Contributions of Ammonia to High Concentrations of PM2.5 in an Urban Area

Atmospheric ammonia (NH3) plays a critical role in PM2.5 pollution. Data on atmospheric NH3 are scanty; thus, the role of NH3 in the formation of ammonium ions (NH4+) in various environments is understudied. Herein, we measured concentrations of NH3, PM2.5, and its water-soluble SO42−, NO3−, and NH4+ ions (SNA) at an urban site in Jeonju, South Korea from May 2019 to April 2020. During the measurement period, the average concentrations of NH3 and PM2.5 were 10.5 ± 4.8 ppb and 24.0 ± 12.8 μg/m3, respectively, and SNA amounted to 4.3 ± 3.1, 4.4 ± 4.9, and 1.6 ± 1.8 μg/m3, respectively. A three-dimensional photochemical model analysis revealed that a major portion of NH3, more than 88%, originated from Korea. The enhancement of the ammonium-to-total ratio of NH3, NHX (NHR = [NH4+]/[NH4+] + [NH3]) was observed up to ~0.61 during the increase of PM2.5 concentration (PM2.5 ≥ 25 μg/m3) under low temperature and high relative humidity conditions, particularly in winter. The PM2.5 and SNA concentrations increased exponentially as NHR increased, indicating that NH3 contributed significantly to SNA formation by gas-to-particle conversion. Our study provided experimental evidence that atmospheric NH3 in the urban area significantly contributed to SNA formation through gas-to-particle conversion during PM2.5 pollution episodes.


Introduction
Global emissions of NH 3 have annually increased from an estimated 1.9 Tg in the 1960s to 16.7 Tg in 2010 [1]. Reports have indicated that the main source of atmospheric NH 3 at the global scale is agricultural activities involving livestock, fertilizers, soil, and crops [2][3][4][5]; these activities accounted for approximately 60% of the total NH 3 emitted from Asia in the 2000s [1]. NH 3 is important because it can contribute to the acidification of ecosystems [6,7]. Moreover, it plays a critical role in chemical reactions in the atmosphere, where its conversion to particulate ammonium can lead to high concentrations of particulate matter [8][9][10][11][12][13][14]. These particulate ammonium can influence air quality, visibility, and human health [15][16][17].
Field measurements have shown that concentrations of atmospheric NH 3 generally vary depending on the season and location [9,[18][19][20][21][22][23][24][25][26][27]. NH 3 concentrations are temperaturedependent; they increase in summer and decrease in winter [9,20]. For example, an average ambient NH 3 concentration of~36.2 ppb, with variations ranging from~73.9 ppb in July to 13.5 ppb in September, was detected in the Northern Plains of China in 2013 [8]. Only a few studies have been conducted in Korea, which showed the average NH 3 concentration in Seoul of~6.0 ppb from 1996 to 1997 and~11.0 ppb from 2010 to 2011 [23,27], and 10.5 ppb in Jeon-ju from 2019 to 2020, with higher concentrations occurring during the summer [19].
NH 3 in the atmosphere can react with acidic species, such as sulfuric acid (H 2 SO 4 ), nitric acid (HNO 3 ), and hydrochloric acid (HCl), which lead to the production of secondary inorganic aerosols (SIAs) including ammonium sulfate ((NH 4 ) 2 SO 4 ), ammonium nitrate (NH 4 NO 3 ), and ammonium chloride (NH 4 Cl) [28,29]. Previous studies have shown that these SIAs can account for up to~70% of the mass of PM 2.5 , depending on the location and season [30][31][32][33][34]. Moreover, recent studies have shown that high conversion ratios of ammonium from the gas to particle phase can significantly promote high PM 2.5 concentration [8,9,25,[35][36][37][38][39]. In rural areas of Shanghai, China, PM 2.5 concentrations were found to be influenced by secondary NH 4 + from NH 3 at a conversion ratio of up to~0.8 during periods of high PM 2.5 pollution in October 2013 [9]. In Delhi, India, the conversion ratio from NH 3 to NH 4 + increased up to~0.6 during PM 2.5 pollution episodes from 2013 to 2015 [36]. Increases of SIAs following increasing water content of aerosols result in various aqueous phase reactions and high concentration of PM 2.5 [40][41][42]. Although atmospheric NH 3 is one of the key species for the formation of SIAs, which cause aerosol pollution, studies on evaluations of the impacts of NH 3 on the formation of PM 2.5 are still limited. In addition, characteristics of atmospheric NH 3 and its impact on PM 2.5 pollution are scarce in urban areas In this study, atmospheric NH 3 and water-soluble ions, including SO 4 2− , NO 3 − , and NH 4 + (SNA) concentrations were measured over one year from May 2019 to April 2020 in an urban area, Samcheon-dong, Jeonju, South Korea. Using the dataset, we explored how the ambient NH 3 contributes to high concentrations of PM 2.5 . Moreover, we applied a threedimensional photochemical model to identify the origin of the ambient NH 3 . Altogether, our results provide a more comprehensive understanding of the gas-to-particle conversion process in the atmosphere and the role of NH 3 in the formation of aerosol pollution.

Monitoring Site
The concentrations of NH 3 (Figure 1). The monitoring site can be considered as a representative urban site for Jeollabuk-do because the area is surrounded by residential clusters, business buildings, and roads. It is located~50 km from agricultural areas consisting of large-and small-scale livestock farms (pigs, cows, and chickens) and other types of farmlands,~75 km from the Yellow Sea,~190 km from Busan, and~200 km from Seoul.

Measurements
The method of NH 3 measurements has been previously described by Park et al. [19]. Briefly, atmospheric NH 3 concentrations were measured with cavity ring-down spectroscopy (CRDS) (Picarro Inc., model G2103, Santa Clara, CA, USA) on a 1 s basis from 4 May 2019 to 15 April 2020, and the 1-h-averaged data were used for the analyses. The NH 3 analyzer has an average precision of 0.03 ppb for 300 s, with a response time of less than 1 s and a detection limit below 0.09 ppb [43]. Additionally, the analyzer has a low drift value of 0.15 ppb over 72 h [43]. Theoretically, atmospheric NH 3 absorbs light of a characteristic wavelength from within the cavity; when the laser is turned off, the concentration can be calculated using the attenuation curves that disappear. No additional external calibration is required, according to the manual for the Model G-2103 CRDS analyzer [44]. However, in this study, calibration was conducted to confirm the performance of the analyzer by using a mixture of standard NH 3 (11.9 ppm, Airkorea, Korea) and N 2 gas. The calibration was repeated three times with four points at 25, 20, 15, and 5 ppb, and the resulting R 2 was 0.997 ( Figure S1). During the measurement period, perfluoroalkoxy (PFA) tubing (internal Atmosphere 2021, 12, 1676 3 of 15 diameter 4 mm) was used for the inlet, and the inlet length was as short as possible (~1.5 m) to limit the residence time to shorter than 1 s during the measurement period [8]. PM 2.5 was collected on Teflon filters (PTFE, R2PJ047, PALL, New York, USA) over a 24-h period from 09:00 am to 09:00 am at a flow rate of 16.67 L/min using a sequential low volume sampler (APM, PMS-104, Bucheon, Korea) at the monitoring site. A total of 118 PM 2.5 filters were collected during the measurement period (Table S1). The mass concentration of PM 2.5 was determined using the method recommended by the USA. Environmental Protection Agency (EPA) (Compendium of Methods for the Determination of Inorganic Compounds in Ambient Air, Methods), (EPA, https://www.epa.gov) (accessed on 15 September 2021). The concentrations of ion species such as NO 3 − , SO 4 2− , and NH 4 + , as well as minor ions in the PM 2.5 , were analyzed by ion chromatography (AQUION, Thermo Scientific, Massachusetts, USA).
Hourly averaged meteorological parameters of temperature, relative humidity (RH), wind speed, and wind direction were collected from the Jeollabuk-do Institute of Health and Environment Research. The hourly average temperature during the measurement period was 13.6 ± 9.5 • C, and the relative humidity was 67.6 ± 18.6%. To reduce the uncertainties from measurements and instruments during high-precipitation events [45,46], data were excluded from the analyses when the hourly amount of precipitation exceeded 5 mm. These excluded data (~10.5% of the original measured NH 3 data) were mostly from July-August and September (~10.3% of the original measured NH 3 data) due to the monsoon and frequent occurrence of typhoons, respectively. teristic wavelength from within the cavity; when the laser is turned off, the concentration can be calculated using the attenuation curves that disappear. No additional external calibration is required, according to the manual for the Model G-2103 CRDS analyzer [44]. However, in this study, calibration was conducted to confirm the performance of the analyzer by using a mixture of standard NH3 (11.9 ppm, Airkorea, Korea) and N2 gas. The calibration was repeated three times with four points at 25, 20, 15, and 5 ppb, and the resulting R 2 was 0.997 ( Figure S1). During the measurement period, perfluoroalkoxy (PFA) tubing (internal diameter 4 mm) was used for the inlet, and the inlet length was as short as possible (~1.5 m) to limit the residence time to shorter than 1 s during the measurement period [8].
PM2.5 was collected on Teflon filters (PTFE, R2PJ047, PALL, New York, USA) over a 24-h period from 09:00 am to 09:00 am at a flow rate of 16.67 L/min using a sequential low volume sampler (APM, PMS-104, Bucheon, Korea) at the monitoring site. A total of 118 PM2.5 filters were collected during the measurement period (Table S1). The mass concentration of PM2.5 was determined using the method recommended by the USA. Environmental Protection Agency (EPA) (Compendium of Methods for the Determination of Inorganic Compounds in Ambient Air, Methods), (EPA, https://www.epa.gov) (accessed on 15 September 2021). The concentrations of ion species such as NO3 − , SO4 2− , and NH4 + , as well as minor ions in the PM2.5, were analyzed by ion chromatography (AQUION, Thermo Scientific, Massachusetts, USA).
Hourly averaged meteorological parameters of temperature, relative humidity (RH), wind speed, and wind direction were collected from the Jeollabuk-do Institute of Health and Environment Research. The hourly average temperature during the measurement period was 13.6 ± 9.5 °C, and the relative humidity was 67.6 ± 18.6%. To reduce the uncertainties from measurements and instruments during high-precipitation events [45,46], data were excluded from the analyses when the hourly amount of precipitation exceeded 5 mm. These excluded data (~10.5% of the original measured NH3 data) were mostly from July-August and September (~10.3% of the original measured NH3 data) due to the monsoon and frequent occurrence of typhoons, respectively.

Modelling
Identifying the origins of the NH3 is challenging because the airborne NH3 concentration is affected by various physical and chemical processes, including emission, transport, deposition, and chemical transformation. In this study, we simulated NH3 and NH4 + concentrations in Northeast Asia using the community multiscale air quality

Modelling
Identifying the origins of the NH 3 is challenging because the airborne NH 3 concentration is affected by various physical and chemical processes, including emission, transport, deposition, and chemical transformation. In this study, we simulated NH 3 and NH 4 + concentrations in Northeast Asia using the community multiscale air quality (CMAQ) model version 4.7.1 [47]. It is a three-dimensional photochemical model. To operate CMAQ, the meteorological inputs were prepared using Weather Research and Forecasting (WRF) version 3.5.1 [48] with final operational global analysis data. The Sparse Matrix Operator Kernel Emission (SMOKE; version 3.1) [49] was applied to process the KORUSv5 emissions inventory [50,51] for regional emissions, excluding South Korea. For South Korea, CAPSS 2016, developed and released by the National Air Emission Inventory and Research Center (NAIR), was utilized [52].
In the WRF-SMOKE-CMAQ simulation, the brute force method (BFM) was applied to identify the relative contribution of NH 3 and NH 4 + from China and South Korea to the downwind area. The BFM estimates the sensitivity of pollutant concentrations to change in targeted emissions [53]. Kim et al. [54] showed that the estimated contributions using  [55]. The emission perturbation rate has been used in previous air quality modeling studies over the region [56][57][58]. We assumed that the change of NH x concentrations in South Korea to the NH 3 emission perturbations in China shows a low nonlinear response based on a previous NH 3 sensitivity conducted by Kim et al. [55].
where C B is the NH 3 (or NH x ) concentration simulated using the base run, C c,50% is the NH 3 (or NH x ) concentration simulated using a 50% reduction in Chinese NH 3 emissions, and E 50% is the emission perturbation rate (0.5 in this study). Then, the zero-out contribution (ZOC) of Chinese NH 3 emissions to NH 3 concentrations in South Korea was calculated by dividing the NH 3 sensitivity by the perturbation rate (0.5 in this study) [56,58] as shown in Equation (2). Additionally, the difference between the NH 3 concentration in the base run and the ZOC of Chinese NH 3 in the downwind area was considered as the ZOC of South Korea.

Characteristics of Atmospheric NH 3 in Urban Area
During the entire measurement period at the urban site, the hourly averaged concentration of atmospheric NH 3 was 10.5 ± 4.8 ppb, ranging from 2.0 ppb to 54.5 ppb. These atmospheric NH 3 concentrations are comparable to those measured in other regions ( Table 1). The NH 3 level in Jeonju was similar in Seoul in 2010-2011 [23], which was higher than that of the Shanghai urban area in China in 2013 [9]. Moreover, the NH 3 concentration in Jeonju was higher than that of Osaka, Japan, and Ho Chi Minh, Vietnam in 2015, also in Asia [22]. Further, the NH 3 concentration in the urban area was up to 3-4 times higher than that of North America and Europe (Table 1) [24,25,59]. Shown in Figure 2 is the seasonal diurnal variation of the atmospheric NH 3 observed at the urban site. In spring and summer, the diurnal variation of the ambient NH 3 concentration was low, from 00:00 to 12:00 local time, and then the concentration increased and reached a maximum concentration of more than~14 ppb at 20:00. Previous studies reported that such variations, with high concentrations observed in late afternoon, are caused by NH 3 transport to urban area from the vicinity of rural areas by agricultural sources, expansion of a planetary boundary layer, and wind directions [60,61]. This could explain the high concentration in the study site, which is surrounded by agricultural lands (rice fields, and large and small livestock farms)~10 km to the west and southwest ( Figure S2). A recent study simultaneously measured atmospheric NH 3 concentration from rural and urban areas, which are close to the present study site, and showed that the NH 3 concentrations at both sites (rural and urban areas) were significantly higher in summer, particularly in June, than in other seasons [62]. When the highest atmospheric NH 3 concentrations occurred in June in the urban area, elevated NH 3 concentrations were also observed in the adjacent rural area [62]. They suggested that the enhanced ambient NH 3 concentrations observed in the urban area were influenced by high NH 3 concentrations from the rural area located to the west [62]. Contrastingly, bimodal peaks in the morning and late afternoon determined in autumn and winter were likely due to the impact of traffic in the urban areas. Seasonally, the NH3 concentration in Jeonju was observed as follows: summer (13.3 ± 5.8 ppb) > spring (12.1 ± 5.1 ppb) > winter (9.2 ± 4.3 ppb) > autumn (8.9 ± 3.1 ppb) ( Figure  3). In this study, the NH3 concentration showed a strong correlation with ambient temperature ( Figure S3). The atmospheric NH3 concentration in the urban site was enhanced as temperature was increased, which is consistent with previous studies [8,9,19,20,37,38,62]. However, the concentrations decreased when the temperature was above 30 °C ( Figure  S3), because of the wet deposition and removal effects that occur in monsoon ( Figure S4) [9,19,36]. Seasonally, the NH 3 concentration in Jeonju was observed as follows: summer (13.3 ± 5.8 ppb) > spring (12.1 ± 5.1 ppb) > winter (9.2 ± 4.3 ppb) > autumn (8.9 ± 3.1 ppb) (Figure 3). In this study, the NH 3 concentration showed a strong correlation with ambient temperature ( Figure S3). The atmospheric NH 3 concentration in the urban site was enhanced as temperature was increased, which is consistent with previous studies [8,9,19,20,37,38,62]. However, the concentrations decreased when the temperature was above 30 • C ( Figure S3), because of the wet deposition and removal effects that occur in monsoon ( Figure S4) [9,19,36].
3). In this study, the NH3 concentration showed a strong correlation with ambient temperature ( Figure S3). The atmospheric NH3 concentration in the urban site was enhanced as temperature was increased, which is consistent with previous studies [8,9,19,20,37,38,62]. However, the concentrations decreased when the temperature was above 30 °C ( Figure  S3), because of the wet deposition and removal effects that occur in monsoon ( Figure S4) [9,19,36].

Contribution of NH3 to PM2.5 Pollution
To explore the effects of NH3 on aerosol pollution, which have never been studied in Korea through field measurements, we measured PM2.5 and its water-soluble ions, and investigated how NH3 contributes to NH4 + formation and the production of PM2.5. Throughout the measurement period, the monthly average PM2.5 concentration was 24.0

Contribution of NH 3 to PM 2.5 Pollution
To explore the effects of NH 3 on aerosol pollution, which have never been studied in Korea through field measurements, we measured PM 2.5 and its water-soluble ions, and investigated how NH 3 contributes to NH 4 + formation and the production of PM 2.5 . Throughout the measurement period, the monthly average PM 2.5 concentration was 24.0 ± 12.8 µg/m 3 , and NO 3 − was the most abundant (4.4 ± 4.9 µg/m 3 ), followed by SO 4 2− (4.3 ± 3.1 µg/m 3 ) and NH 4 + (1.6 ± 1.8 µg/m 3 ) in the PM 2.5 (Figure 4b,c, Table S1). The NO 3 − in the PM 2.5 was significantly enhanced in winter time. Previous studies also reported that PM 2.5 concentrations were elevated, particularly in winter, with a remarkable increase in the NO 3 − concentrations at the measurement site during winter [63][64][65][66]. In particular, in January, high concentrations of PM 2.5 were observed, with a monthly average of 38.1 ± 20.3 µg/m 3 (Table S1), and an average NO 3 − concentration of 11.8 µg/m 3 ( Figure S5).  Table S1). The NO3 − in the PM2.5 was significantly enhanced in winter time. Previous studies also reported that PM2.5 concentrations were elevated, particularly in winter, with a remarkable increase in the NO3 − concentrations at the measurement site during winter [63][64][65][66]. In particular, in January, high concentrations of PM2.5 were observed, with a monthly average of 38.1 ± 20.3 μg/m 3 (Table S1), and an average NO3 − concentration of 11.8 μg/m 3 ( Figure S5). In this study, PM2.5 pollution was defined as a daily average of PM2.5 ≥ 25 μg/m 3 , based on the daily mean PM2.5 guideline value recommended by the World Health Organization [67]. During the entire period, 47 d (spring: 13 d, summer: 6 d, autumn: 10 d, and winter: 18 d) out of a total of 118 d showed PM2.5 pollution. Figure 5 presents a comparison of the SNA concentrations, NH3, and the ratio of NH4 + to total ammonia, NHx (where NHR = [NH4 + ]/([NH4 + ] + [NH3])) [68], for clean days (PM2.5 < 25 μg/m 3 ) versus polluted days (PM2.5 ≥ 25 μg/m 3 ). On the polluted days, the NO3 − and NH4 + mass fraction significantly increased to 46% and 18%, respectively, while the SO4 2− fraction was reduced to 36% in the SNA fraction (Figure 5a). The NH3 concentration was slightly higher (12.6 ppb) during PM2.5 pollution than during clean days (10.7 ppb) (Figure 5b). Moreover, on the PM2.5 pollution, the daily average NHR increased dramatically to 0.24 (Figure 5c), with a maximum daily ratio of 0.61 in January ( Figure S5). It was comparable with the NHR of only 0.06 during the clean days.  In this study, PM 2.5 pollution was defined as a daily average of PM 2.5 ≥ 25 µg/m 3 , based on the daily mean PM 2.5 guideline value recommended by the World Health Or-ganization [67]. During the entire period, 47 d (spring: 13 d, summer: 6 d, autumn: 10 d, and winter: 18 d) out of a total of 118 d showed PM 2.5 pollution. Figure 5 presents  mass fraction significantly increased to 46% and 18%, respectively, while the SO 4 2− fraction was reduced to 36% in the SNA fraction (Figure 5a). The NH 3 concentration was slightly higher (12.6 ppb) during PM 2.5 pollution than during clean days (10.7 ppb) (Figure 5b). Moreover, on the PM 2.5 pollution, the daily average NHR increased dramatically to 0.24 (Figure 5c), with a maximum daily ratio of 0.61 in January ( Figure S5). It was comparable with the NHR of only 0.06 during the clean days. Illustrated in Figure 6a is the relationship between NHR and NH3 on PM2.5 pollution. The NHR was inversely proportional to the atmospheric NH3 concentrations (Figure 6a), and the atmospheric NH3 decreased as the NHR increased. These data reflect the interconversion between atmospheric gases and particles [9], thus, suggesting that NH3 was converted to NH4 + on PM2.5 pollution, resulting in high PM2.5 concentration. Moreover, as the NHR increased, the PM2.5 and SNA concentrations increased exponentially with R 2 values of 0.49 and 0.73, respectively (Figure 6b). This indicates that the increase in PM2.5 concentration was facilitated by the reactions of gaseous NH3 with acidic species that converted the NH3 to particulate NH4 + [8,9,36]. Illustrated in Figure 6a is the relationship between NHR and NH 3 on PM 2.5 pollution. The NHR was inversely proportional to the atmospheric NH 3 concentrations (Figure 6a), and the atmospheric NH 3 decreased as the NHR increased. These data reflect the interconversion between atmospheric gases and particles [9], thus, suggesting that NH 3 was converted to NH 4 + on PM 2.5 pollution, resulting in high PM 2.5 concentration. Moreover, as the NHR increased, the PM 2.5 and SNA concentrations increased exponentially with R 2 values of 0.49 and 0.73, respectively (Figure 6b). This indicates that the increase in PM 2.5 concentration was facilitated by the reactions of gaseous NH 3 with acidic species that converted the NH 3 to particulate NH 4 + [8,9,36].  NH3 is considered to be neutralized by sulfuric acid to form (NH4)2SO4, and then the excess NH3 reacts with other gaseous acidic species (i.e., HNO3 and HCl) to form NH4NO3 and NH4Cl [69]. Sung et al. [18] Figure S6), again suggesting NH4NO3 formation during the SIA formation. NH4NO3 is a semi-volatile species; thus, it can exist in different phase states depending on the temperature and humidity [69]. As shown in Figure 5c and Figure 7, a high NHR (>0.3) was found under NH4 + -rich, low temperature (7.9 ± 7.6 °C) and high RH (71.7 ± 7.0%) conditions, which are the conditions of higher deliquescence RH of NH4NO3 [71]. These data indicate that in the study site, NH4NO3 was likely present in mainly the aqueous phase. NH 3 is considered to be neutralized by sulfuric acid to form (NH 4 ) 2 SO 4 , and then the excess NH 3 reacts with other gaseous acidic species (i.e., HNO 3 and HCl) to form NH 4 Figure S6), again suggesting NH 4 NO 3 formation during the SIA formation. NH 4 NO 3 is a semi-volatile species; thus, it can exist in different phase states depending on the temperature and humidity [69]. As shown in Figures 5c and 7, a high NHR (>0.3) was found under NH 4 + -rich, low temperature (7.9 ± 7.6 • C) and high RH (71.7 ± 7.0%) conditions, which are the conditions of higher deliquescence RH of NH 4 NO 3 [71]. These data indicate that in the study site, NH 4 NO 3 was likely present in mainly the aqueous phase.

Origin of Total NH3
To examine the origin of NH3 during the measurement period, we used air quality simulation with the photochemical model. The simulated NH3 and NH4 + were evaluated with the observations (Figures S7-S9 and 2). The simulated NH4 + concentrations agreed well with the observations in the urban site. Moreover, the simulated NH3 concentrations were overestimated by 2-6 ppb for spring, autumn, and winter in the site, which can be attributable to the uncertainty in the NH3 emissions inventory [72]. Figure 8 shows the monthly ZOC of Chinese NH3 emissions in Northeast Asia. The ZOC averaged over China was as high as ~5.3 ppb. For South Korea, however, the ZOC was as low as ~0.5 ppb, except during spring, when NH3 emissions increased due to agricultural activities. It is known that transboundary transport of air pollutants from China to South Korea increases during spring compared to the other seasons [73,74]. However, the calculations yielded an NH3 concentration of just ~2 ppb, which is significantly lower than the measured value during spring (~12 ppb) (Figure 3). This suggests that domestic influences remain strong even during the spring. Figure 9 shows the simulated monthly NH3 concentrations and the relative contributions of NH3 emissions released from China and South Korea, respectively, in Samcheondong, South Korea. NHx was also added because NH3 can be converted into NH4 + during the long-range transport. During the study period, the relative NH3 contributions from South Korea were dominant, ranging from 88% to 99%, despite the uncertainties that still existed in the simulation results associated with the input emissions and meteorology data. This is because most NH3 originating from China is converted into NH4 + after the long-range transport, considering the short residence time of NH3 in the atmosphere (one day or less) [75,76]. Although the simulations overestimated NH3 concentrations in Samcheon-dong ( Figure S8), they clearly confirmed that most NH3 originated from South Korea, rather than China, during the measurement period.

Origin of Total NH 3
To examine the origin of NH 3 during the measurement period, we used air quality simulation with the photochemical model. The simulated NH 3 and NH 4 + were evaluated with the observations (Figures S7-S9 and Figure 2). The simulated NH 4 + concentrations agreed well with the observations in the urban site. Moreover, the simulated NH 3 concentrations were overestimated by 2-6 ppb for spring, autumn, and winter in the site, which can be attributable to the uncertainty in the NH 3 emissions inventory [72]. Figure 8 shows the monthly ZOC of Chinese NH 3 emissions in Northeast Asia. The ZOC averaged over China was as high as~5.3 ppb. For South Korea, however, the ZOC was as low as~0.5 ppb, except during spring, when NH 3 emissions increased due to agricultural activities. It is known that transboundary transport of air pollutants from China to South Korea increases during spring compared to the other seasons [73,74]. However, the calculations yielded an NH 3 concentration of just~2 ppb, which is significantly lower than the measured value during spring (~12 ppb) (Figure 3). This suggests that domestic influences remain strong even during the spring. Figure 9 shows the simulated monthly NH 3 concentrations and the relative contributions of NH 3 emissions released from China and South Korea, respectively, in Samcheondong, South Korea. NH x was also added because NH 3 can be converted into NH 4 + during the long-range transport. During the study period, the relative NH 3 contributions from South Korea were dominant, ranging from 88% to 99%, despite the uncertainties that still existed in the simulation results associated with the input emissions and meteorology data. This is because most NH 3 originating from China is converted into NH 4 + after the long-range transport, considering the short residence time of NH 3 in the atmosphere (one day or less) [75,76]. Although the simulations overestimated NH 3 concentrations in Samcheon-dong ( Figure S8), they clearly confirmed that most NH 3 originated from South Korea, rather than China, during the measurement period.

Conclusions
In this study, we measured the concentrations of NH 3 , PM 2.5 , and its water-soluble SNA to determine the effect of NH 3 on PM 2.5 pollution at an urban area, Jeonju, South Korea from May 2019 to April 2020. During the entire period, the hourly average concentration of atmospheric NH 3 was 10.5 ± 4.8 ppb and the daily average concentration of PM 2.5 was 24.0 ± 12.8 µg/m 3 with 4.4 ± 4.9 µg/m 3 for NO 3 − , 4.3 ± 3.1 µg/m 3 for SO 4 2− , and 1.6 ± 1.8 µg/m 3 for NH 4 + . Seasonal variations showed that the atmospheric NH 3 was enhanced in summer, while the PM 2.5 was increased in winter at the monitoring site. Further, when the level of atmospheric NH 3 enhanced, the concentration showed a late afternoon peak due to the influence of nearby rural areas by agricultural activities. This was evident in spring and summer; on the other hand, in winter, the two peaked during high traffic times.
During PM 2.5 pollution episodes (daily PM 2.5 average ≥ 25 µg/m 3 ), we observed a remarkable increase in the fraction of NH 4 + and NO 3 − in PM 2.5 . In addition, the daily average NHR increased dramatically to 0.24 (with a maximum ratio of~0.61 in January) when high PM 2.5 concentration was observed. This was comparable to the result of the NHR-value of 0.06 for clean days (PM 2.5 < 25 µg/m 3 ). We also observed an inversely proportional correlation between the NHR and NH 3 , and a strong positive exponential correlation between the NHR and PM 2.5 and SNA, suggesting that NH 3 contributed significantly to SNA formation by gas-to-particle conversion. To explore the origin of the NH 3 at the monitoring site, we performed three-dimensional photochemical models using CMAQ and BFM. The modeling results showed that most of the NH 3 originated from South Korea, rather than China, during the studied period. The simulations proved that most NH 3 originated from South Korea, rather than China, during the measurement period. Overall, our results provided an in-depth understanding of the chemistry and origin of PM precursors and aerosol pollution in the atmosphere. This knowledge can further contribute to the development of effective air quality improvement strategies, such as regulation policies for air pollutants.
Supplementary Materials: The following are available online at https://www.mdpi.com/article/ 10.3390/atmos12121676/s1, Figure S1: Calibration of the NH 3 analyzer using a diluted standard gas mixture of NH 3 and N 2 . Figure S2: A result of polar plot during the measuring period in Jeonju. Figure S3: Relationship between hourly NH 3 concentrations and ambient temperature at the Samcheon-dong monitoring station during May 2019 to April 2020. Figure S4: Hourly averaged concentration of NH 3 versus the precipitation per hour during the measuring period (temperature > 30 • C). Figure S5: Monthly variation in (a) temperature and the ratio of NH 4 + to total NH 3 (NHR) and (b) concentrations of NH 3 , SNA, and PM 2.5 at the monitoring site in January 2020. Figure S6: NO 3 − to SO 4 2− molar ratio and NH 4 + to SO 4 2− molar ratio used for the analysis of NH 4 + conditions during polluted periods. Figure S7: Cluster analysis for 72 h backward trajectories at the monitoring site from May 2019 to April 2020. Figure S8: Seasonally observed (left) and simulated (right) ammonia (NH 3 ) concentrations in Samcheon-dong. Figure S9: Time series for the observed and simulated NH 4 + concentrations in the six supersites (Baengnyeong, Seoul, Daejeon, Gwangju, Jeju, and Ulsan) in South Korea from May 2019 to April 2020. Table S1: Monthly average concentrations of PM 2.5 and its ionic species, and NH 3 , and meteorological conditions in Jeonju from May 2019 to April 2020.  Institutional Review Board Statement: Not applicable.

Informed Consent Statement: Not applicable.
Data Availability Statement: The publicly available Hybrid Single-Particle Lagrangian Integrated Trajectory (HYSPLIT) model can be found at https://www.ready.noaa.gov (accessed on 1 May 2021) and run either online or offline. The data can be found data from the link: ftp://arlftp.arlhq.noaa.gov/ pub/archives/gdas1/ (accessed on 1 May 2021). In this study, the PM 2.5 mass concentration analysis method used the methodology provided by EPA, which can be found from the link: https://www. epa.gov/amtic/compendium-methods-determination-inorganic-compounds-ambient-air (accessed on 15 November 2021).