Source Apportionment of Ambient Aerosols during a Winter Pollution Episode in Yinchuan by Using Single-Particle Mass Spectrometry

Abstract

For a winter pollution episode in Yinchuan, a city in Northwestern China, ambient aerosols were characterized using a real-time single-particle aerosol mass spectrometer (SPAMS). More than 160,000 individual particles analyzed with the SPAMS were classified into eight major categories on the basis of their mass spectral patterns: traffic emissions, biomass burning, dust, coal burning, industrial emissions, secondary inorganic, cooking, and others, all of which contribute to fine particles. The results revealed that coal burning (29.6%) and traffic emissions (23.4%) were the main sources during the monitoring period. Industrial emissions and secondary inorganic aerosols accounted for 16.6% and 14.0%, respectively. The SPAMS data indicated that the number concentration of the eight types of particles was markedly different in the different pollution cases, and higher number concentrations were discovered more often during pollution episodes. The three pollution cases were mainly caused by the accumulation of fine particles, mainly from traffic emissions, industrial emissions, and increased secondary inorganic conversion.

1. Introduction

Aerosols are uniformly dispersed in a relatively stable suspension system formed by gases. They strongly influence the global climate, regional environmental pollution, and human health by scattering and absorbing sunlight, ground radiation, and cloud condensation nuclei, which are an essential component of the earth’s atmosphere [1,2,3]. Aerosols are fine particles (aerodynamic particle size < 2.5 μm (PM_2.5)) with a large specific surface area, long atmospheric residence time, and high scattering efficiency; they therefore exert extremely severe effects on the atmospheric environment and human health. Aerosol pollution has consequently attracted extensive attention [4,5,6].

Many methods exist for studying aerosol particles. Early studies used the filter sampling method. At present, the aerosol time-of-flight mass spectrometer (ATOFMS) and single-particle aerosol mass spectrometer (SPAMS) are widely used to analyze aerosols’ chemical composition and source apportionment [7,8,9,10,11,12,13]. SPAMSs have the advantage of high time resolution and an ability to continuously analyze the size and chemical composition of individual particles in the atmosphere in real time [14,15,16]. The University of California has one of the most representative research teams worldwide, studying atmospheric particles by using single-particle technology [17]. They developed an ATOFMS to study the type and mixing states of atmospheric particles in northern Mexico [17]. The SPAMS produced by Hexin Analytical Instruments, Guangzhou, is widely used in China [14,18,19,20]; it can detect the chemical and physical properties of single particles. Numerous studies have been conducted in Chinese cities—including Beijing [16], Shanghai [21,22], Guangzhou [23,24], and Chongqing [25]—to determine the size and chemical composition of local atmospheric particles and comprehensively understand the source and mixing state of individual particles [26]. For example, Ma et al. used a SPAMS to assess the chemical composition and source appointment of aerosols in Beijing on clear, hazy, and dusty days; their results revealed that carbonaceous and K-rich particles were dominant under different conditions and that the contribution of industrial metal particles was higher during hazy periods [16]. Mu et al. used a SPAMS to analyze the mixing state of fine particles during pollution periods in Shanghai and reported that EC (Elemental Carbon) particles tended to mix with secondary ions in the atmosphere [22]. Bi et al. determined the chemical composition of fog droplet residue particles by using a SPAMS at the ground level in an urban area. The results revealed that the effects of organic particles and ammonium on fog formation were weak in Guangzhou [24]. A SPAMS-based analysis of the mixing state of particles from meat smoking activities in Chongqing revealed that smoked meat contained numerous biomass-burning and ECOC (EC mixed with organic carbon) particles [25].

In Yinchuan, a city located in northwest China, ambient air quality is automatically monitored under the orders of the local government; the annual average PM_2.5 concentration has been generally declining, but the regional PM_2.5 pollution is exacerbated in winter, which cannot be ignored [27]. In recent years, Yinchuan has become a complex pollution area, with this pollution comprising PM_2.5 and gaseous pollutants. Because of local pollutant emissions and long-range transport of pollution from other heavily polluted areas, the physicochemical properties of ambient aerosols in Yinchuan vary greatly.

To understand the characteristics, composition, and source appointment of fine particles in Yinchuan, we used a SPAMS to conduct real-time online source analysis of the main components of PM_2.5 and to analyze the influence of various pollution sources on PM_2.5. By using meteorological data and source analysis results, we analyzed the temporal distribution of various pollution sources with the aim of providing basic data for the source analysis of PM_2.5 and a technical reference for effective formulation of strategies for controlling the atmospheric environment in Yinchuan.

2. Methods

In situ measurements of ambient aerosols were conducted from 18 to 29 December 2020, by using a SPAMS (Hexin Analytical Instrument, Guangzhou, China). Descriptions of the SPAMS and its operational principles can be found in our previous articles [28,29,30]. In brief, the air first passed through a 0.1 mm orifice at 0.08 L/min and was sucked into a series of aerodynamic lenses in the SPAMS, where the aerosol particles were focused and accelerated. Next, the particles entered a vacuum dual-beam diameter measuring system. The number of particles was determined and their size was measured using the different flight times required by two consecutive but separate 532 nm laser beams. When the particles entered the ionization zone, positive and negative ions were generated through laser ionization. Finally, the chemical composition of the particles was detected using a bipolar time-of-flight mass analyzer. Polystyrene latex spheres (PLSs) (0.2–2.0 μm) and standard metal solutions were used to calibrate the particle size and mass before sampling [7,31]. The aerodynamic diameter measurement was calibrated with a curve generated by monodisperse PSL with known aerodynamic diameters. The particle sizes measured by SPAMS are in the range of 200 to 2000 nm. In this study, the ionization energy was set at 0.7 mJ/Pulse.

Yinchuan is located in the temperate continental climate, with sparse rain and snow, intense evaporation, dry climate, and large temperature difference between day and night. It is one of the regions with the largest solar radiation and sunshine hours in China. Yinchuan is divided into mountainous areas and plains. It is higher in the west and south, and lower in the north and east. Yinchuan has the Helan Mountain in the west, north to east.

The sampling site (Figure 1) was located in the Linhe Industrial Park of Ningdong Base (106°32′12″ E, 38°13′4″ N); it is surrounded by Hengshan Road and Chaoyang Road, which have a lot of traffic flow in the morning and evening rush hours. To the north is the Ningxia Baofeng Energy Zone, to the southwest is the Ningdong Aluminum Branch of Qingtongxia Aluminum, and to the southeast is Ningxia Jingneng Ningdong Power Generation.

3. Results and Discussion

3.1. Air Quality and Meteorological Conditions

Meteorological and air quality information were obtained simultaneously at an air quality monitoring site close to our sampling site. The temporal profiles of SO₂, CO, O₃, NO₂, wind direction, and wind speed as well as PM_2.5 and particulate matter of size < 10 μm (PM₁₀) mass concentrations are illustrated in Figure 2. The PM_2.5 mass concentration was discovered to peak three times—on 21, 22, and 27 December—and its maximum was 148 µg/m³ (at 0:00 local time (LT) on 22 December); the average mass concentration of PM_2.5 was 53 μg/m³ during the period. The PM₁₀ concentration exhibited a similar trend to that for PM_2.5. It varied from 14 to 256 µg/m³, with an average of 88.9 µg/m³. Only one major peak was found for SO₂: that at 350 µg/m³ at 21:00 LT on 23 December. The CO concentration exhibited a similar trend to the SO₂ concentration. The opposite trends in O₃ and NO₂ are illustrated in Figure 2. During the monitoring period, the average wind speed was 2.3 m/s, the average temperature was −5.7 °C, the relative humidity (RH) was 19% to 68%, and no precipitation occurred. Therefore, the monitoring period was predominantly characterized by stagnant weather conditions, highly elevated PM_2.5 levels, and low RH.

3.2. Overall Source Apportionment

In this study, 167,000 individual particle mass spectra were analyzed, accounting for approximately 58% of all the particles that were sized in the SPAMS. The mass spectra were classified into several types on the basis of their similarities by using a clustering algorithm called adaptive resonance theory (ART-2a) [10]. Similar to in previous studies [29,30,32,33], the vigilance factor, learning rate, and iterations for ART-2a were set to 0.85, 0.05, and 20, respectively. Finally, fine particles were manually categorized into eight types on the basis of their chemical nature: dust, biomass burning, traffic emissions, coal burning, industrial emissions, secondary inorganic emissions, cooking, and other. The average mass spectral profiles for eight pollution source types are shown in Figure 3.

The number fractions for each group are illustrated in Figure 4. During the sampling period, the atmosphere was mainly affected by coal burning (29.6%) and traffic emissions (23.4%). Industrial emissions and secondary inorganic particles accounted for 16.6% and 14.0% of aerosols, respectively.

According to national ambient air quality standards (NAAQS) for particulate matter (PM) in 2012 (GB3095-2012), air quality is categorized on the basis of PM_2.5 values as follows: 0–35 μg/m³, excellent; 35–75 μg/m³, good; 75–115 μg/m³, slight pollution; 115–150 μg/m³, moderate pollution; and 150–250 μg/m³, severe pollution. In our sample, the air quality was divided into four grades: excellent, good, slight pollution, and moderate pollution. Figure 5 presents the number fractions of the eight particle types under various pollution cases. As the average PM_2.5 mass concentration increased from 28 to 95 μg/m³, the air quality changed from excellent to slight pollution, the number fraction of traffic emissions increased from 21.1% to 25.3%, and the number fraction of secondary inorganic sources increased from 12.2% to 16.3%. As the average PM_2.5 mass concentration increased from 95 to 125 μg/m³, the air quality changed from slight pollution to moderate pollution, the number fraction of coal burning increased from 26.8% to 27.3%, and the number fraction of dust increased from 4.2% to 5.4%. These findings indicated that the increase in PM_2.5 during the sampling period was influenced mainly by the increase in coal burning, traffic emissions, secondary inorganic aerosols, and dust.

3.3. Temporal Variation in Ambient Aerosols

The number concentrations of ambient aerosols detected using the SPAMS changed rapidly and in a complex pattern (Figure 6). The high-resolution temporal variation in the eight types of number concentration indicated several peaks, which may have been due to the typical fresh plume or municipal air pollution. As illustrated in Figure 7, strong correlation (R² = 0.51109) between the number concentration of ambient particles collected using a SPAMS and PM_2.5 mass concentration indicated that the concentration of ambient particles determined using the SPAMS can reflect the atmospheric pollution status of fine particles. To illustrate this point, we defined the periods from 11:00 to 24:00 on 21 December, 09:00 to 17:00 on 22 December, and 10:00 on 27 December to 04:00 on 28 December, as Cases 1, 2, and 3, respectively. From 0:00 on 24 December to 10:00 on 25 December, the PM_2.5 mass concentration remained at a low level (<35 μg/m³), so we defined this period as the “Clean Day”. Figure 8 presents the temporal variation in the number fraction of ambient particles and in the PM_2.5 mass concentration with 1 h resolution. To better illustrate the phenomenon, the three cases (Cases 1, 2, and 3) are colored yellow, whereas the “Clean Day” is colored blue. In this study, we will focus on three pollution cases. The contributions of the eight particle types in the three cases are summarized in Figure 9.

For Cases 1, 2, and 3, both the particle number concentration detected by SPAMS and PM_2.5 mass concentration increased sharply with respect to time (PM_2.5 = 87.75, 92.38, and 88.32 µg/m³, respectively; Figure 7). The PM_2.5 mass concentrations for the three cases, represented by the height of the bar chart, were higher than 75 µg/m³ (Figure 9), indicating slight pollution. The number fraction of traffic emissions on the Clean Day and in Cases 1, 2, and 3 was 21.14%, 24.24%, 26.11%, and 25.95%, respectively. The number fraction of industry emissions on the Clean Day and in Cases 1, 2, and 3 was 13.76%, 19.97%, 16.8%, and 17.93%, respectively. The number fraction of secondary inorganic pollution on the Clean Day and in Cases 1, 2, and 3 was 10.47%, 16.72%, 15.85%, and 17.8%, respectively. All three of these parameters were higher for the pollution cases than on the Clean Day, indicating that slight pollution was caused by traffic emissions, industry emissions, and increased secondary inorganic conversion.

Compared with other cities, higher PM_2.5 number concentrations often have been identified during haze episodes in Shanghai, and the PM_2.5 mainly consisted of traffic emission particles, biomass burning particles and heavy-duty diesel engine emission particles [30]. In Beijing, carbonaceous and K-rich particles were dominant under different conditions and the contribution of industrial metal particles was higher during hazy periods [16]. In our work, higher number concentrations were also discovered more often during pollution episodes in Yinchuan. The pollution cases were mainly caused by the accumulation of fine particles, mainly from traffic emissions, industrial emissions, and increased secondary inorganic conversion.

Taken together, these data suggest that traffic emissions, industry emissions, and increased secondary inorganic conversion contributed to a major fraction of ambient aerosols measured at the sampling site.

4. Conclusions

In this study, we characterized ambient particles during both clean and polluted periods in winter (December 2020) in Yinchuan, China; this was achieved by analyzing the particles’ source apportionment and mixing state by using a SPAMS, which can detect the physical and chemical properties of ambient aerosols at the resolution of a single particle. The mixing state of the particles was drastically different in the different pollution cases, and higher number concentrations were discovered to occur too often during pollution episodes. Overall, source apportionment analysis revealed that the high PM_2.5 concentrations during the sampling period were influenced mainly by considerable coal burning, traffic emissions, secondary inorganic aerosols, and dust. The real-time variation in ambient aerosols measured using a SPAMS indicated that at our sample site, slight pollution was caused by traffic emissions, industry emissions, and increased secondary inorganic conversion.