Hygroscopic properties of newly formed ultrafine particles at an urban site 1 surrounded by deciduous forest (Sapporo, northern Japan) during the 2 summer of 2011 3

20 To investigate the hygroscopic properties of ultrafine particles during new particle 21 formation events, the hygroscopic growth factors of size-segregated atmospheric 22 particles were measured at an urban site in Sapporo, northern Japan, during the summer 23 of 2011. The hygroscopic growth factor at 85% relative humidity [ g (85%)] of freshly 24 formed nucleation mode particles was 1.11 to 1.28 (average: 1.16 ± 0.06) at a dry 25 particle diameter (D p ) centered on 20 nm, which is equivalent to 1.17 to 1.35 (1.23 ± 26 0.06) at a dry D p centered on 100 nm after considering the Kelvin effect. These values 27 are comparable with those of secondary organic aerosols, suggesting that low-volatility 28 organic vapors are important to the burst of nucleation mode particles. The equivalent 29 g(85%) at a dry D p of 100 nm for nucleated particles that have grown to Aiken mode 30 sizes (1.24 to 1.34; average: 1.30 ± 0.04) were slightly higher than those of newly 31 formed nucleation mode particles, suggesting that the growth of freshly formed 32 nucleation mode particles to the Aitken mode size can be subjected to condensation of 33 not only low-volatility organic vapors, but also water-soluble inorganic species. Based 34 on this result, and previous measurement of radiocarbon in aerosols, we suggest that the 35 burst of nucleation mode particles and their subsequent growth were highly affected by 36 biogenic organic emissions at this measurement site, which is surrounded by deciduous 37 forest. Gradual increases in mode diameter after the burst of nucleation mode particles 38 were observed under southerly wind conditions, with a dominant contribution of 39 intermediately-hygroscopic particles. However, sharp increases in mode diameter were 40 observed when the wind direction shifted to northwesterly or northeasterly, with a sharp 41 increase in the highly-hygroscopic particle fraction of the Aitken mode particles, 42 indicating that the hygroscopic growth factor of newly formed particles is perturbed by


Introduction
New particle formation (NPF) in the atmosphere is an important factor governing the radiative forcing of aerosols and cloud formation that frequently occurs globally (1) 141 where D p (dry) is the dry particle diameter under RH <5%, and D p (RH) is its diameter at 142 a specific RH. The g(RH) of pure (NH 4 ) 2 SO 4 was measured as 1.56 ± 0.01 (n = 3) at 85% 143 RH, which agrees well with the g(85%) of 1.56 predicted using the thermodynamic 144 aerosol inorganic model (AIM;Clegg et al., 1998). The predicted g(RH) was calculated 145 by considering the Kelvin effect and assuming a density of 1.76 g cm -3 and dynamic 146 shape factor (χ) of unity for dry particles. Meteorological data (temperature, wind speed, wind direction, RH, and rainfall) for the 150 measurement period in Sapporo were obtained from a station located around 2.6 km 151 south of our study site (43º3´56″ N, 141º21´27″ E, 17 m above ground level) operated 152 by the Japan Meteorological Agency.

153
The average ambient temperature and RH were measured as 23 ± 3 C and 72 ± 154 11%, respectively. Although most measurements were carried out under clear weather 155 conditions, three rainfall events occurred on 27 July (13:00-14:00 LT), 5 August (16:00 156 LT), and 6 August (16:00 LT). The prevailing local wind directions during the urban backward trajectory analysis method (Draxler and Rolph, 2012;Rolph, 2012). All 164 calculated backward trajectories extended backwards for 96 h, with a 1-h interval. 165 Errors of up to 20% of the traveled distance are typical for those trajectories computed 166 from analyzed wind fields (Stohl, 1998). Thus, calculated air mass pathways indicate 167 the general airflow pattern rather than the exact pathway of an air mass.  which were approximately 2-3 times higher than those (1-48 μg m -3 , 14 ± 9 μg m -3 ) 182 obtained during the clean period.
183 Figure 2b and c shows the air mass backward trajectories for the measurement site 184 during the polluted and clean periods, respectively. Local wind direction and wind 185 speed during both periods are also shown in Fig. 2d and e, respectively. These backward 186 trajectories indicate that during the polluted period air masses originated from the downwind areas of the Asian continent (Fig. 2b), whereas during the clean period the 188 air masses originated from the northwest Pacific (Fig. 2c). Local wind direction also 189 clearly differed during the two periods, with a dominance of northwesterly winds during  The g(85%) of particles at a dry D p of <100 nm is less than that of equivalent larger were converted to those of equivalent larger particles assuming a dry D p of 100 nm.

225
Hereafter, the converted terms are denoted as Equiv. g(85%) at dry D p = 100 nm. As the 226 g(85%) of a dry D p <100 nm before correction was similar to that of water-soluble 227 organic aerosols, but much lower than that of inorganic compounds such as ammonium 228 sulfate (  highly-hygroscopic particles, respectively (Fig. 6). g(85%) at dry D p = 20 nm in Fig. 6 14 is Equiv. g(85%) at dry D p = 100 nm. The less-hygroscopic mode was frequently 284 observed when elevated highly-hygroscopic mode distributions were observed as seen Peak number concentrations of highly-hygroscopic particles were obtained at a g(85%) 308 of about 1.4. As this value is lower than that of water-soluble inorganic species such as 309 ammonium sulfate, highly-hygroscopic particles can be attributed to a mixture of 310 secondary organic aerosols and inorganic ions.  As only unimodal size distributions were observed for freshly formed nucleation 330 mode particles (Fig. 3b) and the Aiken mode particles (Fig. 3c), the g(85%) values for these two types of particles were obtained from a Gaussian fit of the number size radiocarbon in the WSOC fraction of the aerosols collected at the same site.  demonstrate that the subsequent growth of freshly formed nucleation mode particles and 385 their hygroscopic properties were highly affected by the local wind direction (Fig. 7).

386
Under southerly wind conditions on 31 July and 5 August, gradual increases in Mode D p 387 occurred with a dominant contribution from intermediately-hygroscopic particles.

388
However, sharp increases in Mode D p occurred when the wind direction shifted to 389 northwesterly or northeasterly on 1, 2, and 6 August, with a sharp increase in highly- highly-hygroscopic particles to total particles at dry D p = 120 nm as a typical example 399 of the large Aitken to small accumulation mode particles. The number fractions of the 400 less-, intermediately-, and highly-hygroscopic particles at dry D p = 120 nm were found 401 to be 18 ± 9%, 14 ± 8%, and 69 ± 14%, respectively, during the polluted period.