Source Quantification of PM10 and PM2.5 Using Iron Tracer Mass Balance in a Seoul Subway Station, South Korea

In this study, we simultaneously measured the PM10 and PM2.5 mass concentrations and their heavy metal content for three days at a subway station in Seoul to investigate the airborne PM flows. The average concentrations were 59 μg m–3, 37 μg m–3, 111 μg m–3, and 369 μg m–3 for the PM10 and 43 μg m–3, 28 μg m–3, 58 μg m–3, and 132 μg m–3 for the PM2.5 at the outdoor air inlet, in the concourse, on the platform, and in the tunnel, respectively. We also found strong correlations between the temporal variations at adjacent sampling locations for both fractions, although they were higher for the PM2.5. Additionally, of the airborne trace metals detected at the sampling locations inside the station (the concourse, platform, and tunnel), iron (Fe) displayed the highest concentration and was thus selected as a tracer of PM. Applying a simple mass balance model to the Fe concentrations and ventilation rates revealed that 78% of the PM10 and 62% of the PM2.5 on the platform emanated from the tunnel, whereas 84% of the PM10 and 87% of the PM2.5 in the concourse originated outdoors (and arrived in the filtered air). These results further confirm that reducing PM emission from the tunnel is the most effective strategy for improving air quality on the platform and achieving compliance with the national guideline.

The subway is a major public transportation in most megacities throughout the world. For 36 convenience, the subway is usually located in high-density traffic areas with large numbers of 37 pedestrians. People in subway are more prone to be exposed harmful levels of air pollutants if 38 indoor air quality in subway system is not properly managed. Shen and Gao (2019) investigated a 39 personal exposure to PM during four transportations (subway, bicycle, bus and walking) 40 commuting in Nanjing and found that passengers in subway station are exposed to highest PM2.5. 41 The air quality of subway station is largely dependent to characteristics of location and space.

A C C E P T E D M A N U S C R I P T
4 As shown in Fig. 1(b), we collected samples in four major sectors of the station: outdoor, 83 concourse, platform, and tunnel. The outdoor sampling point was located at the outdoor air inlet 84 of the air filtration system to insure the same characteristics of the inflow air into the subway 85 station. In concourse and platform, measurements were taken in the middle sections and at 1.5 m 86 from the ground. Tunnel PM inflowing to the platform was sampled at the rear end of the 87 platform. 88 PM2.5 and PM10 were sampled at 6 L min -1 and 5 L min -1 , respectively, using a portable PM 89 sampler (minivol portable air sampler, Airmetrics, USA). The sampling filter was weighed by the 90 auto-weighing system (Chabal-500, C2K Creative, Korea) after filter conditioning for 24 hr 91 (temperature 20 ± 2 ℃, relative humidity 35 ± 5%). 92 As shown in Fig. 1(b), filtrated outdoor air was supplied with rate of 2,384 m 3 min -1 in the 93 platform and 1,634 m 3 min -1 in concourse. This air filtration system was operated continuously 94 during train operation hours (05:30 -24:40). The indoor volume of the platform and concourse 95 was 7,193 m 3 and 11,853 m 3 , respectively. The filtrated air was designed to refresh the concourse 96 with 20 times hr -1 and platform with 8 times hr -1 . The air filtration device efficiency of PM10 and 97 PM2.5 were found to be 37% and 35% considering to reduction of PM concentration in filtrated 98 airs. This efficiencies were consistent with other studies for PM10 of 30 -60%, and PM2.5 of 20 -99 40% . 100

PM concentration and Metal content analysis 102 103
We measured the real-time PM concentration using a light scattering analyzer (model 1.180, 104

A C C E P T E D M A N U S C R I P T
5 measurements were corrected by the gravimetric method in daily basis. 108 As shown in Fig. 2, PM2.5 and PM10 sampled indoor and outdoor in the Subway station were 109 extracted using microwaves (QWAVE2000, Questron Tech. Corp., Canada) with a 10 ml acid 110 solution (16.7% HCl + 2.5% HNO3). The extract was filtered using a Teflon syringe filter (0.45 111 um) and mass up to 25 mL. The pretreatment solution was analyzed using ICP-OES (Ciros vision, 112 Specctro, Germany) to determine the heavy metal content in the PM. Target  platform and tunnel. In concourse, concentration of PM10 and PM2.5 were 37% and 35% lower 124 than in outdoor. However, in platform, the PM10 and PM2.5 concentrations were higher by 88% 125 and 35% than the outdoor, despite with larger the filtered air supply. This occurred because of the 126 significant PM entrainment from the tunnel even with screen installed. The concentrations of 127 PM10 and PM2.5 in tunnel were more than 3 times higher than those in platform, which indicated 128 that considerable portions of PM prevented from entering through platform from tunnel loaded 129 with heavy PM. 130 The ratios of PM2.5/PM10 on measurement sites in subway station are shown in table 1. The ratio

A C C E P T E D M A N U S C R I P T
6 Qioa et al., 2015). In concourse, level of PM2.5/PM10 concentration ratio was similar to those of 133 outdoor due to low influence from tunnel PM. As the platform air was affected by both outdoor 134 and tunnel, the PM2.5/PM10 concentration ratio in platform was placed half-value between them. 135 136 Temporal correlation of PM among subway sampling locations 137 138 Table 2 shows the temporal correlations of PM10 and PM2.5 among sampling sites, namely 139 outdoor, concourse, platform and tunnel. All correlation coefficients of PM between sampling 140 locations were very large (r > 0.8). Particularly, correlations between adjacent locations were 141 close to 1 with exception between platform and tunnel, which indicated active air and aerosol 142 exchanges between different subway locations. Noticeably, the correlations of PM2.5 between 143 different sampling locations was slightly higher than that of PM10, which may indicate PM2.5 144 penetration rate was likely higher during the air transfer because of its smaller deposition rate. 145 Therefore, PM2.5 would be more suitable to trace its mass budget than PM10 with safely ignoring 146 its sinks. 147 148 149  150  Table 3 shows the heavy metal content rate in PMs, Fe is the most abundant which were 48% of 151 PM 10 and 44% of PM 2.5 in the tunnel. The Fe content of PM 10 was higher than that of PM 2.5 , 152 which was consistent with the characteristics of relatively higher coarse PM concentration in 153 tunnel. The Fe and Cu contents of PM decreased rapidly in the platform, concourse, and outdoor 154 sampling locations as they were generated in the tunnel, inflowing to the platform and then to the 155 concourse. The ratios of Cu and Fe contents in PM were approximately ~ 0.03 in regardless of 156 the measurement points, which confirmed that they were originated from common source. To simply this model in estimating four unknown airflow rates (Qpc, Qtp, Qc_out, Qp_out), the 177 following assumptions were made. First, aerosol from outdoor and railway tunnel are only 178 sources of PMs and Fe. This assumption is particularly true for Fe, and therefore Fe was used 179 specifically as a tracer for this mass balance models. Second, airflow rates between sectors and 180 aerosol concentrations are steady state throughout the observation period. The aerosol and Fe

A C C E P T E D M A N U S C R I P T
8 concentrations of each sector was safely assumed to be a steady state because the residence times 182 of air in concourse and platform are short enough (less than 7 minutes) to achieve steady state. 183 Aerosol mass balance and air flow balance within concourse were determined as Eq. (1) and Eq. 184 (2), respectively. Also, corresponding balances within platform were determined as Eq. (3)  Co・(1-ɳ)・Qfoc + Cp・Qpc = Cc・Qc_out (1) 191 Employing Fe concentrations in PM2.5 and known η, Qfoc and Qfop, four unknown flow rates 197 (Qpc, Qtp, Qc_out and Qp_out) were calculated and listed in Table 4. The result shows that the 94% 198 (1,634 m 3 min -1 ) of concourse air was originated from ventilation system with outdoor air 199 filtration device and the rest 6% (102 m 3 min -1 ) was originated from platform air. Although the 200 airflow into concourse from platform was very limited, aerosol mass contribution to concourse 201 from platform were relatively high with 16% (0.7 of total 4.3 g hr -1 ) for PM10 and 13% (0.4 of 202 total 3.1 g hr -1 ) for PM2.5 due to their higher concentrations in platforms than outdoor air. In case 203 of platform, contributions of airflow rate from ventilation system was lower (74%) than those in 204 concourse although its flowrate of 2,384 m 3 min -1 was 46 % higher. Increased airflow rate from 205 tunnel with very high PM concentrations, aerosol mass contribution rates from tunnel to platform 206 was very high for 78% (18.4 of total 23.7 g hr -1 ) for PM10 and 62% (6.6 of total 10.6 g hr -1 ) for

Effective reduction of subway PM concentrations 228
Using the mass balance equations earlier stated, we could assess how to effectively reduce the 229

A C C E P T E D M A N U S C R I P T
system. Especially, the PM reduction in concourse was steeply linear to PM filtration efficiency 233 by ventilation system. If the filtration efficiency is improved from current 35% to 70%, the PM10 234 and PM2.5 levels in concourse are reduced by 45% -50%, which is quite significant. However, 235 the PM10 and PM2.5 in platform is to reduce only 12% and 20%, respectively along with the same 236 amount of filtration efficiency improvement. The obvious reason was that air in platform was 237 more influenced by highly loaded aerosols from tunnel airs. If the tunnel PM concentration is 238 reduced by half, platform PM10 and PM2.5 concentrations are reduced to approximately 39% and 239 31%, respectively ( Fig.6 (b)). Consequently, the effective way to reduce PM levels in platform is 240 to regulate subway tunnel aerosol sources. As full-height enclosed screen doors are installed in 241 the platform of studied subway station, platform air is only intermittently exposed from tunnel The model shows that 94% of concourse air mass was originated from filtered outdoor air and 272 the rest 6% was originated from platform air. In case of platform, outdoor air contribution was 273 relatively low (74%) compared to concourse, although outdoor ventilation rate to platform was 274 46% higher than those to concourse. Compared with air mass contributions, 78% of PM10 and 275 62% of PM2.5 in platform were contributed from tunnel sources, while 84% of PM10 and 87% of 276 PM2.5 in concourse were originated from outdoor airs. Although indoor PM level was below 277 Korean indoor air quality guideline level, its concentrations in this studied subway platform 278 exceeded guideline level occasionally. It required additional PM control strategy to fully lower 279

A C C E P T E D M A N U S C R I P T
12 such as extra cleaner outdoor ventilation is not so effective to reduce PMs in the platform. In this 282 analysis, we assumed steady state of flowrates and PM concentrations within an hourly time scale. 283 However, they may vary significantly as trains pass every few minutes. Also, as the sampling 284 time was limited to three days, further verification is needed for the longer sampling period. 285 Nevertheless, this mass balance approach to resolve air quality behaviors in the subway spaces 286 and would be adequate to apply the specific mitigation measures to reduce air pollutants in other 287 places and locations.

A C C E P T E D M A N U S C R I P T
22

A C C E P T E D M A N U S C R I P T
23