Compositional Characteristics of Atmospheric Aerosols during a Consecutive High Concentration Episode in Seoul, Korea

: This study focuses on the temporal variation in the compositional characteristics of atmospheric aerosols in Seoul, South Korea, during the consecutive high aerosol concentration episode from 30 December 2013 to 2 January 2014. The temporal variations in the observed physical, optical, and chemical properties show that there were three distinct episodes during the period: haze, mixed haze / Asian dust, and Asian dust episodes. For the haze period, the concentration of secondary inorganic aerosols increased and both secondary inorganic aerosols and calcium species exhibited simultaneously high concentrations during the mixed haze / Asian dust period. The neutralization factors by ammonia in the haze periods were higher as 1.03 than the other periods, meanwhile the neutralization contribution by calcium carbonate was relatively higher as 1.39 during the Asian dust episode. The backward trajectory analysis showed that concentrations of SO 42 − , NO 3 − , and NH 4 + were relatively high when air masses moved over East China. Principal component analysis showed that water-soluble components originated from soil dust / incineration, secondary aerosols / biomass burning, and road dust from the haze aerosol. For the mixed haze / Asian dust episode, the major source of aerosols was estimated to have originated from soil dust, pollutants from fossil fuel combustion, biomass burning, and sea-salt emissions. Furthermore, the main sources of ionic species in the Asian dust aerosols were estimated to be sea-salt / soil dust, secondary aerosols / coal combustion, and road dust. the ratio of we investigated the neutralization factors, and


Introduction
Aerosol particles play important role in the ambient air related to air quality, urban and regional air pollution, visibility, global climate, biogeochemical cycles, human health, etc. [1,2]. Aerosols can offset regional greenhouse warming effects directly by scattering sunlight back into space and indirectly by enhancing cloud albedo, thereby cooling the climate [3]. However, aerosol components such as soot heat the atmosphere by absorbing sunlight and thereby augmenting the warming produced by the greenhouse effect [3,4]. Over polluted regions, long-range transport of mineral dust mixed with anthropogenic species could induce substantial changes in the optical and chemical properties of aerosols [3,5].
inorganic species (Na + , NH 4 + , K + , Mg 2+ , Ca 2+ , SO 4 2− , NO 2 − , NO 3 − , Cl − ). Eluent for the cationic analysis was 20 mM methane sulfonic acid with a flow rate of 0.5 mL min −1 using CG12A/CS12A column and CSRS-ULTRA II suppressor. The analysis of anionic species was performed with an ion chromatograph equipped using AG19/AS19 column (13 mM KOH eluent) and ASRS-ULTRA II suppressor. The standard calibration curve for the analysis of cations and anions showed linearity with R 2 = 0.9999 or better. During the analysis period and depending on the species, the limit of detection (LOD) for IC analyses of the ionic species ranged from 0.05 to 0.1 µg m −3 .

Meteorological Conditions and Mass Concentrations
Surface weather maps for 29 December 2013 to 1 January 2014 are shown in Figure 1. The atmosphere around the Korean peninsula was stable due to the effect of high pressure from 29 December to the morning of 31 December (Figure 1a, Figure 2b). Haze was observed in Seoul on 30 and 31 December. On the afternoon of 31 December the Korean peninsula was affected by a low-pressure system over northeastern China. The 850 hPa wind speed increased, and the wind direction changed from southwest to northwest after January 1, 2014. Asian dust was observed from 02:30 KST 1 January 2014, and the ratios of PM 2.5 to PM 10 were mainly below 0.5 after this time.
into two ion chromatography systems every 20 min to analyze the ionic concentrations for the major water-soluble inorganic species (Na + , NH4 + , K + , Mg 2+ , Ca 2+ , SO4 2− , NO2 − , NO3 − , Cl − ). Eluent for the cationic analysis was 20 mM methane sulfonic acid with a flow rate of 0.5 mL min −1 using CG12A/CS12A column and CSRS-ULTRA II suppressor. The analysis of anionic species was performed with an ion chromatograph equipped using AG19/AS19 column (13 mM KOH eluent) and ASRS-ULTRA II suppressor. The standard calibration curve for the analysis of cations and anions showed linearity with R 2 = 0.9999 or better. During the analysis period and depending on the species, the limit of detection (LOD) for IC analyses of the ionic species ranged from 0.05 to 0.1 μg m −3 . Surface weather maps for 29 December 2013 to 1 January 2014 are shown in Figure 1. The atmosphere around the Korean peninsula was stable due to the effect of high pressure from 29 December to the morning of 31 December (Figure 1a, 2b). Haze was observed in Seoul on 30 and 31 December. On the afternoon of 31 December the Korean peninsula was affected by a low-pressure system over northeastern China. The 850 hPa wind speed increased, and the wind direction changed from southwest to northwest after January 1, 2014. Asian dust was observed from 02:30 KST 1 January 2014, and the ratios of PM2.5 to PM10 were mainly below 0.5 after this time.  December, the ratio of PM2.5 to PM10 exceeded 0.6, thus indicating that fine particles were predominant. The maximum concentration of PM10 during this period was 156 μg m -3 . After 12:00 KST on 31 December, the concentration began to increase again and reached the highest value of 279.7 μg m −3 between 00:00 -03:00 KST 1 January (Figure 2a). Figure 2c, 2d shows that the wind speed increased while humidity decreased from 09:00 KST on 31 December 2013. In addition, the mass concentration both PM10 and PM2.5 were increased, and the concentration ratio of PM2.5/PM10 showed a tendency to increase after 15:00 KST on 31 December Therefore, it was determined that the mass concentration of PM10 was increased relative to PM2.5 and PM1.0 after 09:00 KST on 31 December 2013. Figure 3a shows the means of AOD (500 nm), AE, and SSA (500 nm) retrieved from the sky radiometer data for 31 December and 1 January. There was no sky radiometer observation on 30 December due to clouds. The daily mean AOD was 0.57 on 31 December 2013, which was higher by 80% as compared to that on 1 January. The AE values were 1.03 and 0.48 on 31 December and 1 January, respectively, indicating that coarse particles dominated on 1 January. The average values for SSA were 0.90 and 0.93 for 31 December and 1 January, respectively. Aerosol types can be classified using AE and SSA, which determine the size of aerosols and distinguish absorbing/nonabsorbing aerosols, respectively. Finer particles have a higher AE, and more absorbing aerosols have a lower SSA. Black carbon is a well-known absorbing aerosol, but dust particles also appeared to be solar absorbing aerosols [17]. Aerosol type classification for columnar aerosols was suggested by   Figure 2c,d shows that the wind speed increased while humidity decreased from 09:00 KST on 31 December 2013. In addition, the mass concentration both PM 10 and PM 2.5 were increased, and the concentration ratio of PM 2.5 /PM 10 showed a tendency to increase after 15:00 KST on 31 December Therefore, it was determined that the mass concentration of PM 10 was increased relative to PM 2.5 and PM 1.0 after 09:00 KST on 31 December 2013. Figure 3a shows the means of AOD (500 nm), AE, and SSA (500 nm) retrieved from the sky radiometer data for 31 December and 1 January. There was no sky radiometer observation on 30 December due to clouds. The daily mean AOD was 0.57 on 31 December 2013, which was higher by 80% as compared to that on 1 January. The AE values were 1.03 and 0.48 on 31 December and 1 January, respectively, indicating that coarse particles dominated on 1 January. The average values for SSA were 0.90 and 0.93 for 31 December and 1 January, respectively. Aerosol types can be classified using AE and SSA, which determine the size of aerosols and distinguish absorbing/non-absorbing aerosols, respectively. Finer particles have a higher AE, and more absorbing aerosols have a lower SSA. Black carbon is a well-known absorbing aerosol, but dust particles also appeared to be solar absorbing aerosols [17]. Aerosol type classification for columnar aerosols was suggested by previous studies [18][19][20]. This study adopted one of these aerosol type classification criteria, suggested in Mielonen et al. [20]. Fine-mode aerosols are defined as having an AE greater than 1.2, while coarse-mode Atmosphere 2020, 11, 310 5 of 13 aerosols are defined as having an AE less than 0.8. Aerosols between these threshold values are classified as a "Mixture" mode. The SSA threshold of 0.95 is used to distinguish absorbing from non-absorbing aerosols. Values below 0.95 in the coarse mode indicate "Dust" and values above 0.95 indicate "Uncertain" SSA values below 0.95 in the fine mode indicate "black carbon (BC)," and values above 0.95 indicate "non-absorption (NA)" The dominant aerosol types on two days are shown in Figure 3b. The means of SSA for both days were similar, but SSAs on 1 January were more scattered (0.73-0.95) than on 31 December. All data on 31 December were in the "Mixture" category indicating that both coarse and fine particles exist, whereas "Dust" was most frequently detected on 1 January. 5 previous studies [18][19][20]. This study adopted one of these aerosol type classification criteria, suggested in Mielonen et al. [20]. Fine-mode aerosols are defined as having an AE greater than 1.2, while coarse-mode aerosols are defined as having an AE less than 0.8. Aerosols between these threshold values are classified as a "Mixture" mode. The SSA threshold of 0.95 is used to distinguish absorbing from non-absorbing aerosols. Values below 0.95 in the coarse mode indicate "Dust" and values above 0.95 indicate "Uncertain" SSA values below 0.95 in the fine mode indicate "black carbon (BC)," and values above 0.95 indicate "non-absorption (NA)" The dominant aerosol types on two days are shown in Figure 3b. The means of SSA for both days were similar, but SSAs on 1 January were more scattered (0.73-0.95) than on 31 December. All data on 31 December were in the "Mixture" category indicating that both coarse and fine particles exist, whereas "Dust" was most frequently detected on 1 January.

Ionic Concentrations of Aerosols
The concentrations of the major water-soluble ionic species in PM10 were analyzed semicontinuously using the PILS-IC system to examine the chemical characteristics of the aerosol particles. Time series variations in the concentrations of ionic species (Na + , NH4 + , K + , Mg 2+ , Ca 2+ , SO4 2− , NO2 − , NO3 − , Cl − ) are shown in Figure 4a. Here, the concentrations of calcium carbonate and magnesium carbonates were calculated as the soluble (sol) fractions of Ca and Mg and were added to the calculated CO3 2− concentrations [21]. The CO3 2− was calculated using Equation (1): The concentrations of the ionic species were ordered as NO3 . The concentrations of NO3 − , SO4 2-, Ca 2+ , and NH4 + were 2.28-25.37, 1.71-12.65, 0.03-8.26, and 0.03-11.67 μg m -3 , respectively. In addition, the concentrations of NO3 − , SO4 2− , and NH4 + were relatively high from 02:00 KST on 30 December to 08:00 KST on 31 December. The concentrations of NO3 − and CO3 2− were high from 09:00 KST on 31 December to 08:00 KST on 1 January. The concentrations of Ca 2+ and CO3 2− were high from 09:00 KST on 1 January to 10:00 KST on 2 January. Based on this analysis, the study period was divided into three episodes: the haze period (02:00 KST on 30 December to 08:00 KST on 31 December), the mixed haze/Asian dust period (09:00 KST on 31 December to 08:00 KST on 1 January), and the Asian dust period (09:00 KST on 1 January to 10:00 KST on 2 January). This chemical characteristic was similar to the PM10 mass concentration, the ratio of the PM2.5/PM10, and optical characteristics. Using these results, we investigated the chemical compositions, neutralization factors, and pollution sources of the three episodes.

Ionic Concentrations of Aerosols
The concentrations of the major water-soluble ionic species in PM 10 were analyzed semi-continuously using the PILS-IC system to examine the chemical characteristics of the aerosol particles. Time series variations in the concentrations of ionic species (Na + , NH 4 + , K + , Mg 2+ , Ca 2+ , Figure 4a. Here, the concentrations of calcium carbonate and magnesium carbonates were calculated as the soluble (sol) fractions of Ca and Mg and were added to the calculated CO 3 2− concentrations [21]. The CO 3 2− was calculated using Equation (1): The concentrations of the ionic species were ordered as NO 3 from 09:00 KST on 1 January to 10:00 KST on 2 January. Based on this analysis, the study period was divided into three episodes: the haze period (02:00 KST on 30 December to 08:00 KST on 31 December), the mixed haze/Asian dust period (09:00 KST on 31 December to 08:00 KST on 1 January), and the Asian dust period (09:00 KST on 1 January to 10:00 KST on 2 January). This chemical characteristic was similar to the PM 10 mass concentration, the ratio of the PM 2.5 /PM 10 , and optical characteristics. Using these results, we investigated the chemical compositions, neutralization factors, and pollution sources of the three episodes. The composition ratios of the ionic species in PM 10 during the study period are shown in Figure 4b. The composition ratios of secondary inorganic aerosols (NH 4 + , NO 3 − , and SO 4 2− ), soil (Ca 2+ and CO 3 2− ), sea-salt (Na + , Cl − , and Mg 2+ ), and other species in PM 10 during the study periods were 55.7%, 28.8%, 11.5%, and 4.10%, respectively. Their species ranges were 28.3-85.8%, 0.3-53.5%, 7.9-16.3%, and Atmosphere 2020, 11, 310 6 of 13 0.9-12.6%, respectively. The composition ratios of the secondary inorganic aerosol species were 80.2%, 57.0%, and 36.5%, and those of the soil species were 7.9%, 28.1%, and 44.6% of the total ionic species for the haze, mixed haze/Asian dust, and Asian dust episodes, respectively. Sea-salt species accounted for 9.9%, 10.8%, and 13.2%, while the others accounted for 1.9%, 4.2%, and 5.6% of the haze, mixed haze/Asian dust, and Asian dust episodes, respectively. The composition ratios of secondary inorganic aerosol species were higher during the haze episode, whereas those of the soil and other species were higher during the Asian dust episode. In particular, CO 3 2− was the most dominant anion, and Ca 2+ and Mg 2+ were major cations during the mixed haze/Asian dust and Asian dust episodes. 6 The composition ratios of the ionic species in PM10 during the study period are shown in Figure  4b. The composition ratios of secondary inorganic aerosols (NH4 + , NO3 − , and SO4 2− ), soil (Ca 2+ and CO3 2− ), sea-salt (Na + , Cl − , and Mg 2+ ), and other species in PM10 during the study periods were 55.7%, 28.8%, 11.5%, and 4.10%, respectively. Their species ranges were 28.3-85.8%, 0.3-53.5%, 7.9-16.3%, and 0.9-12.6%, respectively. The composition ratios of the secondary inorganic aerosol species were 80.2%, 57.0%, and 36.5%, and those of the soil species were 7.9%, 28.1%, and 44.6% of the total ionic species for the haze, mixed haze/Asian dust, and Asian dust episodes, respectively. Sea-salt species accounted for 9.9%, 10.8%, and 13.2%, while the others accounted for 1.9%, 4.2%, and 5.6% of the haze, mixed haze/Asian dust, and Asian dust episodes, respectively. The composition ratios of secondary inorganic aerosol species were higher during the haze episode, whereas those of the soil and other species were higher during the Asian dust episode. In particular, CO3 2− was the most dominant anion, and Ca 2+ and Mg 2+ were major cations during the mixed haze/Asian dust and Asian dust episodes.    the Asian dust episode, the air masses passed over Mongolia and Inner Mongolia, the major Asian Dust source regions, and Ca 2+ mainly originating from soil dust showed higher concentrations.  Figure 5 shows the 72 h air mass backward trajectories at a height of 500 m colored according to the concentration of the ionic component pollutants. The backward trajectories were similar for the haze and mixed haze/Asian dust episodes, while air masses for the Asian dust case seemed to move faster than those for the other cases. NO3 − , SO4 2− , NH4 + , i.e., the anthropogenic pollutants, showed elevated concentrations when the trajectories passed over the industrialized area in East China. During the Asian dust episode, the air masses passed over Mongolia and Inner Mongolia, the major Asian Dust source regions, and Ca 2+ mainly originating from soil dust showed higher concentrations.

Ammonium Salt Formation and Neutralization
The secondary inorganic aerosol species, including SO 4 2− , NO 3 − , and NH 4 + , often exist in various forms, such as (NH 4 ) 2 SO 4 , NH 4 HSO 4 , and NH 4 NO 3 [22]. The ratios for (NH 4 ) 2 SO 4 and NH 4 NO 3 that existed mainly in the ambient air could be calculated using Equations (2) and (3) Atmosphere 2020, 11, 310 8 of 13 Figure 6 shows the concentrations of (NH 4 ) 2 SO 4 and NH 4 NO 3 estimated from the ionic analysis and the neutralization factors caused by the reaction between inorganic acids and bases, such as ammonia and calcium carbonate in PM 10 aerosols. The estimated concentrations of (NH 4 ) 2 SO 4 in Seoul were 12.6, 10.6, and 4.7 µg m -3 , and those of NH 4 NO 3 were 24.7, 16.6, and 1.5 µg m −3 for the haze, mixed haze/Asian dust, and Asian dust episodes, respectively, indicating relatively high concentrations in the haze period but lower concentrations in the Asian dust period. The estimated concentrations of (NH 4 ) 2 SO 4 in PM 10 at the Gosan, South Korea site were 22.9, 10.5, and 8.0 µg m −3 , and those of NH 4 NO 3 were 24.3, 0.1, and 0.09 µg m -3 for the haze, mixed haze/Asian dust, and Asian dust, respectively [24].  Table 1 shows the correlation coefficients between the water-soluble ions in PM10 aerosols. In the results, NH4 + was strongly correlated with SO4 2− in all haze (r = 0.880), mixed haze/Asian dust (r = 0.848), and Asian dust (r = 0.911) episodes. Strong correlations were also found between NH4 + and NO3 − in haze (r = 0.783), mixed haze/Asian dust (r = 0.971), and Asian dust (r = 0.969). Ca 2+ was strongly correlated with CO3 2− for all three episodes. These results suggest that (NH4)2SO4, NH4HSO4, NH4NO3, and CaCO3 were the major chemical fractions in all three episodes.  Generally, the acid compounds in the air are neutralized by basic ammonia and carbonate compounds through the reactions between inorganic acids and base species, which are mostly ammonia or soil components, such as calcium carbonate and magnesium carbonate [25,26]. The neutralization factors are presented in Figure 6. The neutralization factors by ammonia were 1.03, 0.81, and 0.41 for the haze, mixed haze/Asian dust, and Asian dust episodes, respectively. Meanwhile, the neutralization factors by calcium carbonate were 0.16, 0.77, and 1.39, for haze, mixed haze/Asian dust, and Asian dust episodes, respectively.

Emission Sources of Ambient Aerosols
It has been shown that the neutralization by ammonia was noticeably higher during the haze episode but decreased as the Asian dust episode began to occur. Conversely, the neutralization by calcium carbonate increased during the Asian dust episode. Interestingly, the neutralization by ammonia and calcium carbonate were reversed during the haze and mixed haze/Asian dust episodes. Therefore, it is thought that the concentrations of ammonium sulfate and ammonium nitrate would increase during the haze episode, while the soil-originated CaCO 3 would increase during the Asian dust episode. Table 1 shows the correlation coefficients between the water-soluble ions in PM 10 aerosols. In the results, NH 4 + was strongly correlated with SO 4 2− in all haze (r = 0.880), mixed haze/Asian dust Principal component analysis (PCA) was applied to estimate the major emission sources of aerosol components [27]. PCA is the most common multivariate statistical method applied in environmental studies. The Statistical Package for the Social Sciences (SPSS) software (version 18.0) was used for the multivariate analysis. We applied PCA for the water-soluble ions analyzed during the episodes for source attribution. The varimax rotated factors of the loading matrix are presented in Table 2. In the haze episode, the first three principal components with eigenvalues greater than one were extracted; they accounted for 89.0% of the total variance. The first principal component (PC1), which explained 36.9% of the total variance with an eigenvalue of 3.7 was mostly dependent on Ca 2+ , Cl − , SO 4 2− , and CO 3 2− . The Ca 2+ and CO 3 2− species are considered to be soil dust sources [28], while SO 4 2− and Cl − are typically produced by waste incinerators [29]. This indicates that PC1 is associated with soil dust and emissions from waste incinerators. The second principal component, PC2 accounted for 30.5% of the total variance and showed high loadings for NH 4 + , SO 4 2− , NO 3 − , and K + . Because K + is regarded as a tracer for biomass burning [30], PC2 can be associated with secondary aerosols mixed with biomass-burning aerosols [31]. SO 4 2− showed a significant loading on both PC1 and PC2 indicating composite sources. Finally, PC3 explained 21.6% of the total variance, which was dominated by Na + and Mg 2+ . A strong correlation between Na + and Mg 2+ indicates that Na + and Mg 2+ had the same origin Table 1a. Thus, the major sources of PC3 were natural minerals characteristics (Figure 7a) [32].

Emission Sources of Ambient Aerosols
In the mixed haze/Asian dust episode, four principal components were identified and accounted for 94.2% of the total variance. PC1, accounting for 36.8% of the total variance was heavily loaded by Na + , Mg 2+ , Ca 2+ , and CO 3 2− , with loadings of 0.702, 0.866, 0.985, and 0.987, respectively. Mg 2+ , Ca 2+ , and CO 3 2− are regarded as tracers for soil dust sources from natural and a variety of crustal minerals [33]. The high correlation between Na + and Mg 2+ (r = 0.777) indicates that Na + and Mg 2+ share the same origin Table 1b. PC2 showed high loadings of the three secondary ions as well as Clindicating that the secondary pollutants originated from fossil fuel combustion and other anthropogenic sources [31]. The loading of K + on PC3 (16.6% of the total variance) mainly originated from biomass burning. PC4, which explained another 8.0% of the total variance grouped Na + and Cl − and appears to represent the sea-salt source (Figure 7b) [34]. Three principal components that explained 96.2% of the total variance were identified in the Asian dust episode. PC1, with 46.7% of the total variance, was loaded with Na + , Ca 2+ , Cl − , and CO 3  tracers of secondary origins, and Cland K + are emitted mainly from coal combustion and biomass burning [33]. PC3 accounted for 9.4% of the total variance and was strongly loaded with K + and Mg 2+ . These are regarded as tracers for crustal and construction dusts (Figure 7c).

Conclusions
Through optical analyses using the sky radiometer, aerosol types were classified as "Mixture" and "Dust" particles for 31 December 2013 and 1 January 2014, respectively. The secondary inorganic aerosol species (NH 4 + , NO 3 − , and SO 4 2− ) during the haze episode had relatively higher concentrations (80.2%) as compared to the other episodes. In addition, the composition ratios of soil (Ca 2+ and CO 3 2− ) were 3.6 and 5.6 times increased in both the mixed haze/Asian dust and Asian dust episodes, respectively. The NH 4 NO 3 content was higher than (NH 4 ) 2 SO 4 during the haze, but it was relatively low during the Asian dust episode. Comparing the high aerosol concentration episodes, the neutralization by ammonia was noticeably higher during the haze episode, and calcium carbonate was more prominent during the Asian dust episode. The backward trajectories analysis indicated that the concentrations of the secondary inorganic aerosol species (NH 4 + , NO 3 − , and SO 4 2− ) were higher when the air masses moved over the industrialized area in East China. However, when the air mass was transported over Mongolia and Inner Mongolia during the Asian dust episode, the concentration of Ca 2+ species was relatively higher. The neutralization factors of ammonia were 1.03, 0.81, 0.41, respectively, for the haze, mixed haze/Asian dust, and Asian dust episodes. Moreover, the neutralization factors of calcium carbonate were 0.16, 0.77, 1.39, respectively. Based on PCA, the haze episode compositions might be influenced by a mixture of soil dust and emissions from waste incinerators, a mixture of secondary aerosols and biomass burning, and crustal mineral dust. The compositions of the mixed haze/Asian dust period were most affected by a mixture of soil and crustal dusts, followed by secondary pollutants from fuel combustion, biomass burning, and sea-salt emissions. Moreover, a mixture of sea-salts and soil dust, a mixture of secondary aerosols with biomass burning and coal combustion, and mineral dust were the potential sources in the Asian dust episode.