Atmospheric Behaviour of Polycyclic and Nitro-Polycyclic Aromatic Hydrocarbons and Water-Soluble Inorganic Ions in Winter in Kirishima, a Typical Japanese Commercial City

Kirishima is a typical Japanese commercial city, famous for frequent volcanic activity. This is the first study to determine the characteristics of PM2.5-bound polycyclic and nitro-polycyclic aromatic hydrocarbons (PAHs and NPAHs) and water-soluble inorganic ions (WSIIs) in this city. In this study, the non-volcanic eruption period was taken as the target and daily PM2.5 samples were collected from 24 November to 21 December 2016. The daily concentrations in PM2.5 of ƩPAHs, ƩNPAHs, and ƩWSIIs ranged from 0.36 to 2.90 ng/m3, 2.12 to 22.3 pg/m3, and 1.96 to 11.4 μg/m3, respectively. Through the results of the diagnostic ratio analyses of the PAHs, NPAHs, and WSIIs and the backward trajectory analysis of the air masses arriving in Kirishima, the emission sources of PAHs, NPAHs, and WSIIs in PM2.5 in Kirishima were influenced by the coal burning that came from the East Asian continent, although there was no influence from volcanic emission sources during the sampling period. The total benzo[a]pyrene (BaP)-equivalent concentration was lower than many other cities but the health risks in Kirishima were nonetheless notable. These findings are very important for future research on PM samples during the inactive Asian monsoon and volcanic eruption periods, to further understand the characteristics of air pollutants in Kirishima, and to contribute to the improvement in health of residents and a reduction in the atmospheric circulation of air pollutants in East Asia.


Introduction
Many studies have highlighted that atmospheric particulate matter (PM) is closely related to human health [1][2][3]. The potential toxicity of PM depends on its size and chemical composition [1, 4,5]. Regarding the size of PM, fine particles (PM 2.5 ) penetrate the respiratory system more easily than coarse particles and become deposited deeply in the alveoli, causing various diseases in humans [6,7]. Regarding the chemical composition of PM, polycyclic aromatic hydrocarbons (PAHs) and their nitrated derivatives (NPAHs) are

PM 2.5 Sampling
As shown in Figure 1, PM 2.5 sampling was performed at the National Institute of Technology, Kagoshima College (31 • 43 48 N, 130 • 43 12 E), which is located in Kirishima city, Kagoshima Prefecture, Japan. The air sampler was set up on the rooftop of a building 5 m above the ground and approximately 500 m from the road. Daily PM 2.5 samples were collected from 24 November to 21 December 2016, using a high-volume air sampler (HV-1000F, Sibata Scientific Technology Ltd., Sibata, Japan) equipped with quartz fiber filters (2500QAT-UP, 8 × 10 inches, Pall Co., Port Washington, NY, USA) at a flow rate of 1000 L/min. The sampling started at 16:30 and continued for 24 h, and the filters were changed the next day at 16:30. After sampling, PM 2.5 sample filters were packaged in aluminum foil, sealed in plastic bags, and stored at −25 • C until experimental analysis. the next day at 16:30. After sampling, PM2.5 sample filters were packaged in aluminum foil, sealed in plastic bags, and stored at −25 °C until experimental analysis.

PAH, NPAH, and WSII Analyses and Chemical Compositions
Descriptions of the PAH and NPAH pretreatment methods can be found in our previous study [29]. The PM2.5 samples were cut into small pieces and combined with internal standards (pyrene-d10 (Pyr-d10), benzo[a]pyrene-d12 (BaP-d12), and 2-fluoro-7-nitrofluorene (FNF)) in flasks. After two ultrasonic extractions by adding benzene:ethanol (3:1, v/v), the extracted solution was washed successively with sodium hydroxide solution (5%; w/v), sulfuric acid solution (20%; v/v), and water. Then, the solution was concentrated to 100 μL, and ethanol was added to bring the concentrated residue up to 1 mL. After pretreatment, nine PAHs and three NPAHs were separately detected by a high-performance liquid chromatography (HPLC) fluorescence detection system [30]. The WSIIs in PM2.5 samples were ultrasonically extracted in ultrapure water. The extracted solution was divided into two parts, and eight WSIIs, including five cations and three anions, were separately detected by the ion chromatography (IC) system [14].
PAH, NPAH, and WSII standard solutions were injected into the HPLC and IC systems to check the HPLC and IC analysis methods before actual sample detection. PAH, NPAH, and WSII standard solutions with different concentration gradients were tested at least three times. The relative standard deviations of all species were within 5%. The calibration curves of all PAHs, NPAHs, and WSIIs exhibited good linearity (r > 0.998). Three blank filters were also analysed under the same pretreatment as the samples. The analyses of the blank filters showed that none of the target chemicals were detected, indicating that there was no background contamination during the transport process. The recoveries of the internal standards were used to calibrate the concentrations and to determine whether errors occurred during the experiment. Pyr-d10 was used for the 4-ring PAHs (fluoranthene (FR), pyrene (Pyr), benz[a]anthracene (BaA), and chrysene (Chr)), and BaP-d12 was used for the 5-ring PAHs (benzo[b]fluoranthene (BbF), benzo[k]fluoranthene (BkF), and benzo[a]pyrene (BaP)) and 6-ring PAHs (benzo[ghi]perylene (BgPe) and indeno[1,2,3cd]pyrene (IDP)). FNF was used for the three NPAHs (1-NP, 2-NP, and 2-NFR). The recoveries of internal standards in all samples in this study were within 20%. Table 1 shows the name and abbreviation of the target species of PAHs, NPAHs, and WSIIs.

PAH, NPAH, and WSII Analyses and Chemical Compositions
Descriptions of the PAH and NPAH pretreatment methods can be found in our previous study [29]. The PM 2.5 samples were cut into small pieces and combined with internal standards (pyrene-d 10 (Pyr-d 10 ), benzo[a]pyrene-d 12 (BaP-d 12 ), and 2-fluoro-7nitrofluorene (FNF)) in flasks. After two ultrasonic extractions by adding benzene:ethanol (3:1, v/v), the extracted solution was washed successively with sodium hydroxide solution (5%; w/v), sulfuric acid solution (20%; v/v), and water. Then, the solution was concentrated to 100 µL, and ethanol was added to bring the concentrated residue up to 1 mL. After pretreatment, nine PAHs and three NPAHs were separately detected by a high-performance liquid chromatography (HPLC) fluorescence detection system [30]. The WSIIs in PM 2.5 samples were ultrasonically extracted in ultrapure water. The extracted solution was divided into two parts, and eight WSIIs, including five cations and three anions, were separately detected by the ion chromatography (IC) system [14].
PAH, NPAH, and WSII standard solutions were injected into the HPLC and IC systems to check the HPLC and IC analysis methods before actual sample detection. PAH, NPAH, and WSII standard solutions with different concentration gradients were tested at least three times. The relative standard deviations of all species were within 5%. The calibration curves of all PAHs, NPAHs, and WSIIs exhibited good linearity (r > 0.998). Three blank filters were also analysed under the same pretreatment as the samples. The analyses of the blank filters showed that none of the target chemicals were detected, indicating that there was no background contamination during the transport process. The recoveries of the internal standards were used to calibrate the concentrations and to determine whether errors occurred during the experiment. Pyr-d 10 was used for the 4-ring PAHs (fluoranthene (FR), pyrene (Pyr), benz[a]anthracene (BaA), and chrysene (Chr)), and BaP-d 12 was used for the 5-ring PAHs (benzo[b]fluoranthene (BbF), benzo[k]fluoranthene (BkF), and benzo[a]pyrene (BaP)) and 6-ring PAHs (benzo[ghi]perylene (BgPe) and indeno[1,2,3-cd]pyrene (IDP)). FNF was used for the three NPAHs (1-NP, 2-NP, and 2-NFR). The recoveries of internal standards in all samples in this study were within 20%. Table 1 shows the name and abbreviation of the target species of PAHs, NPAHs, and WSIIs. Table S1 (Supplementary Materials) lists the limit of determination (LOD) of each PAH, NPAH, and WSII species. In this study, the PAH standard solution (US EPA 610 PAH mix) was purchased from Supelco Park (Bellefonte, PA, USA); 1-nitropyrene (1-NP), 2-NP, and FNF were purchased from Aldrich Chemical Company (Osaka, Japan); 2-NFR was purchased from Chiron AS (Trondheim, Norway); and Pyr-d 10 , BaP-d 12 , and WSII standard solutions were purchased from Wako Pure Chemicals (Osaka, Japan). All other analytical reagent-grade reagents used in the HPLC and IC analyses were purchased from Wako Pure Chemicals (Osaka, Japan).

Meteorological Conditions
Meteorological data, including the average temperature, precipitation, relative humidity, sunshine hours, wind speed, and prevailing wind direction in Kirishima during the sampling period shown in Table S2 (Supplementary Materials), were obtained from the Japan Meteorological Agency (http://www.jma.go.jp/jma/menu/menureport.html).

Cation Equivalent (CE), Anion Equivalent (AE), and Non-Sea Salt (nss-) WSIIs
The WSII acid-base balance at Kirishima during the sampling period was analysed by using CE and AE, which were calculated as follows [

Backward Trajectory
The air masses that flowed into Kirishima during the sampling period were analysed by their backward trajectories, which were downloaded from the U.S. National Oceanic and Atmospheric Association's HYSPLIT4 model (WINDOWS-based). In this study, each backward trajectory was calculated every hour at a sampling point height of 500 m above ground level, with a tracking time of 72 h. All backward trajectories during the sampling period were classified into four clusters according to their characteristics.

Health Risk Assessment
The BaP-equivalent (BaP eq ) concentrations were expressed as follows: where C i is the concentration of each species (pg/m 3 ) and TEF i is the toxic equivalency factor of each species relative to BaP, obtained by referring to previous studies [8,9,32]. The inhalation lifetime cancer risk (ILCR) from exposure to PAHs and NPAHs was expressed as follows: where UR BaP is the unit cancer risk from BaP, which was assigned a value of 8.7 × 10 −5 per ng/m 3 in this study [33].

Statistical Analysis
Statistical analysis of the data was performed using IMB SPSS version 25.0. Spearman correlation analysis was used to determine the correlation between PAHs, NPAHs, and WSIIs. Differences in the results were considered significant at p values of less than 0.05 or 0.01. Table 1 summarizes the concentrations of PAHs, NPAHs, and WSIIs in Kirishima during the sampling period. The daily concentrations of ΣPAHs ranged from 0.36 to 2.90 ng/m 3 , with an average of 1.32 ± 0.71 ng/m 3 ; this level is comparable to those from other Japanese commercial cities such as Sapporo (1.79 ng/m 3 ) and Sagamihara (1.83 ng/m 3 ) in winter 2013 [34] and Kanazawa (1.00 ng/m 3 ) in winter 2018 [35], but lower than those from other Asian cities such as Shanghai, China in winter 2018 (7. The daily concentrations of ΣNPAHs ranged from 2.12 to 22.3 pg/m 3 , with an average of 9.98 ± 5.75 pg/m 3 . The concentration level of ΣNPAHs was much lower than that of ΣPAHs in this study, which is also consistent with results from the urban cities listed above [30,[34][35][36].

Concentrations of PAHs, NPAHs, and WSIIs
The daily concentrations of ΣWSIIs ranged from 1.96 to 11.4 µg/m 3 , with an average of 5.74 ± 2.59 µg/m 3 ; this level is slightly lower than those of other cities of the same type in Japan, such as Osaka in 2015 (8.1 µg/m 3 ) [39] and Yokohama from 1999 to 2005 (9.83 µg/m 3 ) [40], and much lower than those of other Asian cities, such as Zhengzhou  [43]. The Spearman correlation analysis showed that there were strong positive correlations among PAHs, NPAHs, and WSIIs (p < 0.01), indicating that there were some internal connections between these species, although the main sources of emissions of these species were not entirely the same.

Composition of PAHs, NPAHs, and WSIIs
As shown in Table 1, FR had the highest average concentration (0.31 ± 0.20 ng/m 3 ) during the sampling period. The average concentrations of Pyr, BbF, BgPe, and IDP ranged from 0.16 to 0.18 ng/m 3 , higher than those of the other PAHs. The average proportions of 4-, 5-and 6-ring PAHs during the sampling period accounted for approximately 47.0%, 25.7%, and 27.4% of the ΣPAHs, respectively. Four-ring PAHs that originated mainly from coal and biomass burning made up a relatively large proportion of the ΣPAHs, which is consistent with other reports [44][45][46]. This phenomenon occurs because 4-ring PAHs can be transferred from the gaseous phase to the particle phase easily at low ambient temperatures in winter due to vapor pressure [47] and may also be related to the emission sources that will be discussed in Section 3.3. Among the three NPAHs, 2-NFR had the highest average concentration (7.75 ± 4.59 pg/m 3 ), making up 72% to 84% of the daily ΣNPAHs, followed by 1-NP (1.77 ± 1.02 pg/m 3 ), which constituted 11% to 21% of the daily ΣNPAHs during the sampling period (Table 1). Additionally, consistent with previous studies, the concentration of 2-NFR was higher than those of 1-NP and 2-NP; of these NPAH types, 2-NFR and 2-NP are secondarily generated [30,48,49].
Among the eight WSIIs, SO 4 2− had the highest average concentration (3.78 ± 1.77 µg/m 3 ), followed by NH 4 + (1.35 ± 0.59 µg/m 3 ). Moreover, the concentration of NO 3 − (0.28 ± 0.23 µg/m 3 ) was also higher than those of the other WSIIs (Table 1). These three species constituted at least 85% of the daily ΣWSIIs and are therefore the main WSII species of PM 2.5 , which is consistent with previous studies [24,41,50]. Because anions can increase the acidity of PM and cations can increase the alkalinity of PM, the AE/CE ratio is a good indicator for determining the acidity or alkalinity of PM [23]. As shown in Figure 2, AE/CE was 1.06, with good linearity (r = 0.97), indicating that PM 2.5 was relatively neutral at Kirishima during the sampling period. Moreover, the AE/CE value, which was close to 1, also corroborates the validity of the WSII measurements, indicating that most WSII species were quantified [51].    [52,53], the potential source of the emissions in this study was coal burning. Moreover, the [1-NP]/[Pyr] ratios, which ranged from 0.005 to 0.019 and had an average of 0.008, was also close to that of coal burning emissions [21].  [37,53,54]. Of the three NPAHs, 1-NP is primarily formed, and 2-NFR and 2-NP are secondarily formed [48]. As shown in Figure 3, the [2-NFR]/[1-NP] ratios ranged from 3.50 to 7.68, with an average of 4.51, and its values were   [52,53], the potential source of the emissions in this study was coal burning. Moreover, the [1-NP]/[Pyr] ratios, which ranged from 0.005 to 0.019 and had an average of 0.008, was also close to that of coal burning emissions [21]. However, the [ these values are between those for coal burning and traffic emissions, indicating mixed sources for the PAHs in this study [37,53,54]. Of the three NPAHs, 1-NP is primarily formed, and 2-NFR and 2-NP are secondarily formed [48]. As shown in Figure 3, the [2-NFR]/[1-NP] ratios ranged from 3.50 to 7.68, with an average of 4.51, and its values were mostly lower than 5. These values indicate the higher contribution of direct emissions such as coal burning during the sampling period [55]. The [NO 3 − ]/[SO 4 2− ] ratio is usually used to estimate the relative importance of traffic emissions and coal burning sources [42]. Table 2 shows that these ratios ranged from 0 (NO 3 − concentration was less than the LOD) to 0.32, with an average of 0.11. Ratios lower than 1.0 indicate that the emission sources were more likely related to coal burning [42]. Consequently, the diagnostic ratios of PAHs, NPAHs, and WSIIs indicated that the main sources at Kirishima during the sampling period were mixed but that coal burning made a higher contribution than traffic emissions.

Potential Emission Sources
On the other hand, Figure 3 also shows that the [2-NFR]/[2-NP] ratios ranged from 7.61 to 32.0, with an average of 18.4. Values of this ratio near 10 indicate that 2-NFR is mainly secondarily formed by the OH radical-initiated reaction rather than formed through the NO 3 radical-initiated reaction, which was similar to the results for 2-NP [55]. Moreover, some WSII species had both sea-salt sources and non-sea-salt sources. According to the calculations from Equations (3)-(5), the concentrations of [nss-SO 4 2− ], [nss-K + ], and [nss-Ca 2+ ] accounted for 90% to 99% of the total SO 4 2− , K + , and Ca 2+ , indicating that these species were mostly emitted from non-sea-salt sources. Table 3 shows the concentration ratios of these species to Na + during the sampling period. According to the reference data [56], the ratios of [SO 4 2  Table 3); both of these results suggest that sea salt had a relatively large impact as a source of Mg 2+ [56].      Figure 4 shows the source areas of the air masses that arrived at Kirishima during the sampling period determined by performing a cluster analysis of their tracked 72-h backward trajectories. Of the four clusters, clusters 2, 3, and 4 constituted approximately 74% of all the air masses and all came from the northwest direction, consistent with the prevailing wind direction at Kirishima (NNW; Table S2, Supplementary Materials). Specifically, cluster 1 contained 26% of the air masses, which came from the Sea of Japan and then moved across domestic Japan and from the Pacific Ocean to Kirishima. Clusters 2 and 3 contained 32% and 20% of the air masses, respectively, which came from different source areas in Russia and moved across both Mongolia and North China. Cluster 4 contained 22% of the air masses, which came from North China and then passed across the Yellow Sea to Kirishima. These results were consistent with previous studies showing that the air masses that arrived in Japan in the wintertime came mostly from the Asian continent [57,58]. The source areas of Mongolia and northern China, which the air masses came from or passed through, contained high concentrations of air pollutants during the sampling period because biomass burning for warmth is common in Mongolia [43] and coal burning in heating systems is common in North China [46]. In addition, Figure 4 shows that the height ranges of the air masses that came from local Japan and the ocean (cluster 1) were lower than 1000 m, which is much lower than those that came from the Asian continent, including Russia, Mongolia, and China (clusters 2, 3, and 4). These results suggest that the air pollutants originating from emission areas in Japan and the sea are very likely to sink during the long-range transport process [26] and thus are less likely to arrive at Kirishima than those originating from the Asian continent. The diagnostic ratios discussed in Section 3.3 suggested that coal burning had a larger impact than other sources and that the direct emission of NPAHs made a high contribution. These results may be because the air masses from the Asian continent arriving at Kirishima contained these species emitted from combustion sources that did not undergo substantial degradation during the long-range transport process; this phenomenon has also been reported in previous studies [26,27,48].

Backward Trajectory Analysis
According to the daily concentrations of each species shown in Tables S3 and S4 Table S5 (Supplementary Materials), the main source areas for the air masses on these two days were both domestic across the Pacific Ocean and Kirishima, and the height ranges of the air masses were both lower than 500 m; these air masses were in cluster 1. Previous studies have reported that air masses coming from or passing through the ocean contain a relatively low concentration of air pollutants because the sea has a diluting effect on air pollutants [27,29]. Moreover, the meteorological conditions shown in Table S2 (NNE), suggesting that the ground source areas were different. This difference may have led to the difference in WSII concentrations between these two days because WSIIs can not only be emitted from combustion sources but can also come from non-combustion sources such as road dust [31,56].
burning in heating systems is common in North China [46]. In addition, Figure 4 shows that the height ranges of the air masses that came from local Japan and the ocean (cluster 1) were lower than 1000 m, which is much lower than those that came from the Asian continent, including Russia, Mongolia, and China (clusters 2, 3, and 4). These results suggest that the air pollutants originating from emission areas in Japan and the sea are very likely to sink during the long-range transport process [26] and thus are less likely to arrive at Kirishima than those originating from the Asian continent. The diagnostic ratios discussed in Section 3.3 suggested that coal burning had a larger impact than other sources and that the direct emission of NPAHs made a high contribution. These results may be because the air masses from the Asian continent arriving at Kirishima contained these species emitted from combustion sources that did not undergo substantial degradation during the long-range transport process; this phenomenon has also been reported in previous studies [26,27,48].  Table S5 (Supplementary Materials), the main source areas for the air masses on these two days were both domestic across the Pacific Ocean and Kirishima, and the height ranges of the air masses were both lower than 500 m; these air masses were in cluster 1. Previous studies As shown in Table 1, the median concentrations of most PAHs, NPAHs, and WSIIs were lower than their average concentrations. In particular, the median concentration of ∑PAHs was 18.9% lower than the average level, suggesting that high concentrations had relatively large impacts on the total concentration during the whole period. According to the daily concentrations shown in Table S3 (Supplementary Materials), there were nine days, including 25 and 27 to 29 November and 1, 5, 6, 8, and 9 December 2016, on which the concentrations of ∑PAHs were higher than the average level (1.32 ng/m 3 ); the daily concentrations of NPAHs on those nine days were also higher than the average level (9.98 pg/m 3 ). In addition, there were six days among these nine days in which the daily concentrations of ∑WSIIs were higher than the average level (5.74 µg/m 3 ). Table S5 (Supplementary Materials) shows that all air masses on these days came from the Asian continent, with high height ranges, except those on 29 November and 8 December 2016, which had relatively low heights. This suggests that the air masses arriving at Kirishima contained relatively high concentrations of air pollutants [21]. Therefore, the impact of air masses from the Asian continent in winter on Kirishima cannot be ignored, although not all air masses from the Asian continent showed high concentrations of PAHs, NPAHs, and WSIIs. Table 4 summarizes the BaP eq concentrations of nine PAHs, 1-NP, and 2-NFR (2-NP had no available TEF value) at Kirishima during the sampling period. The ΣBaP eq concentrations ranged from 31.2 to 302 pg/m 3 , with an average of 142 pg/m 3 , and the nine PAHs contributed mostly to ΣBaP eq concentrations. In addition to BaP (90.9 pg/m 3 ), BbF (18.7 pg/m 3 ) and IDP (16.3 pg/m 3 ) had the highest BaP eq concentrations, indicating that they represented higher health risks than the other species. On the other hand, although the 1-NP and 2-NFR concentrations were much lower than those of the nine PAHs (Table 1), the BaP eq concentrations of 1-NP (0.18 pg/m 3 ) and 2-NFR (0.39 pg/m 3 ) were comparable to those of FR (0.31 pg/m 3 ) and Pyr (0.18 pg/m 3 ) because the TEF values of 1-NP and 2-NFR were higher than those of FR and Pyr (Table 4) [8,9,32]. The ILCR at Kirishima during the sampling period was 1.22 × 10 −5 , indicating that approximately 12 cancer cases may occur among 10 6 people due to PAHs and NPAHs exposure. The ILCR in this study was much lower than that in other Asian cities that used the same UR BaP value [30,37,60]. However, it was one order of magnitude over the acceptable level established by the US EPA (10 −6 ), indicating that exposure to PAHs and NPAHs at the levels observed in this study has adverse effects on human health. On the other hand, the UR BaP value used to calculate the ILCR in this study was obtained from an epidemiological study of coke oven workers whose ILCR was very high [33]; this created some uncertainty in determining the risk of exposure to PAHs and NPAHs for the non-professional population.

Conclusions
The characteristics of PM 2.5 -bound PAHs, NPAHs, and WSIIs in Kirishima, Japan, in an urban context without specific natural activity were investigated in this study. The concentrations of PM 2.5 -bound PAHs, NPAHs, and WSIIs at Kirishima were comparable to those in other Japanese cities and lower than those in many other Asian cities in a similar period. Meteorological conditions such as precipitation can have a strong impact on the concentrations of air pollutants. The air masses that arrived at Kirishima came mostly from the Asian continent and may have contained high levels of air pollutants emitted from coal burning. Sea salt had a larger impact on Mg 2+ than on other WSII species. Moreover, PAHs contributed the majority of the ΣBaP eq concentration, and the main contributors to the ΣBaP eq concentration in this study were BaP, BbF, and IDP; however, the health risks of NPAHs could not be ignored.
Kirishima is one of the typical commercial cities in the Kyushu area, Japan. The emission source of PAHs was not complex, and the urban background concentration was low. This study found that the air pollutants in Kirishima were also influenced by the air masses long-range transported from the East Asian continent during the East Asian winter monsoon period, similar to the region of the Sea of Japan. This finding is not only a reminder for the operation of environmental protection policies, but it can also be a reminder for other similar areas. Although the PM 2.5 samples were only collected at one site, this study is the first to determine the atmospheric behaviour of PAHs, NPAHs, and WSIIs in PM 2.5 in Kirishima, leading to a basic understanding. In the future research, we need to collect PM 2.5 samples simultaneously at different sites in Kirishima in other seasons and specific periods of natural activity such as volcanic eruptions, and to analyse other pollutants such as gaseous pollutants, to further determine their atmospheric behaviours in Kirishima.