NO
2 0.64 0.70 0.27 0.26 0.12 0.13 0.40 0.42 0.41 0.51
NO
3 11.92 11.79 11.18 10.37 9.14 10.27 12.29 12.12 11.52 11.37
2

SO4 13.52 13.95 18.42 15.28 12.69 12.91 20.96 19.72 17.07 16.52
C2O2
4 0.43 0.36 0.25 0.16 0.32 1.01 0.36 0.28 0.35 0.44
PO3

4 0.17 0.24 0.43 0.41 0.28 0.25 0.21 0.16 0.26 0.29

No. means the number of samples measured.

 ARTICLE IN PRESS
3776 Y. Wang et al. / Atmospheric Environment 39 (2005) 3771–3784

PM2.5 SO2 PM2.5 O3 VP

500 300 3.0
400 240 2.5
2.0
300 180
1.5
200 120
1.0
100 60 0.5
0 0 0.0
01/12/01
01/12/03
01/12/05
01/12/07
01/12/09
01/12/11
01/12/12
01/12/14
01/12/16
01/12/18
01/12/20
01/12/29
01/12/31
02/01/03
02/01/06
02/01/09
02/12/01
02/12/05
02/12/07
02/12/11
02/12/13
02/12/17
02/12/19
02/12/25
02/12/29
02/12/31
03/01/05

01/07/09
01/07/12
01/07/14
01/07/16
01/07/20
01/07/27
01/08/03
01/08/09
02/06/18
02/06/21
02/06/27
02/06/30
02/07/03
02/07/07
02/07/10
02/07/13
02/07/16
02/07/19
Fig. 3. The correlations between PM2.5 and SO2 in winter (left), and between PM2.5, O3, vapor pressure (VP) in summer (right) at BNU
during 2001–2003. (left axis: PM2.5, SO2 and O3 in mg m
3, right axis: vapor pressure in kPa).

there might be many processes contributing to the sphere (Arimoto et al., 1996; Yao et al., 2002; Xiao and
2

formation of PM2.5, and the measured PM2.5 mass Liu, 2004). The seasonal variation of NO
3 /SO4 shown
measured could be the combination of all these in Fig. 4 ranged from 0.01 to 2.94 (mean ¼ 0.71,
processes. However, they showed clear correlations SD ¼ 0.48), comparable with the value of 0.67 measured
when the data were investigated seasonally. For during 2001–2003 in Beijing, higher than the value of
example, PM2.5 was well correlated with SO2 in winter, 0.58 measured during 1999–2000 in Beijing (Yao et al.,
with O3 and vapor pressure in summer. The high PM2.5 2002), much higher than that in Shanghai (0.43; Yao et
mass concentration in winter was most likely due to the al., 2002), Qingdao (0.35; Hu et al., 2002), Taiwan (0.20;
combination of increased emissions from heating Fang et al., 2002), and Guiyang (0.13; Xiao and Liu,
2

sources and the meteorological factors that limited 2004). Arimoto et al. (1996) ascribed high NO
3 /SO4
dispersion; while in summer PM2.5 might be partially mass ratios to the predominance of mobile source over
from the photochemical transformation and its low stationary source of pollutants (Arimoto et al., 1996). In
concentration was likely related to the high wet China, gasoline and diesel fuel contain 0.12% and 0.2%
deposition. Besides, several severe high PM2.5 episodes sulfur (by weight), respectively (Kato, 1996). The
occurred in spring due to the intrusion of dust from the estimated ratios of NOx to SOx from the emission of
west or northwest of China. For example, the highest gasoline and diesel fuel burning are 13:1 and 8:1,
value of daily average PM2.5 mass (1393.0 mg m
3) respectively. The sulfur content in coal is 1% and the
occurred when a super dust storm attacked Beijing on estimated ratio of NOx to SOx is 1:2 from coal burning.
20 March 2002. This was verified by the backward It is reasonable to use SO2

4 as an indicator of stationary
trajectory shown in Fig. 1(b), in which the air mass was emission and NO
3 of mobile emission. The results
shown to come from the northwest of China. Therefore, indicated that air pollution from vehicle emissions
it must be noted that the air pollution in Beijing might (mobile source) in Beijing has become sever compared
2

be under the influence of local emissions as well as long- with that to a few years ago. Seasonally NO
3 /SO4
range transport from outside areas. showed low values in winter (0.49) and summer (0.63)
compared with that in autumn (0.93) and in spring
3.2. Ion composition and speciation of PM2.5 (0.84). As shown in Fig. 2, NO2 was almost evenly
distributed all the year round, while SO2 was higher in
2

3.2.1. Ionic composition of PM2.5 winter. The lower NO
3 /SO4 value in winter could be
Water-soluble ions comprise a large part of aerosol due to the high SO2 emission. In summer, high
particles and play an important role in the atmosphere. temperature, high RH, and high radiation are more
The concentrations of various ionic species in the whole favorable for the formation of SO2
4 , and hypothetically
2

sampling period are shown in Table 1. The sum of ions the lower ratio of NO
3 /SO4 could be due to the higher
contributed an average of 30% of PM2.5 mass concen- SO2
4 in summer.
tration. The sum of the major cations (ammonium Fig. 5 showed the comparison of the ion concentra-
(19%), calcium (6%), potassium (4%)) and the major tions in different locations. The y-axis represented the
anions (sulfate (34%), nitrate (23%), chloride (7%)) ratio of the concentrations in the other sites to those we
contributed more than 90% of the total ion concentra- got in Beijing. It could be seen clearly that the aerosol
tion, which were the most important alkaline and acidic pollution was more serious in Beijing than those sites in
species in PM2.5, respectively. East Asia and Africa. Crustal species (Ca2+ as a
2

The mass ratio of NO
3 /SO4 has been used as an representative) exhibited higher concentrations in
indicator of the relative importance of mobile vs. Beijing than most of other cities. Pollution aerosol, such
stationary sources of sulfur and nitrogen in the atmo- as sulfate, in Beijing was the highest amongst those
 ARTICLE IN PRESS
Y. Wang et al. / Atmospheric Environment 39 (2005) 3771–3784 3777

3.0
2.5 0.49 0.93 0.63 0.84
2.0
NO3 /SO4 1.5
1.0
0.5
0.0

8 0.76 2.12 1.27 0.82

6
Ca2+/Al

0
Winter Autumn Summer Spring
2
2+
Fig. 4. Seasonal variations of NO

3 /SO4
and Ca /Al ratios at BNU during 2001–2003. (All the samples were grouped to the four
seasons. The last five points denoted samples collected during dust storm. The average values in each season were shown in the figure).

2.0
Cl- NO3-
2- +
Mass SO4 NH4 K+ Ca2+
Conc. ratio (other site/BJ)

1.5

1.0

0.5

0.0
BJ NJ SH QD XM HK TW Seoul Hanoi Cario

Fig. 5. Comparison of concentrations of PM2.5 and the major ions in various urban sites. (BJ: Beijing, China, (Yao et al., 2002); NJ:
Nanjing, China, (Wang et al., 2003); SH: Shanghai, China, (Yao et al., 2002); QD: Qingdao, China, (Hu et al., 2002); XM: Xiamen,
China, (Gao et al., 1996); HK: Hong Kong, (Ho et al., 2003); TW: Taiwan, (Fang et al., 2002); Seoul, Korea, (Park et al., 2004); Hanoi,
Vietnam, (Hien et al., 2004); Cario, Egypt, (Mahmoud et al., 2002))

urban areas reported in literature. The higher coal parameter used to denote the acidity of the aerosols
consumption combined with poor dispersion in winter, directly. It is well known that low pH values of aerosols
and the high photochemical reaction in summer could result from acidic matters such as sulfate, nitrate,
account for the very high concentrations of sulfate. chloride, and carboxylic anions, whereas basic water-
Levels of nitrate and ammonium in Beijing were higher soluble matters such as ammonium, calcium, and
by a factor of 3–4, compared to those at other sites, magnesium result in the increase in pH values. The
which was likely due to heavy traffic and other local mean value of 5.57, a little lower than the blank value,
anthropogenic activities in Beijing. Such high levels of indicated that although Beijing was not in the state-
ammonium, nitrate, and sulfate indicated that pollution controlled zone of acid rain, it would face the acid
from acidic and secondary aerosol pollutants in Beijing problem in the near future.
has been a critical issue in the control of air quality. The ion balance expressed by the sum of the
equivalent concentration (meq m
3) ratio of cation to
3.2.2. Acidity of PM2.5 anion (C/A) was an indicator to study the acidity of the
The pH of PM2.5 aerosols at Beijing ranged from 3.94 environment. The ratio calculated from all the measured
to 7.65 (mean ¼ 5.57, SD ¼ 0.84) with a blank value of ionic species ranged from 0.63 to 3.28 (mean ¼ 1.09,
5.74 as shown in Table 1. pH of the aerosol filtrate was a SD ¼ 0.48). The mean ratio of 1.09, which was close
 ARTICLE IN PRESS
3778 Y. Wang et al. / Atmospheric Environment 39 (2005) 3771–3784

to 1, indicated that almost all of the ion components had correlation coefficients decreased from 0.92 for
been quantified. When the total equivalents of anions NH+ 2
+

4 –SO4 , 0.88 for NH4 –NO3 to 0.71 for Na –F .
+

were plotted against the total equivalents of cations, the Therefore, the (NH4)2SO4 concentration was calculated
slope of the regression line was slightly lower than unity first, then to NH4NO3 and NaF, successively. The
(slope ¼ 0.87, R ¼ 0:92), which might be attributed to concentrations of the succeeding species were calculated
H+, which was not counted in the calculation, or due to based on the concentrations of the formed ions and the
parts of NH+ 4 that was probably vaporized into the gas foregoing species. Table 3 showed the calculated
phase. The highest values of both pH and C/A were on concentrations of those major species in PM2.5. Without
20 March 2002 when a super dust storm attacked counting the samples from the period of dust storm, the
Beijing, which indicated clearly that the dust from West/ average concentrations of (NH4)2SO4 and NH4NO3
Northwest China brought a large amount of crustal ranged from 16 to 30 and from 5 to 10 mg m
3,
species and could alleviate the acidity of the local respectively. The total secondary aerosol (TSA, refer
environment. With a closer consideration of the regres- to (NH4)2SO4 and NH4NO3 here only) comprised a
sion between cations and anions, an anion de major part of the total water-soluble ions (TWSI), from
(slope ¼ 1.61, R ¼ 0:80) was found in the dust storm 52.3% in spring to 72.0% in summer. However, during
period, which might be attributed to the presence of dust days, TSA only comprised 11.8% of TWSI. This
bicarbonate or carbonate in the samples that was not indicated that the formation of the secondary aerosols
measured with IC in this study. A positive correlation due to the conversion of gaseous precursors was
(R ¼ 0:90) was found when Ca2+ was plotted against significant in summer, but inhibited under dry weather
the anion de, indicating that bicarbonate or carbonate in the dust season.
could indeed account for the missing anion during the Sulfur oxidation ratio defined as SOR ¼ n-SO2
4 /(n-
dust period. The interaction between carbonate and the SO2
4 +n-SO2) and nitrogen oxidation ratio defined as
pollutant gases (SO2, NOx, HCl) or acidic particles NOR ¼ n-NO


3 /(n-NO3 +n-NO2) could be the indica-
(SO2



4 , NO3 , Cl ) might be a major route to alleviate tion of the secondary transformation processes. Fig. 6
the acidifying processes. showed the SOR and NOR obtained for PM2.5 aerosols
in Beijing. The average values of SOR were 0.08, 0.12,
3.2.3. The speciation of major ions 0.39, 0.19, and 0.07 in the dust storm, spring, summer,
The chemical forms of those major ions, i.e. SO2
4 ,
autumn, and winter periods, respectively. The corre-
NO

+
3 , Cl , NH4 , Ca
2+
, and K+, in aerosols in Beijing sponding values for NOR were 0.00, 0.05, 0.08, 0.04,
were identified by bivariate correlations. Table 2 showed and 0.05, respectively. SOR expresses the degree of
the correlation coefficients and linear regression equa- oxidation of sulfur in terms of the ratio of the sulfur in
tions among these major ions. It could be seen that sulfate to the total sulfur (in sulfate and sulfur dioxide).
NH+ 2

4 was closely correlated with SO4 . The slope of the
Similarly, the NOR expresses the degree of oxidation of
+ 2

regression between NH4 and SO4 (meq vs. meq) for the nitrogen in terms of the ratio of the nitrogen in nitrate to
whole data set was 1.14, which indicated the complete the total nitrogen (in nitrate and nitrogen dioxide).
neutralization of SO2
4 by NH+ 4 , and suggested that
Higher SOR and NOR suggest that the oxidation of
(NH4)2SO4, instead of NH4HSO4, was the major species gaseous species would occur and more secondary
formed by SO2
+
4 and NH4 . In addition, NH4NO3, NaF,
aerosols can exist in the atmosphere. Earlier studies
NaCl, and KCl were found to be the major chemical (Pierson et al., 1979; Truex et al., 1980) have reported
species in the aerosol particles based on their correlation that in the primary pollutant the value of SOR is lower
coefficients. The concentrations of these species have than 0.10. Ohta and Okita (1990) suggested that when
been calculated based on the individual ion concentra- the ratio value was greater than 0.10, photochemical
tions and their mutual relationships. For example, the oxidation of SO2 would occur in the atmosphere. SOR

Table 2
The correlation coefficients (R) and the linear regression equations between major ions

R F
C1
NO

3 SO2

4 NH+ 2

4 ¼ 0.08+1.14SO4

Na+ 0.71 0.55 0.27 0.35 NH+

4 ¼ 0.11+2.03NO3

NH+4 0.43 0.65 0.88 0.92 NH4 ¼ 0.06+0.78(SO2

+

4 +NO3 )
K+ 0.50 0.52 0.51 0.53 Na+ ¼ 0.01+1.02F

Mg2+ 0.45 0.30 0.18 0.22 Na+ ¼ 0.01+0.15Cl

Ca2+ 0.26 0.08 0.02 0.02 K+ ¼ Mg2++Ca2+ ¼ 0.10+0.37Cl

 ARTICLE IN PRESS
Y. Wang et al. / Atmospheric Environment 39 (2005) 3771–3784 3779

Table 3 Table 4
The concentrations (mg m
3) of major species in particles in The correlation coefficients between SOR, NOR, and other
four seasons in Beijing parameters

Spring Summer Autumn Winter Dust NH+

4 O3 Temperature Atmospheric RH
pressure
No.a 96 86 40 107 5
(NH4)2SO4 17.190 25.847 15.769 27.983 4.181 SOR 0.43 0.47 0.64
0.56 0.38
NH4NO3 8.683 9.527 5.284 9.566 0.000 NOR 0.76 0.08 0.13
0.18 0.38
NaF 0.512 0.183 0.178 1.113 0.536
Clorideb 4.157 2.378 1.844 6.616 2.983 Note: boldface indicates that the correlation coefficient is
TSAc/TWSId 0.523 0.720 0.648 0.591 0.118 higher than 0.60.

a
No.: number of samples measured.
b
Cloride: NaCl, KCl, MgCl2 and CaCl2.
c
TSA: total secondary aerosol ((NH4)2SO4 plus NH4NO3). (Seinfeld, 1986), the positive correlation of SOR with
d
TWSI: total water soluble ions. temperature would suggest a possible oxidation mechan-
ism of SO2 to SO2
4 as that the local gas phase oxidation
of SO2 by OH radical followed by the condensation or
0.5
SOR NOR
45 absorption into the particle phase. The positive correla-
Temp. NH4 tion of NOR with NH+ 4 would suggest that NH4NO3
35
0.4 might be the main chemical form of NO
3 . The
25 correlation among those factors shown in Table 4
0.3
15
indicated evidently that strong solar radiation (high
0.2 temperature), high O3 concentration, high RH, low
5 atmospheric pressure, along with high ammonia emis-
0.1 -5 sion in summer might accelerate the secondary conver-
sion, while the dry weather during the dust period and
0.0 -15
Dust Spring Summer Autumn Winter high atmospheric pressure during the winter might drive
this process to the lowest degree. The dominant
Fig. 6. Seasonal variations of SOR, NOR, and their relations
controlling factors for the oxidation of SO2 and NO2
to temperature (temp.) and ammonium. (Left axis: SOR, NOR;
might be temperature and NH+ 4 , respectively.
Right axis: temperature 1C, ammonium mg m
3).

of Beijing PM2.5 aerosols were higher than 0.10 in 3.3. The spatial distribution of ions
summer and autumn, comparable or lower than 0.10 in
dust period, spring, and winter. The result clearly The five sampling sites in studying the spatial
showed that secondary formation of SO2
4 from SO2 distribution of aerosols in Beijing were classified into
occurred in summer and autumn, but inhibited in dust two groups, the urban sites of BNU, CS, and YH and
period and winter. The distribution of NOR in the four the rural sites of MY and PG. The paired samples t-test
seasons was similar to those of SOR. NOR were (SPSS 11.0 for Windows) and a parameter of CD
generally lower than SOR. The lower NOR in Beijing (coefficient of divergence) were used to judge the
PM2.5 aerosols suggested that the secondary formation similarity or difference because there was a large spread
of NO
3 from NO2 weakly occurred in Beijing compared
in the concentrations of the various chemical compo-
to that of SO2
4 .
nents. CD was deas follows (Park and Kim, 2004):
To identify the factors controlling the formation of vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
u p 
secondary ions (SO2
and NO
u1 X xij
xik 2
3 ), the correlation
4
CDjk ¼ t ,
coefficients between SOR, NOR and other parameters p i¼1 xij þ xik
were calculated and the results were shown in Table 4
and Fig. 6. SOR positively correlated with temperature where, xij represents the average concentration for a
and negatively correlated with atmospheric pressure. chemical component i at site j, j and k represent two
NOR showed correlations with NH+ 4 . There are many sampling sites, and p is the number of chemical
mechanisms reported for the formation of SO2
4 from components. The chemical components include particle
SO2, such as gas phase reaction of SO2 and OH radical, mass concentration, pH, and 15 ionic species. If the CD
aqueous transformation processes (metal catalyzed approaches zero, the data from the two sites are
oxidation or H2O2/O3 oxidation), and in-cloud pro- considered to be similar. If the CD approaches one,
cesses, etc. Since the gas-phase oxidation of SO2 to SO2

4 they are considered to be very different. Based on the
by OH radical is a strong function of temperature results from t-test for paired samples used in this
 ARTICLE IN PRESS
3780 Y. Wang et al. / Atmospheric Environment 39 (2005) 3771–3784

comparison, the concentrations of all those components (0.85, 0.67, 0.62, and 0.49 meq m
3 for winter, summer,
in PM2.5 were not statistically different at the po0:05 spring, and autumn, respectively). Compared with the
level for urban and rural sites. Concentrations of PM2.5 seasonal trend of PM2.5 mass concentrations as win-
chemical components for the paired rural–urban sam- ter4spring4autumn4summer, it was clear that more
ples showed similarities with CD of 0.40, indicating that ions were formed and attached to particles in the
the concentrations of the chemical species in PM2.5 were summer period.
similar among the sites, i.e. there was no obvious Fig. 7 showed the seasonal variations of several
regional difference between urban and rural areas over typical ions in PM2.5. As the tracers of soil/dust, Ca2+
the entire city. As fine particles could be transported to a and Mg2+ were frequently observed in lower concentra-
much longer range for its longer residence time in the tions in summer (average: 0.73 mg m
3 for Ca2+,
atmosphere, PM2.5 fine particles could likely inter-mix 0.10 mg m
3 for Mg2+), which could be due to more
or intra-mix well with each other in the atmosphere and precipitation in this season, while in spring they were in
lead to a more even spatial distribution. higher concentrations (average: 2.54 mg m
3 for Ca2+
and 0.24 mg m
3 for Mg2+), which could be due to the
3.4. Seasonal variation of ions intense episodic peaks of mineral dust that originated
from west–northwest often observed as shown in the air
The seasonal variations for water-soluble ionic mass trajectories. The concentrations of Ca2+ and
components were shown in Table 1. Clear seasonal Mg2+ in non-dust peaks days in spring (average:
patterns for all the ionic species in PM2.5 were observed. 2.22 mg m
3 for Ca2+, 0.22 mg m
3 for Mg2+) were
The sum of the concentrations for cations was in the comparable to those days of a relative high level in
order of 0.81, 0.65, 0.60, and 0.43 meq m
3 for winter, winter (average: 1.68 mg m
3 for Ca2+, 0.20 mg m
3 for
summer, spring, and autumn, respectively. Concentra- Mg2+), as the dry season was favorable for the re-
tions dropped in a similar extent for the sum of anions suspension of soil particles.

2 - 15
F Ca 2+
10
1
5

0 0
25 - 2
Cl Mg2+
20
15
1
Concentrations (µ g m-3)

10
5
0 0
3 40
NO2- NH4+
30
2
20
1
10
0 0
60 100
NO3- SO42-
75
40
50
20
25
0 0
Winter Autumn Summer Spring Winter Autumn Summer Spring

Fig. 7. Seasonal variations in the concentrations of representative ions of different sources (soil: Mg2+, Ca2+; secondary: NH+

4 , NO3 ,
SO2




4 ; combustion: F , Cl ; traffic activities: NO2 ). (All the samples in 3 years were grouped to the four seasons. The last five points
denoted samples collected during the dust storm).
 ARTICLE IN PRESS
Y. Wang et al. / Atmospheric Environment 39 (2005) 3771–3784 3781

NH+ 2


4 , SO4 , and NO3 did not show the same magnitude of eigenvalues (greater than one), and these
seasonal variations as those of Ca2+ and Mg2+, six factors accounted for 80.03% of total variance in the
indicating that they were of different sources. Concen- entire set of the data. The factor was highly loaded in
trations of NH+ 4 in spring, summer, autumn, and winter NH+ 2


4 , S, SO4 , NO3 , and moderately loaded with Cl ,

were 6.47, 10.10, 6.33, 10.64 mg m
3, of SO2

4 were 13.52,
+

K , Se, F , and it explained 26.24% of the total system
18.42, 12.69, 20.96 mg m
3, of NO
3 were 11.92, 11.18, variance. This factor appears to represent the secondary
10.27, 12.12 mg m
3, respectively. The results showed pollution sources from the fossil fuel combustion, traffic
clearly that the concentrations of these secondary ions emission, plus the possible biomass burning. Se, Cl
,
were generally higher in summer and winter and lower in and F
might be mainly from coal burning. K+ was
spring and autumn. As the average concentration of SO2 likely from biomass burning, as in Beijing suburban
was 163.65 mg m
3 in winter, 11 times than that of areas, wheat straws and maize stalks are burned in field
14.84 mg m
3 in summer, the high SO2
4 concentration in during the autumn and winter. The second factor had
winter could be attributed to the higher concentration of strong factor loadings for Fe, Al, Mn, Ca2+, Mg2+, and
SO2 in this season, which was due to the higher coal Na+, which accounted for 22.00% of the total variance.
burning combined with the poor dispersion and the This factor could be identified as a crustal source,
lower rate of removal via wet deposition. The good including air-borne road dust, construction dust, and
correlation between SO2 and SO2
4 (R ¼ 0:35, n ¼ 107) fugitive dust. The third factor was highly loaded in Ni,
in winter supported such an explanation. In summer, Cu, and Zn, which explained a further 12.20% of the
there were higher SOR and lower SO2, which would variance. This factor could be attributed to the source
suggest that the high SO2
4 could be mainly from the from the smelting industry and partially from incinera-
secondary transformation instead of the direct local tion. The fourth factor grouped MSA and F
together,
emissions. The average concentrations of NO2 were which explained another 7.29% of the total variance. As
92.16 and 64.22 mg m
3 in winter and summer, respec- a considerable amount of MSA had been detected in
tively. NO
3 is from NH4NO3 that is generally formed by Beijing aerosol, which has been demonstrated not to be
the oxidation of NOx to form the gas phase of nitric acid from the sea source (Yuan et al., 2004), but was
and then with the reaction with NH3. The lower suspected to be from the unknown industrial emissions.
temperature in winter would favor the shift from the The fifth factor grouped C2O2
4 and NO
2 together.
2

gas phase of nitric acid to the particle phase of C2O4 was well correlated with O3 in spring (R ¼ 0:40,
ammonium nitrate, which could lead to the high n ¼ 49) and summer (R ¼ 0:31, n ¼ 86), which indicated
concentration of NO
3 in winter. In summer, both the that the photochemical formation was a source of
higher NOR and the higher emission of NH3 contrib- C2O2
4 (Kawamura and Ikushima, 1993). It was also
uted to the high concentration of NO
3. reported that C2O2
4 could be originating directly from
High Cl
concentrations averaged of 5.28 mg m
3 vehicle exhausts (Kawamura and Kaplan, 1987). NO
2
were observed in winter, which was 4 times higher could be one of the oxidation products of NOx that was
than that in summer of 1.41 mg m
3. As Beijing is mostly from vehicular emissions. Thus, this factor could
150 km from the sea, it has been demonstrated that the be attributed to the photochemical products related to
contribution to the aerosols from the sea could be vehicle emissions. The sixth factor only showed a high
ignored here (Yuan et al., 2004). The major source of loading of PO3
4 , whose source was not clear at this
Cl
in Beijing might be coal burning, which could stage and further study is needed. Overall, the secondary
reasonably explain the higher Cl
concentration in formation of coal/biomass burning products, crust,
winter, because of the enhanced burning activities in this industrial and traffic emissions were the major sources
season. The seasonal variation of F
was similar to Cl
, of the fine aerosols in Beijing.
indicating that both F
and Cl
were primarily related
to the burning activity. Low concentration of NO
2 was 3.5.2. The source identification by Ca2+/Al ratio
observed in summer, which might be related to the faster Crustal components have been a major part of PM2.5
decay rate of NO
2 under high solar radiation plus more aerosol in Beijing. Al is a typical crustal component and
precipitation in summer, as the traffic exhaust was likely it has been used widely as a tracer for suspended soil and
to be the major source of nitrite at urban locations. long-range-transported mineral aerosols. Ca2+ could be
derived from soil dust and construction materials in
3.5. Source of PM2.5 fine particles urban aerosols. Ca2+ was used as an indicator for
construction dust in Beijing (Zhang and Iwasaka, 1999).
3.5.1. Source apportionment by factor analysis Ca2+/Al ratio could be seen as a good tracer for the
Table 5 presents the results of the factor loadings from mixing of soil dust with the suspended construction
the factor analysis (SPSS for window 11.0). Six of the materials in urban aerosols. Fig. 4 showed the seasonal
major components of the variation in the ionic variation of Ca2+/Al ratio at one of the sampling sites,
concentrations were identified on the basis of the BNU, in Beijing. The ratio of Ca2+/Al was low in winter
 ARTICLE IN PRESS
3782 Y. Wang et al. / Atmospheric Environment 39 (2005) 3771–3784

Table 5
Varimax rotated factor matrix for the complete data set

Factor 1 2 3 4 5 6 Communality

NH+
4 0.92
0.09 0.23
0.06 0.05 0.14 0.93
S 0.90 0.06 0.24
0.08
0.01 0.12 0.90
SO2

4 0.90 0.02 0.20
0.06 0.17 0.12 0.90
NO
3 0.87
0.04 0.21
0.04 0.17 0.11 0.84
Cl
0.75 0.10 0.22 0.26
0.01
0.37 0.83
K+ 0.66 0.15 0.08 0.26
0.13
0.07 0.55
Se 0.59 0.09 0.30 0.34
0.26
0.01 0.62
F
0.58 0.24 0.10 0.56
0.10
0.32 0.83
Fe
0.07 0.95
0.03
0.13
0.06 0.04 0.92
Al
0.12 0.94
0.03
0.12
0.02 0.03 0.92
Mn 0.22 0.90 0.12
0.06
0.11 0.00 0.89
Ca2+
0.04 0.86 0.10 0.16 0.14
0.01 0.80
Mg2+ 0.16 0.80 0.18 0.29 0.15
0.06 0.82
Na+ 0.37 0.66 0.18 0.48 0.01
0.19 0.87
Ni 0.29 0.18 0.89 0.09 0.11 0.05 0.93
Cu 0.30 0.19 0.87 0.11 0.15 0.01 0.92
Zn 0.41
0.05 0.75
0.06
0.16 0.00 0.76
MSA
0.03
0.05 0.03 0.70 0.17 0.22 0.58
C2O2
4 0.02 0.10 0.06 0.19 0.80 0.07 0.70
NO
2 0.08
0.06 0.03
0.10 0.64
0.46 0.64
PO3

4 0.16
0.01 0.06 0.10
0.06 0.79 0.67
Eigenvalue 5.51 4.62 2.56 1.53 1.35 1.23
% of variance 26.24 22.00 12.20 7.29 6.44 5.86
Cumulative % 26.24 48.24 60.44 67.73 74.17 80.03

Extraction method: principal component analysis.

Rotation method: varimax with Kaiser normalization.
Note: Boldface indicates that the factor loading is higher than 0.50.

and spring, and high in autumn and summer. The low NO


3 , Cl ) for each sector identified in Fig. 1(c). The
ratio in winter could be because that the construction results showed that a large number of air parcels were
activities decreased in the cold season. In spring, the low from the W–NW sector in the spring and winter season,
ratio of Ca2+/Al denoted the great increase of soil dust, while more samples prevailed in the SW sector in the
which was transported to Beijing from the west or the summer and autumn season. It showed clearly that in
northwest of China. The high ratio of Ca2+/Al in spring, which are characterized by Asian dust outbreaks,
autumn and summer could attribute to increasing the Ca2+ levels associated with the long-range transport
construction activities in warm days as well as increased from W–NW sector were higher, while the NH+

4 , NO3 ,
removal of suspended soil dust by precipitation in the Cl
levels were lower, compared to other sectors. This
wet season. indicated that the dust from the long-range transport
could bring lots of crustal species to Beijing in spring. In
3.5.3. Influence of air mass parcels on the levels of fine summer, the frequencies of air masses from each sector
particles were comparable, but the highest concentrations of ionic
Air mass origin could influence the levels of the fine species (NH+ 2



4 , SO4 , NO3 , Cl ) were associated with
particles because different air masses traveling through the transport from SW sector, which could be due to
different regions bring aerosols with different chemical more industrial factories located in the SW sector of
components. Dust air parcels, which often attack Beijing Beijing.
in spring, are mainly from the west or northwest of
China, so air masses from W–NW sector could bring a
large amount of crustal species. As the industrial 4. Conclusions
factories are largely located in the south and southwest
areas of Beijing, the air masses from S–SE–E and SW A 3-year campaign from 2001 to 2003 measuring
sectors could bring much pollution. Table 6 presented organic and inorganic constituents of PM2.5 was
the distribution of five species (NH+ 4 , Ca
2+
, SO2

4 , conducted at five sites in Beijing, China. Concentrations
 ARTICLE IN PRESS
Y. Wang et al. / Atmospheric Environment 39 (2005) 3771–3784 3783

Table 6 intrusion of dust from west and northwest of China. The

Concentrations of ions (NH+ 4 , Ca
2+
, Cl
, NO
2

3 and SO4 ) Ca2+/Al ratio indicated the mixing of different dust
from different air-mass origin in four seasons (mg m
3) sources. Factor analysis showed that the secondary
Season Ion species
formation of coal/biomass burning products, crust,
industrial and traffic emissions were the major sources
No. NH+
4 Ca2+ SO2

4 NO

3 Cl
of the fine aerosols in Beijing. Traffic sources became
more significant with the motorization in recent years.
Spring Compared with studies in other areas, it could be seen
Local 12 5.25 1.45 8.62 10.30 3.49 clearly that aerosol pollution is a more serious issue in
N–NE 7 7.87 1.06 17.79 13.52 3.94
China than other East Asian sites.
S–SE–E 9 11.89 1.63 21.90 22.33 4.05
SW 24 10.21 2.26 19.12 18.86 4.04
W–NW 49 3.75 3.32 9.83 6.78 1.88

Summer Acknowledgements
Local 14 9.85 0.90 17.78 11.28 1.35
N–NE 11 4.53 0.39 7.05 5.74 0.90 This work was supported by the National Natural
S–SE–E 13 13.64 0.68 24.56 14.92 1.41 Science Foundation of China (Grant nos. 29837190,
SW 28 14.83 0.81 27.44 16.52 2.10 30230310, 20077004, and 20477004), Beijing Natural
W–NW 20 4.40 0.72 8.49 4.19 0.77
Science Fund (Grant nos. 8991002 and 8041003), the
Autumn special fund for the doctoral study of the Education
Local 2 2.88 0.08 5.59 3.77 0.86 Ministry of China, and also in part supported by LAPC,
N–NE 6 2.33 0.23 5.33 3.58 0.67 The Institute of Atmospheric Physics, CAS, and the
S–SE–E 6 8.39 1.27 18.95 9.71 1.28 Swedish International Development Cooperation
SW 12 10.91 1.07 22.33 16.00 1.70
Agency (SIDA) through the Asian Regional Research
W–NW 14 3.72 1.76 5.92 6.16 0.70
Program on Environmental Technology (ARRPET) at
Winter the Asian Institute of Technology.
Local 19 9.32 1.40 16.71 11.72 5.84
N–NE 1 7.27 1.09 3.36 1.09 1.69
S–SE–E 1 8.41 0.50 17.55 10.36 5.06
SW 12 19.19 1.60 41.77 22.03 5.63 References
W–NW 74 9.66 1.79 18.96 11.04 5.13
Arimoto, R., Duce, R.A., Savoie, D.L., Prospero, J.M., Talbot,
No. means the number of samples measured. R., Cullen, J.D., Tomza, U., Lewis, N.F., Ray, B.J., 1996.
Relationships among aerosol constituents from Asia and the
North Pacific during Pem-West A. Journal of Geophysical
Research 101, 2011–2023.
of the water-soluble ions (SO2




4 , NO3 , Cl , F , PO4 ,
3

Chen, Z.L., Ge, S., Zhang, J., 1994. Measurement and analysis
2
+ 2+
NO
2 , CH 3COO

, HCOO

, MSA, C O
2 4 , NH 4 , Ca , for atmospheric aerosol particulates in Beijing. Research of
K+, Mg2+, Na+) and 23 elements were measured for a Environmental Sciences 7 (3), 1–9.
total of 334 samples. A relatively even spatial distribu- Cheng, Z.L., Lam, K.S., Chan, L.Y., Wang, T., Cheng, K.K.,
tion throughout Beijing and a significant seasonal 2000. Chemical characteristics of aerosols at coastal station
in Hong Kong. I. Seasonal variation of major ions, halogens
variation were observed. SO2



4 , NO3 , Cl , NH4 ,
+
2+ + and mineral dusts between 1995 and 1996. Atmospheric
Ca , and K were the major ions and existed mainly Environment 34 (17), 2771–2783.
in the form of (NH4)2SO4, NH4NO3, NaCl, KCl, and Draxler, R.R., Rolph, G.D., 2003. HYSPLIT (HYbrid Single-
CaCl2 in aerosol particles. Most ions showed high Particle Lagrangian Integrated Trajectory) Model Access
concentrations in winter and low in summer. Secondary via NOAA ARL READY Website. NOAA Air Resources
ions, mainly SO2


4 , NO3 , exhibited high concentrations Laboratory, Silver Spring, MD http://www.arl.noaa.gov/
in both summer and winter due to the secondary ready/hysplit4.html.
transformation that is accelerated under high humidity Fang, G., Chang, C., Wu, Y., Fu, P.P., et al., 2002. Ambient
and strong solar radiation in summer and the higher suspended particulate matters and related chemical species
study in central Taiwan, Taichung during 1998–2001.
concentrations of SO2 from coal burning and the lower
Atmospheric Environment 36, 1921–1928.
removal rate in winter. The formations of SO2
4 and
+ Gao, J., Wang, W., Du, J., Liu, H., Pang, Y., Tang, D., 1996.
NO
3 were determined largely by temperature and NH4 , Preliminary Study on the aerosol characteristics of Xiamen
respectively. Temperature, relative humidity, rainwater in Spring. Research of Environmental Sciences (in Chinese)
frequency, and air mass origin might be the main factors 9 (5), 33–37.
regulating the aerosol distribution. Crustal ions exhib- He, K., Yang, F., Ma, Y., Zhang, Q., Yao, X., Chan, C.K.,
ited sporadic but high peaks in spring due to the Cadle, S., Chan, T., Mulawa, P., 2001. The characteristics of
 ARTICLE IN PRESS
3784 Y. Wang et al. / Atmospheric Environment 39 (2005) 3771–3784

PM2.5 in Beijing, China. Atmospheric Environment 35, Seinfeld, J.H., 1986. Atmospheric Chemistry and Physics of Air
4959–4970. Pollution. Wiley, New York 348pp.
Hien, P.D., Bac, V.T., Thinh, N.T.H., 2004. PMF receptor Shwartz, J., Dockery, D.W., Neas, L.M., 1996. Is daily
modeling of and coarse PM10 in air masses governing mortality associated specially with the particles? Air and
monsoon conditions in Hanoi, northern Vietnam. Atmo- Waste Management Association 46, 927–939.
spheric Environment 38, 189–201. Sun, Y., Zhuang, G., Wang, Y., Han, L., Guo, J., Dan, M.,
Ho, K.F., Lee, S.C., Chan, C.K., Yu, J.C., Chow, J.C., Yao, Zhang, W., Wang, Z., 2004. The air-borne particulate
X.H., 2003. Characterization of chemical species in PM2.5 pollution at Beijing—concentrations, composition, distribu-
and PM10 aerosols in Hong Kong. Atmospheric Environ- tion, and sources of Beijing aerosol. Atmospheric Environ-
ment 37, 31–39. ment 38 (35), 5991–6004.
Hu, M., He, L., Zhang, Y., Wang, M., Kim, Y.P., Moon, K.C., Truex, T.J., Pierson, W.R., Mckee, D.E., 1980. Sulfate in diesel
2002. Seasonal variation of ionic species in fine particles at exhaust. Environmental Science & Technology 14,
Qingdao, China. Atmospheric Environment 36, 5853–5859. 1118–1121.
Hughes, L.S., Cass, G.R., Gone, J., Ames, M., Olmez, I., 1998. US Environmental Protection Agency (US EPA), 1997.
Physical and chemical characterization of atmospheric ultra National ambient air quality standards for particulate
fine particles in the Los Angles area. Environmental Science matter, final rule. Federal Register Vol. 62(138), Part II,
& Technology 32, 1153–1161. EPA, 40 CFR Part 50, 18 July.
Kato, N., 1996. Analysis of structure of energy consumption Wang, H., Wang, G., Gao, S., Wang, L., 2003. Characteristics
and dynamics of emission of atmospheric species related to of atmospheric particulate pollution in spring in Nanjing
the global environmental change (SOx, NOx, and CO2) in City. China Environmental Science (in Chinese) 23 (1),
Asia. Atmospheric Environment 30, 2757–2785. 55–59.
Kawamura, K., Ikushima, K., 1993. Seasonal changes in the Wang, J., Xie, Z., Zhang, Y., Shao, M., Zeng, L., Cheng, C.,
distribution of dicarboxylic acids in the urban atmosphere. Xu, X., Zhao, X., Meng, Y., 2004. Study on the
Environmental Science and Technology 27, 2227–2235. characteristics of mass concentration of atmospheric fine
Kawamura, K., Kaplan, I.R., 1987. Motor exhaust emission as particles in Beijing. Acta Meteorologic Sinica (in Chinese)
a primary source for dicarboxylic acids in Los Angeles 62 (1), 104–111.
ambient air. Environmental Science and Technology 21, Winchester, J.W., Lu, W., Ren, L., Wang, M., Maenhaut, W.,
105–110. 1981. Fine and coarse aerosol composition from a
Kim, K., Lee, M., Lee, G., Kim, Y., Youn, Y., Oh, J., 2002. rural area in north China. Atmospheric Environment 15,
Observations of aerosol-bound ionic compositions at Cheju 933–937.
Island, Korea. Chemosphere 48, 317–327. Xiao, H., Liu, C., 2004. Chemical characteristics of water-
Mahmoud, A., Alan, W., Douglas, H., 2002. A preliminary soluble components in TSP over Guiyang, SW China, 2003.
apportionment of the sources of ambient PM10, PM2.5 and Atmospheric Environment 38, 6297–6306.
VOCs in Cairo. Atmospheric Environment 36, 5549–5557. Yao, X., Chan, C.K., Fang, M., Cadle, S., Chan, T., Mulawa,
Michelozzi, P., Forastiere, F., Fusco, D., Perucci, C.A., Ostro, P., He, K., Ye, B., 2002. The water-soluble ionic composi-
B., Ancona, C., Pallotti, G., 1998. Air pollution and daily tion of PM2.5 in Shanghai and Beijing, China. Atmospheric
mortality in Rome, Italy. Occupational Environment and Environment 36, 4223–4234.
Medicine 55 (9), 605–610. Ye, B., Ji, X., Yang, H., Yao, X., Chan, C.K., Cadle, S., Chan,
Ohta, S., Okita, T., 1990. A chemical characterization of T., Mulawa, P., 2003. Concentration and chemical compo-
atmospheric aerosol in Sapporo. Atmospheric Environment sition of PM2.5 in Shanghai for a 1 yr period. Atmospheric
24A, 815–822. Environment 37, 499–510.
Ostro, B., 1995. Fine particulate air pollution and mortality in Yuan, H., Wang, Y., Zhuang, G., 2003. The simultaneous
two southern California counties. Environmental Research determination of organic acid, MSA with the inorganic
70 (2), 98–104. anions in aerosol and rain-water by ion chromatography.
Park, S.S., Kim, Y.J., 2004. PM2.5 particles and size-segregated Journal of Instrumental Analysis (in Chinese) 6, 11–14.
ionic species measured during fall season in three urban sites Yuan, H., Wang, Y., Zhuang, G., 2004. MSA in Beijing
in Korea. Atmospheric Environment 38, 1459–1471. aerosol. China Science Bulletin 49 (10), 1020–1025.
Pierson, W.R., Brachaczek, W.W., Mckee, D.E., 1979. Sulfate Zhang, D., Iwasaka, Y., 1999. Nitrate and sulphate in
emissions from catalyst equipped automobiles on the individual Asian dust-storm particles in Beijing, China in
highway. Journal of Air Pollution Control Association 29, spring of 1995 and 1996. Atmospheric Environment 33,
255–257. 3213–3223.
Raizenne, M., Neas, L.M., Damokosh, A.L., Dockery, D.W., Zhuang, G., Guo, J., Yuan, H., Zhao, C., 2001. The
Spen-gler, J.D., Koutrakis, P., Ware, J.H., Speizer, F.E., compositions, sources, and size distribution of the dust
1996. The effects of acid aerosols on North American storm from China in spring of 2000 and its impact on the
children: pulmonary function. Environmental Health Per- global environment. China Science Bulletin 46 (11),
spectives 104, 506–514. 895–901.