Introduction
Atmospheric particles are vital players in tropospheric chemistry, regional
air pollution and climate change. High concentrations of fine particles
(i.e., PM2.5) can reduce visibility (Xu and Penner, 2012), deteriorate
air quality (Huang et al., 2014) and are harmful to human health (Xie et
al., 2016). They play an essential role in Earth's radiation balance and
hence affect climate change, directly by scattering and absorbing the
incoming solar radiation (IPCC, 2013) and indirectly by modifying the cloud
properties (Ding et al., 2013; Fukushima et al., 2016). Aerosol particles can
also serve as a medium for reactive gases to undergo heterogeneous and
aqueous-phase reactions (Chang et al., 2011). Understanding the chemical
composition and sources of atmospheric particles is crucial for quantifying
their environmental and health consequences and formulating science-based
mitigation strategies.
Particulate nitrate (NO3-) is a principal chemical component of
atmospheric fine particles. It is an oxidation product of nitrogen oxides
(NOx = NO + NO2) in the ambient
atmosphere. During the day, the oxidation of NO2 by the hydroxyl
radical (OH) produces gaseous nitric acid (HNO3), which then reacts
with ammonia (NH3) or other alkaline compounds to form nitrate
aerosol (Calvert and Stockwell, 1983). The partitioning of HNO3
between gas and aerosol phases is dependent on ambient temperature, humidity
and the abundances of alkaline species (Song and Carmichael, 2001; X. Wang et
al., 2009; Yao and Zhang, 2012). In dark conditions, the reaction of
NO2 and O3 produces the nitrate radical (NO3),
which forms an equilibrium with N2O5 that can be subsequently
taken up onto particles to enhance nitrate aerosol (Pathak et al., 2009,
2011; Brown and Stutz, 2012). The contribution from this pathway is minimized
during the day by the rapid photolysis of NO3 and thermal
decomposition of N2O5. The HNO3 partitioning and
N2O5 hydrolysis reactions have been recognized as the major sink
pathways of NOx in the troposphere (Dentener and Crutzen, 1993; Liu et al.,
2013). There are some other formation routes of nitrate aerosol, such as the
uptake of NO3 radicals onto particles and their subsequent aqueous
reactions with some water-soluble species (Hallquist et al., 1999; see also
Table S1 in the Supplement). The ambient formation of nitrate aerosol highly
depends on the chemical mix of NOx, O3 and
NH3. To date the detailed relationship between nitrate formation
and the chemical mix of NOx, O3 and NH3
is still poorly understood. The contribution of the N2O5 hydrolysis
pathway tends to show a seasonal dependence on the largest influence in the
winter season (Brown and Stutz, 2012; Baasandorj et al., 2017) but will be
dependent on the rate of NO3 formation and reaction, the
N2O5 uptake coefficient (γ (N2O5)) and
formation yield of ClNO2. Field measurement studies have shown that
the γ (N2O5) is highly variable and disagrees with the
laboratory-derived parameterizations (Brown and Stutz, 2012; McDuffie et al.,
2018; Tham et al., 2018). Vertical mixing of air aloft in the residual
layer may also contribute to the surface nitrate pollution (Watson et al.,
2002; Brown et al., 2006; Pusede et al., 2016; Baasanforj et al., 2017;
Prabhakar et al., 2017). Consequently, there are still some remaining
questions for better understanding the nitrate formation mechanisms.
China has been suffering from severe haze pollution as a result of its fast
urbanization and industrialization processes in the past decades. The North
China Plain (NCP), covering the Beijing–Tianjin–Hebei area and surrounding
Shandong and Henan provinces, is the most polluted region with the highest
annual concentrations of PM2.5 in China
(http://www.cnemc.cn/kqzlzkbgyb2092938.jhtml). Previous air pollution
control in China primarily focused on the reduction in anthropogenic
emissions of sulfur dioxide (SO2), given the dominant contributions
of sulfate (SO42-) to the PM2.5 and acid deposition (Hao et
al., 2000, 2007). In the last decade, a circa 75 % reduction in
SO2 emissions in China has successfully resulted in decreases in
the ambient levels of both SO2 and aerosol SO42- in
fast-developing regions including the NCP (Wang et al., 2013; Li et al.,
2017). In comparison, several recent observational studies have indicated an
increasingly important role of aerosol nitrate, which may even dominate
summertime haze formation in the NCP region (Wen et al., 2015; Li et al.,
2018). A recent modeling study has predicted a significant increase in aerosol nitrate along with the decrease in sulfate during 2006–2015 over
eastern China (Wang et al., 2013). To the best of our knowledge, there are no
previous observational reports of increasing nitrate aerosol over northern
China. Long-term measurements are necessary to confirm and quantify this
trend and better understand the nitrate formation mechanisms in China.
To achieve a better understanding of the summertime nitrate pollution and its
formation mechanism in the NCP region, four phases of intensive observations
were conducted at three different sites covering urban, rural and remote
areas in 2014 and 2015. The spatial distribution and temporal variation in
nitrate aerosol pollution were examined. The data were combined with previous
measurements to derive the trends of the nitrate / PM2.5 and
nitrate / sulfate ratios, confirming the statistically significant
increase in regional nitrate pollution during 2005–2015. A multiphase
chemical box model, constrained by in situ observations, was then deployed to
unravel the formation mechanisms of fine particulate nitrate. The impacts of
NO2, O3 and NH3 on the regional nitrate
formation were finally quantified. Overall, the present study provides the
first piece of observational evidence for the increasing trend of nitrate
aerosol in northern China, and our findings have important implications for
the future control of regional haze pollution in the NCP region.
Map showing the study region and three measurement sites. Panel (a)
is color coded with the OMI (Ozone Monitoring Instrument)-retrieved tropospheric NO2 column
density in July 2014, and panel (b) is color coded with the topographic
height (the pink regions denote urban areas).
Materials and methods
Study sites
To better understand the regional-scale nitrate pollution and formation
processes, four phases of intensive field campaigns were conducted at three
sites in the central part of the North China Plain in the summers of
2014–2015. Considering that southerly/southeasterly winds generally prevail
in summertime, the three study sites were carefully selected to lie on a
southeast–northwest transect (see Fig. 1) and to represent the typical urban,
rural and remote atmospheres of the region. A summary of the measurement
locations and periods is shown in Table S2 in the Supplement.
The urban site (36.67∘ N, 117.06∘ E; ∼ 50 m above
sea level, a.s.l.) was located in the downtown area of Jinan, the capital
city of Shandong province, accommodating more than 7 million inhabitants,
∼ 1.7 million automobiles and many factories. Jinan is one of the
largest cities in the central NCP and has been frequently ranked among the
worst ten key cities of China in terms of air quality
(http://www.cnemc.cn/kqzlzkbgyb2092938.jhtml). The site is built on the
rooftop of a six-floor building in the Central Campus of Shandong University,
which is situated in the residential and commercial areas. Details of this
site have been provided in our previous publications (Gao et al., 2011; Wang
et al., 2015). Two intensive campaigns took place during 5–17 May 2014 and
23 August–21 September 2015. In addition, measurements of
aerosol ionic components have been made previously at this site in selected
years since 2005 (Yang et al., 2007, 2012; Gao et al., 2011; Zhu et al.,
2015).
The rural site (36.87∘ N, 116.57∘ E; ∼ 23 m a.s.l.)
was set up at the Chinese Academy of Sciences Comprehensive Station in
Yucheng. Although Yucheng belongs to Dezhou city, it serves as a satellite
town of Jinan. The measurement site is located about 50 km northwest
(normally downwind in summer) of downtown Jinan (Fig. 1) and can thus be
regarded as a receptor site of urban pollution. The instruments were housed
in a temperature-controlled container that was placed in open cropland
with few anthropogenic emissions nearby (Wen et al., 2015; Zhu et al., 2016).
A 6-week campaign was carried out here from 2 June to 16 July 2014.
The remote site (36.26∘ N, 117.11∘ E; 1465 m a.s.l.) was
installed at the summit of Mt. Tai. Mt. Tai is the highest mountain over the
North China Plain (with a peak of 1534 m a.s.l.), and has been widely
deployed as the sampling platform to investigate regional air pollution (Gao
et al., 2005; Sun et al., 2016). The station was set up in a hotel to the
north of the mountain peak with a little lower elevation. It is located
approximately 15 km north of Tai'an city (with a population of 5.6 million)
and 40 km south of urban Jinan (Fig. 1). Detailed descriptions of this site
can be found elsewhere (Guo et al., 2012; Shen et al., 2012). In the present
study, the measurements were conducted from 23 July to 27 August 2014.
Previous data collected at Mt. Tai in summer 2007 are also analyzed to
examine the long-term change in the regional nitrate pollution (Zhou et al.,
2010).
Measurement techniques
A Monitor for AeRosols and GAses (MARGA; ADI20801, Applikon-ECN,
Netherlands) was deployed in the present study to measure continuously, at a
time resolution of 1 h, inorganic water-soluble ions (i.e.,
NO3-, SO42+, NH4+, Cl-, etc.) in
PM2.5 together with acid and alkaline gases (i.e., HNO3, HCl,
NH3, etc.). The target gases and ions are collected and dissolved
by a wet rotating denuder (WRD) and a steam jet aerosol collector (SJAC),
respectively (Brink et al., 2009). The dissolved components are then analyzed
by a cationic and an anionic ion chromatography with eluent solutions of
methane sulfonic acid (308 mg L-1) and NaHCO3
(672 mg L-1)–Na2CO3 (742 mg L-1). An internal
standard solution of LiBr (4 mg L-1) was added automatically to the
collected sample solutions to calibrate the detection in each analytic
process. Multipoint calibration was performed before and after the field
campaigns to examine the sensitivity of the detectors. The detection limits
were evaluated as 0.05, 0.04 and 0.05 µg m-3 for particulate
NO3-, SO42- and NH4+ and 0.01, 0.01
and 0.07 ppbv for gaseous HNO3, SO2 and NH3,
respectively. The MARGA instrument has been deployed in many field studies in
the high aerosol loading environment in China (e.g., Wen et al., 2015; Xie et
al., 2015).
To achieve a detailed analysis of nitrate formation processes, a large suite
of ancillary measurements were concurrently made during the field studies.
PM2.5 mass concentrations were quantified in situ by a SHARP monitor
(Model 5030, Thermo Scientific, USA); particle size and counts in
the range of 5–10 000 nm were monitored by a wide-range particle spectrometer
(WPS; Model 1000XP, MSP Corporation, USA); NO and NO2 were monitored by a
chemiluminescence instrument equipped with an internal molybdenum oxide (MoO)
catalytic converter (Model 42C, Thermo Electron Corporation, USA);
O3 was monitored by an ultraviolet absorption analyzer (Model 49C, Thermo
Electron Corporation, USA); CO was monitored by a nondispersive infrared analyzer
(Model 300EU, API, USA); SO2 was monitored by an ultraviolet
fluorescence analyzer (Model 43C, Thermo Electron Corporation, USA);
meteorological parameters including temperature, relative humidity (RH) and
wind sectors were monitored by commercial automatic weather stations. All of these
techniques have been widely applied in many previous studies, for which there
is detailed information (Gao et al., 2011; Wang et al., 2012;
Xue et al., 2014).
Multiphase chemical box model
A zero-dimensional chemical box model was configured to simulate the in situ
formation of fine nitrate aerosol. It couples the regional atmospheric
chemistry mechanism version 2 (RACM2; including 363 chemical reactions) and
the chemical aqueous-phase radical mechanism version 2.4 (CAPRAM 2.4;
including 438 chemical reactions) to account for gas- and aqueous-phase
atmospheric chemistry (Goliff et al., 2013; Herrmann et al., 2000, 2005). The
gas–aqueous-phase transfer processes were adopted from the resistance scheme
of Schwartz (1986). This model explicitly describes the gas-to-aqueous phase
partitioning of various chemical species, which connects the detailed
chemical reactions in both gas and aqueous phases. The chemical reactions
representing nitrate formation in the model are outlined in Table S1 in
the Supplement. Briefly, these reactions can be divided into three major
formation pathways, namely, the partitioning of gaseous HNO3 to the
aerosol phase, hydrolysis reactions of N2O5 and aqueous-phase
reactions of NO3 radicals. The HNO3 partitioning is
largely affected by the availability of NH3, since the partitioning
of NH3 would decrease the aerosol acidity and hence enhance the
partitioning of HNO3 to the aerosol phase (see Table S1 in the
Supplement). For the N2O5 hydrolysis process, the uptake
coefficient of N2O5 on particles (γN2O5) is
the parameter with large uncertainty in modeling studies. Recent studies have
shown that γN2O5 tends to be largely variable, and
significant discrepancy exists between field-derived laboratory-derived
parameterizations (Chang et al., 2011; McDuffie et al., 2018; Tham et al.,
2018). The RACM–CAPRAM model does not take γN2O5 into
account but explicitly describes the N2O5 gas-to-aqueous phase
partitioning as well as its subsequent aqueous-phase reactions. See Table S1
in the Supplement for the detailed treatment of the N2O5
hydrolysis processes in the model. We estimated the
γ(N2O5) from the reaction rate for the N2O5
gas-to-particle partitioning and the measured aerosol surface area
concentrations and derived a γ(N2O5) value of 0.018 for
our selected cases. Such a level is well within the reported range of
γ(N2O5) derived from the field observations in other
locations worldwide (e.g., 0.001–0.1), including several polluted areas in
northern China (Tham et al., 2018; and references therein). This model has
been utilized previously to simulate the nighttime nitrate formation in
Beijing and Shanghai (Pathak et al., 2011).
The model calculation requires a large number of variables and parameters,
including
gas phase concentrations of NO, NO2, O3,
SO2, HCl, HNO2, HNO3, NH3, CO and
VOCs (volatile organic compounds), etc.;
particulate (or aqueous) phase concentrations of
NO3-, SO42-, Cl-, HSO4-,
NH4+, H+, OH-, etc.;
other auxiliary
parameters such as temperature, RH, pressure, boundary layer height, particle
radius and aerosol water content.
Most of the above parameters were
observed in situ during our intensive measurement campaigns, and the
available data were directly used to constrain the model. The measured
aerosol ions data such as nitrate, sulfate and ammonium were only used as
initial conditions of the model simulation. The model was initialized with
the measured nitrate concentration at the beginning of the episodes, and then
simulated its formation with constraints of other relevant species. The
particle radius was calculated from the measured aerosol number and size
distribution with an assumption that all particles were spherical. A
hygroscopic growth factor obtained from the NCP region by Achtert et
al. (2009) was adopted to take into account the effect of hygroscopic growth
on the particle size and surface. Aerosol H+, OH-,
HSO4-, and water content were simulated by a thermodynamic model
(E-AIM; http://www.aim.env.uea.ac.uk/aim/aim.php) based on the measured
aerosol chemistry data (Clegg et al., 1998; Zhang et al., 2000). The VOC
measurements were not made during the present study, and we used the campaign
average data previously collected in the same areas during summertime for
approximation (Wang et al., 2015; Zhu et al., 2016, 2017). The detailed VOC
species and their concentrations as model input are documented in Table S3 in
the Supplement. We conducted sensitivity tests with 0.5 or 1.5 times of the
initial VOC concentrations and found that the model simulation was somewhat
insensitive to the initial VOC data (the differences in the model-simulated
nitrate formation between sensitivity tests and base run were within
12 %; see Fig. S1 in the Supplement). This likely to be mainly due to the
low levels of biogenic VOCs in the study area. Given the lack of in situ VOC
measurements, however, the treatment of VOC data presents a major uncertainty
in the present modeling analyses. The boundary layer height, which affects
the dry deposition of various constituents, was estimated by the Nozaki
method (Nozaki, 1973). The dry deposition velocity of HNO3 was set
to 2 cm s-1 in the model. With such a configuration, dry deposition only
presents a minor fraction of the daytime HNO3 sink
(< 1 %), compared to the HNO3 gas-to-particle
partitioning.
Simulations were conducted for selected nighttime or daytime nitrate
formation cases. The starting time and simulation periods depended on the
individual cases. The output data included particulate nitrate concentrations
and reaction rates of the major aerosol formation pathways. In addition, a
number of sensitivity simulations were performed to examine the relationships
between nitrate formation and its precursors (see Sects. 3.3 and 3.4).
Statistics (average ± standard deviation) of the measured
aerosol chemical properties, trace gases and meteorological conditions in
urban Jinan, rural Yucheng and Mt. Tai.
Site
Jinan (urban)
Jinan (urban)
Yucheng (rural)
Mt. Tai (remote)
Period
May 2014
Aug–Sep 2015
Jun–Jul 2014
Jul–Aug 2014
NO3- (µg m-3)
8.8 ± 8.2
7.4 ± 5.1
13.6 ± 10.3
6.0 ± 4.6
SO42- (µg m-3)
12.2 ± 7.5
12.7 ± 7.9
23.6 ± 13.4
14.7 ± 8.9
NH4+ (µg m-3)
6.8 ± 5.3
11.1 ± 8.2
11.9 ± 7.7
7.3 ± 5.0
Cl- (µg m-3)
1.3 ± 2.1
1.3 ± 1.7
1.2 ± 1.2
0.7 ± 0.5
PM2.5 (µg m-3)
68.4 ± 41.7
59.3 ± 31.8
97.9 ± 53.0
50.2 ± 31.7
NO3- / PM2.5
0.12 ± 0.06
0.14 ± 0.07
0.14 ± 0.07
0.11 ± 0.05
SO42- / PM2.5
0.18 ± 0.06
0.24 ± 0.10
0.27 ± 0.12
0.30 ± 0.11
NH4+ / PM2.5
0.10 ± 0.05
0.18 ± 0.11
0.13 ± 0.06
0.15 ± 0.07
Cl- / PM2.5
0.015 ± 0.016
0.024 ± 0.027
0.012 ± 0.010
0.019 ± 0.025
[NO3-]/[SO42-]
1.04 ± 0.46
0.98 ± 0.49
0.93 ± 0.53
0.62 ± 0.33
NO2 (ppb)
20.5 ± 9.0
14.1 ± 4.5
16.6 ± 10.7
3.0 ± 2.3
O3 (ppb)
31 ± 19
43 ± 36
38 ± 26
75 ± 21
SO2 (ppb)
10.4 ± 11.1
7.1 ± 4.6
4.2 ± 7.4
2.4 ± 2.8
CO (ppb)
1835 ± 2046
–
622 ± 280
609 ± 214
Particle diameter (nm)
41 ± 12
55 ± 50
78 ± 34
66 ± 21
Particle number (103 # cm-3)
7.1 ± 4.1
12 ± 8.3
3.0 ± 3.8
3.4 ± 2.8
NORa
0.11 ± 0.07
0.16 ± 0.08
0.24 ± 0.13
0.39 ± 0.20
Excess NH4+ (µg m-3)b
2.0 ± 2.4
4.3 ± 6.4
0.9 ± 2.5
1.0 ± 1.8
T (∘C)
22.2 ± 4.2
23.6 ± 3.4
25.4 ± 4.7
18.0 ± 2.7
RH (%)
38.8 ± 19.7
66.0 ± 21.0
70.3 ± 19.8
86.9 ± 12.8
a NOR (nitrate oxidation ratio) =
[NO3-] / ([NO3-]+[NOx]);
b Excess NH4+ =
([NH4+] - 1.5 ⋅ [SO42-] -
[NO3-] - [Cl-]) ⋅ 18.
Note that [NO3-], [NOx], [NH4+],
[SO42-] and [Cl-] are molar concentrations of
NO3-, NOx, NH4+,
SO42- and Cl-, respectively.
Average diurnal variations in fine particulate NO3-,
NO2 and meteorological conditions in (a) urban Jinan in May 2014,
(b) rural Yucheng and (c) Mt. Tai. Error bars stand for the standard
deviation of the measurements. The shaded area denotes the nighttime period.
Results and discussions
Temporal and spatial variations
Table 1 summarizes the statistics of the aerosol chemical properties, trace
gases and meteorological parameters measured at three study sites. It clearly
shows the spatial distribution of regional aerosol pollution though elevated
levels of PM2.5, and major ions were observed at all three sites. The
highest PM2.5 levels were recorded at the receptor rural site (Yucheng;
with a campaign average ±SD of 97.9 ± 53.0 µg m-3),
followed by the urban (Jinan; 68.4 ± 41.7 and
59.3 ± 31.8 µg m-3 in 2014 and 2015, respectively) and
mountain sites (Mt. Tai; 50.2 ± 31.7 µg m-3). Nitrate
shows a similar gradient, with average concentrations ranging from
6.0 ± 4.6 µg m-3 at Mt. Tai to
13.6 ± 10.3 µg m-3 at Yucheng. In comparison,
SO42- shows a slightly different pattern with the lowest levels
found in Jinan (12.2 ± 7.5 and
12.7 ± 7.9 µg m-3) and then Mt. Tai
(14.7 ± 8.9 µg m-3) and Yucheng
(23.6 ± 13.4 µg m-3). Chloride showed comparable levels
in urban Jinan (1.3 ± 2.1 and 1.3 ± 1.7 µg m-3)
and rural Yucheng (1.2 ± 1.2 µg m-3), with a relatively
lower level at Mt. Tai (0.7 ± 0.5 µg m-3). For
NO2, an anthropogenic emission indicator and a major precursor of
NO3-, the highest mixing ratios were determined in urban Jinan,
followed by Yucheng and Mt. Tai. The nitrate oxidation ratio (NOR), defined
as the molar ratio of NO3- to
NO3- + NOx, shows an opposite pattern with
the lowest values in Jinan (0.11 ± 0.07 and 0.16 ± 0.08 in 2014
and 2015) and the highest levels at Mt. Tai (0.39 ± 0.20). This indicates
the different extent of chemical processing of air masses in different types
of areas. The air masses sampled at Mt. Tai were more aged and longer air
transport allowed more time for chemical processing. The above regional
gradients of air pollution are mainly due to the spatial distribution of
anthropogenic emissions and different chemical aging of air masses. It should
be noted that these measurements were not conducted simultaneously, and thus
differences in the reported data at the three study sites can be expected in view
of the potential differences in the meteorological conditions which affect
atmospheric mixing and transport processes. However, the spatial
distributions of emissions, atmospheric chemical and physical processes are
still believed to be the major factor shaping the observed regional pattern
of aerosol pollution.
Long-term trends of (a) mass ratio of NO3-/PM2.5
and (b) molar ratio of NO3- / SO42- in urban Jinan
and at Mt. Tai in summertime from 2005 to 2015. The fitted lines are derived
from the least square linear regression analysis, with the slopes and p values
(99 % confidence intervals) denoted.
Table 1 also illustrates some homogeneity of the regional aerosol pollution
and chemistry in the NCP region. First, secondary inorganic ions (i.e.,
SO42-, NO3- and NH4+) accounted for, on
average, 41–56 % of PM2.5 at three sites, indicating their
dominant roles in the aerosol composition and regional haze. Second,
NO3- alone presented an important fraction of fine particles,
and the NO3- / PM2.5 ratios were nearly uniform over the
region, with average values of 11–14 % at all three sites. Based on
field measurements in January 2013, Huang (2014) also reported that
NO3- accounted for 12, 14 and 13 % of PM2.5
in Beijing, Shanghai and Guangzhou, with a smaller ratio (7 %) recorded
in a western city (Xi'an). At the surface sites (i.e., Jinan and Yucheng) in
the present study, the molar concentrations of NO3- were even
comparable to SO42-, with mean
NO3- / SO42- ratios of 0.93–1.04. In
comparison, the NO3- / SO42- ratio was lower
(0.62 ± 0.33) at Mt. Tai, which is likely to be due to the longer lifetime of
sulfate aerosol and frequent transport of power plant plumes to the mountain
site (Z. Wang et al., 2017). Finally, NH4+ was generally in excess
in PM2.5. The average excess NH4+ (excess
NH4+ = 18 ⋅ ([NH4+] -
1.5 ⋅ [SO42-]-[NO3-]-[Cl-])) was
calculated to be in the range of 0.9–4.3 µg m-3 at our three study
sites. This highlights the NH3-rich chemical environment of the NCP
region, and the abundant NH3 may significantly affect the formation
of nitrate aerosol (see Sect. 3.4).
Figure 2 clearly shows the distinct diurnal variation patterns of
NO3- and NO2 at the three different types of sites. In
urban Jinan, NO3- showed a maximum level in the early morning
(08:00 local time, LT) with a secondary peak in the afternoon (15:00 LT). At
Yucheng, the average diurnal profile displays a continuous nitrate formation
process throughout the nighttime with a NO3- increase of
16.9 µg m-3 from 16:00 to 08:00 LT, followed by a sharp
decrease during the day with a trough in the late afternoon (16:00 LT). The
absolute nighttime NO3- levels were higher than the daytime
concentrations at Jinan and Yucheng, owing to the dilution within the
developed planetary boundary layer (PBL) and thermal decomposition of aerosol
in high temperature conditions during the day. An opposite diurnal profile
was observed at Mt. Tai, which showed an NO3- increase
throughout the daytime and high concentrations remaining in the early
evening. The daytime increase was due to the development of the PBL and the
mountain-valley breeze, both of which could carry the boundary layer
pollution aloft, and the elevated evening levels should be ascribed to the
regional transport of polluted plumes to the mountain top (Sun et al., 2016).
Overall, NO2 showed similar diurnal variations with
NO3-, and the NO3- concentration peaks generally
lagged behind NO2, suggesting the role of NO2 in nitrate formation as a precursor. The inspection of diurnal variations day by day
also revealed frequent nitrate formation at all three sites during nighttime
and also during the day (especially at Mt. Tai). We then selected a dozen of
nitrate formation cases for detailed modeling analyses in Sect. 3.3.
Trend over 2005–2015
Figure 3 shows the increasing trends of NO3- in PM2.5 in
the past decade over the NCP region. Intensive measurements of aerosol ionic
species have been made by our group in urban Jinan in selected years since
2005 (Yang et al., 2007, 2012; Gao et al., 2011; Zhu et al., 2015) and at
Mt. Tai in 2007 (Zhou et al., 2010), and these previous data were combined
with the more recent observations in the present study to derive the decadal
trends. To eliminate the interference of interannual variation in weather
conditions on the absolute concentrations, we focused on the ratios of
NO3- / PM2.5 and
NO3- / SO42- for the trend analysis. In urban
Jinan, the fraction of NO3- in PM2.5 has increased at a
rate of 0.9 % per year over 2005–2015 (p < 0.01). A similar
increasing rate (0.7 % per year) was also derived at Mt. Tai from data
collected in 2007 to 2014, affirming the statistically significant increase in fine particulate nitrate over the region. At the same time, the
SO42- in PM2.5 has statistically significantly declined in
urban Jinan (-0.7 % per year) and at Mt. Tai (-1.3 % per year;
figures not shown), as a result of the strict control of SO2
emissions in China. As a result, the molar ratio of
NO3- / SO42- has increased at a rate of
0.09 per year in Jinan during 2005–2015 (p < 0.01) and 0.05 per
year at Mt. Tai from 2007 to 2014 (Fig. 3b).
Comparison of the model-simulated versus observed nitrate
enhancement (a) as well as the contributions from the major three
formation pathways (b) for the daytime cases in urban Jinan, rural
Yucheng and Mt. Tai. The error bars are the standard error of the differences
between simulated and observed increase in nitrate aerosol.
We also examined the trends in the absolute concentrations of PM2.5,
nitrate and sulfate in urban Jinan and at Mt. Tai (see Fig. S2 in the
Supplement). As expected, the ambient concentrations of PM2.5 (-6.3
and -1.4 µg m-3 yr-1) and SO42- (-2.1
and -1.2 µg m-3 yr-1) have rapidly decreased at both
locations during the past decade, which should be largely attributed to the
stringent control of SO2 emissions and primary particles. In
comparison, the absolute concentrations of NO3- showed an
increasing trend with average rates of change of 0.39 and
0.29 µg m-3 yr-1 at Jinan and Mt. Tai, respectively.
This confirms the increase in absolute nitrate aerosol pollution in the NCP
region. Nevertheless, the available observations since 2011 also showed a
decrease in the absolute levels of nitrate aerosol in Jinan. This trend may
be true considering the strict NOx emission control of China
since 2011, but it may be also partly interfered by the higher aerosol
pollution observed during the campaign of 2011 with unfavorable
meteorological conditions. More measurement efforts are urgently needed to
further examine the recent trend of nitrate aerosol after 2011 and evaluate
the impact of the NOx emission control of China.
Our observations provide direct evidence of a statistically significant
increase in summertime nitrate aerosol in the NCP region along with a
decrease in sulfate in the last decade. The comparable contributions of
NO3- and SO42- to PM2.5 suggest the gradual
shift of the secondary inorganic aerosol type from
SO42--dominant to
NO3--and-SO42--dominant. A recent modeling study
also predicted an increase in nitrate with a decrease in sulfate from 2006 to
2015 over all of eastern China (Wang et al., 2013). A more recent
observational study at two sites (Beijing and Xinxiang) in the NCP region
indicated the important contributions of nitrate in PM1 and its driving
role in the summertime haze pollution (Li et al., 2018). Overall, nitrate has
been playing an increasingly important role in the haze pollution in northern
China. In recent years, the strict antipollution measures implemented by the
central government have led to a significant reduction in the primary
PM2.5 in the NCP, while secondary aerosols such as nitrate are still at
high levels and present the major challenge for further mitigation of haze
pollution (http://www.cnemc.cn/kqzlzkbgyb2092938.jhml). Nitrate and its
precursors should be the next major target for the future control of regional
haze pollution in China.
The same as Fig. 4 but for the selected nocturnal nitrate
formation cases.
Nitrate formation mechanisms
Multiphase chemical modeling was then conducted for typical nitrate
formation events to understand the formation mechanisms of fine particulate
nitrate at three study sites. The selected cases met the following criteria:
The nitrate formation (accumulation) process should last for a
considerable time period (i.e., at least 3 h).
The observed NOR
(NOR = [NO3-] / ([NO3-] +
[NOx]) increased throughout the event.
The
meteorological conditions were stable with a constant wind direction or calm
conditions and without wet deposition.
The data in the early morning
period (i.e., 06:00–09:00 LT) were excluded from analyses to roughly
eliminate the potential influence from downward mixing of air aloft to the
surface sites. A total of 21 nitrate formation events were finally selected, including 10 daytime cases (3, 3 and 4 in Jinan, Yucheng and Mt. Tai)
and 11 nighttime ones (3, 5 and 3 in Jinan, Yucheng and Mt. Tai). Details of
these selected cases are provided in the Supplement (see Table S4).
Figure 4 compares the model-simulated versus observed nitrate enhancements
for the daytime cases and also presents the contributions of the major
nitrate formation pathways. Generally, the model reproduced the observed
nitrate formation well, with a strong positive correlation between simulations and
observations (with a reduced major axis (RMA) slope of 0.90 and r2 of
0.60; see Fig. S3 in the Supplement). The partitioning of HNO3 gas
to the particulate phase was clearly the predominant daytime formation
pathway of nitrate aerosol, with average contributions of 96, 95 and 94 %
at the urban, rural and remote sites, respectively. Hydrolysis of
N2O5 contributed to the remaining nitrate (4–6 %), and the direct
uptake and aqueous-phase reactions of NO3 radicals were negligible.
The modeling results for the nighttime cases are shown in Fig. 5. The model
also worked reasonably well for the simulation of nitrate formation at night,
as indicated by the strong positive correlation between the simulated and
observed NO3- enhancements with a RMA slope
(simulation/observation) of 1.60 and r2 of 0.93 (see Fig. S3 in the
Supplement). Figure S4 shows that the model reproduced the absolute
concentrations of nitrate for two specific typical cases. The hydrolysis
reaction of N2O5 turned out to be the overwhelming formation
pathway at nighttime, with mean contributions of 94, 98 and 91 % at the
urban, rural and remote sites, respectively. Other processes such as the
HNO3 partitioning and aqueous reactions of NO3 radicals
were minor routes. These results are in line with the previous studies that
have assessed the nitrate formation pathways. For example, Pathak et
al. (2011) found that the N2O5 hydrolysis contributed to
50–100 % of the nocturnal nitrate formation in Beijing and Shanghai.
Based on the field measurements of N2O5 and related species,
H. Wang et al. (2017) suggested that the N2O5 hydrolysis
contributed comparably to or even higher than the partitioning of
HNO3 to nitrate formation in Beijing on a daily basis. Overall, the
significant roles of HNO3 partitioning and N2O5
hydrolysis in nitrate formation have been outlined well (Brown and Stutz,
2012).
Model-simulated daytime average NO3- enhancements as a
function of the X times of the base concentrations of NH3,
NO2 and O3 in (a) urban Jinan,
(b) rural Yucheng and (c) Mt. Tai. The results are
the average of sensitivity analyses for all selected daytime cases.
The same as Fig. 6 but for the nocturnal nitrate formation cases.
The budgets of nitrate formation were almost the same among the three study
sites. This indicates the regional homogeneity of formation mechanism of fine
nitrate aerosol over the NCP region. The formation of HNO3 and its
subsequent partitioning to the aerosol phase is the principal formation route
during the day, while the hydrolysis reactions of N2O5 on the
particles play a dominant role during the night. This is in line with the
current understanding that the oxidation of NO2 by OH forming
HNO3 and heterogeneous reactions of N2O5 present the
major NOx sinks during the daytime and nighttime,
respectively (Liu et al., 2013).
According to the above identified major formation pathways, nitrate formation can be influenced by the availability of NOx,
O3 and NH3. NOx are direct precursors of
nitrate formation. O3 is a major oxidant and supplier of OH
radicals during the day and is also a precursor of N2O5 at
night. NH3 may prompt the partitioning of HNO3 to the
aerosols and alter the aerosol acidity that affects not only the
partitioning of HNO3 but also the hydrolysis of N2O5.
Therefore, we further examined the dependence of nitrate formation on
NO2, O3 and NH3 at the three sites by
sensitivity analyses. Sensitivity modeling calculations were conducted by
adjusting the concentrations of the target species (NO2 or
O3 or NH3) by X times (i.e., 0, 0.1, 0.2, 0.3, 0.4,
0.5, 0.8, 1.2 and 1.5), and the other settings remained unchanged with the
base simulations. The difference in the simulated NO3-
concentrations between base and sensitivity runs is like to reflect the impact of
the change in the target species on nitrate formation. The sensitivity
modeling results for the daytime cases are documented in Fig. 6. Similar
results were derived from the three different study areas. During the day,
nitrate formation was the most sensitive to NO2, a necessary
precursor of NO3- aerosol. It was also sensitive to a lesser
extent to O3, which is a major OH source and thus affects the
gaseous HNO3 formation. An interesting finding was the dependence
of nitrate formation to the abundance of NH3. Adjusting (neither
increasing nor decreasing) the currently measured NH3
concentrations by up to 50 % would not lead to significant changes in the
model simulated NO3-, whilst further reduction in NH3
(more than ca. 50–80 %) would result in a significant decrease in NO3-. This indicates that NH3 plays an important role
in nitrate formation, but it is now highly in excess in the NCP region so
that nitrate formation is somewhat insensitive to NH3.
Figure 7 presents the dependence of nitrate formation to NO2,
O3 and NH3 for the nighttime cases. Again, the results
obtained from the three study sites were similar. Nitrate formation was very
sensitive to both NO2 and O3. Adjusting the abundances
of NO2 or O3 would lead to almost a linear response in
the model-simulated nitrate formation. As discussed above, the nocturnal
nitrate formation was mainly controlled by the hydrolysis reactions of
N2O5, which is the product of the reactions of NO2
with O3. In comparison, nitrate formation was not sensitive to
NH3 at all three sites. Interestingly, large reductions in NH3 (ca. > 60 % at Yucheng and
> 90 % in Jinan) would result in a slight increase in the
NO3- aerosol formation. This is likely to be due to the increase in aerosol acidity by reducing the NH3 levels, which could change the
partitioning of the formation of both nitrate and ClNO2 from the
N2O5 hydrolysis. Increasing the aerosol acidity would restrict
the reaction of NO2+ with Cl- yielding ClNO2 and
hence enhance the formation of nitrate aerosol (Roberts et al., 2008). We
conducted sensitivity tests without the inputs of Cl-, and the
results did not show any increase in nitrate formation with reduction in
NH3 (figures not shown).
It should be noted that the Mt. Tai site is located at around
1465 m a.s.l., which is almost near the top of the PBL in summer. Thus, the
Mt. Tai data can provide insights into the chemical conditions in the top
boundary layer during the day and in the residual layer during the night. Our
observations at Mt. Tai demonstrate the serious nitrate aerosol pollution
throughout the PBL in the NCP region. Furthermore, the nitrate formation
mechanisms, including the major formation routes and sensitivities to NOx,
O3 and NH3, were fairly consistent between Mt. Tai and
the surface sites. This implies the regional homogeneity in the in situ
formation of fine nitrate aerosol within the PBL over the NCP region.
Implications for control policy
The above analyses revealed the important roles of NO2 and
O3 in nitrate formation in three different types of areas.
Although NH3 can facilitate the partitioning of HNO3 to
the aerosol phase, it seems that the summertime nitrate formation is less
sensitive to NH3 due to the NH3-rich environments in the
NCP region. To achieve a comprehensive understanding of the effect of
NH3 on nitrate formation, a large set of theoretical simulations
were designed with varying initial concentrations of NO2 and
NH3. The multiphase chemical box model was initialized by a
typical pollution and meteorological condition in the NCP region (see Table
S5 for the detailed modeling setup) and was run to simulate the daytime
nitrate formation from 08:00 to 19:00 LT. The initial concentrations of
NO2 and NH3 were set to vary in wide ranges of
0–200 and 0–40 ppbv to cover a variety of real atmospheric
conditions. The dependence of the model-simulated nitrate increment
(ΔNO3-) on the pair of NO2 and NH3
can be established.
Contour plot of the model-simulated daytime NO3-
formation as a function of the initial concentrations of NO2
(0–200 ppbv) and NH3 (0–40 ppbv). Note that the dashed lines are
artificially drawn to separate the three zones with a different sensitivity of
nitrate formation to NO2 and NH3.
Figure 8 shows the contour plot of the model-simulated daytime
ΔNO3- as a function of NO2 and NH3
concentrations. Several interesting aspects are noteworthy from the figure.
First, NH3 does indeed play a very important role in prompting
nitrate formation. A relatively small amount of NH3 could
significantly enhance the nitrate formation efficiency of
NOx. For example, the formation of 25 µg m-3
of NO3- would consume 116 ppbv of NO2 in the absence of
NH3 but only need 16 ppbv of NO2 in the presence of 10 ppbv
of NH3. Second, in high NH3 conditions, nitrate formation
becomes insensitive to NH3. Nitrate formation is mainly limited by
NO2 when NH3 is in excess. Third, the nitrate formation
regimes can be divided into three types, namely,
“NOx-limited in NH3-deficient conditions”,
“NH3-controlled” and “NOx-limited in
NH3-rich conditions”, according to the concentration ratios of
NO2 and NH3. The identification of the nitrate formation
regime is a fundamental step towards the formulation of a science-based
control policy of nitrate pollution.
Contour plot of the model-simulated nighttime NO3-
formation as a function of the initial concentrations of NO2
(0–80 ppbv) and O3 (0–80 ppbv). Note that the dashed lines are
artificially drawn to separate the three zones with a different sensitivity of
nitrate formation to NO2 and O3.
Similarly, we also performed theoretical simulations to examine the detailed
dependence of nocturnal nitrate formation on both NO2 and
O3. The detailed model configuration is given in Table S6 in the
Supplement. The initial concentrations of NO2 and O3 were
set to vary in the range of 0–80 ppbv to represent various nocturnal
environments. The contour plot is shown in Fig. 9. It clearly shows the three
categories of nighttime nitrate formation regimes, i.e.,
“NOx-limited” under high O3 and low
NO2 conditions, “O3-limited” in low O3 and
high NO2 conditions, and “mixed-limited” by both
NOx and O3. The ambient pollution conditions
measured at the three study sites in the present study were generally in
the mixed-limited regime. Effective control measures could be established
based upon the diagnosis of the nitrate formation regimes.
Our findings have important implications for the control policy of regional
aerosol pollution. Our observations demonstrate the increasing trend and
serious situation of nitrate pollution over the NCP region. Given the decline in sulfate and primary particles in the recent decade (Wang et al., 2013),
nitrate should be a major target for the future control of haze pollution in
China. The observation-based modeling analyses in this study suggest that the
summertime nitrate formation in the NCP region is mainly controlled by
NOx and O3 (particularly in nighttime). Recent
studies have also confirmed the increasing trends of surface O3
levels in the past decades in several major fast-developing regions of
eastern China (Ding et al., 2008; Xu et al., 2008; T. Wang et al., 2009; Xue et
al., 2014; Sun et al., 2016). Therefore, further reduction in anthropogenic
NOx emissions and mitigation of regional O3
pollution should be an efficient way to alleviate the nitrate-driven haze
pollution in China. NH3 also plays a very important role in the
nitrate aerosol formation, as a relatively small amount could efficiently
prompt HNO3-to-NO3- partitioning and nitrate formation.
However, the summertime nitrate formation seems to be less sensitive to
NH3 in the NCP region, where ambient NH3 is generally in
excess. Indeed, the available field observations of ambient NH3
confirmed the widespread NH3-excess chemical environments in
polluted regions of northern China (Meng et al., 2018 and references
therein). Thus, it looks like cutting down the NOx
emissions should be more efficient for the current control of nitrate
pollution in the NH3-rich environments. Nevertheless, reduction in
NH3 emissions is still very important for the future aerosol
pollution control in North China from a long-term perspective in light of
the fact that nitrate formation would be largely restricted in NH3-poor conditions (see Fig. 8).
It is worth noting that in addition to NOx, O3 and NH3,
there are also some other factors that influence nitrate formation. For
example, VOCs are principal O3 precursors, and regulate the
abundances of OH and losses of NO3 (and N2O5). Thus, VOCs can affect the daytime HNO3 formation and nocturnal
N2O5 hydrolysis, which in turn affect nitrate formation. In
addition, the increasing nitrate aerosol may reduce the N2O5
uptake and restrict the nocturnal nitrate formation (Chang et al., 2011). The
results in Fig. 9 only hold if the sensitivity of nitrate production to
N2O5 uptake does not change under different
NOx and O3 conditions. Furthermore, the model
simulations are constrained to ground-based observations and the chemistry
aloft may show a different sensitivity than in Figs. 7 and 9. These aspects
were not quantified in this study. Further studies are needed to explore the
detailed dependence of nitrate formation on a variety of factors including
NOx, O3, NH3, VOCs, aerosol composition and
meteorological conditions.