Evidence of haze-driven secondary production of 1 supermicrometer aerosol nitrate and sulfate in size distribution 2 data in South Korea 3 4

Abstract. This study reports measurements of size-resolved aerosol
composition at a site in Incheon along with other aerosol characteristics
for contrast between Incheon (coastal) and Seoul (inland), South Korea,
during a transboundary pollution event during the early part of an intensive
sampling period between 4 and 11 March 2019. Anthropogenic emissions were
dominant in the boundary layer over the study region between 4 and 6 March,
with much smaller contributions from dust, smoke, and sea salt. The
meteorology of this period (shallow boundary layer, enhanced humidity, and
low temperature) promoted local heterogeneous formation of secondary
inorganic and organic species, including high nitrate (NO3-)
relative to sulfate (SO42-). Seoul exhibited higher PM2.5
levels than Incheon, likely due to local emissions. The following findings
point to secondary aerosol formation and growth sensitivity to water vapor
during this pollution event: (i) significant concentrations of individual
inorganic and organic acids in the supermicrometer range relative to their
full size range (∼40 %) at higher humidity; (ii) high
correlation (r=0.95) between oxalate and SO42-, a marker of
secondary aqueous production of oxalate; (iii) increased sulfur and nitrogen
oxidation ratios as a function of humidity; and (iv) matching composition
apportionment (for soluble ions) between the PM1 and PM2.5−1 size
fractions. The last finding confirms that PM1 aerosol composition
measurements fully capture PM2.5 composition apportionment (for soluble
ions) during haze events and may therefore be reliably applied in modeling
studies of such events over the full PM2.5 size range. However, the
differences evident in the periods following the haze event imply that under
other atmospheric conditions PM1 composition measurements will not
fully reflect the apportionment of PM2.5 aerosols. The study period was
marked by relatively low temperatures that made NO3- the most
abundant species detected, pointing to the sensitivity of PM2.5 levels
and composition as a function of season during such transboundary events.
For instance, other such events in previous studies exhibited more
comparable levels between SO42- and NO3- coincident with
higher temperatures than the current study. This dataset can contribute to
future evaluation of model PM2.5 composition to better support
regulatory efforts to control PM2.5 precursors.



Introduction
Seoul (Fig. S11), which allows for a critical look at how PM2.5 levels compare between the sites and if Seoul has 106 higher levels (suggesting local PM2.5 production) as described by Eck et al. (2020) and Jordan et al. (2020). This 107 study examines the role for meteorological parameters like humidity in impacting size-resolved aerosol 108 composition. This is especially important since past studies have mainly focused on bulk PM1, PM2.5 or PM10 109 (Park et  regions are labeled with individual DLPI + sets (see Table 2) overlapping in time. 139 140

Aerosol and Gas Monitoring 141
The focus of this study is on three specific monitoring sites in Incheon and Seoul, (~30 km apart), which 142 were compared to a wide network of other stations in those cities to confirm agreement in temporal variability 143 and concentrations. Incheon has a population of 2,936,367 in contrast to Seoul having 9,565,990 (Korea, 2021). 144 The three primary sites include Inha University and Sungi in Incheon, and the Seoul Intensive Monitoring Station 145 (hereafter referred to as Seoul site) in Seoul (locations in Fig. S11). Measurements at these sites are described 146 below and summarized in Table 1. Figure S11 shows the location of the other 17 and 40 National Ambient air 147 quality Monitoring Information System (NAMIS) stations in Incheon and Seoul, respectively, which were 148 operational during the study period with data provided online by AirKorea; Sungi is one of those sites in Incheon 149 but the Inha and Seoul sites are not part of that network. The rationale for including two primary sites in Incheon 150 is because Inha provided unique data not typically measured by NAMIS, and Sungi was closest to Inha among 151 the NAMIS options. Furthermore, theThe Seoul site had very comprehensive measurements itself and thus it was 152 sufficient for this study. the LOD are replaced with LOD/2, with the percent of such samples per species provided in Table S1. For anion 190 analysis, an AS11-HC 250 mm column and potassium hydroxide eluent were used with the following gradient 191 program with a suppressor (AERS 500e) current of 28 mA: begin at 2 mM, increase to 8 mM from 0 to 20 minutes, 192 and then increase from 8 to 28 mM from 20 to 30 minutes. Cation analysis involved a CS12A 250 mm column 193 and methanesulfonic acid eluent with the following program with a suppressor (CERS 500e) current of 22 mA: 194 start at 5 mM, isocratic from 0 to 13 minutes, increase from 5 to 18 mM from 13 to 16 minutes, and then isocratic 195 at 18 mM from 16 to 30 minutes. Sample concentrations were corrected using background sample concentrations 196 for individual species, which included sodium (Na + ), ammonium (NH4 + ), potassium (K + ), magnesium (Mg 2+ ), 197 calcium (Ca 2+ ), chloride (Cl -), NO3 -, SO4 2-, methanesulfonate (MSA), adipate, maleate, oxalate, and phthalate.

198
The latter five species were summed in parts of the analysis and referred to as "organic acids" with the caveat that 199 they represent the dissociation anion of either sulfonic or organic acids.

200
Concentrations for SO4 2-, Mg 2+ , K + , and Ca 2+ refer to their non-sea salt (NSS) values based on 201 calculations relying on the measured concentrations of Na + and its ratio with these species in pure sea salt (Seinfeld 202 and Pandis, 2016). The mean percentage of sea salt mass removed for those four species relative to total speciated 203 mass of each filter set (excluding the sea salt portion of those four species) was 2%. Those species for which more 204 than 15% of samples were below the LOD (i.e., Mg 2+ , maleate, phthalate, adipate, MSA) are not discussed on an 205 individual basis in this study but are used in calculations dependent on the cumulative dataset such as the overall 206 charge balance; the exception is bromide (Br -), which is fully removed from calculations as it was always below 207 the LOD.  Meteorological data for winds, temperature (T), pressure (P), relative humidity (RH), and rain were 232 used from monitoring stations in Incheon and Seoul. The Korea Meteorology Administration provided data from 233 an Automated Synoptic Observing System (ASOS) for the Incheon station (70 m ASL; 37° 28' 39.85"N, 126° 37' 234 28.40"E) (kma.go.kr), which is located ~5 km northwest of the Inha University sampling site. Meteorological data 235 were collected at the same site described for Seoul in Sect. 2.1.3. Specific humidity (q) was calculated using 236 measured values of T, P, and RH (Bolton, 1980). using a terrain-following sigma-pressure coordinate system that provides data in 1° × 1° grids every 6 hours. The and "anthropogenic and biogenic fine (ABF)" that includes secondarily produced species (i.e., SO4 2-,and SOA) 265 and primary organic aerosols, and SOA) mainly confined to the fine mode (< 1 µm). average; species with concentrations below their respective LOD in more than 50% of samples (see Table  282 S1) are not shown.   in pollution levels, which is consistent with HYSPLIT data showing back-trajectories shifting from northerly for 292 most of the clean period to northwesterly (i.e., from the Beijing area) (Fig. 23). The R 2 value between PM2.5 293 hourly data was 0.75 and 0.82 for Seoul-Inha and Seoul-Sungi, respectively. Although there was decent 294 agreement, differences are apparent in the PM2.5 hourly time series (Fig. 1c2c) with Seoul's levels being 295 significantly enhanced during parts of the polluted and transition periods. The difference in PM2.5 between Seoul 296 and Incheon suggests enhanced local production promoting large differences between hourly PM2.5 over Seoul 297 versus Incheon, with the maximum difference observed being nearly 50 µg m -3 on 5 March, with an average 298 difference that day of 37 ± 24 µg m -3 . For example, the maximum/mean ± standard deviation in the PM2.5 299 difference (µg m -3 ) between Seoul and Sungi were as follows for the three periods: polluted = 59/10 ± 26; 300 transition = 36/6 ± 28; clean = 42/-6 ± 14. 301 302  The PM10:PM2.5 ratio is helpful to examine whether a divergence in values occurs that would suggest a 314 strong source of dust as compared to typical background conditions. More specifically, higher values of this ratio 315 could potentially suggest enhanced dust influence owing to mass concentrations of dust being abundant above 2.5 316 µm. As PM10 was only measured at Seoul, Table 4 compares the mean (± standard deviation) value of this ratio 317 for the three time periods: polluted = 1.39 ± 0.09, transition = 1.27 ± 0.12, clean = 1.69 ± 0.38. As will be discussed 318 more subsequently, dust did not drive the high PM concentrations during the polluted period, but rather it was 319 driven in large part by secondarily produced species (i.e., SO4 2-, NO3 -, NH4 + ) that grew into the 1 -2.5 µm range. 320 321 4 PM Composition 322 4.1 PM2.5 Composition 323 Speciated PM2.5 data from Seoul (  Table 4, SO4 2exhibited the highest relative enhancement during the polluted period versus the 328 clean period (factor of 7.9), whereas NH4 + and NO3were enhanced by factors of 6.8 and 6.1, respectively. OA 329 was only enhanced by a factor of 1.7. In terms of mass concentrations, the difference between the polluted and 330 clean periods for the sum of nitrate, sulfate, and ammonium was 72 µg m -3 versus the difference in OA of 7 µg 331 m -3 (Table 4). The change in OA is < 10% of the change in the three major inorganic ions. 332 The strong enhancement of the inorganic constituents owes most likely to rapid production (both 333 locally and in transport) in contrast to transported PM that was already produced upwind; the latter would tend 334 to increase OA along with inorganic constituents more comparably than what was observed. In lesser abundance 335 were Cl -, Na + , K + , Mg 2+ , and Ca 2+ , which are linked to sea salt and dust (Seinfeld and Pandis, 2016) and thus 336 expected to have appreciable concentrations above 2.5 µm. In terms of the elemental species, the most prevalent 337 species in all three periods were the crustal tracer species Si and Fe, which were 5.0 and 2.4 times higher in 338 concentration, respectively, during the polluted period versus the clean period (Table 4). Most of the crustal 339 tracer species showed enhancements ranging from 1.8 -2.8. Only Si, Se, and V showed greater enhancement 340 ratios with the latter two enhanced by factors of 4 and 8, respectively. 341 The sum of the PM2.5 components using OC in Table 4 accounted for 79% (polluted), 75% 342 (transition), and 68% (clean) of the total PM2.5, which is partly due to unmeasured species and that OC includes 343 only the carbon mass and not other elements associated with organic compounds. Using OA instead of OC 344 yields improved PM2.5 closure: 85% (polluted), 85% (transition), 83% (clean). This decent level of closure may 345 be largely attributed to the high relative abundance of more easily measured inorganic species, predominantly 346 NO3 -, SO4 2-, and NH4 + . 347 348

Mass Size Distributions 349
The mass size distribution measurements in Incheon provide a unique look into a typical transboundary 350 pollution event, with the ability to contrast it to the subsequent transition and clean periods. Insights gathered from 351 this analysis have direct relevance to Seoul owing to close proximity to Incheon (~30 km) with the exception of 352 any additional aerosol processing and formation that took place between Incheon and Seoul, including especially 353 Seoul itself. Charge balance details can be found in Sect. S1 and Fig. S3S2, with a general anion deficit during 354 the study period, including anions not speciated with the IC technique such as various types of organics. 355 Figure 34 summarizes size-resolved composition for the polluted, transition, and clean periods of this 356 study. Ions typically associated with primary natural aerosol sources such as sea salt and dust (Arimoto et al.,357 1992;Seinfeld and Pandis, 2016), including Ca 2+ , Na + , and Cl -, did not exhibit any significant enhancement during 358 the polluted period (cumulative mass concentrations in Table 3), with varying size distribution peaks based on the 359 species and period during the study. In contrast, the ions linked to secondary formation via gas-to-particle 360 conversion processes (i.e., SO4 2-, NO3 -, NH4 + , and organic acids) were dramatically enhanced during the polluted 361 period compared to the clean period ( Fig. 34)  clean period was generally marked by these species exhibiting peak concentrations for particles between 0.25 and 367 0.38 µm with a secondary peak from 0.6 and 0.94 µm, albeit neither is pronounced. 368 The likely formation pathway for SO4 2-, NO3 -, NH4 + , and organic acids in the polluted period was 369 secondary production, which was assisted in part by high humidity as discussed in more detail in Sect. 6.2. Their 370 common formation mechanism is supported by significant correlations (r ≥ 0.94; see time series in Fig. S5) during 371 the polluted period between SO4 2-, NO3 -, NH4 + , and oxalate, with the latter being the most abundant organic acid 372 during the entire study period but especially in the polluted period (~70% of organic acid mass). Oxalate is that oxalate exhibits a greater enhancement ratio in Table 3 than that of OA in Table 4 is not surprising since not 378 all OA is produced via aqueous processing and even components that are may be produced at different rates. Thus, 379 it is cautioned that oxalate is not a good proxy for OA overall in haze. Oxalate is produced efficiently via aqueous- Species that were below LOD in more than 50% of samples (see Table S1) are not shown. 387

388
We next examine if NH3 was completely neutralized by HNO3 and H2SO4. A charge balance between 389 SO4 2-, NO3 -, and NH4 + can indicate complete neutralization with a slope of unity, as has been tested by other 390 studies (e.g., Lee et al., 2003). A charge balance using individual stages of the three polluted sets collected at Inha 391 University (NH4 + on y-axis) between these three species (Fig. S4S3) yielded a best-fit line slope and y-intercept 392 of 1.34 and −0.02, respectively (R 2 = 0.99, n = 20). Our data reveal an anion deficit and that there was sufficient 393 NH3 to fully neutralize HNO3 and H2SO4. Furthermore, there likely was limited interaction of these acidic gases 394 with coarse particles (e.g., dust and sea salt) that were relatively low in abundance. 395 396 5 Regional Conditions Influencing PM 397

Atmospheric Circulation and Meteorology 398
According to previous meteorological analysis (Park et al., 2020a), the airmass history for the polluted 399 period was influenced by the interaction between a Siberian high-pressure system and a migratory anticyclone 400 system, which were located over the Korean peninsula. From 28 February to 1 March 2019, a high-pressure system 401 was located to the west and low-pressure systems were located to the east, all conspiring to yield westerly and suggested that pollution between 3 -5 March was transported to the Seoul area at low and high altitudes from the 404 Shandong Peninsula and Beijing, respectively. Pollution levels were exacerbated over South Korea due to a weak 405 high pressure system lingering over the Korean peninsula between 2 and 5 March (Park et al., 2020a). This is 406 evident from the HYSPLIT back-trajectories (Fig. 2a3a) showing airmasses moving slower in the polluted period 407 versus the subsequent clean period (Fig. 2b3b). Following the lingering period, a low-pressure system moved in 408 over Russia associated with more clouds and rain upwind of the sampling sites (Fig. S5S4). There were stronger 409 winds that were northerly between 7 and 11 March. In both the polluted and clean periods, air masses descended 410 from ~2.5 -4 km four days earlier to within the mixing layer (Fig. S5S4). The NAAPS spatial maps of speciated 411 optical depths (Fig. S6) and surface mass concentrations (Fig. S7) confirm the spatial extent of the regional haze 412 event extending from the Korean peninsula to areas like Beijing with the ABF (Anthropogenic and Biogenic Fine 413 = SO4 2-, primary organic aerosols, and SOA) component being most prominent during the study period and the 414 driver of the polluted period enhancements in PM, consistent with the dominance of SO4 2-, NO3 -, and NH4 + from 415 the in-situ measurements.

416
The local weather conditions at both Incheon (Table 3) and Seoul (Table 4) exhibited the following 417 common characteristics: (i) generally low average temperatures (< 10°C) that decreased after the polluted period; 418 (ii) low average wind speeds (2.5 -3.7 m s -1 in Incheon, < 2 m s -1 in Seoul); iii) highest mean PBLH during the 419 transition period (1.87 mkm compared to ≤ 0.53 mkm for the other periods); (iii) lowest average humidity in the 420 clean period (RH ≤ 51% and q ≤ 2.95 g kg -1 ); and (iv) negligible rain. Although locally there was negligible rain, 421 there was some precipitation along the trajectories arriving at these sites in the polluted and clean periods (Fig.  422 S5S4), with some potential to reducedreduce aerosol concentrations via wet removal. The time series of these 423 various environmental conditions at Incheon and Seoul are shown in (Fig. S8 and) demonstrate high temporal 424 similarity and the similar characteristics noted above. The relatively low regional wind speeds during the polluted 425 period (Figs. 23 and S8) indicate that transport was slow allowing for the accumulation and persistence of the 426 haze over this whole domain. 427 428

Gas Concentrations 429
Gases (CO, O3, NO2, SO2) exhibited fairly similar values (Tables 3-4) and temporal patterns (Fig. S9) 430 between Sungi and Seoul, with CO exhibiting the largest relative reduction between successive periods, followed 431 by O3. PM2.5 only exhibited a strong relationship with CO at Sungi and Seoul based on correlation coefficients 432 during the full study period (r = 0.84 and 0.87, respectively). As CO is a tracer for anthropogenic emissions 433 (Fishman and Seiler, 1983), its high correlation with PM2.5 at both sites supports the strong influence of 434 anthropogenic aerosol at both sites. The reduction of CO from the polluted to the clean period is possibly due to 435 Chinese influence; in contrast to South Korea, combustion efficiency has been shown to be worse in China 436 (Halliday et al., 2019), which supports the high CO levels during the polluted period with air masses coming from 437 China. COCarbon monoxide concentrations in Seoul exceeded Incheon by 177 ppb on average during the polluted 438 period, suggestive of added influence from local emissions in Seoul superimposed on top of the transported 439 pollution. Carbon monoxide is commonly used in calculations related to aerosol transport studies (e.g., 440 Dadashazar et al., 2021;Hilario et al., 2021b) as it is relatively insensitive to wet scavenging processes with a long 441 lifetime in the atmosphere (~1 month) compared to aerosol particles (Weinstock, 1969). 442 443 6 Evidence for Enhanced Local Secondary Aerosol Production 444

Differences Between Seoul and Incheon 445
We now consider PM2.5 differences between Incheon and Seoul, where the latter exhibits elevated levels 446 during most of the polluted period (Fig. 12). KORUS-AQ research highlighted that while transport brings aerosol 447 particles from upwind sources, high humidity and cloudiness concentrates local pollution in a shallow stable that Seoul exhibited PM2.5 levels that were on average ~10 µg m -3 higher than those at Incheon during the 450 transport/haze period of KORUS-AQ that persisted for 7 days. The maximum daily mean enhancement of PM2.5 451 in Seoul from Incheon over that period was 24 µg m -3 , while the peak hourly mean enhancement reached 32 µg 452 m -3 . These observations were attributed plausibly to more local emissions in Seoul. The same explanation arguably 453 applies to a large extent in our study period too, where the mean difference during the polluted period between 454 Seoul and Incheon PM2.5 was 10 µg m -3 for Sungi and 17 µg m -3 for Inha (Tables 3 and 4) with the peak hourly 455 enhancement between Seoul and Sungi being nearly 60 µg m -3 (Fig. 12). Here again, humidity (both q and RH) is 456 elevated and PBLH is low compared to the subsequent clean period when PM levels were drastically lower. As 457 the two sites are quite close to one another (~ 30 km), the most likely explanation for higher PM levels at Seoul 458 is more local emissions rather than additional aging to produce secondary aerosol species via transport. 459 460 6.2 Role of Humidity 461 TheOne finding of this work is the significant amount of secondarily produced species in the 462 supermicrometer range. More specifically, the relative fraction of SO4 2-, NO3 -, NH4 + , and organic acids in the 463 supermicrometer range (i.e., technically D ≥ 0.94 µm) as compared to all sizes sampled at Inha (D ≥ 0.016 µm) 464 during the polluted period was 43%, 44%, 42%, and 36%, respectively, which is appreciable and potentially 465 influenced by the humid conditions. More specifically, it is hypothesized that in the polluted period there was both 466 hygroscopic growth of particles and additional chemical uptake in those swollen particles with enhanced aerosol-467 laden water to promote higher concentrations of these secondary species.

468
Of the meteorological parameters shown in Tables 3-4, PM2.5 levels at Inha University, Sungi, and Seoul 469 were best correlated with q (r = 0.66, 0.64, and 0.78, respectively) across the entire study period. The second-best 470 relationship was with RH (r = 0.53, 0.47, and 0.53, respectively) with minimal relationships with either 471 temperature (r: 0.02 -0.17) or wind speed (-0.29 ≤ r ≤ 0.13). This motivates an examination of the relationships 472 between PM and humidity to assess the plausibility of a role for heterogeneous secondary aerosol production from 473 local and transported gas-phase precursors. 474 One metric used to quantify such enhanced aerosol production is the oxidation ratio, specifically the 475 sulfur and nitrogen oxidation ratios (SOR and NOR, respectively) where SOR = SO4 2-/(SO4 2-+ SO2) and NOR = 476 NO3 -/(NO3 -+ NO2) (Colbeck and Harrison, 1984). Higher values of SOR and NOR indicate that the gaseous 477 precursors form higher relative amounts of SO4 2and NO3 -, respectively (Kaneyasu et al., 1995 Given our finding that q is better correlated with PM2.5 than RH, we compare SOR and NOR to q (Fig. 45). We 483 evaluate these ratios for our Seoul data only as all the requisite data were measured at the same site. The mean (± 484 standard deviation) of NOR and SOR during the three time periods of the study werewas as follows (NOR/SOR):: 485 polluted = 0.39 ± 0.1/; transition = 0.22 ± 0.10; clean = 0.09 ± 0.04. Similarly, the mean (± standard deviation) of 486 SOR was as follows: polluted = 0.51 ± 0.06; transition = 0.22 ± 0.10/0.44 ± 0.14; clean = 0.09 ± 0.04/0.14 ± 0.07.

487
We find a positive relationship between NOR and SOR with q(RH), with R 2 values being 0.58(0.24) and 2021). While the average q and RH were slightly higher for the transition period in Seoul relative to the polluted 491 period (Table 4), the peak values of q, RH, SOR, and NOR all occurred during the beginning of the polluted period 492 (Fig. S10). 493 This study's results suggest there was significant heterogeneous processing to produce species like SO4 2-, 494 NO3 -, NH4 + , and organic acids above ~1 µm. These species accounted for ~93% of the total speciated ion mass at 495 Inha during the polluted period (Table 3) and are strongly correlated with one another (0.93 ≤ r ≤ 1.00). 496 Heterogeneous production of inorganic species such as SO4 2in cloudy and humid conditions is common for the 497 study region Jordan et al., 2020;Park et al., 2020b). Furthermore, significant secondary 498 production of SO4 2above 1 um at high RH has been noted in Beijing (Wang et al., 2020). Mechanisms potentially 499 responsible include aqueous oxidation by O3, H2O2, and transition-metal ion-catalyzed O2, and also heterogeneous 500 oxidation on surfaces of aerosol particles and droplets via the same oxidants (Li et al., 2020). Table 4 shows that 501 at Seoul the concentrations of elements such as Fe, Cu, Zn, and Pb were higher during the polluted period, which 502 is assumed to be similar at Incheon, supporting the possibility of transition metal-catalyzed secondary production 503 of secondary SO4 2-. Enhanced aerosol liquid water in more humid conditions also promotes partitioning of species 504 to the aerosol-phase as has been documented for NO3in the study region (Seo et al., 2020) and is common for The size-resolved composition data in Incheon allows for a comparison of how the composition of 516 PM0.94 differs from the fraction of remaining material contributing to PM2.5 (denoted PM2.5-0.94). Figure 56 shows 517 that while the transition and clean periods exhibit differences between the apportionment of the mass between 518 PM0.94 and PM2.5-0.94, during the polluted period the composition is essentially the same. This supports the 519 argument that the presence of supermicrometer secondary inorganic species derives from the same processes the 520 give rise to those compounds in PM0.94. Hence, composition measurements using instruments that exclude 521 supermicrometer particles can be used to investigate the composition and evolution of East Asian haze events. 522 Further, models can reliably apply PM1 composition apportionment to the full PM2.5 size range in their assessments 523 of sources and mitigation strategies for these events. A cautionary note is that these implications apply when PM 524 is dominated by inorganics, as with our case, with a limitation of our analysis being the lack of comprehensive 525 size-resolved OC measurements. However, the differences evident in the transition and clean periods imply that 526 under other atmospheric conditions PM1 composition measurements will not fully reflect the apportionment of 527 PM2.5 aerosols. Another important conclusion from Fig. 56 is that the relative amount of PM2.5-0.94 versus PM0.94 528 was highest in the polluted period (39% of speciated PM2.5 vs. 28% for transition and 21% for clean), further 529 reinforcing that there was increased production of secondarily formed inorganic species in the coarse mode. is about an order of magnitude larger in this study, while SO4 2is only about a factor of two greater. Gas-phase 543 SO2 is comparable between the two events. The difference is most likely explained by the much lower 544 temperatures in the March 2019 event relative to the other study, which is thermodynamically more favorable for 545 HNO3 partitioning to the particle phase to increase NO3levels (Seinfeld and Pandis, 2016). The ratio NO3 -:SO4 2-546 based on the full size distribution of the Inha filter sets was ~2.1 during the polluted period, where for the Seoul 547 PM2.5 data the ratio was ~2.2. The relative production of NO3relative to SO4 2likely varies seasonally with colder 548 temperatures and higher humidity more conducive to higher NO3and thus PM2.5 levels. High NO3events arewere 549 not particularly common, as shown especially for Beijing (Yang et  This work relies on a unique set of datadataset collected during a major transboundary pollution event 557 that impacted the Korean peninsula in March 2019. In-situ gas, aerosol, and meteorological data are compared 558 between Incheon (coastal) and Seoul (inland) along with the use of HYSPLIT and NAAPS reanalysis data. The 559 results reveal notable features that are important for both regulatory purposes and general understanding of aerosol 560 transport and formation processes. 561 562 • The pollution event stemmed from westerly transport under meteorological conditions that promoted 563 secondary inorganic aerosol production in Incheon and Seoul. 564 • Seoul exhibited significantly higher PM2.5 levels than Incheon during the polluted period with the 565 difference arising from some combination of local emissions and extensive secondary aerosol formation 566 due to favorable environmental conditions: low temperatures, elevated q and RH, and a shallow boundary 567 layer. 568 • Secondarily produced inorganic and organic acids exhibited significant mass concentrations above 0.94 569 µm during the polluted period (~40% of total mass), and their size-resolved concentrations were highly 570 correlated (0.94 ≤ r ≤ 1.00). The lack of coarse particle influence in promoting concentrations of species 571 like SO4 2-, NO3 -, NH4 + , and oxalate provided added support for the role of secondary aerosol formation 572 assisted by high humidity. PM2.5 at Seoul and Incheon were best related to q and RH as compared to 573 other examined meteorological parameters. The higher humidities during the polluted period were 574 coincident with increased sulfur and nitrogen oxidation ratios. This highlights the importance of 575 heterogeneous processing and hygroscopic growth in contributing to the high supermicrometer 576 concentrations of inorganic and organic acids in the polluted period. Increased particle size with 577 hygroscopic growth in the humid conditions likely led to increased chemical uptake. 578 • Atypically highHigh values of both NO3mass and NO3 -:SO4 2mass ratios were observed at both Incheon 579 and Seoul, likely due to low temperatures promoting efficient NO3formation. 580 581 The size distribution information from this work addresses specific concerns that have been raised about 582 the applicability of PM1 composition datasets to understanding PM2.5 air quality exceedances in East Asian haze 583 events. Here, we show that PM2.5 composition apportionment for water-soluble ions is fully captured by the PM1 584 fraction for this haze pollution event. Greater differences in the composition apportionment were observed for 585 other atmospheric conditions. after the haze period of this study. The contrast in the dominance of NO3here 586 (March 2019, T ≈ 9°C) versus the comparable amounts of NO3and SO4 2observed during the KORUS-AQ 587 campaign haze event (May 2016, T ≈ 20°C) points to the importance of conducting measurements at different 588 times of the year to more fully understand haze formation and its impacts on air quality. Equally important is the 589 use of these data to rigorously test apportionment of PM2.5 composition in air quality models to ensure that the 590 integrated impacts of transport and enhanced chemistry are adequately represented. This work contributes to the 591 growing body of data required for ongoing model assessments of PM2.5 that will inform mitigation strategies to 592 improve air quality in South Korea. 593 594 Data Availability. 595 The sampled aerosol and meteorological data used in this study can be accessed at