Introduction
Particulate pollution in China is a serious environmental problem,
influencing air quality, regional and global climate, and human health.
Especially during recent winters, large-scale and severe haze pollution has
brought China's particulate pollution at the forefront of world-wide media
and has evoked great scientific interest in air pollution studies. Measurements
at a number of major cities showed that the wintertime daily average mass
concentrations of PM2.5 (particulate matter with an aerodynamic
diameter <2.5 µm) are approximately 1–2 orders of magnitude
higher than those observed in urban areas in the US and Europe
(Huang et al., 2014). Severe particulate pollution is often accompanied by
extremely poor visibility and poor air quality leading to a sharp increase in
respiratory diseases. Long-term exposure to high levels of particulate
pollution was estimated to have resulted in 1.1 million deaths in China in 2015,
ranking China first in the world with respect to air-pollution-related mortality (Cohen et al., 2017).
The Beijing–Tianjin–Hebei (BTH) region is one of the important city
clusters in China, but also suffers from serious air pollution. Seven cities
in this region ranked the top 10 most polluted cities in China in the year
2014–2015 (http://www.zhb.gov.cn, last acces: 7 February 2017). The urgent need for an air quality improvement in this region has
been recognized by central and local governments as well as the public,
which has led to mitigating actions being undertaken by the authorities. In
particular, various emission control measures have been implemented in this region
to clean Beijing's air, e.g., during the 2014 Asia-Pacific Economic
Cooperation (APEC) summit. These temporal measures include the odd–even ban
on vehicles and shutdowns of factories and construction sites, which have led to
serious side effects on daily life and economic growth. Therefore, the
identification of the major sources and atmospheric processes producing
airborne particles is required for implementing targeted and optimized
emission control strategies.
The first step for quantifying PM sources requires the measurement of
inorganic and organic tracers and/or mass spectrometric fingerprints of
ambient PM samples. This can be realized by online ambient measurements
using aerosol mass spectrometric (AMS) techniques to determine aerosol
composition (Jimenez et al., 2009; Ng et al., 2011b; Elser et al., 2016b). In
particular, the quadrupole aerosol chemical speciation monitor (Q-ACSM) and,
recently, the time-of-flight aerosol chemical speciation monitor (TOF-ACSM) have
been developed for long-term continuous measurements of non-refractory
submicron aerosols (Ng et al., 2011a; Fröhlich et al., 2013). Aerosol
sources have been successfully identified from AMS measurements using
positive matrix factorization (PMF) analysis (Ulbrich et al., 2009; Crippa et
al., 2013; Elser et al., 2016a). In terms of Q-ACSM data sets, the use of PMF
often fails to resolve sources with similar mass spectral profiles, e.g., the
mixing of cooking organic aerosol with traffic organic aerosol in Nanjing
(Zhang et al., 2015), or those present in low contributions, e.g., the lack
of success in resolving a factor related to biomass burning in Beijing (Jiang
et al., 2015). It has also been pointed out that PMF cannot separate the aerosol
sources of temporal covariations driven by low temperature and periods of
strong inversions (Canonaco et al., 2013; Reyes-Villegas et al., 2016). Several source
apportionment studies (in which PMF did not find optimal results) have
utilized the multilinear engine (ME-2) solver, which enables the constraint of
the factor profiles/time series, providing a superior separation of the PM
sources (e.g., Canonaco et al., 2013, 2015; Fröhlich et
al., 2015a, b; Minguillón et al., 2015; Petit et al., 2015; Ripoll et
al., 2015; Reyes-Villegas et al., 2016; Bressi et al., 2016; Schlag et al., 2016; Wang
et al., 2017; Zhu et al., 2018). However, studies using ME-2 to resolve OA
sources from the ACSM measurements are scarce in the BTH region.
Apart from the lack of applications of ME-2 for OA source apportionment,
most field studies have mainly focused on the aerosol pollution in
Beijing (Sun et al., 2013, 2014, 2016; Jiang et al.,
2015; Xu et al., 2015; Elser et al., 2016a; Hu et al., 2016a). These and
related studies have clearly shown that Beijing is sensitive to the regional
transport of aerosols from its surrounding areas (Xu et al., 2008; Zhang et
al., 2012; P. Li et al., 2015). For example, Guo et al. (2010) estimated that
the regional pollutants accounted for 69 % of PM10 and
87 % of PM1.8 on average in Beijing during summer, with sulfate, ammonium, and
oxalate mostly formed regionally (regional contributions >87 %). Sun et
al. (2014) reported that 66 % of NR-PM1 was from regional transport in
Beijing during the 2013 winter haze event. Among the surrounding areas of
Beijing, the Hebei Province is the main source area leading to high aerosol
loadings in Beijing (Chen et al., 2007; Xu et al., 2008; Lang et al., 2013;
P. Li et al., 2015).
Shijiazhuang, the capital of Hebei Province, is located ∼270 km south
of Beijing and has a population approximately half that of Beijing. P. Zhao et
al. (2013) and P. S. Zhao et al. (2013) characterized the spatial and seasonal variations of
the PM2.5 chemical composition in the BTH region, and Shijiazhuang was
selected as the representative of the polluted cities in Hebei Province. The
off-line analysis results showed that organic carbon (OC) and elemental
carbon (EC) concentrations in Shijiazhuang were lower in the spring and
summer than those in the autumn and winter. The sum of secondary inorganic
species (SO42-, NO3-, and NH4+) was
highest in the autumn. However, the temporal profiles of PM composition cannot be
captured by off-line analyses, hindering more detailed study of the sources
and formation of PM. In this work, we present for the first time the 30 min
time resolved NR-PM1 measurements in Shijiazhuang during the winter
heating season. The characteristics of NR-PM1 are analyzed, which
include the following: (1) time series, mass fraction, and diurnal variation of NR-PM1
species; (2) multilinear engine (ME-2)-resolved OA sources and their mass
fraction as well as their diurnal variation; and (3) the characteristics and
atmospheric evolution of aerosol composition and sources under different
aerosol loadings and meteorological conditions.
Results and discussion
Concentration and chemical composition of NR-PM<inline-formula><mml:math id="M61" display="inline"><mml:msub><mml:mi/><mml:mn mathvariant="normal">1</mml:mn></mml:msub></mml:math></inline-formula>
Figure 1 shows the time series of NR-PM1 species, trace gases, and
meteorological conditions during the entire measurement period. The measured
mass concentrations of NR-PM1 for the entire campaign period ranged from
a few to 508.4 µgm-3, with an average of 178±101 µgm-3. That was much higher than the
wintertime/summertime concentrations measured in many other cities (see
Table 1). The mass concentration of NR-PM1 correlated strongly with that
of PM2.5 (R2=0.76) with a regression slope of 0.72, indicating
that NR-PM1 represents a majority of PM2.5 mass. The NR-PM1
concentrations exceeded the Chinese PM2.5 limit of 75 µgm-3 for 90 % of days during the measurement period, showing the
severity of particulate air pollution at Shijiazhuang.
The fine PM mass concentrations and fractional contribution of
different compositions at different locations.
City
Season
NR-PM1
OA %
SO42- %
NO3- %
NH4+ %
Cl- %
Ref.
(µgm-3)
Beijing
Winter, 2010
60
54
14
11
12
9
Hu et al. (2016a)
Beijing
Winter, 2011
59
51
13
17
14
5
Sun et al. (2015)
Beijing
Winter, 2012
66.8
52
14
16
13
5
Sun et al. (2013)
Beijing
Winter, 2012
79
52
17
14
10
7
Wang et al. (2015)
Beijing
Winter, 2013
77
50
19
16
12
3
Sun et al. (2014)
Beijing
Winter, 2013
13.0
52
17
14
10
7
Jiang et al. (2015)
Beijing
Winter, 2013
64
60
15
11
8
6
Sun et al. (2016)
Beijing
Winter, 2014
75*
56
16
10
7
11
Elser et al. (2016a)
Beijing
Summer, 2011
80
32
28
21
17
2
Hu et al. (2016a)
Beijing
Summer, 2012
52
41
14
25
17
3
Sun et al. (2015)
Lanzhou
Winter, 2014
57.3
55
13
18
11
3
Xu et al. (2016)
Lanzhou
Summer, 2012
24
53
18
11
13
5
Xu et al. (2014)
Ziyang
Winter, 2012
60
40
24
15
17
4
Hu et al. (2016b)
Handan
Winter, 2015
178
47
16
15
13
9
Li et al. (2017)
Shenzhen
Autumn, 2009
38.3
46
29
12
11
2
He et al. (2011)
Shanghai
Summer, 2010
27
31
36
17
14
2
Huang et al. (2012)
Nanjing
Summer, 2013
36.8
42
14
24
19
1
Zhang et al. (2015)
Hong Kong
Winter, 2012
14.5
33
40
10
16
1
Y. J. Li et al. (2015)
Hong Kong
Summer, 2011
8.7
26
56
3
15
0.1
Y. J. Li et al. (2015)
Paris
Winter, 2010
16.7
35
16
33
15
1
Crippa et al. (2013)
Fresno, California
Winter, 2010
11.8
67
3
20
8
2
Ge et al. (2012)
Shijiazhuang
Winter, 2014
178
50
21
12
11
6
This study
* NR-PM2.5
Similar to measurements at other urban sites, OA was the dominant fraction of
NR-PM1, with an average of 50 % (31 %–80 %), followed by 21 %
of sulfate (4 %–36 %), 12 % of nitrate (2 %–26 %), 11 % of
ammonium (4 %–21 %), and 6 % of chloride (2 %–20 %). The dominant
contribution of organics in NR-PM1 is also consistent with measurements
from other urban sites in the BTH region during winter heating seasons (see
Table 1). Sulfate was the second largest contributor to NR-PM1. The
large fraction of sulfate was likely associated with the large consumption of
coal in Hebei Province, i.e., 296 million tons was used in
coal-fired power plants and steel industry (producing ∼11 % of global
steel output) in 2014. The enhancement of chloride fraction from >1 % to 4 % in other Chinese cities in summer (see Table 1) to 6 % in
Shijiazhuang in winter (within the range of >2 %–7 % in other Chinese
cities in winter, see Table 1) can be attributed to the substantial emissions
from coal and/or biomass burning activities.
Figure 2a shows the diurnal variations of NR-PM1 components, which were
affected by the evolution of the planetary boundary layer (PBL) height that
governed the vertical dispersion of pollutants and by the diurnal cycle of
the emissions and atmospheric processes. The concentrations of pollutants
increased at night as a result of enhanced emissions from residential heating
(in particular, for organics and chloride) and a progressively shallower PBL.
During daytime the PBL height was developed by solar radiation; thus, the
pollutants became diluted resulting in the decrease of organics, sulfate,
ammonium, and chloride in the afternoon. In contrast, the concentrations of
nitrate increased after sunrise but then remained rather constant throughout the
afternoon, suggesting a strong source or production of nitrate which offsets
the dilution from PBL development. To minimize the effects of PBL heights,
data were normalized by ΔCO. CO is often used as an emission tracer
to account for dilution on timescales of hours to days because of its
relatively long lifetime against oxidation by OH radicals (approximately
1 month) (DeCarlo et al., 2010). After offsetting the PBL dilution effect,
sulfate, nitrate, and ammonium showed clear increases from 07:00 to 15:00 LT
(local time; Fig. 2c), indicating the efficient daytime production of these secondary
inorganic species. It should be noted that the increase of nitrate (about 2
times, from ∼6 to ∼12 µgm-3ppm-1) is
slightly larger than that of sulfate (about 1.6 times, from ∼11 to
∼17.5 µgm-3ppm-1), indicating more efficient
photochemical production of nitrate than sulfate, given that the loss rate of
sulfate could not be higher than that of nitrate as nitric acid is
semi-volatile and may be further lost by evaporation. Furthermore, the continuous
increase of organics after sunrise suggested efficient photochemical
production of secondary organic aerosol (SOA).
Diurnal variations of NR-PM1 composition (a), OA sources (b),
NR-PM1 species/ΔCO (c), and OA sources/ΔCO (d).
![]()
Sources of organic aerosol
From the PMF analysis, we first examined a range of solutions with three to
eight factors. The solution that best represents the data is the five-factor solution
(Fig. S1 in the Supplement). The solutions with factor numbers more than five
provide no new meaningful factors (see Fig. S2 and more details in the
Supplement).
Although the five-factor solution can reasonably represent the data, HOA is
still mixed with BBOA because the HOA profile contains a higher than expected
contribution from m/z 60. In addition, COA contains no signal at m/z 44,
which might indicate a suboptimal splitting between the contributing sources.
To better separate HOA from BBOA, we constrained the HOA profile from Ng et
al. (2011b), which is an average profile over 15 cities from China, Japan,
Europe, and the US. Although gasoline vehicles dominate in China
while diesel vehicles dominate in Europe, HOA mass spectra do not show
significant variability when compared to different sites in China and Europe
(Ng et al., 2011b; Reyes-Villegas et al., 2016; Bozzetti et al., 2017), indicating
that traffic emissions from different types of vehicles have similar
profiles. To avoid the influences of other sources on COA, the COA profile
from Paris (Crippa et al., 2013) was used as a constraint because high
similarities were found between the COA profile from Paris and four COA
profiles from different types of Chinese cooking activities (He et al., 2010;
Crippa et al., 2013). However, the constraint on HOA and COA profiles still
seems to sub-optimally resolve the apportionment of BBOA from CCOA, as one
unconstrained factor contains high contributions from both m/z 60 and
PAH-related m/z's (m/z 77, 91, and 115, as shown in Fig. S3) which
indicates the mixing between BBOA and CCOA. To separate BBOA and CCOA, we
constrained BBOA using the average of BBOA profiles from the five-factor
unconstrained PMF solutions.
To explore the solution space, an a value of 0–0.5 with an interval of 0.1
was used to constrain both the HOA and COA reference profiles from literature
while BBOA was constrained with an a value of 0 because the BBOA profile was
resolved from an unconstrained PMF solution which is not expected to vary
significantly. Thirty-six possible results were obtained by limiting the range of a
values. Three criteria for optimizing OA source appointment are as follows:
The diurnal pattern of COA. The diurnal cycle of COA should have
higher concentrations during mealtimes.
Minimization of m/z 60 in HOA. The upper limit of m/z 60 in the
HOA profile is 0.006, which is the maximal fractional contribution derived
from multiple ambient data sets in different regions (mean +2σ) (Ng
et al., 2011b).
The rationality of unconstrained factors. OOA should have
abundant signal at m/z 44 and should contain much lower signals at PAH-related ion
peaks compared with CCOA.
Nine solutions match the criteria above. Therefore, the final time series and mass
spectra are the averages of these nine solutions. The diurnal
variations of mass concentrations of the OA factors and their PBL-corrected
results are shown in Fig. 2b and d, respectively. The mass spectra and time
series of the OA factors and their correlation with external tracers are
shown in Fig. 3. The relative contributions of each OA source to the m/z's
are shown in Fig. S4. Potential source contribution function (PSCF) analysis
was also performed and the result is shown in Fig. S5.
Mass spectrums (left) and time series (right) of five OA sources.
Error bars of the mass spectrums represent the standard deviation of each m/z over
all accepted solutions.
![]()
OOA is characterized by high signals at m/z 44 (CO2+) and
m/z 43 (C3H7+ or C2H3O+) and accounts for
85 % of m/z 44 signal, which is much higher than other OA sources. The time series
of OOA is highly correlated with that of sulfate (R2=0.70), nitrate
(R2=0.75), and ammonium (R2=0.76), confirming the secondary nature
of this factor. The diurnal cycle of OOA shows an increase from 07:00 to
11:00 LT, followed by a decrease in the afternoon due to the PBL evolution
effect. After normalizing the PBL effect, OOA increased continuously from
07:00 to 15:00 LT, indicating the importance of photochemical oxidation. This
diurnal feature in combination with the PSCF results indicated that a large
fraction of OOA was produced locally and/or produced from the highly
populated and industrialized surrounding areas, consistent with the sulfate
production discussed below.
The mass spectrum of CCOA is featured by prominent contributions of
unsaturated hydrocarbons, particularly PAH-related ion peaks (e.g., 77, 91,
and 115). The CCOA profile shows a weaker signal at m/z 44 than that
observed in Beijing (Hu et al., 2016a) and Lanzhou (Xu et al., 2016). This
difference can be caused by the difference in coal types, burning conditions,
and aging processes (Zhou et al., 2016). CCOA accounts for 42 %–66 % of
PAH-related ion peaks, much higher than those in other OA sources. This
result suggested that the major source of PAHs was coal combustion in
Shijiazhuang in wintertime. The campaign-averaged mass concentration of CCOA was
23.2 µgm-3, which is higher than that in Xi'an
(10.1 µgm-3) but is similar to that in Beijing
(23.5 µgm-3) observed in the same winter (Elser et al.,
2016a). Nevertheless, during haze extremes, the average CCOA concentration
was 77.5 µgm-3 in Shijiazhuang, much higher than that in
Beijing (48.2 µgm-3, Elser et al., 2016a). CCOA showed
distinct diurnal variations with low concentrations (down to 12.6 µgm-3) during the day and high concentrations (up to 37.6 µgm-3) at night, corresponding to 19 % and 35 % of OA, respectively.
The elevated CCOA concentrations at night suggested a large emission from
residential heating activities using coal as the fuel compounded by the
shallow PBL. The average contribution of CCOA to the total OA was 27 %,
which is consistent with studies in Beijing and Handan (∼160 km south
to Shijiazhuang) where CCOA was found to be the dominant primary OA (Elser et
al., 2016a; Sun et al., 2016; Li et al., 2017). Given this large fraction of
OA from coal combustion, mitigating residential coal combustion is
of significant importance for improving air quality in the BTH regions.
The BBOA mass spectrum is featured by prominent m/z 60 (mainly
C2H4O2+) and 73 (mainly C3H5O2+)
signals (He et al., 2010). These two ions (C2H4O2+ and
C3H5O2+) are fragments of anhydrous sugars produced from
the incomplete combustion and pyrolysis of cellulose and hemicelluloses
(Alfarra et al., 2007; Lanz et al., 2007; Mohr et al., 2009). Consistently,
BBOA accounts for 50 % of m/z 60 and 56 % of m/z 73, which is much higher than
those in other sources. In addition, BBOA accounts for 9 %–27 % of the
PAH-related m/z's (i.e., m/z 77, 91, and 115), which is lower than those in CCOA but
higher than those in other primary OA sources. This suggested that BBOA was
also an important PAH source in Shijiazhuang in wintertime. A high correlation
was found between the time series of BBOA and that of chloride
(R2=0.75), the latter of which was suggested to be one of the tracers of
biomass burning. BBOA accounted for 17 % of OA on average, which is higher
than those (9 %–12 %) observed in Beijing during wintertime heating
seasons (Elser et al., 2016a; Hu et al., 2016a; Sun et al., 2016). The higher
BBOA contribution in Shijiazhuang in wintertime is likely associated with the
widespread use of wood and crop residuals for heating and cooking in
Shijiazhuang and surrounding areas, as supported by the PSCF results
(Fig. S5).
The COA profile is characterized by a high m/z 55/57 ratio of 2.7, which is much
higher than that in non-cooking POA (0.6–1.1) but within the range of
2.2–2.8 in COA profiles reported by Mohr et al. (2012). COA shows a clear
diurnal cycle with distinct peaks at lunchtime (between 11:00 and 13:00 LT) and dinnertime (between 19:00 and 21:00 LT). A small peak was also
observed in the morning between 06:00 and 07:00 LT, consistent with
breakfast time. COA accounted for 16 % of total OA on average with the
highest contribution of 24 % during dinnertime.
The HOA mass spectrum is dominated by hydrocarbon ion series of
[CnH2n+1]+ and
[CnH2n-1]+ (Canagaratna et al., 2004; Mohr et
al., 2009). The diurnal variation of HOA is featured by a high concentration at
night, likely due to enhanced truck emissions (only allowed to drive on the road
between 23:00 and 06:00 LT) and a shallow PBL at night. Similar diurnal cycles
were found in Beijing and Xi'an in wintertime (Sun et al., 2016; Elser et al.,
2016a). HOA, on average, accounted for 13 % of total OA for the entire
observation period, which was higher than that in Beijing (9 %–10 %) but
lower than that in Xi'an (15 %) measured in the same winter (Elser et al.,
2016a; Sun et al., 2016).
Chemical nature and sources at different PM levels
Figure 4 shows the mass fractions of NR-PM1 species and OA sources on
reference days and extremely polluted days. Here, the days with a NR-PM1
daily average mass concentration higher than the 75th percentile (i.e., ≥238 µgm-3) are denoted as extremely polluted days and
the rest of days are as designated as reference days. The average concentration of NR-PM1
was 310 µgm-3 during extremely polluted days, about 2 times
higher than that during reference days (162 µgm-3). The
average concentration of secondary inorganic aerosol was 65 µgm-3 (40 % of NR-PM1 mass) during reference days and increased
to 143 µgm-3 (46 % of NR-PM1 mass) during extremely
polluted days. Secondary organic aerosol also increased from 19 µgm-3 (22 % of OA) during reference days to 40 µgm-3
(26 % of OA) during extremely polluted days. The enhanced mass
concentrations (∼2 times) of both secondary inorganic aerosol and
secondary organic aerosol during extremely pollution days suggested strong
secondary aerosol production during pollution events. Such enhancement was
likely confounded by stagnant weather conditions (e.g., average wind speed
of 0.9 ms-1) and a high RH of 69.4 % which facilitated the
production and accumulation of secondary aerosol. Note that it was already
very polluted during the reference days with an average NR-PM1 concentration
of 162 µgm-3, which may explain the relatively
small increase in the fractional contribution of secondary aerosol from reference
days to extremely polluted days.
Relative contributions of NR-PM1 species and OA sources (OOA,
CCOA, BBOA, COA, and HOA) on reference days (a) and extremely polluted days (b).
Extremely polluted days are defined as days with a NR-PM1 daily average mass
concentration higher than the 75th percentile (237.3 µgm-3), and the remaining days are referred to the reference days. Data
collected during the Chinese Spring Festival are excluded to eliminate the influence of the change
in emission patterns during the holiday.
![]()
Figure 5a and b show the factors driving the pollution events by binning the
fractional contribution of each chemical species and OA source to total
NR-PM1 and OA mass, respectively. The data clearly show that high
pollution events are characterized by an increasing secondary fraction,
reaching ∼55 % at the highest NR-PM1 mass bin
(300–360 µgm-3). In particular, from the lowest NR-PM1
bin to the highest NR-PM1 bin, the fractional contribution increases
from 14 % to 25 % for sulfate in NR-PM1 and from 18 % to 25 %
for OOA in OA, demonstrating the importance of secondary aerosol formation in
driving particulate air pollution (Huang et al., 2014; Elser et al., 2016a;
Wang et al., 2017). To investigate the oxidation degree of sulfur at
different NR-PM1 mass, the sulfur oxidation ratio (FSO4) was
calculated according to Eq. (1):
FSO42-=n[SO42-]n[SO42-]+n[SO42-],
where n is the molar concentration. As can be seen from Fig. 6,
FSO4 shows a clear increasing trend with NR-PM1 mass,
climbing from 0.08 in the lowest mass bin to 0.21 in the highest mass bin.
However, the highest FSO4 value is still much lower than that
reported in previous studies, e.g., 0.62 in Xi'an (Elser et al., 2016a),
suggesting a low atmospheric oxidative capacity during the measurement period
in Shijiazhuang. This may also explain the relatively low OOA fraction (see
Fig. 5b). Certainly, it should be noted that the mass concentration of
sulfate may also be affected by other parameters, including aerosol liquid
water content, aerosol, or cloud water pH, besides atmospheric oxidative
capacity.
Relative contributions of NR-PM1 species (a) and OA sources (b)
as a function of the daily average NR-PM1 mass concentrations. The numbers
above the bars refer to the OA mass concentration (µgm-3). Data
collected during the Chinese Spring Festival are excluded to eliminate the influence of the change
in emission patterns during the holiday.
![]()
Evolution of aerosol composition and sources at different RH
levels
Figure 7a and b show the mass concentrations of the NR-PM1 species and
of the OA sources as a function of RH, with RH bins of 10 % increments. The
absolute mass concentrations of secondary inorganic species increased as RH
increased from 60 %, whereas chloride showed a decreasing trend. Among the OA
sources, OOA and HOA were enhanced with RH – increasing from <60 % to
90 % – whereas other OA sources did not show a clear trend. As RH increased
gradually with the decrease in wind speed (Fig. 6a), the development of
stagnant weather conditions (including a shallower PBL) promoted both the
accumulation of pollutants and the formation of secondary aerosol (Tie et
al., 2017). To minimize the effects of PBL variations, the NR-PM1
species and OA fractions were normalized by the sum of the POA, as a
surrogate of secondary aerosol precursors. The resulting ratios were further
normalized by the values at the first RH bin (<60 %) for better
visualization. As shown in Fig. 7c, when RH increased from 60 % to 100 %,
the normalized sulfate increased by a factor of ∼1.7, suggesting the
importance of aqueous-phase SO2 oxidation in the formation of
sulfate at high RH. The enhancements for nitrate and ammonium were slightly
lower (∼1.2) compared to that of sulfate, because NH4NO3 is
thermally labile and its gas–particle partitioning is affected by both
temperature and RH. The importance of aqueous-phase chemistry is further
supported by the increase of FSO4 as a function of RH (Fig. 6b).
At RH <60 %, FSO4 was rather constant, with an average of
0.09, indicating a low sulfur oxidation degree. At RH >60 %,
FSO4 increased rapidly with the increase in RH, reaching a maximal
average of 0.18 at the last RH bin (90 %–100 %). Note that the sulfur
oxidation degree at high RH (>60 %) was much lower compared with those
measured in Xi'an during the same winter (average FSO4 0.62 at a
RH of 90 %–100 %, Elser et al., 2016a). The low sulfur oxidation degree
observed in Shijiazhuang (i.e., >80 % of sulfur is still not oxidized)
indicated insufficient atmospheric processing and also suggested a large
fraction of pollutants in Shijiazhuang was likely emitted locally and/or
transported from the heavily populated and industrialized surrounding areas.
With a longer atmospheric processing time in the downwind region, e.g.,
Beijing, higher secondary aerosol fractions are expected, as observed in
previous studies (e.g., Huang et al., 2014). Similar to sulfate, the
normalized OOA increased by a factor of ∼1.2 when RH increased from
60 % to 100 % (Fig. 7d). The mass fraction of OOA increased from 29 %
to 41 % when RH increased from 70 % to 100 %, whereas POA contribution
decreased correspondingly from 71 % to 59 % (Fig. 6d). These results
support the above discussion that aqueous-phase chemistry also plays an
important role in the formation of OOA under high-RH conditions during haze
pollution episodes.
Variations of wind speed as a function of RH (a),
FSO4 as a function of RH (b) and of the
NR-PM1 mass concentrations (c), and the mass fraction
of OA as a function of RH (d).
![]()
The average mass concentration of NR-PM1 species (a) and OA
sources (b) as a function of RH. The average mass concentration of
NR-PM1 species (c) and OA sources (d) normalized to the sum of primary
sources (HOA, COA, BBOA, and CCOA) as a function of RH. All ratios are
further normalized to the values at the first RH bin (<60 %) for better illustration.
![]()
Primary emissions versus secondary formation
Frequent changes between clean and polluted episodes were observed in this
study. To get a better insight into aerosol sources and atmospheric
processes, four clean periods (C1–C4) with a daily average NR-PM1 mass concentration
lower than the 25th percentile, six high-RH (>80 %) polluted
episodes (H1–H6), and four low-RH (<60 %) polluted episodes (L1–L4) with
daily average NR-PM1 mass concentration higher than the 75th
percentile were selected for further analysis. As shown in Fig. 8, the
chemical composition and sources differed during different episodes. The
contributions of organics showed a decreasing trend, from 54 %–64 %
during C1–C4 to 49 %–58 % during L1–L4, and to 35 %–44 % during
H1–H6, while the corresponding contributions of secondary inorganic species
increased. This indicated a notable production and accumulation of secondary
inorganic aerosol during severe haze pollution events. For example, the mass
fraction of sulfate in NR-PM1 was much higher during high-RH pollution
events (H1–H6, 27 %–30 %) compared to those during low-RH pollution
events (L1–L4, 11 %–18 %) and clean events (C1–C4, 11 %–17 %). OOA
also showed a much higher contribution to OA during high-RH pollution events
(H1–H6, 29 %–50 %) than during low-RH pollution events (L1–L3,
17 %–26 %) and clean events (C1–C4, 10 %–34 %). Interestingly, when
comparing high-RH and low-RH pollution events of similar PM levels (Fig. 8),
secondary inorganic species and OOA dominated the particulate pollution
during high-RH pollution events, which was likely due to enhanced secondary formation, similar
to previous studies (e.g., Wang et al., 2017), whereas POA dominated the
particulate pollution at low RH and under stagnant conditions. The
concentrations of POA are determined by both emissions and meteorological
conditions. The different significance of primary aerosol and secondary
aerosol in low- and high-RH pollution events highlights the importance of
meteorological conditions in driving particulate pollution.
Summary of relative humidity and temperature, gaseous species,
organic sources, and NR-PM1 chemical composition for high-RH (H1–H6)
polluted, low-RH (L1–L4) polluted, and clean (C1–C4) episodes.
![]()
Time series of meteorological factors (relative humidity,
temperature, wind speed, and wind direction), gaseous species, OA factors, and
NR-PM1 chemical composition for the first period (average RH <50 %) (a) and the second period (average RH >80 %) (b).
![]()
Figure 9 shows the evolution of aerosol species in two cases with different RH
levels. The first case had an average RH <50 % from 20 to 24 January (C2 and
L3 episodes). The high wind speed (>6 ms-1) from the
northwest before the L3 episode led to a significant reduction of air
pollutants (the C3 episode, a clean-up period). When the wind direction
switched from northwest to the 90–270∘ sector and the wind speed
decreased to <3 ms-1, the measured pollutants (except
O3 which was reacted out by increasing NO emissions) started to
build up. Specifically, NR-PM1 showed a dramatic increase by a factor of
19 over the first 11 h (from 20 January 16:00 to 21 January 03:00 LT) from
12 to 233 µgm-3. During this process POA contributed to an
average 69 % of NR-PM1 mass. The other three processes were also
characterized by a rapid increase of NR-PM1 mass (39–50 µgm-3 h-1) and a high contribution of POA, i.e., from 22 January
00:00–03:00, 22 January 16:00–20:00, and 23 January
12:00–19:00 LT. Such rapid increases in NR-PM1 mass under low
RH were associated with stagnant weather conditions (e.g., low wind speed)
which promoted the accumulation of pollutants. The second case had an average RH
>80 % from 5 to 8 February (H3 and H4 episodes). In this case, the wind
speed was low (<3 ms-1) throughout the 4-day period. Under
these very stagnant weather conditions, POA accumulated continuously (Fig. 9).
However, unlike from the low-RH case, the concentration of secondary
species also showed continuous increases in this high-RH case. The
enhancement of secondary aerosol formation was likely driven by aqueous-phase
chemistry at high-RH levels (Elser et al., 2016a; Wang et al., 2017) and the
accumulation of pollutants under stagnant weather conditions (Tie et al.,
2017) which further promoted the formation of secondary species.