Hygroscopic properties and CCN activity of atmospheric aerosols under 1 the influences of Asian continental outflow and new particle formation at a 2 coastal site in East Asia 3 4

Abstract. The chemical composition of fine particulate matters (PM2.5), the size distribution and number concentration of aerosol particles (NCN) and the number concentration of cloud condensation nuclei (NCCN) were measured at the northern tip of Taiwan Island during a campaign from April 2017 to March 2018. The parameters of aerosol hygroscopicity (i.e. activation ratio, activation diameter and kappa) were retrieved from the measurements. Significant variations were found in the hygroscopicity of aerosols, which were suggested be subject to various pollution sources, including aged air pollutants originating in the eastern/northern China and transported on the Asian continental outflows, fresh particles emitted from local sources and distributed by land-sea breeze circulations as well as produced by new particle formation (NPF) processes. Cluster analysis was applied to the backward trajectories of air masses to investigate their respective source regions. The results showed that the aerosols associated with Asian continental outflows were characterized with higher kappa values, whereas higher NCCN and NCN with lower kappa values were found for aerosols in local air masses. The distinct features in hygroscopicity were consistent with the characteristics in the chemical composition of PM2.5. Moreover, this study revealed that the nucleation mode particles from NPF could have participated in the enhancement of CCN activity, most likely by coagulating with sub-CCN particles, although the freshly produced particles were not favored for CCN activation due to their smaller sizes. Thus, the results of this study suggested that the NPF coupling with coagulation processes can significantly increase the NCCN in atmosphere.



Introduction 1
Aerosols suspended in the atmosphere allow condensation of water vapor under certain super-2 saturation conditions and subsequently evolve into cloud droplets. Activation of cloud 3 condensation nuclei (CCN) depends on the size and chemical composition of aerosol particles, 4 as well as on the meteorological conditions (i.e. water vapor supersaturation (SS), and uplift 5 force for air parcels) (Seinfeld and Pandis 1998). Among the chemical and physical properties 6 of aerosols, hygroscopicity plays critical roles in the complex aerosol-cloud interactions 7 (McFiggans et al., 2006;Lee et al., 2010). Atmospheric aerosols are a mixture of different 8 chemical species rather than a single compound and exist in various size ranges and mixing 9 states. A single parameter called kappa (κ) has been developed to evaluate hygroscopicity of 10 aerosols, which represents a scaled volume fraction of soluble materials in particles and 11 provides a theoretical framework to derive bulk hygroscopicity for aerosols with internal 12 mixtures (Petters and Kreidenweis, 2007). However, while the hygroscopicity and CCN 13 activity of a single component can be characterized in laboratories, the properties of their 14 mixture in ambient air are difficult to estimate owing to the complexity in physiochemical 15 characteristics of aerosols. Thus, field investigations have been conducted to study aerosol 16 hygroscopicity and CCN activity in various environmental settings including rural, urban, 17 forest and marine boundary layer (Ehn et al., 2007;Massling., 2007;Gunthe et al., 2009;Wu 18 et al., 2016;Schmale et al., 2017;Park et al., 2018). Furthermore, in-situ measurements of 19 physicochemical properties of aerosols and CCN in critical geographical areas in climate 20 system could provide a means of constraining representation of relevant schemes in global 21 climate models (Khairoutdinov and Randall, 2001;Betancourt and Nenes, 2014;Seinfeld et al., 22 2016). 23 24 Due to the rapid industrialization and economic development in the East Asia (EA) during the 25 past few decades, the EA has become one of the most polluted regions in the world where 26 significant amount of particulate matters (PM) and their precursors were emitted (Streets et al., 27 2003;Dentener et al., 2006;Zhang et al., 2009). Taiwan is located in the downwind area of the 28 EA continental outflows, and thereby is influenced by the pollution outbreaks during the winter 29 monsoon seasons. Besides, the air quality in Taiwan is also known to be affected by the 30 photochemical production of secondary aerosols. The geographical location thus provides a 31 strategic platform to investigate the CCN activation of aerosols influenced by a complex 32 mixture of pollutants (Chou et al., 2005(Chou et al., , 2017Chang et al., 2010;Cheung et al., 2013Cheung et al., , 201633 Li et al., 2016;Lee et al., 2019). Cheung et al. (2013) reported that new particle formation interpretation, the hourly average mass concentration of PM2.5, the mixing ratio of trace gases 1 (i.e. CO, O3, SO2 and NO2) and the meteorological parameters (i.e. wind direction/speed) 2 reported from the air quality station of Taiwan EPA that collocated with the CAFÉ station were 3 analyzed in this study. 4 5 2.2 Data processing and analysis for aerosol hygroscopicity 6 Firstly, the NCCN and NCN data were synchronized into 5 mins averaged data which matched 7 the time interval for PSD data measured by SMPS. The CCN activation ratio (AR), i.e. the ratio 8 of NCCN to NCN, was calculated for a given SS condition. Given the assumption that the 9 particles are homogeneously internally mixed and larger particles are activated first. Also, the 10 number concentration of particles out of the measured particle size range is assumed negligible. 11 The minimum diameter (Dss) required for the CCN activation with the AR value at a given SS 12 condition was calculated according to equation (1) (Hung et al., 2014). The hygroscopicity parameter () was then calculated as the followings : 20 where Sc is the water saturation (= SS + 1), Dd is the dry particle diameter and equivalent to 26 Dss calculated by equation (1), ⁄ is the solution surface tension (0.072 J m -2 ), is the 27 water density (997 kg m -3 ), is the molecular weight of water (0.018 kg mole -1 ), R is the 28 universal gas constant (8.314 J K -1 mole -1 ) and T is ambient temperature. 29

30
The kappa value is used to describe the hygroscopicity of the aerosols; for example, ammonium continents. The standardized κ values at SS of 0.5 % were found to be 0.48, 0.41, 0.55, and 8 0.30 for rural background, alpine, coastal background, and urban environmental settings, 9 respectively. The estimated κ value at SS of 0.5 % was 0.31 for this study, which was 10 significantly lower than that for coastal background and was more similar to that of urban 11 aerosols. This is likely because the aerosol composition at CAFÉ station were frequently 12 influenced by urban air pollution, as indicated in previous studies (Chou et al., 2008(Chou et al., , 2010(Chou et al., , 13 2017. 14 15 It is noteworthy that both the κ and DSS decrease with the SS, which implies some small and 16 less hygroscopic particles getting activated at higher SS. Previous studies on size-resolved 17 chemical composition of PM2.5 at northern Taiwan reported that the size distribution of 18 aliphatic carbons peaked at 0.12-0.15 µm and 0.62-0.87 µm while that for carbonyl carbons 19 peaked only at 0.6-0.64 µm (Chou et al., 2005). Cheung et al. (2016) showed that the ultra-fine 20 particles (i.e. d < 100 nm) collected from Taipei City, an urban site in northern Taiwan, 21 consisted mostly of organic matters. Moreover, Salvador et al. (2016) revealed that low-22 molecular-weight organic acids were abundant in the submicron aerosols in Taipei, Taiwan. In 23 this context, the low hygroscopicity of small aerosols found in this study was consistent with 24 the results of investigations upon aerosol chemical composition. In addition, up-taking hygroscopic species during particle growth and coagulation processes 7 may influence the hygroscopicity of aerosols, which will be discussed in further details later 8 on. organic species, which showed that significantly higher κ values were found for major 16 inorganics species in aerosols, such as ammonium sulfate, ammonium nitrate, sodium chloride 17 (kappa: 0.61-1.28), while κ values for organic species were usually lower than 0.2. Thus, 18 relatively lower kappa values observed during Jul. -Aug. 2017 were consistent to the PM2.5 19 chemical composition data in which a higher mass fraction of organic carbon was found. 20 21

Implications of different types of air masses 22
The air masses reaching this study site are known to be associated with the Asian continental 23 outflows and/or with local pollution in northern Taiwan (Cheung et al., 2016). Since CO has 24 longer atmospheric lifetime than NO2, a higher ΔCO/ΔNO2 can be used to indicate influences 25 of aged regional air pollutants. The averaged median ΔCO/ΔNO2 ratios for the 5 trajectory 26 clusters were 76, 75, 32, 60 and 33, respectively. A higher ΔCO/ΔNO2 ratio was found in 27 Clusters 1, 2 and 4, whereas ΔCO/ΔNO2 of Cluster 4 was found slightly lower than that of 28 Cluster 1 and 2. This was attributed to the differences in air mass history; the air masses of 29 both Clusters 1 and 2 were originating in the inland areas of the Asian Continent, whereas the 30 air masses of Cluster 4 passed through the south of Korea and Japan and came from the east of 31 CAFÉ station and, thereby, was occasionally impacted by some fresh emissions. The mixing 32 ratio of O3, a typical secondary pollutant, provided further information about the sources of air 33 plumes. The results showed that higher O3 levels (43-46 ppb) were found in continental outflows (i.e. Cluster 1, 2 and 4) as compared to those of marine air masses (i.e. 26-28 ppb for 1 Cluster 3 and 5). 2 3 Furthermore, higher κ values were found for CCN transported with the continental outflows, 4 which ranged from 0.19 to 0.69 for SS of 0.15-0.86 %. On the contrary, lower κ values (0.14 -5 0.56) were found for the CCN in air mass of Clusters 3 and 5, which originated in the remote 6 Pacific region and passed through Taiwan Island during summertime. This result was 7 reasonable since aged polluted air masses contained more inorganic species (with higher κ 8 values), while the organic species (with lower κ values) contributed a higher fraction to the 9 aerosol mass loading in urban areas of Taiwan (Chou et al., 2010(Chou et al., , 2017. On the other hand, 10 higher NCCN and NCN were found in Clusters 3 and 5 compared to that in Clusters 1 and 2 (see 11   Table 3). This could be due to the substantial production of new particles during warmer 12 seasons (Cheung et al., 2013(Cheung et al., , 2016. 13 14

Implications of New Particle Formation 15
As described in Section 3.1, large variations in NCCN and kappa values were found in summer 16 during which NPF events occurred frequently. A NPF event is defined as the increase of the 17 number concentration of nucleation mode particles, and those particles are growing into Aitken 18 and/or accumulation mode size range (≥ 25nm) and last for a few hours until they disappear 19 into the atmosphere by condensation/coagulation sinks (Dal Maso et al., 2005). In total 53 NPF 20 events were observed during the entire study period and among which 31 were observed in 21 warm months (from June to September 2017), representing an occurrence frequency of 58.5%. 22 Investigations reported that NPF occurred more frequently during summer (34.6 -42.8%) and 23 occasionally during spring (11.5%) in urban areas of northern Taiwan (Cheung et al., 2013(Cheung et al., , 24 2016. Figure 6 illustrates the median particle size distribution for NPF and non-NPF days as 25 well as the quartiles. The particle number concentration for NPF events was significantly higher 26 than that for non-NPF case. In addition, large variations were associated with the particle size 27 below 100 nm in NPF events, suggesting that a large amount of ultra-fine particles formed. 28 29 In Figure 7, diurnal variations in particle size distribution for NPF and non-NPF cases are 30 presented along with the aerosol hygroscopic parameters DSS, κ and AR at SS = 0.29 %. In the 31 plot of particle size distribution for NPF events, a banana feature (growth of particle diameter 32 indicated by the geometric mean diameter, GMD) is obviously illustrated, which is typical for NPF process (Dal Maso et al., 2005;Cheung et al., 2011), while relatively stable particle size 1 distribution exhibites for non-NPF periods with the particles of 50-60 nm dominate throughout 2 a day. 3 4 On NPF days, a nucleation burst as indicated by a surge in nucleation mode particles (N30, 5 number concentration of particle size ≤ 30 nm) from 06:00 to 10:00 LT was observed (as shown 6 in Figure 7). Note that the number concentration of Aitken mode particles (indicated by N30-7 100, for particle size between 30 to 100 nm) increased consistently, implying coagulation was 8 active during the period. NCCN started to increase significantly around 07:00 LT. It was found 9 that the increasing rate of NCN was higher than that of NCCN, which in turn resulted in the 10 decreases in AR. The increases in NCCN were attributed to coagulation processes. Figure 8  11 illustrates schematically the CCN enhancement by coagulation processes at the initial stage of 12 a NPF event. Once the NPF process starts, the freshly formed nucleation mode particles could 13 get coagulated with the pre-existing particles, with either CCN or sub-CCN sizes. The 14 preexisting sub-CCN particles coagulate with the newly formed particles and, as a result, grow 15 rapidly into CCN and thereby increase the NCCN. On the other hand, the new particles could 16 also coagulate with the preexisting CCN, which should not increase the NCCN but will result in 17 an increase in the size of CCN. The observation of this study (see Figure 7) showed that DSS 18 slightly increased from about 80 nm at 04:00 LT to 87nm at 08:00 LT, suggesting that 19 preexisting CCN particles were still predominant in this stage despite the production of "new 20 CCN" has resulted in the increases in NCCN. Several observational studies reported that 21 enhancement of CCN number concentrations were associated with NPF process (Sihto et al., 22 2011;Yue et al., 2011;Leng et al., 2014;Wu et al., 2015). However, the time for the newly 23 formed nano-particles growing to CCN sizes ranges from a few hours to more than a day 24 (Keriminen et al., 2018). Hence the enhancement of CCN at the initial stage of a NPF event as 25 observed in this study cannot be explained by the growth of new particles, and was most likely 26 due to coagulation among the newly formed particles and pre-existing particles. 27 28 At a later stage, because the coagulation sink exceeds the production rate of new particles, the 29 NCN turn to decrease, whereas the NCCN keeps the increasing trend for the production of new 30 CCN by coagulation among particles. As a result of increases in NCCN and decreases in NCN, a 31 significant increase in AR is expected. This has been observed in this study. Figure 7 illustrates 32 that the AR on NPF days increased rapidly since 10:00 LT, in phase with the drastic decreases 33 in the number density of N30 and N30-100. Agreeing with the earlier stage, the increased NCCN was suggested a result of the coagulation between nucleation mode particles and the Aiken 1 mode particles in sub-CCN size range, which thus grew into CCN size range. However, as 2 more and more "new CCN" formed along with the NPF processes, which would become 3 majority in the CCN population and thereby shift the size distribution of CCN to the left (as 4 shown in Figure 8). This inference was evidenced by the observation in this study, where the 5 decrease in DSS from 87 nm at 08:00 LT to 74 nm at 15:00 LT was found. It should be noted 6 that the transport of external CCN during the particle growth process could also increase the 7 CCN concentration; however, this influence should be minor because it cannot explain the 8 simultaneous changes in NCN and AR. In contrast, the increases in kappa during the later NPF course suggested that the "new CCN" 24 were dominated by hygroscopic species. The field studies at North China Plain found two types 25 of NPF events (Yue et al., 2010, Ma et al. 2016, including sulfur-rich NPF, i.e., condensation 26 and neutralization of sulfuric acid contributed most to the growth of the new particles with high 27 particle hygroscopicity, and sulfur-poor NPF, i.e., condensation of organic compounds had a 28 higher contribution to the growth with a lower particle hygroscopicity. Our results showed that 29 the NPF events in northern Taiwan were characterized by elevated levels in both sulfur and 30 organic matters (as shown in Figure 5). In particular, the submicron particles in northern 31 Taiwan were found enriched in sulfate (Cheung et al., 2016) and organic acids (Salvador et al., 32 2016). Thus it was inferred that the preexisting sub-CCN particles were more hygroscopic, 33 which resulted in the increases in kappa when they evolved into CCN through coagulation with ultrafine particles. The growth of sub-CCN particles at the study site could also be due to 1 condensation of organic compounds. However, as illustrated in Figure 8, the composition of 2 "new CCN" are dominated by the preexisting sub-CCN particles and thereby characterized 3 with high kappa values. In this context, the increases in NCCN during the NPF events were 4 unlikely contributed from the condensation growth of newly formed particles. 5 6 The result of this study is similar to the previous studies which indicated that an enhancement 7 of CCN number was associated with the NPF process. The increase of CCN was observed in a 8 few hours (Yue et al., 2011;Wu et al., 2015;Leng et al., 2014) to a few days after the start of 9 the NPF (Sihto et al., 2011). However, distinct responses of the CCN activation diameter were 10 observed. Sihto et al. (2011) indicated that DSS increased gradually with the increased of NCCN, 11 whereas Wu et al. (2015) showed a decrease in DSS once NPF process occurred and increased 12 in the later stage. In the present work, DSS slightly increased once NPF started, and then 13 decreased in later stage of the NPF event (see Figure 7 and 8). The discrepancy observed in 14 respective studies showed the complexity in the particle growth processes. Nevertheless, the 15 results of this study suggest that NPF coupling with coagulation is an important process to 16 enhance the number of CCN in the study region. Asian continental outflows contained more inorganic species and thereby were characterized 28 with higher κ values, as comparing to those associated with local urban pollution which 29 consisted substantially of organic matters. 30

31
The higher levels of NCCN and NCN found in spring and summer were attributed mainly to the 32 NPF events occurred frequently during warm months. A two-stage hypothesis was proposed 33 according to the results of this study for the implications of NPF for CCN activity. At the early stage of a NPF event, new particles formed and resulted in increases in N30 and thereby NCN, 1 which was followed immediately by increases in the number density of Aiken mode particles 2 (N30-100). The new particles coagulated with preexisting sub-CCN particles, which thereby 3 evolved into "new CCN" and resulted in the increases in NCCN. The new particles coagulated 4 also with the preexisting CCN and resulted in increases in Dss before the "new CCN" became 5 predominant. At the later stage, along with the NPF and coagulation processes, N30, N30-100 and 6 NCN decreased for the larger coagulation sink, whereas generation of "new CCN" continued 7 and resulted in increases in NCCN and a significant enhancement in AR. The activation diameter 8 got smaller (Dss) as the "new CCN" overwhelming in the CCN population at this stage. 9 Moreover, the investigation results showed that the kappa of CCN exhibited a decrease at the 10 early stage and an increasing trend during the second stage. It was inferred accordingly that the 11 newly formed particles were composed mostly of organic matters that "diluted" the 12 hygroscopicity of preexisting CCN at the early stage, whereas the sub-CCN particles consisted 13 of highly hygroscopic components dominated in the later stage of the event. 14 15 The seasonal characteristics of hygroscopicity and CCN activity under the influences of a 16 complex mixture of pollutants from different regional and/or local pollution sources have been 17 illustrated in this study, and the impacts of NPF was demonstrated. Nevertheless, the mixing 18 state and chemical composition of the aerosols, in particular the organic content of the sea spray 19 aerosols, would critically influence the aerosol hygroscopicity in coastal areas. Hence further 20 investigations are necessitated to understand the atmospheric processing involved in the CCN 21 activation which would in turn affect cloud formation and the regional climate. and cloud condensation nucleus activity. Atmos. Chem. Phys., 7, 1961-1971 https://doi.org/10.5194/acp -7-1961-2007, 2007.  Environ., 140, 565-575, https://10.1016/j.atmosenv.2016.06.029, 2016 and Gysel, M.: Collocated observations of cloud condensation nuclei, particle size 12 distributions, and chemical composition. Sci. Data, 4, https://doi.org/10.1038Data, 4, https://doi.org/10. /sdata.2017 2017. 14 Schmale, J., Henning, S., Decesari, S., Henzing, B., Keskinen,H.,Sellegri,K.,Ovadnevaite,15 J.,Pöhlker,M.L.,Brito,J.,Bougiatioti,A.,Kristensson,A.,Kalivitis,N.,Stavroulas,I.,16 Carbone,S.,Jefferson,A.,Park,M.,Schlag,P.,Iwamoto,Y.,Aalto,P.,Aijälä,M.,17 Bukowiecki,N.,Ehn,M.,Frank,G.,Frohlich,R.,Frumau,A.,Herrmann,E.,Herrmann,H.,18 Holzinger,R.,Kos,G.,Kulmala,M.,Mihalopoulos,N.,Nenes,A.,O'Dowd,C.,Petäjä,T.,19 Picard,D.,Pöhlker,C.,Pöschl,U.,Poulain,L.,Prévôt,A.S.H.,Swietlicki,E.,Andreae,M.O.,20 Artaxo, P., Wiedensohler,A.,Ogren,J.,Matsuki,A.,Yum,S.S.,Stratmann,F.,Baltensperger,21 U., and Gysel, M.: Long-term cloud condensation nuclei number concentration, particle    were originating in the inlands of the Asian Continent, but the movement of cluster 2 air masses 4 was faster and from higher elevation. Air masses in cluster 4 were pushed by high pressure system 5 towards the south of Korea and Japan, then moved along marine boundary slowly before reaching 6 CAFÉ station, while Cluster 3 and 5 represent air masses originated in the South China Sea and 7 remote Pacific region, respectively. were measured under SS = 0.29%. GMD were calculated based on the multiple curves fitting result 5 by DOFIT model which one to three modes were defined depends on the particle size distribution 6 data. Nucleation mode particles formed once NPF started, ii) DSS increased slightly while NCCN 5 increased in Stage I when existing CCN particles grew into larger size, and iii) DSS decreased while 6 NCCN continued to increase in Stage II when sub-CCN particles grew to sufficiently large to act as 7