Qi Jiang1, Fei Wang2, Yele Sun 3

National Meteorological Centre, Beijing 100081, China
Chinese Academy of Meteorological Sciences, Beijing 100081, China
State Key Laboratory of Atmospheric Boundary Layer Physics and Atmospheric Chemistry, Institute of Atmospheric Physics, Chinese Academy of Sciences, Beijing 100029, China


 

Received: December 26, 2018
Revised: March 22, 2019
Accepted: April 2, 2019

Download Citation: ||https://doi.org/10.4209/aaqr.2018.12.0480  

  • Download: PDF


Cite this article:

Jiang, Q., Wang, F. and Sun, Y. (2019). Analysis of Chemical Composition, Source and Processing Characteristics of Submicron Aerosol during the Summer in Beijing, China. Aerosol Air Qual. Res. 19: 1450-1462. https://doi.org/10.4209/aaqr.2018.12.0480


HIGHLIGHTS

  • The physicochemical properties of PM1 species were analyzed.
  • The source identification of OA showed the different mechanisms on the SPM and PPM.
  • The formation and conversion of the SPM play the key role in the formation of haze.
  • Aerosol chemical composition information is linked with meteorological conditions.
  • The conversion of SO2 to SO42– is effective through the aqueous-phase oxidation of SO2.
 

ABSTRACT


In this study, an aerosol chemical speciation monitor (ACSM) and various collocated instruments are used to observe and analyze the chemical compositions, sources and extinction characteristics of submicron aerosol (PM1; aerodynamic diameter < 1 µm) in Beijing from July to September 2012. The results show that the average mass concentration of the PM1 for the entire observation period is 53.8 µg m−3, accounting for 70–85% on average of the PM2.5, and the average mass concentration of the non-refractory submicron aerosol (NR-PM1) declines monthly from July to September as the fraction of organic aerosol (OA) in it increases. During clean days, OA forms the largest mass fraction of the PM1, and the fraction of inorganics shows a significant increasing trend as pollutants accumulate. The effects of meteorology on PM pollution and aerosol processing are also explored. In particular, the SOR increases significantly during periods of elevated relative humidity (RH), suggesting that SO2 is more efficiently converted to SO42− during pollution episodes via aqueous-phase oxidation than gas-phase oxidation. In addition, the effect of wind speed is significantly weaker on primary species (PPM) than secondary species (SPM). Furthermore, the mass concentration of the SPM (or organics) is more sensitive than that of the PPM (or inorganics) to changes in wind speed. The proportion of oxygenated OA (OOA) is significantly higher than that of hydrocarbon-like OA (HOA) in the OA, and as the proportion of OA in the PM1 increases, the mass fraction of OOA in the OA gradually decreases. Moreover, the aerosol acidity in Beijing is essentially neutral during the observation period. The total extinction coefficient of the particulate matter (PM) correlates well with the mass concentration of the PM1 (r2 = 0.72), and the extinction efficiency of the secondary particulate matter (SPM) (r2 = 0.92) is significantly higher than that of the primary particulate matter (PPM) (r2 = 0.58). Meanwhile, the correlation is weaker between the OA and the extinction coefficient (r2 = 0.56) than between the inorganic aerosol and the extinction coefficient (r2 = 0.86).


Keywords: ACSM; NR-PM1; SPM; Extinction coefficient.


INTRODUCTION


Aerosols arise from direct emissions of particles and the conversion of certain gases in the air, which could inhibit visibility and threaten human health (Seinfeld and Pandis, 2006). With the rapid economic development and population explosion, haze pollution has been a serious environmental problem throughout the year in the megacity of Beijing, China (Sun et al., 2014). The nature of the pollution is presented by the intertwining of the sources and sinks of the pollutants, the coupling effect of pollution conversion processes and the synergistic effect of environmental impacts (Jiang et al., 2017). The development of haze is mainly caused by the highly concentrated fine particles (PM2.5) in regional scale (Zheng et al., 2015). Particulate matter (PM), especially submicron particulates (PM1), with mass median aerodynamic diameter less than 1 µm (Qiao et al., 2015), have been found to play a crucial role in human health (Wang et al., 2007) and urban air pollution throughout the world, due to their high deposition efficiency in the human respiratory system, lengthy atmospheric residence time (Chatterjee et al., 2013), large specific surface area and high light scattering efficiency (Mazidi et al., 2018). Human exposure to high concentrations of fine particles can result in higher morbidity and mortality (Chen et al., 2016).

PM1 is mainly produced by volatile gas through gas-to-particle conversion processes. The chemical composition of PM1 is complex, and the components of different sources vary greatly. The major chemical constituents of PM1 include organics, sulfate, ammonium, nitrate, chloride and a variety of minerals (Takami et al., 2005; Canagaratna et al., 2007; Sun et al., 2013). Among them, organic aerosol is the major fraction, accounting for 45% of PM1 on average, and its sources include both primary organics discharged directly and secondary organics generated by photochemical or liquid-phase chemical reactions (Song et al., 2013; Zhang et al., 2007). The sources of PM1 depend on sites and seasons, mainly including secondary inorganics species (sulfate + nitrate + ammonium salt), mineral dust, motor vehicle exhaust, biomass burning and coal combustion (Sun et al., 2012a). According to Liu et al. (2013), because of the rapid hygroscopic effect under high humidity conditions, the secondary inorganics account for a large proportion of PM2.5, especially in the formation period of haze. Based on aerosol mass spectrometers (AMS) observations, Sun et al. (2011) and Huang et al. (2010b) believed that secondary OA (SOA) is more significant in summertime, while primary OA (POA) is more significant during winter. Jiang et al. (2013) observed the autumn aerosol characterization in Beijing by an Aerodyne ACSM (aerosol chemical speciation monitor), and the results showed that there were significant differences in aerosols’ chemical composition between pollution days and clean days. The average contribution of OA to PM1 was about 70% during clean days, while the proportion of secondary inorganics species increased significantly during pollution days, more than ~50%. Simultaneously, the increased contributions of secondary species during pollution days resulted in a higher mass extinction efficiency of PM1 (Jiang et al., 2015). The impact of PM1 on regional and global climate change is mainly reflected in both direct and indirect radiation effects. The indirect effect is of the most uncertainties on radiative forcing of climate change, and one reason is that previous studies could not accurately measure the optical properties of aerosols. The optical properties of atmospheric aerosol are closely related to their chemical compositions and aging processes (Garland et al., 2008; He et al., 2009; Huang et al., 2010a). Han et al. (2015) showed that the extinction coefficient is linearly related to the mass fraction of secondary species and inorganic components in PM1.

Observation study of the physicochemical properties of submicron aerosol is indispensable in understanding the formation and evolution of secondary aerosol and exploring the origins of haze. Due to the complex sources of the aerosol particles and the variability of its evolution in atmosphere, a systematic understanding of the concentration, evolution processes, composition and the extinction characteristics of PM1 is helpful to further study the environmental behavior and climate effects of air pollution. To this end, in this study submicron aerosol species (organics, nitrate, sulfate, ammonium, chloride, and black carbon (BC)), PM2.5, extinction characteristics and gaseous species (O3, SO2, NOx and CO, etc.) were measured in situ by ACSM along with a range of collocated instruments in Beijing from July to September 2012. Then we could obtain a detailed characterization of PM1 species, including composition, concentration, diurnal cycles, extinction characteristics, monthly variation and its relationship with meteorological factors. Moreover, source apportionment of organic is done by using positive matrix factorization (PMF), so as to illustrate the different influence mechanisms of the primary and secondary particulate matters.


MATERIALS AND METHOD



Observation Platform and Data Acquisition

An ACSM equipped with a PM2.5 inlet impactor was used to continuously measure the mass concentration of non-refractory submicron aerosol (NR-PM1) from July to September 2012. The time interval of observation is about 15 min. The sampling site is located on the roof of a second-story building (~8 m altitude) at the Institute of Atmospheric Physics (IAP), Chinese Academy of Science (39°58ʹ28ʺ, 116°22ʹ16ʺE). The meteorological data, including temperature, relative humidity, wind speed, wind direction etc., were obtained from the meteorological tower of IAP which is ~30 m far from the sampling site. In addition to ACSM deployment, the mass concentration of BC was simultaneously measured by an Aethalometer (Model AE22, Magee Scientific) at the wavelength of 880 and 370 nm. In order to do the data quality control, an external flow meter is used to make flow calibration and dynamic zero calibration for noise and leak detection. The PM2.5 mass was measured by a heat Tapered Elemeant Oscillating Microbalance (TEOM series 1400a, Thermo Scientific), and collocated gaseous species (including CO, SO2, NO, NOx and Ox) were measured by various gas analyzers (Thermo Scientific). Simultaneously, a cavity attenuated phase shift-based particle extinction monitor and cavity attenuated phase shift-based NO2 monitor were used to determine the total extinction coefficient of PM2.5 at the wavelength of 630 nm and NO2 mass concentration respectively. The descriptions of specific measuring instruments are shown in Table 1.


Table 1. Instruments and observation projects.


Introduction of ACSM

Aerosol chemical speciation monitor (ACSM) has the same aerosol sampling, vaporization and ionization modules as the previous version of Aerosol Mass Spectrometers (AMS) (Canagaratna et al., 2007). In this study, with the time resolution of 15 minutes and a scan rate of mass spectrometer at 500 ms amu−1 from m/z 10 to 150, the ACSM is able to continuously characterize and monitor the mass and chemical of non-refractory submicron aerosol species (including organic, sulfate, nitrate, ammonium, chloride) in real time and long-term. Through a 100 um critical orifice, aerosol particles with vacuum aerodynamic diameter ~40–1000 nm enter into aerodynamic particle focusing lens. The particle beam is transmitted through the first two chambers into the final detection chamber where the non-refractory particulate material is flash vaporized at the oven with a temperature of about 600°C, and then it is ionized by 70 eV electron impact. The positive ions are analyzed by a quadrupole mass spectrometer. At vaporizer temperature of ~600°C, the ACSM cannot detect refractory materials, e.g., mineral dust and BC, thus, a two-wavelength Aethalometer (Model AE22) is used to measure BC in PM2.5. The detection limits (DLs) of organics, sulfate, nitrate, ammonium, chloride for 30-min average data are 0.54, 0.07, 0.06, 0.25 and 0.03 µg m3 for ACSM. The detailed description and calibration of ACSM has been given by Ng et al. (2011). In addition, due to the orifice blockage the data from 02:00 on August 18 to 15:00 on August 20 is removed. The data after 13:00 on August 27 is rejected due to the air leakage. Other data missing is due to the instrument maintenance and power outages. For ACSM, a collection efficiency (CE) was introduced to compensate for the particle loss, mostly due to particle bounce at the vaporizer (Matthew et al., 2008). CE varies with three main factors: the acidity of aerosol particles, the contribution of ammonium nitrate (ANMF) and the particle phase water (Matthew et al., 2008). To reduce the uncertainties of collection efficiency (CE) due to particle phase water, a silica gel diffusion dryer was used to keep the relative humidity (RH) in sampling line below 40%. In addition, CE is less affected by the acidity with basically neutral atmospheric aerosol (see “Aerosol acidity” section for details). Given high contribution of ammonium nitrate is often observed (> 40%) (Matthew et al., 2008) in this study, the relationship between ANMF and CE is parameterized as: CE = max(0.45, 0.0833 + 0.9167 × ANMF) (Middlebrook et al., 2012).

The positive matrix factorization (PMF) method is performed on ACSM organic aerosol mass spectra for source apportionment, following the procedures described by Ulbrich et al. (2009). Relative to the AMS, the OA mass spectra m/z > 125 of ACSM has greater uncertainty due to the impact of the low transmission efficiency (TE). In addition, m/z 127–129 is affected by the internal standard naphthalene signals. Considering the low contribution of m/z 125–150 to the total signal, we limit PMF analysis to m/z 12–125 in this study. Then, the results of PMF are further evaluated with an Igor Pro-based PMF Evaluation Tool (PET, v 2.04) (Ulbrich et al., 2009). At last, we choose two OA factors, i.e., a hydrocarbon-like OA (HOA) and an oxygenated OA (OOA). 


Cavity Attenuated Phase Shift Spectrometer Particle Extinction Monitor (CAP-PMext) and C
avity Attenuated Phase Shift-based NO2 Monitor (CAPS-NO2)

The CAPS-PMext uses a light-emitting diode (LED) as a light source, a 26 cm optical cavity as a sampling chamber, and a vacuum photodiode to detect light waves. Since the high concentration of particles has a strong scattering and absorption effect on the incident light, the presence of the particles could cause a change in the effective optical path length. When the square wave modulated light emitted by the LED enters the sampling cavity through the first high reflectivity lens, the change in the effective optical path length could cause the square wave to be distorted. This change can be detected as a phase shift by the vacuum photodiode. The signal is then converted to an extinction coefficient value output. The measuring wavelength is 630 nm and the flow rate is ~0.85 L min−1. The CAPS-PMext is measured per second with a precision of 1 Mm1.

CAPS-NO2 is an instrument that can directly measure the absorption spectrum of NO2 in the 450 nm band and then deduces the concentration of NO2 in the air. When the light passes through the absorption cell, due to the presence of NO2, the phase of the light wave will deform due to the absorption of NO2. Therefore, by measuring the magnitude of the phase shift, the NO2 concentration can be directly measured. A detailed instrument introduction has been published in the paper of Kebabian et al. (2008). The CAPS-NO2 is measured per second with a precision of µg m−3


RESULTS AND DISCUSSION


General Descriptions of the Overall Data during July–September 2012

Fig. 1 shows the time series of NR-PM1 and PM2.5 mass concentration, chemical components in PM1 (organics (Org), SO42, NO3, NH4+, Cl and BC), extinction coefficients and meteorological elements from July to September in 2012. The data gap is caused by data culling due to power off, instrument failure or data quality control. During the period, the average temperature of the site is 25.1°C, and the lowest and highest temperatures are 12.7°C and 36.3°C, respectively. The relative humidity ranges from 14% to 91%, with an average value of 59.9%. The average wind speed is low (< 1 m s−1), and the precipitation is mainly in July and August. From July to September, the relative humidity and temperature gradually decrease, while the pressure increases progressively. The statistical values of each physical quantity are shown in Table 2.


Fig. 1. Time series of (a) precipitation, wind speed (WS), wind direction (WD), temperature (T) and relative humidity (RH); (b) NR-PM1 and PM2.5; (c) total extinction coefficients; (d) NR-PM1 species (Org., SO42–, NO3–, NH4+ and Cl–) proportion and (e) PM1 (NR – PM1 + BC) species for the entire period.Fig. 1. Time series of (a) precipitation, wind speed (WS), wind direction (WD), temperature (T) and relative humidity (RH); (b) NR-PM1 and PM2.5; (c) total extinction coefficients; (d) NR-PM1 species (Org., SO42, NO3, NH4+ and Cl) proportion and (e) PM1 (NR – PM1 + BC) species for the entire period.


Table 2. A summary of the average mass concentrations of aerosol species and meteorological elements during July–September 2012.

During the whole observation period, the variation trend of NR-PM1 measured by ACSM is in good agreement with the variation of PM2.5 measured by TEOM, which indicates the good operation state of the instrument and the feasibility of the data. After superimposing the BC concentration data, the average mass concentration of PM1 (NR − PM1 + BC) is 53.8 µg m3 and accounts for 70–85% of PM2.5 on average, with the correlation (r2) reaching 0.8. The mass concentration of submicron aerosols in Beijing fluctuates greatly with time, from 0.63 µg m3 to 285 µg m3. In the 92 days of the observation, the PM2.5 concentration in 33 days exceeds the second grade of NAAQS (75 µg m3) released by the Ministry of Environmental Protection of People’s Republic of China, and the exceeding rate is ~36%. Among the three months, the exceeding rate in July is the highest, reaching 41%, followed by September, and the lowest rate is in August (33%). The daily mean concentration of PM2.5 exceeds the 24-h NAAQS of U.S. EPA (35 µg m3) in 65 days. The exceeding rate is ~71%, and shows a monthly decreasing trend from July to September. During the whole observation period, the pollution episodes (PM1 > 100 µg m3) alternate with the clean periods (PM1 < 20 µg m3), which usually shows a “sawtooth” variation of the asymmetric structure: slow fluctuation accumulation and rapid clearance mechanism. At the same time, the duration of clean periods is usually short (generally less than 3 days), while the days of pollution often last longer, sometimes even more than a week (e.g., the pollution episode on July 16–21). The switch of clean and polluted days is closely related to meteorological elements in Beijing. Comparatively, the serious pollution episodes are usually associated with higher relative humidity (RH > 60%) and easterly or southerly winds less than ~1.5 m s1.

Organics is one of the most important components of submicron aerosols in Beijing during observation, accounting for ~39% of PM1 on average, followed by nitrate (~21%), sulfate (~17%), ammonium (~14%), BC (~8%), and chloride (~1%). Since there is no extinction coefficient data in July, the average values of atmospheric extinction coefficients are 249.2 Mm−1 and 182.8 Mm−1 in August and September, respectively.

The nitrogen oxidation ratio (NOR) and sulfur oxidation ratio (SOR) are good indicators of secondary transformation. The calculation formulas are as follows:

 

S refers to mass concentration (µg m−3). The valid-data periods for SO2 and NO2 are September 2012. Fig. 2 is the time series of SO2, SO42− and NO2, NO3. It was reported that SOR was smaller than 0.1 in primary emissions whereas the SOR was above 0.1 when SO42− was produced by SO2 through the photochemical oxidation (Sun et al., 2006). The SOR value varies from 0.04 to 0.75, with an average of 0.45 in September. This result is higher than the observations in Beijing in winter (Wen et al., 2007) and close to the results of the study in the summer of Beijing by Sun et al. (2006). The rate of heterogeneous oxidation of SO2 increases with the increase of relative humidity (Dlugi et al., 1981), indicating that aqueous processing might have played a more important role in the formation of sulfate. During September, SOR shows strong positive correlations with relative humidity (r2 = 0.86). The correlation between NOR and relative humidity is slightly lower, but the correlation (r2) also reaches 0.52. At the same time, as the concentration of pollutants (NR-PM1) increases, the probability of occurrence of high SOR and NOR values increases significantly. Especially in the pollution process from September 7 to 12, the SOR is always higher than 0.65, and the contribution of sulfate to NR-PM1 has increased by ~10–15%. At the same time, pollutants continue to accumulate in this period due to the higher relative humidity. The results suggest that the conversion from SO2 to SO42− in pollution episodes is more effective through the aqueous-phase oxidation of SO2 by the catalysis of the transition metals (e.g., Fe) or by H2O2/O3 oxidation directly (Yao et al., 2002) instead of the gas-phase oxidation (Sun et al., 2006). Although there is a higher SOR and a higher oxidation rate of SO2 in September, the mass concentration of sulfate is still significantly lower than that of winter observations. The main reason is that the mass concentration of SO2 in winter increases with the amount of coal pollution during heating, and the mass concentration of SO42− is still higher than non-heating season.


Fig. 2. The time series (hourly averaged data) of (a) SO2, SO42–, SOR and (b) NO2, NO3–, NOR. The data of SOR and NOR are binned according to RH (10% increment). The marker sizes are proportional to the mass concentrations of NR-PM1.Fig. 2. The time series (hourly averaged data) of (a) SO2, SO42–, SOR and (b) NO2, NO3, NOR. The data of SOR and NOR are binned according to RH (10% increment). The marker sizes are proportional to the mass concentrations of NR-PM1.


Aerosol Acidity

Aerosol acidity is an important parameter affecting its toxicity, hygroscopicity and heterogeneous reaction. Under the assumption that NH4+ in the submicron aerosol exists in the form of ammonium sulfate, ammonium nitrate and ammonium chloride, the influence of metal ions and organic acids on the concentration of NH4+Predicted is ignored (Zhang et al., 2007). The acidity of aerosols is determined by the ratio of the measured concentration of NH4+measured and the estimated concentration of NH4+Predicted. The formula for estimating the concentration of ammonium particles is as follows:

A ratio of NH4+measured/NH4+Predicted less than 1 indicates that the atmospheric aerosol is acidic, and the lower thevalue, the more acidic of the aerosol. On the contrary, the atmospheric aerosol is considered to be alkaline. If the ratio is close to 1, the atmospheric aerosol is considered to be basically neutral. Since the gasification temperature of ACSM is 600°C, some of the less volatile composition such as SO42, NO3 and Cl components combined with the positive metal particles cannot be detected, and the influence of metal ions can be negligible. Therefore, the method is suitable for this study.

Fig. 3 shows the linear regression analyses between predicted NH4+ (NH4+Predicted) that is needed to fully neutralize anions (i.e., nitrate, sulfate and chloride) and NH4+ measured by ACSM (NH4+Measured) from July to September. As shown in Fig. 2, NH4+Measured and NH4+Predicted have a high positive correlation in the three months, the correlations (r2) are all above 0.98 and the ratio between them is close to 1, indicating that the NH4+Measured can be completely neutralized by the acidic anion in NR-PM1. Thus, similar to the previous research conclusions in Beijing (Sun et al., 2011, 2012b), the atmospheric aerosol observed is basically neutral during observation period in this study.


Fig. 3. Linear regression of predicted NH4+ mass concentration required for complete neutralization of anions (NH4+Predicted) and NH4+ measured by ACSM (NH4+Measured) in (a) July, (b) August and (c) September, respectively (The different colors in the figure correspond to different times.).Fig. 3. Linear regression of predicted NH4+ mass concentration required for complete neutralization of anions (NH4+Predicted) and NH4+ measured by ACSM (NH4+Measured) in (a) July, (b) August and (c) September, respectively (The different colors in the figure correspond to different times.).


Characteristics of NR-PM1 Chemical Components

Variations in meteorological conditions, atmospheric circulation, source and sink in different months could inevitably lead to the change of the atmospheric aerosol composition. Figs. 45 show the variations of mass fraction of submicron aerosol species as a function of total NR-PM1 mass and the average chemical composition of NR-PM1 and OA during the observation period. From July to September, the probability of NR-PM1 shows a decreasing trend with the increasing of NR-PM1 mass loadings, and the period when the highest frequency appears moves to the decreasing direction of NR-PM1 mass concentration. As a comparison, the frequency of pollution episodes in July and August is significantly higher than that in September. In July, the probability of NR-PM1 mass concentration which is larger than 10% is concentrated in the range of 50–70 µg m−3, and that is 20–50 µg m−3 in August. In September, the main probability distributions of NR-PM1 is less than 30 µg m−3. During the whole observation period, the proportion of different chemical components shows different trends with the increase of NR-PM1 mass concentration. The organics dominates the NR-PM1 mass concentration with an average fraction of ~60% at low mass loadings (NR-PM1 < 25 µg m−3). However, the contribution of inorganics species could increase by ~30% when NR-PM1 mass loading is about 120 µg m−3. The similar conclusion was also observed at the rural site Yufa during summer (Huang et al., 2010b). During the whole observation period, the mass fraction of nitrate increases from < 10% at low mass loading to ~30% at high NR-PM1 mass loading, and analogously, the fraction of ammonium also shows an upward trend, but not significantly. The fraction of sulfate shows a different trend with the increase of NR-PM1 mass. Specifically, the fraction of sulfate increases at low mass loadings (NR-PM1 < 50 µg m3), and then shows a decreasing trend when the mass loading exceeds 80 µg m3 as a result of high concentration of nitrate pollution episodes in September. Comparatively, sulfate remains a consistent high contribution of about 15–20% across all mass loadings of NR-PM1 during July and August. Therefore, it appears that the organics dominates the NR-PM1 mass in the cleaning period, and the inorganics (mainly (NH4)2SO4 + NH4NO3) play a more important role with the accumulation of the pollutants.


Fig. 4. Variations of mass fraction of NR-PM1 and variations of NR-PM1 species as a function of total mass concentration of NR-PM1 in (a) July, (b) August and (c) September, respectively.Fig. 4. Variations of mass fraction of NR-PM1 and variations of NR-PM1 species as a function of total mass concentration of NR-PM1 in (a) July, (b) August and (c) September, respectively.


Fig. 5. Average chemical compositions of NR-PM1 and OA components in July, August and September, respectively.Fig. 5. Average chemical compositions of NR-PM1 and OA components in July, August and September, respectively.

The PMF is used to analyze and evaluate the organics mass spectrometry from July to September, and hydrocarbon-like OA (HOA) and oxygenated OA (OOA) are finally selected by comparing with the simultaneous observation of gas components (O3, SO2, NOx and CO, etc.), BC and various organic source spectra. The mass spectrometry of HOA is mainly characterized by the prominent hydrocarbon ion series of CnC2n+1+ (m/z 29, 43, 57, 71, …) and CnC2n-1+ (m/z 27, 41, 55, 69, …) (Ng et al., 2011). HOA is closely related to BC (a tracer for combustion emissions, r2 = 0.59–0.68) and NOx (r2 = 0.52–0.63) from July to September, indicating the important contribution of vehicle sources. At the same time, the common feature of OOA is characterized by a prominent peak of m/z 44 (mainly CO2+). OOA has a high correlation with secondary inorganic species such as sulfate (r2 = 0.71–0.77) and nitrate (r2 = 0.66–0.75), indicating that OOA is driven by regional production mostly and is in the nature of secondary generation. The mass spectrometry characteristics of OOA have more similarities to low-volatility oxygenated organic aerosol (LV-OOA) in this study, indicating the important contribution of secondary sources to OOA (Jiang et al., 2013). During the whole observation period, organics is the dominant component of NR-PM1 (~38–46%), but it is still less than the data observed in winter (50%) with the average mass loading being ~34.4 (± 27.9) µg m3, which is 1.5 times higher than that during July to September (Sun et al., 2013). Nitrate (~22%) and sulfate (~19%) are significantly higher than the observations in winter (~16% and ~14%, respectively). In addition, the average contribution of ammonium is 15–17%, and that of chloride is much smaller, ~1%. Meanwhile, the proportion of OOA in OA is significantly higher than that of HOA during the whole observation period. OOA respectively contributes 68.4%, 68.2% and 60.8% of OA in July–September. With the increase of the proportion of OA in PM1, the mass fraction of HOA in OA increases gradually, while the trend of OOA is opposite (Fig. 6).


Fig. 6. Variations of mass fraction of HOA and OOA with the increase of Org./NR-PM1.
Fig. 6. Variations of mass fraction of HOA and OOA with the increase of Org./NR-PM1.


Diurnal Variations of Submicron Aerosol

The diurnal variations of the aerosol species are affected by a variety of factors, including daytime photochemical reactions, gas-particle partitioning, local source emissions, and diffusion conditions. Therefore, the diurnal cycles vary between different submicron species. The diurnal cycles of submicron species are shown in Fig. 7. From July to September, the diurnal profiles of OA are similar in each month, which shows an obvious bimodal distribution. Two peaks are observed at noontime (~12:00 LT) and nighttime (~20:00 LT), which reflects the influences of cooking emissions by the residents. The height of the two peaks in July and August is approximately the same, and the night peak is slightly higher than the noon peak, which is consistent with the observation in summer (Huang et al., 2010b). In September, however, the night peak of OA is about 1.5 times of the noon peak. He et al. (2010) also pointed out that cooking OA is one of the important sources of organics in the atmosphere. Due to the decrease of the planetary boundary layer (PBL) height and the weakening of the photochemical reaction, the gas phase VOCs are easier to condense on the surface of the particles. In addition, OA is also affected by traffic jam at 20:00 LT, resulting in an increase in the concentration of organics particles at nighttime. The results lead to that the night peak of organics is higher than noon (Allan et al., 2010). In August and September, there is a small peak between 08:00–09:00 LT, which is presumably related to the emissions of cooking breakfast. In addition, the contribution of peak traffic is also one of the factors that cannot be ignored during this period. In addition, the diurnal profile of sulfate is relatively flat with a relatively consistent contribution to the NR-PM1 throughout the day (16–21%). In July and August, the sulfate presents an enhanced noon peak, and the peak in August is significantly lower than that in July. However, such peak is not observed in September. One of the reasons for the peak of SO42 is the photochemical production of H2SO4 from SO2 + OH, which is enhanced with the solar light intensity at noon. In addition, observations show that sulfate production has a close positive correlation with relative humidity (Sun et al., 2013). Aqueous processing has a great significance in the formation of sulfate. The average relative humidity decreases monthly from July to September, and the solar radiation also shows a downward trend. Therefore, the noon peak of sulfate decreases by month.


Fig. 7. Diurnal cycles of submicron aerosol species from July to September.Fig. 7. Diurnal cycles of submicron aerosol species from July to September.

NO3 is a volatile inorganic substance affected by the photochemical oxidation reactions of the gaseous precursors and the temperature dependent gas-particle partitioning. The diurnal cycle of nitrate shows the highest concentration in the morning (~09:00 LT), corresponding to the morning rush-hour traffic peak. The main reason is that vehicles emit a large amount of gaseous precursor such as NOx during the traffic peak, and ammonium nitrate particles are formed by the reaction of ammonium nitrate with gaseous ammonia through homogeneous or heterogeneous reactions. Due to the combining effects between the evaporative loss with temperature increasing and the photochemical production of HNO3, attaching to the dilution of PBL, NO3 shows a downward trend after early morning, with daily minimum appearing at ~16:00 LT. The higher temperature in July–September leads to the volatilization of nitrate in the transportation process. Therefore, NO3 is mainly contributed by local secondary generation.

Influenced by the vaporizer temperature of ACSM (600°C), the non-volatile chloride (e.g., NaCl) cannot be detected and the chloride measured is primarily in volatile form (NH4Cl). The diurnal variation of chloride is similar to that of NO3. The mass concentration of Cl drops rapidly after ~08:00 LT and maintain a relatively low level during the afternoon, declaring that the diurnal variation of Cl is driven by the temperature-dependent gas-particle partitioning of NH4Cl and the change of the PBL. The diurnal variation of chloride shows a small peak at ~20:00 LT in July, which is supposed to be caused by local emissions.

The diurnal variation of the ammonium is not significant, which shows a small peak consistent with the morning rush-hour traffic peak. BC is mainly caused by incomplete combustion of fossil fuels and biomass fuels, and the diurnal cycle of BC shows a higher concentration at night and a lower value during the daytime, which is mainly caused by the effective dilution of the near-surface BC due to the elevation of the PBL at daytime. BC shows a higher concentration at 04:00–08:00 LT, probably because of the high concentrated emissions from large trucks that are not allowed to enter the urban area during the daytime. Overall, the diurnal profiles of PM1 species show similarity in each month from July to September. The concentrations of PM1 species all show a decreasing trend month by month, except that the diurnal curves of chloride in August and September are almost the same.


Correlations between PM1 Species and Mass Extinction Coefficients

The extinction of the particles includes both light scattering and light absorption of the particles, and scattering is dominant. Fig. 8 shows the correlations of total extinction coefficient versus submicron aerosol species, secondary species (SPM = SO42 + NO3 + NH4+ + OOA) and primary species (PPM = BC + HOA + Cl) in PM1 during July–September 2012. The total extinction coefficient correlates highly with PM1 and the correlation coefficient (r2) reaches 0.72. At the same time, the time series of extinction coefficient shows a similar trend to PM1 (Fig. 1), indicating that PM1 particles have a strong extinction efficiency and play a key role in the formation and development of haze. Obtained by linear regression, the average unit mass extinction efficiency of PM1 during observation period is ~5.04 m2 g1, which is close to that obtained by two pollution episodes in Shanghai (4.87 and 5.17 m2 g1) (Huang et al., 2010b). In the calculation process, the extinction efficiency of 630 nm has been converted to that of 532 nm by using an extinction Ångström exponent of 1.4 (Pereira et al., 2011).


Fig. 8. Scatter plots of total extinction coefficient (Ext.) versus submicron aerosol species, PPM and SPM.Fig. 8. Scatter plots of total extinction coefficient (Ext.) versus submicron aerosol species, PPM and SPM.

Various chemical components have significant differences in extinction efficiency due to their different refractive indices. It is worth noting that the correlation coefficient (r2) of extinction coefficient with SPM reaches 0.92, which is much higher than that with PPM (r2 = 0.58). Furthermore, total extinction coefficient shows very similar diurnal pattern to SPM, however, the diurnal variations of PPM and extinction coefficient are significantly different (Fig. 7). Among them, the correlation coefficients (r2) between extinction coefficient and the SPM species including SO42, NO3, NH4+ and OOA are 0.69, 0.72, 0.86 and 0.65, respectively. And the values of r2 between extinction coefficient and SPM involving BC, HOA and Cl are 0.63, 0.14 and 0.42, respectively. The relatively high correlation of BC is due to its absorption of light. In summary, the contribution of SPM to extinction is significantly higher than that of PPM. An important reason for the difference in extinction between PPM and SPM is the difference in particle size distribution (Xu et al., 2014), and the secondary inorganic species are mainly concentrated in the accumulation mode (Massoli et al., 2012; Sun et al., 2011), whose mode peak (~300–700 nm) is closed to the absorption band of visible light, resulting in strong extinction efficiency. The majority of the mass fraction of PPM (e.g., BC) is small particles (diameter < 300 nm) with a weak light scattering rate, and the extinction contribution is also greatly reduced. Comparatively, the correlation between organics and extinction coefficient (r2 = 0.56) is weaker than inorganics (r2 = 0.86) significantly. The proportion of SOA in OA is much higher than that of POA. At the same time, the contribution of POA to extinction is significantly weaker than that of SOA. Therefore, the contribution of OA to extinction mainly comes from the secondary organic component during this observation. This result is consistent with the findings of Han et al. (2017) in Beijing. Fig. 9 shows that the diurnal cycles of extinction coefficient reach a maximum at about 10:00, which corresponds well with the trend of nitrate and ammonium (Fig. 7). Therefore, the peak may be mainly caused by the extinction of ammonium nitrate. In addition, the extinction coefficient is significantly reduced during the daytime for two reasons. One is the effect of evaporation loss of volatile and semi-volatile aerosols such as ammonium nitrate and ammonium chloride for the high temperature during the day. The other is the effect of boundary layer uplift in the daytime.


Fig. 9. Diurnal variations of total extinction coefficient, mass concentration of primary and secondary particulate matter for the entire periodFig. 9. Diurnal variations of total extinction coefficient, mass concentration of primary and secondary particulate matter for the entire period.


Meteorological Effects on PM1 Species

Meteorological factors are essential in the formation, evolution, transmission and dissipation of PM pollution. Fig. 10 shows the variations of PM1 species and OA components with the wind speed during July–September 2012. Without considering the relative humidity, the mass concentration of all species shows a decreasing trend with the increase of wind speed (< 3 m s1) and the reduction rate of PM1 is ~23% per m s1. When the ground wind speed is larger than 3 m s1, the total mass load of all species drop to a lower concentration (< 10 µg m3). The trends of HOA, BC and Cl are similar, showing an almost linear decrease with the wind speed increasing (< 3 m s1). However, SO42, NO3, NH4+ and OOA show a more similar trend with the wind speed changing. When wind speed is smaller than 2.5 m s1, the mass loading of NH4+, SO42 and OOA decreases slightly, and NO3 also shows a slowly decreasing trend at a reduction rate of less than 9% per m s1. When the wind speed is larger than 2.5 m s1, the mass loading of SO42, NO3, NH4+ and OOA decrease significantly. HOA, BC and Cl belong to PPM, and OOA, SO42, NO3 and NH4+ belong to SPM, indicating that the wind’s dilution efficiency of PPM is more significant than that of SPM. This conclusion is consistent with the study by Sun et al. (2013). However, the efficiency in cleaning up the pollution by wind during July to September is lower than that in wintertime to some extent. At the same time, the inorganic species, e.g., SO42, NO3, NH4+, show relatively weaker wind-dependency in comparison to organics when the wind speed is less than 2.5 m s1.


Fig. 10. Variations of PM1 species, OA components, primary and secondary particulate matter, inorganic and organic matter as a function of wind speed. The data is further separated into two groups with an RH threshold of 60%.Fig. 10. Variations of PM1 species, OA components, primary and secondary particulate matter, inorganic and organic matter as a function of wind speed. The data is further separated into two groups with an RH threshold of 60%.

To further explore the effects of wind on PM species, the wind speed is separated into two groups with relative humidity (RH) higher than 60% and lower than 60%. The average mass concentration of each PM1 species at high RH is higher than that at low RH. The result also proves the importance of RH in the accumulation and development of PM pollution. Similar to the statistical result of all relative humidity conditions, when RH is smaller than 60% and the wind speed is smaller than 2.5 m s1, the dilution effects of wind speed on PPM (inorganics) are significantly weaker than SPM (OA).


CONCLUSIONS


The particle composition and light extinction characteristics of the aerosol in the megacity of Beijing, China, from July to September 2012 have been examined in this study. The NR-PM1 species, including sulfate, organics, ammonium, nitrate and chloride, were measured in-situ with an Aerodyne ACSM, and the mass concentration of the BC was simultaneously measured at the wavelengths of 880 and 370 nm by an Aethalometer. The average mass concentration of the PM1 was 53.8 µg m3, accounting for 70–85% on average of the PM2.5 during the observation period, with a correlation coefficient (r2) of 0.8. During clean days, OA contributed the largest mass fraction of the PM1, and the fraction of inorganics showed a significant increasing trend as pollutants accumulated. During the pollution episodes, the SOR and NOR significantly increased during periods of elevated RH, and the mass concentrations of sulfate and nitrate as well as their proportions in the PM1 also significantly increased, suggesting that controlling SO2 and NO2 emissions should be a priority when mitigating air pollution during haze. OOA contributed significantly more to the organic aerosol (~68.4%, ~68.2% and ~60.8% during July, August and September, respectively) than HOA did. As the proportion of organics in the PM1 increased, the mass fraction of HOA gradually increased while the mass fraction of OOA gradually decreased in the OA. During the observation period, the atmospheric aerosols were basically neutral, and, on a monthly basis, the average concentration of the NR-PM1 decreased while the mass fraction of OA increased. When the probability of NR-PM1 concentration exceeded 10%, the mass concentrations primarily remained in the ranges of 50–70 µg m3, 20–50 µg m3 and < 30 µg m3 for July, August and September, respectively. The total extinction was highly correlated with the mass concentration of the PM1, with a maximum correlation coefficient (r2) of 0.72, and, notably, the extinction efficiency of the SOA was significantly higher than that of the POA. In addition, the extinction coefficient exhibited a weaker correlation with the OA (r2 = 0.56) than with the inorganic aerosol (r2 = 0.86). Finally, wind efficiently diluted the PM pollution, but its effect was significantly weaker on the PPM (inorganics) than on the SPM (organics). However, a ground wind speed above 3 m s1 reduced the concentrations of the PM1 species.


ACKNOWLEDGMENTS


The study is supported by National Key Research and Development Plan Project (2016YFC0203301), National Fund Committee Key Research Project (91644223), China Meteorological Administration Forecaster Special (CMAYBY2018-092) and Causes of Atmospheric Heavy Pollution and Key Control Projects (DQGG0104).



Don't forget to share this article 

 

Subscribe to our Newsletter 

Aerosol and Air Quality Research has published over 2,000 peer-reviewed articles. Enter your email address to receive latest updates and research articles to your inbox every second week.

Latest coronavirus research from Aerosol and Air Quality Research

2018 Impact Factor: 2.735

5-Year Impact Factor: 2.827


SCImago Journal & Country Rank

Aerosol and Air Quality Research (AAQR) is an independently-run non-profit journal, promotes submissions of high-quality research, and strives to be one of the leading aerosol and air quality open-access journals in the world.