Environ Eng Res > Volume 24(1); 2019 > Article
Park and Yu: Absorption properties and size distribution of aerosol particles during the fall season at an urban site of Gwangju, Korea

### Abstract

To investigate the influence of pollution events on the chemical composition and formation processes of aerosol particles, 24-h integrated size-segregated particulate matter (PM) was collected during the fall season at an urban site of Gwangju, Korea and was used to determine the concentrations of mass, water-soluble organic carbon (WSOC) and ionic species. Furthermore, black carbon (BC) concentrations were observed with an aethalometer. The entire sampling period was classified into four periods, i.e., typical, pollution event I, pollution event II, and an Asian dust event. Stable meteorological conditions (e.g., low wind speed, high surface pressure, and high relative humidity) observed during the two pollution events led to accumulation of aerosol particles and increased formation of secondary organic and inorganic aerosol species, thus causing PM2.5 increase. Furthermore, these stable conditions resulted in the predominant condensation or droplet mode size distributions of PM, WSOC, NO3, and SO42−. However, difference in the accumulation mode size distributions of secondary water-soluble species between pollution events I and II could be attributed to the difference in transport pathways of air masses from high-pollution regions and the formation processes for the secondary chemical species. The average absorption Ångström exponent (AAE370-950) for 370-950 nm wavelengths > 1.0 indicates that the BC particles from traffic emissions were likely mixed with light absorbing brown carbon (BrC) from biomass burning (BB) emissions. It was found that light absorption by BrC in the near UV range was affected by both secondary organic aerosol and BB emissions. Overall, the pollution events observed during fall at the study site can be due to the synergy of unfavorable meteorological conditions, enhanced secondary formation, local emissions, and long-range transportation of air masses from upwind polluted areas.

### 1. Introduction

In recent years, the Korean peninsula has suffered from severe air pollution every fall due to stable weather conditions and regional transport of natural and anthropogenic sources from China [13]. Stable air conditions such as low wind speed, high surface pressure, and low boundary layer lead to the formation of a low-hanging layer and the poor dispersion of air pollutants. Ambient fine particles (PM2.5) consist of primary particles from incomplete combustion of fossil and biomass fuels, and secondary particles formed through homogeneous and heterogeneous oxidation processes of SO2, NO2, NH3, and volatile organic compounds (VOCs) from anthropogenic and natural emission sources [46]. Haze phenomena, mainly caused by PM2.5, have attracted much attention due to their adverse effect on air quality, public health, and visibility [711]. These studies of haze aimed to investigate source identification, chemical characterization, and formation processes of PM2.5 during haze pollution [5, 6, 1215].
Generally, it was found that the secondary formation of aerosol components in the atmosphere contributes significantly to the increase in PM2.5 concentration [10, 16, 17]. On a seasonal basis, photochemical reactions during summer are important factors accelerating the formation of secondary SO42− particles, which are responsible for summer haze formation [10, 1821]. Meanwhile, air stagnation and high relative humidity (RH) during winter contribute to the accumulation of aerosol particles from anthropogenic sources and enhanced secondary aerosol formation, giving rise to winter haze formation [6, 8, 10, 13, 22, 23]. Moreover, regional transport could play an important role in the PM2.5 increase during a winter haze event [810, 16, 24, 25]. As mentioned above, chemical characteristics of PM2.5 during haze pollution depend on emissions and transport of air pollutants, formation of secondary aerosols, and meteorological conditions. Thus, understanding the characteristics and mechanisms of PM2.5 pollution occurred at a receptor site is very crucial to establish effective control of the pollution level that depends on the seasonal variation of PM2.5 formation at the site and the surrounding areas.
Atmospheric aerosol particles affect climate radiative forcing of the Earth by directly scattering and absorbing solar radiation [26]. There are two types of carbonaceous aerosols interacting with the radiation. One is the organic carbon (OC) and the other elemental carbon (EC) or black carbon (BC) [19, 20, 27, 28]. BC, which comes from incomplete combustion of fossil and biomass fuels, absorbs light across all ranges from ultraviolet (UV) to infrared (IR) with a weak spectral dependence [29, 30]. While OC has typically been known as light-scattering substances, but a certain type of OC, which is referred to as brown carbon (BrC), absorbs strongly solar radiation in the near UV wavelength range with strong spectral dependence [30, 31]. The previous studies have found that BrC is typically emitted from primary sources such as biomass burning, coal combustion, and vehicles, and produced through the formation of secondary organic aerosol [27, 3136].
The size distribution of aerosol particles could provide important clues to trace emission sources of ambient particles and the formation pathways of secondary components [37]. Size distributions of water-soluble organic carbon (WSOC) and inorganic components (NO3 and SO42−) under typical ambient conditions at an urban site showed dominant droplet modes peaking at 0.55 μm [2, 38, 39]. Meanwhile particle size distributions at the urban site during the haze event with air stagnation, high RH, and regional transportation, in the fall season, exhibited dominant droplet modes at a particle size of 1.0 μm [3]. Yu et al. [2] also conducted a source apportionment of size-resolved WSOC for one year at an urban site. In their study, condensation mode WSOC was found to be strongly associated with secondary organic aerosols and biomass burning (BB) emissions, while heterogeneous reactions and long-range transports were found to be important contributors to the droplet mode WSOC formation. Results from previous studies have indicated that the size distributions of major chemical components in ambient aerosol particles depend on season, location, concentration of precursor gases, meteorological conditions, and long-range transport [2, 3, 3842]. Therefore, to investigate the cause of haze formation during fall, size distributions of water-soluble chemical species in ambient aerosol particles need to be further explored.
In this study, 24-h size-segregated aerosol particles were collected during the fall season of 2017, at the urban site of Gwangju in Korea, to investigate the difference in characteristics of size-resolved water-soluble chemical species between non-pollution, pollution, and Asian dust (AD) periods. Sources and formation processes of size-resolved WSOC are also discussed. Moreover, the difference in the absorption properties of aerosol particles was examined among non-pollution, pollution, and AD periods. Further, the size-resolved primary and secondary sources contributing to light absorption by aerosol particles in the near UV range were identified.

### 2.1. Size-resolved Aerosol Particle Measurements

Size-segregated particle samples were collected daily between October 19 and November 9, 2017, at the urban site of Gwangju in Korea. The sampling would begin at around 09:00 AM and last approximately 24-h. The sampling site was on the rooftop of a three-story building at Chonnam National University, approximately 150 m away from a four-lane traffic road. The sampling location is described in detail in our previous publications [25, 39]. Collection of size-segregated aerosol samples was conducted using a 10-stage Micro-Orifice Uniform Deposit Impactor (MOUDI, MSP 110; MSP Corp., MN), having cut-off diameters between 0.055 and 18.0 μm (0.055, 0.095, 0.17, 0.32, 0.55, 1.00, 1.8, 3.1, 6.2, 9.9 and 18.0 μm) at a flow rate of 30 L/min. Prebaked and pre-weighed 47-mm Al foil substrates were used to collect aerosol particle samples. Further, a 47-mm Teflon substrate was used as back-up filter. A total of 21 MOUDI sets were collected during the sampling period and used to quantify the concentrations of mass, WSOC, and water-soluble ionic species. Field blanks were also prepared, and their background concentrations were applied to obtain the real concentrations of WSOC and ionic species in ambient aerosol particles.

### 2.2. Measurements of Aerosol Light Absorption and Estimation of Brown Carbon Absorption

BC concentrations were observed at a time resolution of 1 min using a dual-spot aethalometer (AE33, Aerosol d.o.o., Slovenia) equipped with a PM2.5 impactor operating at a flow rate of 5.0 L/min. The aethalometer instrument measures the attenuation of a light beam at seven wavelengths from 370 to 950 nm by the aerosols loaded on a filter to determine the wavelength-dependent absorption coefficient. Then, the instrument calculates BC concentration by dividing the absorption coefficient by the mass-specific absorption efficiency for each of seven wavelengths. The mass-specific absorption cross-sections used in the BC calculations are 18.47, 14.54, 13.14, 11.58, 10.35, 7.77, and 7.19 m2/g for 370, 470, 520, 590, 660, 880, and 950 nm, respectively, which were provided by the manufacturer. The principle on which the BC concentration measurement is based has been documented in detail by Drinovec et al. [43]. BC concentration is typically reported at 880 nm because organic aerosols as well as BC absorb light at lower wavelengths [30]. Absorption coefficients reported from wavelengths lower than 880 nm are usually used to identify the absorption of organic aerosols.
BC aerosol strongly absorbs light in the UV to IR wavelength range, but exhibits a weak absorption dependence with wavelength [29]; while a certain type of organic aerosol absorbs strongly radiation in the near UV range but exhibits weaker absorption in the visible wavelengths [30]. The light-absorbing organic carbon components are referred to as BrC. Light absorption coefficients of BrC (bBrC,λ) at wavelengths less than the IR were derived by subtracting the absorption coefficient by BC (bBC,λ) from the total light absorption coefficient (babs,λ). In this relationship, BC absorption was estimated based on the assumption that the total aerosol light absorption at 880 and 950 nm is due to BC and the absorption Ångström exponent (AAE) value for BC is 1.0 [29, 30]. The bBC,λ at wavelengths < 880 nm was calculated by averaging the light absorption at 880 and 950 nm and extrapolating with an AAE value of 1.0 for BC [44] (1):
##### (1)
$bBC,γ=(babs,880+babs,9502)×(γ(880+9502))-AAEBC$
Following the approach of previous studies [29, 30], the wavelength dependence of the light absorption coefficient (babs,λ) follows a power law relationship: babs,λ = AAE, where a is the regression constant. Therefore, the aerosol AAE values in the wavelength range of 370-950 nm and 370-660 nm were calculated by performing a linear fitting regression of ln(babs,λ) against ln(λ).

### 2.3. Determination of Water-soluble Organic and Inorganic Species

The size-segregated aerosol samples collected on MOUDI were extracted with 30 mL of ultrapure distilled de-ionized water for 60 min of ultra-sonication at room temperature. Then the water extracts were filtered with a disposable syringe filter (0.45 μm pore size, Millipore) to analyze the WSOC and ionic species (Na+, NH4+, K+, Ca2+, Mg2+, Cl, NO3, SO42−, and oxalate) using a total organic carbon analyzer (TOC, Sievers 5310C, USA) and an ion chromatography (IC) system (Metrohm 861), respectively. In the TOC analyzer, the organic compounds soluble in water are oxidized to form carbon dioxide (CO2) using UV radiation and a chemical oxidizing agent (ammonium persulfate, (NH4)2S2O8). The CO2 concentration is measured using a sensitive membrane-based conductivity detection technique. The IC system has an anion IC (a Metrohm Metrosep A Supp-5, 4×150 mm, anion column) and a cation IC (a Metrohm Metrosep C4, 4×150 mm, cation column). For the anion analysis, a 3.2 mM sodium carbonate/1.0 mM sodium bicarbonate solution was used as the eluent. The cation eluent was 1.7 mM nitric acid/0.7 mM dipicolinic acid. Eluant flow rate for the anion and cation was 0.7 and 1.0 mL/min, respectively. The method detection limit (MDL) of ionic species was calculated as the average blank value plus three times the standard deviation of the field blanks. The MDL for Na+, NH4+, K+, Ca2+, Mg2+, Cl, NO3, SO42−, and oxalate was 0.07, 0.05, 0.11, 0.07, 0.10, 0.09, 0.10, 0.05, 0.05 μg/m3, respectively. Their sample precision was 2.6, 4.1, 4.4, 1.1, 1.3, 4.2, 5.5, 5.2, and 6.7%, respectively. Detail descriptions for analyzing the WSOC and ionic species can be found in our previous publications [27, 35].

### 3.1. General Features of PM10 and PM2.5, and Classification of Study Period

Fig. 1 shows temporal variations of hourly and 24-h averaged PM10, PM2.5, and PM2.5/PM10 as well as O3, NO2, CO, and SO2. Hourly concentrations of PM10, PM2.5, O3, NO2, SO2, and CO were observed at an urban air monitoring station at Duam-dong, which is approximately 2.0 km from the sampling site. The PM10 and PM2.5 were observed by using two beta attenuation mass monitors (BAM-1020), respectively. The gaseous species O3, NO2, SO2, and CO concentrations were measured by using a pulsed ultraviolet photometric ozone analyzer (Thermo 49i), chemiluminescence-based analyzer (Thermo 42CTL), a pulsed ultraviolet fluorescence SO2 analyzer (Thermo 43i), and a non-dispersive infrared CO analyzer (Thermo 48i), respectively. 24-h averaged PM10, PM2.5, and PM2.5/PM10 were in the range of 24–70 μg/m3, 14–62 μg/m3, and 0.36–0.93, respectively. The Korean Meteorological Agency reported that an AD event during the study period occurred on 8 November, 2017 (http://www.weather.go.kr/weather/asiandust). Hence, we classified the sampling period for November 8 as the AD event. During the AD event, the hourly maximum PM10 and PM2.5 concentrations reached 193 and 76 μg/m3 at 14:00 and 11:00 on November 08, respectively. In addition, the PM2.5/PM10 was observed to be the lowest at 17:00 h with a value of 0.30. During that day, the 24-h averaged PM2.5 concentration was 31 μg/m3. Based on the 24-h PM2.5 concentration and the AD event, the entire study period was classified into four periods of which, PM2.5 pollution events were identified as those periods when daily PM2.5 exceeded the 24-h averaged Korean National Ambient Air Quality Standard of 35.0 μg/m3. Daily PM2.5 values were calculated based on the measurement interval of MOUDI (09:00–09:00). Using the results of this calculation, a total of seven sampling days (October 25–27 and November 01–04) were identified as PM2.5 pollution days. The 24-h averaged PM2.5 masses on October 25, 26, 27; November 01, 02, 03, and 04 were 46, 55, 50, 40, 54, 56, and 62 μg/m3, respectively. Hence, sampling periods for October 25–27, November 01–04, and November 08 were designated as pollution event I, pollution event II, and AD event. Those periods not falling within these three designated events were classified as typical period. Table 1 summarizes PM10, PM2.5, criteria gases (O3, NO2, CO, and SO2), aerosol light absorption, and concentrations of water-soluble chemical species from MOUDI measurements for four periods. In addition, meteorological data (wind speed, relative humidity, surface pressure), which were observed at Gwangju regional office of meteorology (http://www.weather.go.kr/weather/observation/aws_table_popup.jsp), are provided in Table 1. Fig. 2 shows temporal profiles of hourly and 24-h averaged meteorological parameters (ambient temperature, relative humidity, wind speed, and sea level pressure).
For typical, pollution event I, pollution event II, and the AD event, average PM2.5 concentrations were 25.6, 50.1, 53.0, and 30.8 μg/m3, respectively. Respective PM2.5/PM10 ratios were 0.62, 0.76, 0.86, and 0.54. As shown in Fig. 2 and Table 1, higher PM2.5 and PM2.5/PM10 during the two pollution events were attributed to low wind speeds (0.6–0.9 m/s) and high RH (74–76%). These meteorological conditions could accelerate the accumulation of primary particles from traffic emissions around the site and the production of secondary aerosol particles through VOC, SO2, NO2, and NH3 oxidation processes, leading to an increase in PM2.5. Results from previous studies have indicated that high RH conditions could lead to enhanced formation of secondary inorganic species through aqueous-phase oxidations of SO2 and NO2 [810, 23, 45, 46]. Higher concentrations of CO from combustion emission sources were observed during the two pollution events than during the other two periods, suggesting greater contributions from combustion sources to aerosol particles during pollution events I and II. As shown in Fig. 1, the average CO concentrations during typical, pollution event I, pollution event II, and AD event were 0.42 (0.20–0.90), 0.56 (0.30–0.80), 0.53 (0.30–0.80), and 0.38 (0.20–0.80) ppm, respectively. As described above and also in Table 1, there was little difference in meteorological variables such as wind speed and RH, and concentrations of O3, NO2, and CO between pollution events I and II. However, the higher PM2.5/PM10 observed during pollution event II (0.86) could be attributed to a greater influx of long-range transported fine aerosols from polluted regions in China, as suggested by the analysis of transport pathways of air masses and MODIS images below.
To investigate the difference in PM2.5/PM10, light absorption by aerosol particles, and its chemical composition between the pollution events I and II, air mass backward trajectories were computed. The transport pathways of air masses were calculated using the Hybrid Single-Particle Lagrangian Integrated Trajectories (HYSPLIT 5.0) model [47]. Three-day backward trajectories with model vertical velocity were calculated for 0900 UTC on each day at three altitudes (500, 1,000, and 1,500 m) above ground level (AGL). Fig. 3 shows transport pathways of air masses arriving at the site during pollution events I (Oct 25, 26, and 27) and II (Nov. 01, 02, 03, 04), and MODIS images of days before these two pollution events began. During pollution event I, haze occurred over the mid-eastern Chinese regions (MODIS images, http://nmsc.kma.go.kr/html/homepage/ko/satellite/searchSatelliteImageN.do?data_type=1012), but most of the air masses originated from northern China and north Korea, indicating that the air masses were likely dominated by local pollution. Similarly, during pollution event II, haze layers occurred over the mid-eastern regions of China (MODIS images). A small amount of the haze passed over the western sea of the Korean peninsula and reached the study site, indicating that the air masses arriving at the site were influenced by a mixture of the haze transported from mid-eastern China as well as the local pollution. In summary, local combustion emissions (e.g., traffic emissions) and enhanced formation of secondary ionic species under stable air conditions (see Table 1) were important contributors to high PM2.5 during pollution event I, while the long range transportation of polluted aerosols from mid-eastern Chinese regions, local emissions, and further processing of air pollutants under stable air conditions could be possible factors contributing to high PM2.5 during pollution event II.

### 3.2. BC Concentration and Light Absorption by BC and Brown Carbon

Temporal profiles of hourly BC880, aerosol absorption coefficients (babs,370 and babs,880), estimated absorption coefficients of BC and BrC (BC babs,370 and BrC babs,370), and BrC contribution to total aerosol absorption at 370 nm (BrC370) are shown in Fig. 4. Subscripts 370 and 880 denote wavelengths of 370 nm and 880 nm, respectively. The 24-h averaged BC370 ranged from 0.8 to 5.6 μg/m3, which were higher than the average BC880. In addition, the average babs,370 ranged from 15.0 to 103.7 Mm−1, which is approximately 2.9–4.7 times the average babs,880 with a range of 5.1–32.3 Mm−1. The large difference in the aerosol absorption coefficients at wavelengths of 370 and 880 nm suggests the presence of light absorbing organic aerosols, as well as the BC aerosols at the site. As shown in Table 1, BC370, BC880, babs,370, and babs,880 were higher during the two pollution events than during the other two periods, with the highest values during pollution event I. However, babs,370/babs,880 was highest during pollution event II at 3.9, suggesting the greatest contribution of light absorbing BrC to total aerosol absorption during pollution event II, which is discussed in detail in the following section. Over the study period, the contribution of BC babs,370 to total aerosol absorption at 370 nm ranged from 26 Mm−1 during the AD event to 65 Mm−1 during pollution event I. On the other hand, the contribution of BrC babs,370 ranged from 11 Mm−1 during the AD event to 30 Mm−1 during pollution event II, accounting for 26–40% of total aerosol absorption at 370 nm. From the BC880, babs,880, and BC babs,370, it was confirmed that the contribution from local traffic, which emits BC particles, was dominant during pollution event I, while contribution by BrC aerosols was dominant during pollution event II. As shown in Table 1, light absorption at 370 and 880 nm during the AD event was observed to be lower than that during typical period. The low absorption was due to low concentration of light-absorbing aerosol particles from anthropogenic emission sources and low concentration of mineral dust particles. Generally, the mineral dust particles have an absorption cross-section smaller than that of BC particles by a factor of from 100 to 1,000. Thus, if the amount of dust particle is 100 to 1,000 times greater than that of BC, a comparable light absorption may be produced [48]. However, the hourly highest concentration of PM2.5 observed during the AD event was 76 μg/m3 at the most. Moreover, the aethalometer instrument measures light absorption by PM2.5. Thus, the contribution of the mineral dust particles to the total aerosol absorption could be low.
The spectral dependence of aerosol and estimated BrC absorption for four sampling periods are shown in Fig. 5. In Fig. 5, the wavelength dependence of aerosol absorption follows a power-law relationship (babs = a×λ−AAE), with decrease in the aerosol absorption with wavelength. Average aerosol AAE370-950 values for typical, pollution event I, pollution event II, and the AD event were 1.3, 1.3, 1.5, and 1.4, respectively. AAE370-950 values greater than 1.0 could indicate that the aerosol particles collected at the site was due to biomass burning (BB) emissions and the BC aerosols from traffic emissions [3, 27, 29, 30]. Moreover, little difference in estimated BrC AAE370-660 values was observed among the four periods, ranging from 4.2 (AD event) to 4.6 (pollution event II), suggesting that organic aerosols contributed to total aerosol absorption throughout the study period. In general, the AAE values are close to 1.0 for BC aerosols from motor vehicle emissions, while organic aerosols from BB emissions have an AAE of approximately 2.0 [29, 30, 49, 50]. Furthermore, AAE values are larger than 2.0 for dust aerosols [50]. The AAE of 2.27 was obtained from the ACE-Asia program during which collected particles had both urban pollution and desert dust components [51]. Sun et al. [52] suggested that the absorption coefficients can be approximated with an AAE of 6 for the humic-like substances, which are a type of BrC aerosols, and an AAE of 4.3 for the combustion substances. Their results indicate that mixtures of these organic materials and BC have AAE values greater than 1 but less than 4–6. Although the AAE can be very high (4–6) for individual organic compounds, these compounds in the atmosphere are usually mixed with strongly absorbing BC particles. Therefore, the average AAE usually lies between 1 (for BC) and 4–6 (for BrC). Based on previous results, AAE greater than 1.0 for the four different periods was associated with the presence of mixtures of BC and light absorbing organic aerosols.
As shown in Fig. 6, PM2.5 ranged from 18 to 32 μg/m3 between 18:00 on Nov 03 and 01:00 on Nov 04 (“A” period) but thereafter maintained relatively high values (51–105 μg/m3) between 02:00 AM and 12:00 PM on Nov. 04 (“B” period). Enhanced concentrations of PM2.5 were observed early in the morning, and maximum PM2.5 concentration was observed between 05:00–06:00 AM on Nov. 04. To investigate the difference in PM2.5 between the two distinct periods, we examined the meteorological parameters (wind speed and RH), concentrations of criteria gases, air mass transport pathways, BC concentrations, and AAE values calculated from absorptions by aerosol and BrC. However meteorological parameters such as wind speed and RH controlling PM concentrations did not significantly differ between the two time intervals. Moreover, no significant difference in O3, NO2, CO, SO2, and transport pathways of air masses was found between the two time intervals. Wind speed and RH were 0.7–1.9 m/s and 52–65% during “A” period, and 1.1–2.4 m/s and 43–56% during “B” period. BC880 and BC370 concentrations were 0.5–0.8 and 0.9–1.7 μg/m3 during “A” period, respectively, but they increased to 1.0–2.1 and 3.4–6.0 μg/m3 during “B” period, respectively, with a rapid increase in BC370. This result suggests that the PM2.5 observed during “B” period was significantly influenced by light absorbing organic aerosols at the site, the presence of which was also demonstrated by enhanced aerosol and BrC absorption coefficients as shown in Fig. 7. The aerosol AAE370-950 and BrC babs,370 were 1.6 and 10.8 Mm−1 during the “A” period, but increased to 2.1 and 62.6 Mm−1 during the “B” period.

### 3.3. Size Distribution of Water-soluble Chemical Species and Its Implication to BrC Absorption

Fig. 8 shows the mass size distributions of PM, NO3, SO42−, WSOC, oxalate, K+, Cl, and Ca2+ concentrations for typical, pollution event I, pollution event II, and AD event. Moreover, average concentrations of WSOC, NO3, SO42−, oxalate, NH4+, and K+ in PM1.8 (calculated from the sum of MOUDI impactor stages up to cut size of 1.8 μm) are summarized in Table 1 for the four different periods. As shown in Table 1, WSOC, NO3, and NH4+ concentrations in PM1.8 were highest during pollution event I, while SO42− and oxalate concentrations were highest during pollution event II. Higher NO3 during pollution event I could be due to aqueous-phase oxidation processes of NO2 from local traffic activity (Fig. 1 and 2). However, higher SO42− during pollution event II suggests the influence of long-range transported aerosols from mid-eastern Chinese regions, as supported by transport pathways of air masses and MODIS images (see Fig. 3). For the four sampling periods, PM had a bi-modal size distribution peaking at 0.32 or 1.0 μm (fine mode) and 3.1 or 6.2 μm (coarse mode). Predominant mode of PM was observed in fine mode during the typical and two pollution events, and in coarse mode during AD event. In particular, PM during pollution event II showed a dominant droplet mode peaking at 1.0 μm, indicating an influence from more atmospheric processed or long-ranged transported aerosol components during this event.
For the pollution event I, NO3, SO42−, WSOC, oxalate, and K+ concentrations exhibited a dominant condensation mode peaking at 0.32 μm, suggesting that NO3, SO42−, WSOC, and oxalate could likely be formed through homogenous oxidation processes of NO2, SO2, and VOCs, followed by the condensation on pre-existing particles [53]. The condensation mode NO3 and SO42− concentrations accounted for 27.6 and 34.0% of their total concentration, respectively. Furthermore, coexistence of condensation mode WSOC and K+ suggests the impact of BB emissions [54, 55]. The condensation mode WSOC and K+ contributed 35.3 and 20.3% to the total WSOC and K+, respectively.
For the pollution event II, NO3, SO42−, and oxalate concentrations exhibited a dominant condensation mode peaking at 0.32 μm, in which they had similar formation pathways, e.g., homogenous gas-phase reactions [2], but the size distributions of WSOC and K+ concentrations showed a similar shape with a broad peak between 0.32 and 1.0 μm, possibly due to homogeneous and heterogeneous reactions and BB emissions. Condensation mode NO3, SO42−, and oxalate concentrations contributed 21.3, 26.2, and 28.3%, respectively, to their total concentrations, respectively. WSOC and K+ concentrations contributed 25.5 and 15.2%, respectively, to their total concentrations at 0.32 μm, as well as 26.8 and 18.2%, respectively, at 0.55 μm, and 21.3 and 18.2%, respectively, at 1.0 μm. WSOC and K+ concentrations at particle size of 0.32–1.0 μm contributed to 73.6 and 51.6% to their total concentrations, respectively.
However, during the AD event, the PM, NO3, SO42−, WSOC, and K+ showed dominant condensation mode size distributions peaking at 0.32 μm and a coarse mode peaking with a particle size of 3.1 or 6.1 μm. Their condensation mode size distributions were likely associated with homogenous reactions of precursor gases. In coarse mode, however, NO3, SO42−, and WSOC were probably associated with soil dust particles (Ca2+) or sea salt (Na+) particles, as discussed in previous studies [39, 41, 56]. As shown in Fig. 7, size distribution of Cl had a predominant coarse mode peaking at 6.1 μm for four sampling periods, but with a prominent mode during the D event. A very similar distribution was also observed for Ca2+. The size distribution of Na+ is not shown here but it was very similar to that of Cl for the four sampling periods, with R2 ranging from 0.70 (pollution episode II) to 0.88 (AD event), indicating that they are from a similar source. Furthermore, the size distribution of Ca2+ was very strongly associated with that of Cl, with R2 values of 0.84 (AD event) - 0.90 (typical period). Generally, the coarse mode Na+ and Cl are mostly from marine aerosols, and Ca2+ is derived from soil CaCO3 particles. Once they are emitted in the atmosphere, thus, the reactions of NaCl and CaCO3 with HNO3 and H2SO4 during transport may produce the coarse mode NO3 and SO42− [40]. A study conducted by our group also indicated that the coarse mode WSOC was possibly condensed on the surfaces of the produced Ca(NO3)2 and CaSO4 particles [39, 56]. As a consequence, the results from previous studies and the coexistence of coarse mode WSOC, Na+, Cl, and Ca2+ suggest that the coarse mode WSOC during AD event was possibly condensed on the surface of the NaNO3 and Na2SO4 and/or the Ca(NO3)2 and CaSO4 particles. Contributions of the coarse mode NO3, SO42−, WSOC, K+, Cl, and Ca2+ during AD event were 34.5, 10.5, 22.7, 24.2, 53.6, and 49.8% of their total concentrations at a particle size bin of 3.1–6.1 μm, respectively.
As shown in Fig. 8, the size distributions of WSOC during the entire study period were similar to those of NO3, SO42−, oxalate, or K+. To further examine size-resolved WSOC sources, the correlation analyses between WSOC and chemical species (oxalate, NO3, SO42−, and K+) at particle sizes of 0.32 μm, 0.55 μm, 1.0 μm, and PM1.8 were performed, and the results are summarized in Table 2. Fine mode WSOC (PM1.8) was strongly correlated with oxalate (R2 = 0.76) and moderately correlated with NO3 (R2 = 0.55) and SO42− (R2 = 0.42). However, R2 values between WSOC and chemical markers depend on particle size and chemical markers, as shown in Table 2. As shown in Table 2, droplet mode WSOC (0.55–1.0 μm size bins) was strongly associated with K+ (R2 = 0.70–0.82) and oxalate (R2 = 0.68–0.64), suggesting that BB emissions and secondary organic aerosol formation could be possible sources of the droplet mode WSOC. Meanwhile the condensation mode WSOC had good-to-strong correlations with NO3 (R2 = 0.68) and oxalate (R2 = 0.59), suggesting that secondary formation processes (e.g., homogenous gas-phase oxidations) similar to those of condensation mode NO3 and oxalate were likely associated with condensation mode WSOC formation during the study period.
It was confirmed in a previous section that the absorption due to aerosol particles was related to BrC. To further demonstrate the relationship between absorption by BrC at 370 nm estimated from the aethalometer (BrC babs,370) and size-resolved water-soluble chemical markers (WSOC, oxalate, and K+), we carried out correlation analyses between estimated BrC babs,370 and size-resolved concentrations of three chemical markers for various particle sizes (PM1.8, 1.00, 0.55, and 0.32 μm). Their results are summarized in Table 3. Organic markers such as WSOC and oxalate were used to investigate the influence of secondary organic aerosols on light absorption by BrC aerosols [4], while the water-soluble K+ was used to examine the influence from BB emissions on the BrC absorption [54, 55, 57]. As shown in Table 3, BrC babs,370 had moderate-to-strong correlations with size-resolved WSOC with R2 values ranging from 0.56 to 0.73 and size-resolved oxalate with R2 values of 0.51–0.81, respectively. This indicates that BrC absorption in the near UV range was related to formation of WSOC and oxalate existing in condensation and droplet particle, e.g., secondary organic aerosols. Furthermore, BrC babs,370 had good correlations with the droplet mode K+ at particle sizes of 0.55 and 1.0 μm, but had low correlations with K+ at sizes of 0.32 μm and PM1.8, suggesting that BrC absorption was likely associated with droplet mode K+ at 0.55 or 1.0 μm from BB emissions.

### 4. Conclusions

Daily averaged size-segregated filter samples were collected at an urban site from October 19 to November 09, 2017, to examine the impact of pollution and AD events on the water-soluble chemical composition and size distribution of aerosol particles in that area. Moreover, light absorption by aerosol and BrC particles was discussed between typical, two pollution events, and an AD event. Based on the 24-h Korean PM2.5 standard and the AD event, the entire study period was classified into four periods; typical, pollution event I, pollution event II, and AD.
Stable atmospheric conditions such as low wind speed and high surface pressure were observed during the two pollution events I and II, leading to enhanced concentrations of PM2.5, BC370, BC880, and CO, and increased aerosol light absorption coefficients. Furthermore, low wind speeds and high RH conditions during the pollution events gave rise to the accumulation of aerosol particles and can promote the formation of secondary organic and inorganic aerosol species (NO3 and SO42−), resulting in dominant condensation and/or droplet mode size distributions for WSOC, NO3, and SO42−. Difference in the accumulation mode size distributions of water-soluble chemical species between pollution events I and II could be attributed to the difference in transport pathways of air masses from high-pollution regions and the formation processes for the secondary chemical species. In addition, the result from this study suggests that BrC absorption in the near UV range during fall at the site was influenced by both secondary organic aerosols and BB emissions. Overall, the pollution events observed during the fall season at an urban site of Gwangju, Korea can be due to the synergy of poor dispersion conditions, enhanced secondary formation, local emissions, and long-range transportation of air masses from upwind polluted areas.

### Acknowledgments

This research was supported by the National Strategic Project-Fine Particle of the National Research Foundation of Korea (NRF) funded by the Ministry of Science and ICT (MSIT), the Ministry of Environment (ME), and the Ministry of Health and Welfare (MOHW) (2017M3D8A1092015). Also this study was supported by the Basic Science Research Programs through the NRF funded by the Ministry of Education (MOE) (NRF-2017R1D1A3A03000511).

### References

1. Bae MS, Shin JS, Lee KY, Lee KH, Kim YJLong-range transport of biomass burning emissions based on organic molecular markers and carbonaceous thermal distribution. Sci Total Environ. 2014;466:56–66.

2. Yu G, Park S, Lee KSource contributions and potential source regions of size-resolved water-soluble organic carbon measured at an urban site over one year. Environ Sci Proc Imp. 2016;18:1343–1358.

3. Yu G, Zhang Y, Cho SY, Park SInfluence of haze pollution on water-soluble chemical species in PM2.5 and size-resolved particles at an urban site during fall. J Environ Sci. 2017;57:370–382.

4. Park S, Schauer J, Cho SSources and their contributions to two water-soluble organic carbons observed at a roadway site. Atmos Environ. 2013;77:348–357.

5. Park S, Cho S, Bae MSource identification of water-soluble organic aerosols at a roadway site using a positive matrix factorization analysis. Sci Total Environ. 2015;533:410–421.

6. Zhao M, Xiu G, Qiao T, Li Y, Yu JCharacteristics of haze pollution episodes and analysis of a typical winter haze process in Shanghai. Aerosol Air Qual Res. 2016;16:1625–1637.

7. Kim KW, He Z, Kim YJPhysicochemical characteristics and radiative properties of Asian dust particles observed at Kwangju, Korea, during the 2001 ACE-Asia intensive observation period. J Geophys Res. 2004;109:D19S02

8. Li H, Ma Y, Duan F, et alTypical winter haze pollution in Zibo, an industrial city in China: Characteristics, secondary formation, and regional contribution. Environ Pollut. 2017;229:339–349.

9. Ma Q, Wu Y, Zhang D, et alRoles of regional transport and heterogeneous reactions in the PM2.5 increase during winter haze episodes in Beijing. Sci Total Environ. 2017;599–600:246–253.

10. Yu GH, Park SS, Park JS, et alPollution characteristics of PM2.5 observed during winter and summer in Baengryeongdo and Seoul. J Korean Soc Atmos Environ. 2018;34:38–55.

11. Zhang QH, Zhang JP, Xue HWThe challenge of improving visibility in Beijing. Atmos Chem Phys. 2010;10:7821–7827.

12. Guo S, Hu M, Zamora MElucidating severe urban haze formation in China. Proc Natl Acad Sci. 2014;111:17373–17378.

13. Huang RJ, Zhang YL, Bozzetti CHigh secondary aerosol contribution to particulate pollution during haze events in China. Nature. 2014;514:218–222.

14. Wang QQ, Sun YL, Jiang QChemical composition of aerosol particles and light extinction apportionment before and during the heating season in Beijing, China. J Geophys Res. 2015;120:708–722.

15. Zhang Y, Huang W, Cai TQConcentrations and chemical compositions of fine particles (PM2.5) during haze and non-haze days in Beijing. Atmos Res. 2016;174–175:62–69.

16. Park S, Jung S, Gong B, Cho SY, Lee SJCharacteristics of PM2.5 haze episodes revealed by highly time-resolved measurements at an air pollution monitoring supersite of Korea. Aerosol Air Qual Res. 2013;13:957–976.

17. Yu GH, Park SS, Kim YS, et alDifference in chemical composition of PM2.5 and investigation of its causing factors between 2013 and 2015 in air pollution intensive monitoring stations. J Korean Soc Atmos Environ. 2018;34:16–37.

18. Chen L, Chow J, Doddridge B, et alAnalysis of a summertime PM2.5 and haze episode in the Mid-Atlantic region. J Air Waste Manage. 2003;53:946–956.

19. Park SS, Harrison D, Pancras JP, Ondov JMHighly time-resolved organic and elemental carbon measurements at the Baltimore Supersite in 2002. J Geophys Res. 2005;110:D07S06

20. Park S, Kleissl J, Harrison D, et alCharacteristics of PM2.5 episodes revealed by semi-continuous measurements at the Baltimore Supersite at Ponca St. Aerosol Sci Technol. 2006;40:845–860.

21. Park S, Kim J, Jeong JAbundance and sources of hydrophilic and hydrophobic water-soluble organic carbon at an urban site in Korea in summer. J Environ Monit. 2012;14:224–232.

22. Zheng GJ, Duan FK, Su H, et alExploring the severe winter haze in Beijing: The impact of synoptic weather, regional transport and heterogeneous reactions. Atmos Chem Phys. 2015;15:2969–2983.

23. Jiang J, Zhou W, Cheng Z, Wang S, He K, Hao JParticulate matter distributions in China during a winter period with frequent pollution episodes (January 2013). Aerosol Air Qual Res. 2015;15:494–503.

24. Sun YL, Wang ZF, Fu PQ, et alAerosol composition, sources and processes during wintertime in Beijing, China. Atmos Chem Phys. 2013;13:4577–4592.

25. Sun YL, Wang ZF, Du W, et alLong-term real-time measurements of aerosol particle composition in Beijing, China: Seasonal variations, meteorological effects, and source analysis. Atmos Chem Phys. 2015;15:10149–10165.

26. Andreae MO, Ramanatham VClimate’s dark forcing. Science. 2013;340:280–281.

27. Park SS, Son SCRelationship between carbonaceous components and aerosol light absorption during winter at an urban site of Gwangju, Korea. Atmos Res. 2017;185:73–83.

28. Park SS, Yu GH, Lee SOptical absorption characteristics of brown carbon aerosols during the KORUS-AQ campaign at an urban site. Atmos Res. 2018;203:16–27.

29. Kirchstetter TW, Novakov T, Hobbs PVEvidence that spectral dependence of light absorption by aerosols is affected by organic carbon. J Geophys Res. 2004;109:D21208

30. Laskin A, Laskin J, Nizkorodov SAChemistry of atmospheric brown carbon. Chem Rev. 2015;115:4335–4382.

31. Andreae MO, Gelencsér ABlack carbon or brown carbon? The nature of light-absorbing carbonaceous aerosols. Atmos Chem Phys. 2006;6:3131–3148.

32. Cheng Y, He K, Du Z, et alThe characteristics of brown carbon aerosol during winter in Beijing. Atmos Environ. 2016;127:355–364.

33. Hecobian A, Zhang X, Zheng M, Frank N, Edgerton ES, Weber RJWater-soluble organic aerosol material and the light-absorption characteristics of aqueous extracts measured over the Southeastern United States. Atmos Chem Phys. 2010;10:5965–5977.

34. Hoffer A, Gelencsér A, Guyon P, et alOptical properties of humic-like substances (HULIS) in biomass-burning aerosols. Atmos Chem Phys. 2006;6:3563–3570.

35. Park S, Yu JChemical and light absorption properties of humic-like substances from biomass burning emissions under controlled combustion experiments. Atmos Environ. 2016;136:114–122.

36. Wu GM, Cong ZY, Kang SC, et alBrown carbon in the cryosphere: Current knowledge and perspective. Adv Climate Change Res. 2016;7:82–89.

37. Ondov JM, Wexler AWWhere do particulate toxins reside? An improved paradigm for the structure and dynamics of the urban mid-Atlantic aerosol. Environ Sci Technol. 1998;32:2547–2555.

38. Park S, Shin DCharacteristic of size-resolved water-soluble organic carbon in atmospheric aerosol particles observed during daytime and nighttime in an urban area. Part Aerosol Res. 2013;9:7–21.

39. Park S, Son SSize distribution and sources of humic-like substances in an urban ambient air during winter. Environ Sci Proc Imp. 2016;18:32–41.

40. Huang XF, Yu JZ, He LY, Yuan ZWater-soluble organic carbon and oxalate in aerosols at a coastal urban site in China: Size distribution characteristics, sources, and formation mechanisms. J Geophys Res. 2006;111:D22212

41. Timonen H, Saarikoski S, Tolonen-Kivimäki O, et alSize distributions, sources and source areas of water-soluble organic carbon in urban background air. Atmos Chem Phys. 2008;8:5635–5647.

42. Xu JZ, Zhang Q, Wang ZB, et alChemical composition and size distribution of summertime PM2.5 at a high altitude remote location in the northeast of the Qinghai-Xizang (Tibet) Plateau: Insights into aerosol sources and processing in free troposphere. Atmos Chem Phys. 2015;15:5069–5081.

43. Drinovec L, Močnik G, Zotter P, et alThe “dual-spot” Aethalometer: An improved measurement of aerosol black carbon with real-time loading compensation. Atmos Meas Tech. 2015;8:1965–1979.

44. Martinsson J, Eriksson AC, Elbæk Nielsen I, et alImpacts of combustion conditions and photochemical processing on the light absorption of biomass combustion aerosol. Environ Sci Technol. 2015;49:14663–14671.

45. Han T, Liu SC, Zhang YHRole of secondary aerosols in haze formation in summer in the Megacity Beijing. J Environ Sci. 2015;31:51–60.

46. Quan JN, Liu Q, Li X, et alEffect of heterogeneous aqueous reactions on the secondary formation of inorganic aerosols during haze events. Atmos Environ. 2015;122:306–312.

47. Rolph G, Stein A, Stunder BReal-time Environmental Applications and Display sYstem: READY. Environ Modell Softw. 2017;95:210–228.

48. Magee Scientific. The AethalometerTM. Berkeley, California: Magee Scientific Company; 2005.

49. Schnaiter M, Horvath H, Möhler O, Naumann KH, Saathoff H, Schöck OWUV-VIS-NIR spectral optical properties of soot and soot-containing aerosols. J Aerosol Sci. 2003;34:1421–1444.

50. Bergstrom R, Pilewskie P, Russell P, et alSpectral absorption properties of atmospheric aerosols. Atmos Chem Phys. 2007;7:5937–5943.

51. Bergstrom RW, Pilewskie P, Pommier J, et alSpectral absorption of solar radiation by aerosols during ACE-Asia. J Geophys Res. 2004;109:D19S15

52. Sun H, Biedermann L, Bond TCColor of brown carbon: A model for ultraviolet and visible light absorption by organic carbon aerosol. Geophys Res Lett. 2007;34:L17813

53. Seinfeld JH, Pandis SNAtmospheric chemistry and physics: From air pollution to climate change. 2nd edHoboken, New Jersey: John Wiley & Sons, Inc; 2006.

54. Ichikawa Y, Naito SChemical compositions of primary PM2.5 derived from biomass burning emissions. Asian J Atmos Environ. 2017;11:79–95.

55. Yu J, Yu GH, Park S, Bae MSChemical and absorption characteristics of water-soluble organic carbon and humic-like substances in size-segregated particles from biomass burning emissions. Asian J Atmos Environ. 2017;11:96–106.

56. Park SS, Kim JHPossible sources of two size-resolved water-soluble organic carbon fractions at a roadway site during fall season. Atmos Environ. 2014;94:134–143.

57. Park SS, Sim SY, Bae MS, Schauer JJSize distribution of water-soluble components in particulate matter emitted from biomass burning. Atmos Environ. 2013;73:62–72.

##### Fig. 1
Temporal profiles of hourly and 24-h average PM10, PM2.5, PM2.5/PM10, O3, NO2, CO, and SO2 over the study period.
##### Fig. 2
Temporal profiles of hourly and 24-h average temperature, relative humidity, wind speed, and sea level pressure.
##### Fig. 3
Air mass transport pathways and MODIS images around the Korean peninsula prior to the pollution events I and II. Red, blue, and green lines on backward trajectories (upper) indicate heights of 500, 1,000, and 1,500 m AGL.
##### Fig. 4
Temporal profiles of hourly BC880, babs,370, babs,880, BC babs,380, BrC babs,370, and BrC370 contribution.
##### Fig. 5
Spectral dependence of light absorption coefficients by aerosol (a) and BrC (b) for four different periods.
##### Fig. 6
Hourly variation of PM2.5 for two distinct periods during pollution event II.
##### Fig. 7
Spectral dependence of light absorption coefficients by aerosol (a) and BrC (b) for two distinct periods.
##### Fig. 8
Mass size distribution of PM, NO3, SO42−, WSOC, oxalate, K+, Cl, and Ca2+ for four different PM pollution cases.
##### Table 1
A Summary of Measurement Data for Four Different Periods
Item Unit Typical Event I Event II AD event
PM10 μg/m3 40.1 ± 14.7 65.7 ± 9.1 58.1 ± 17.9 63.8 ± 10.0
PM2.5 μg/m3 25.6 ± 12.1 50.1 ± 8.5 53.0 ± 19.4 30.8 ± 10.0
PM2.5/PM10 - 0.62 ± 0.19 0.76 ± 0.08 0.86 ± 0.14 0.54 ± 0.12

O3 ppb 24 ± 15 25 ± 18 23 ± 17 31 ± 15
NO2 ppb 25 ± 14 34 ± 13 29 ± 13 21 ± 10
CO ppm 0.42 ± 0.18 0.56 ± 0.14 0.53 ± 0.12 0.38 ± 0.13
SO2 ppb 1.8 ± 0.6 2.4 ± 1.5 1.9 ± 0.5 2.6 ± 0.4

Temperature °C 14.0 ± 3.2 15.2 ± 1.2 11.4 ± 2.4 11.8 ± 4.7
Wind speed m/s 1.3 ± 0.9 0.6 ± 0.2 0.9 ± 0.4 1.9 ± 0.5
Relative humidity % 66 ± 9 74 ± 5 76 ± 10 64 ± 16
Surface pressure hPa 1,021 ± 5 1,021 ± 2 1,024 ± 5 1,023 ± 2

Aethalometer measurements
BC370 μg/m3 2.7 ± 1.9 4.8 ± 1.6 4.2 ± 1.7 2.0 ± 2.5
BC880 μg/m3 2.0 ± 1.4 3.5 ± 1.2 2.7 ± 1.4 1.4 ± 1.9
babs,370 1/Mm 50 ± 35 88 ± 30 77 ± 12 37 ± 47
babs,880 1/Mm 15 ± 11 27 ± 9 20 ± 7 11 ± 15
BC babs,370 1/Mm 37 ± 26 65 ± 22 47 ± 16 26 ± 36
BrC babs,370 1/Mm 13 ± 10 24 ± 11 30 ± 4 11 ± 13
BrC370 % 27 26 40 33
AAE370-950 - 1.3 1.3 1.5 1.4
BrC AAE370-660 - 4.5 4.3 4.6 4.2

MOUDI measurements
PM1.8 WSOC μgC/m3 1.7 ± 0.8 4.2 ± 0.7 4.1 ± 1.6 1.5
PM1.8 NO3 μg/m3 2.4 ± 1.7 8.8 ± 1.0 6.8 ± 3.1 4.5
PM1.8 SO42− μg/m3 2.0 ± 1.1 4.9 ± 0.6 5.6 ± 1.2 4.9
PM1.8 oxalate μg/m3 0.2 ± 0.1 0.5 ± 0.2 0.7 ± 0.2 0.4
PM1.8 NH4+ μg/m3 1.4 ± 0.7 4.1 ± 0.6 3.1 ± 0.2 2.0
PM1.8 K+ μg/m3 0.5 ± 0.1 0.7 ± 0.2 0.7 ± 0.1 0.6
##### Table 2
Coefficients of Determination (R2) between Size-resolved WSOC, Oxalate, K+, NO3, and SO42−
Size (μm) Oxalate K+ NO3 SO42−
WSOC PM1.8 0.76** 0.38* 0.55** 0.42**
1.00 0.64** 0.82** 0.27* 0.47**
0.55 0.68** 0.70** 0.47** 0.50**
0.32 0.59** 0.42** 0.68** 0.43**

[i] Note) * and ** are significant at p values of < 0.05 and < 0.01, respectively.

##### Table 3
Coefficients of Determination (R2) between PM2.5 BrC babs,370 and Size-resolved WSOC, Oxalate, and K+
Size (μm) WSOC Oxalate K+
PM2.5 BrC babs,370 PM1.8 0.73** 0.81** 0.21*
1.00 0.67** 0.51** 0.57**
0.55 0.72** 0.61** 0.64**
0.32 0.56** 0.74** 0.16*

[i] Note) * and ** are significant at p values of < 0.05 and < 0.01, respectively.

TOOLS
Full text via DOI
E-Mail
Print
Share:
METRICS
 1 Crossref
 0 Scopus
 2,086 View