Frequent new particle formation at remote sites in the temperate/boreal forest of North America

The frequency and intensity of new particle formation (NPF) over remote forest regions in the temperate and boreal zones, and thus the importance of NPF for the aerosol budget and life cycle in the pristine atmosphere, remains controversial. Whereas NPF has been shown to occur relatively frequently at several sites in Scandinavia, it was found to be nearly absent at a mid-continental site in Siberia. To explore this issue further, we made measurements of aerosol size distributions between 10 and 420 nm diameter at two remote sites in the transition region between temperate and boreal 15 forest in British Columbia, Canada. The measurements covered 23 days during the month of June 2019, at the time when NPF typically reaches its seasonal maximum in remote mid-latitude regions. These are the first such measurements in a near-pristine region on the North American continent. Although the sites were only 150 km apart, there were dramatic differences in NPF frequency and intensity between them. At the Eagle Lake site, NPF occurred daily and nucleation mode particle concentrations reached above 5000 cm-3. In contrast, at the Nazko River site, there were only 6 NPF events 20 in 11 days and nucleation mode particle concentrations reached only about 800 cm-3. The reasons for this difference could not be conclusively resolved with the available data; they may include airmass origins, pre-existing aerosols, and the density and type of forest cover in the surrounding regions. Our results suggest that measurement campaigns in the remote forest regions of North America to investigate the role of NPF with a more comprehensive set of instrumentation are essential for a deeper scientific understanding of this important process. 25


4
The aerosol inlet was located about 2 m above ground level, and the sample air was brought into the cabin by ca. 2 m of 6.25 mm OD copper tubing. Tests comparing measurements with and without the inlet tubing at times with a pronounced nucleation mode showed no detectable particle loss. The air was sampled without the use of a dryer.

Airmass history 185
The 48-h airmass backtrajectories initialized 100 m above surface level at 12 local time (LT = 19 UTC) are shown for the two sites in Figs. 3a and 3b (separate 72-h backtrajectories for each individual day are available in the Supplement). All trajectories for the Eagle Lake site had crossed the Pacific coast 22 to 48 hours before arriving at the site and then traveled across the densely forested Coast Range. This fetch area has an extremely low population density and is devoid of any industrial activity. All but two of the trajectories travelled in the boundary layer for at least 48 hours before their arrival. 190 The airmasses arriving on 12 and 13 Jun had not made surface contact with the ocean surface; instead, they arrived from the free troposphere over the Pacific, and descended rapidly after having crossed the Coast Range. With two exceptions, the airmasses arriving at the Nazko River site also arrived from the Pacific coast, which they had crossed about 14 to 48 hours before arriving at NR. During and after crossing the Coast Range, they remained in the boundary layer for their entire travel. Similar to the fetch at EL, there is no industrial activity and very little human population in this fetch area. Two airmasses arrived from the North and had not made contact with the ocean surface in 200 the last 72 hours. These airmasses originated in an area that had been influenced by the smoke from fires that had been burning for several weeks in northern Alberta, and about 24 h before arriving at NR, they crossed the small municipality of Vanderhoof, which lies 108 km from the site and has a population of about 10,000 persons.
In summary, our analysis of the history of the sampled airmasses shows that they had no significant input of anthropogenic emissions for at least 3 days before our measurements. An analysis of 10-day backtrajectories showed that even on this 205 time scale, almost all airmasses had remained over the Pacific Ocean, with only a few trajectories crossing over remote regions of British Columbia or Alaska (Supplemental Figs. S1a and S1b).

Aerosol concentrations and size distributions
Figs. 4a and 4b show time series of the number concentrations of aerosol particles in the nucleation mode (Nnuc; 10 to 24 nm diameter) and across the entire size range covered by the Nanoscan SMPS (N420; 10 to 420 nm), as well as the aerosol 210 mass concentration in the 10 to 420 nm range (M420) derived from the number spectra by assuming spherical particles with a density of 1.2 µg cm -3 . The corresponding size distributions are shown in Figs. 5a and b. The mass concentrations of dust, seasalt, and BC from the MERRA-2 models are provided in the supplement (Supplemental Figs. S2a and b).
The measurements at Eagle Lake indicate an extremely clean atmosphere: The time series plot of aerosol number size distributions ( Fig. 5a) is dominated by particles below 100 nm, often with a distinct nucleation mode below 20 nm and a 215 separate Aitken mode between 20 and 80 nm. The accumulation mode tends to be absent or weak, except for a short period on 12 and 13 Jun, when a mode around 100 nm could be interpreted as a more pronounced accumulation mode.

NPF frequency
New particle formation events were identified using the procedure and criteria of . The criteria for an NPF event were: (1) a distinct increase of Nnuc concentrations, (2) formation of a new nucleation mode persisting for 245 more than two hours, and (3) Table 1. 250 Table 1: New particle formation events at the Eagle Lake and Nazko River sites: Timing, condensation sink at the beginning of the event, formation rate at 10 nm, and growth rate. At Eagle Lake, we identified 21 distinct NPF events in a time span of only 12 days, with up to three events within a 24-h period (Table 1 and Fig. 5a). Using the classification of , which focuses on event days rather than 255 single NPF events, all 12 days were event days. We further classified the events into daytime and nighttime events using an adjusted event start time. Since the lower cutoff our instrument is 10 nm, it will detect an event some 3 hours later than an instrument with a cutoff at 3 nm would, given our average GR of 2-3 nm h -1 . For the purpose of classifying events as daytime or nighttime, we thus adjusted the start time by subtracting 3 hours from the observed start times listed in Table   1, which are based on detection with a lower cutoff at 10 nm. Events were classified as nighttime NPF, if the adjusted 260 start time fell between 21:00 and 04:30 LT, i.e., the time when the SW radiation flux was typically below 20 W m -2 .
Using this criterion, we found 11 nighttime events at EL, with nighttime nucleation occurring on 8 of 11 nights, and in some instances with two events in a single night.
The frequency of NPF was much lower at Nazko River (Fig. 5b), with only six events over 10 days, and five days classified as event days. Nighttime NPF also occurred less frequently, with only three events during the study period. During 265 the smoke-affected period of 21-22 Jun, there was no evidence of NPF, and Nnuc levels were very low, with an average of 16 cm -3 . Nucleation activity resumed immediately, however, around 18 LT on 22 Jun, when accumulation mode concentrations had dropped back down to background values.
Given the relatively short duration of our study compared to the long-term studies at some other sites, we can only make limited comparisons regarding NPF frequency at our BC sites. Our study took place during late spring, when NPF events 270 are most frequent at other temperate/boreal sites, e.g., Hyytiälä (Dada et al., 2017) and Vavihill, Sweden (Kristensson et al., 2008). Similarly, at Pallas, a remote site at the northern edge of the boreal forest (68 ºN) in Finland, Asmi et al. (2011) found that NPF event frequency peaked in spring, while the highest growth rates occurred in summer. Taking both our sites together, we had 17 event days out of a total of 22 days of observations, for a frequency of 77%, which is significantly higher than the median frequencies of 20-50% reported for the spring/summer seasons at comparable sites by Nieminen 275 et al. (2018). This is largely a consequence of the frequent nighttime NPF at our sites, because without the nighttime events there would be only 11 NPF days, or a frequency of 50%.

Diurnal behavior
New particle formation events occurred at our sites just as often during nighttime as during daytime (13 out of 27 events), in sharp contrast to what has been observed at most other sites. For example, NPF at Hyytiälä typically takes place 280 between sunrise and noon (Dada et al., 2017), and at Egbert, Ontario, and Whistler Mountain, British Columbia, only daytime NPF was observed (Pierce et al., 2012;Pierce et al., 2014).
The timing of NPF is further illustrated in Figs. 6 and 7, which show the diurnal variation of the aerosol size spectra averaged over the entire measurement period and for an exemplary day at each site, respectively. At EL, Fig. 6a shows a distinct nucleation mode appearing around 23 LT, representing the nighttime events, which intensifies during the night 285 to reach the highest Nnuc concentrations around 06 LT. Growth into the Aitken mode continues throughout the day, and around mid-day the nucleation mode is absent. Another set of NPF events occurs between 16 and 20 LT, also followed by growth into the Aitken mode. At NR (Fig. 6b) we see overall lower total particle concentrations, much lower Nnuc The details of the aerosol evolution are illustrated for exemplary days in Fig. 7. At EL (Fig. 7a), 11 Jun was a day with very light winds, which resulted in a nearly stationary airmass. After midnight, particle concentrations were quite low, with a faint nucleation mode below 20 nm, an Aitken mode around 50 nm, and a weak accumulation mode around 100 nm. A distinct new nucleation mode appears at 02 LT, which grows throughout the day to ca. 50 nm. Additional minor 300 NPF events are seen in the afternoon and around 23 LT. In contrast, at NR, Fig. 7b shows the near-complete absence of nucleation mode particles at night, followed by a daytime NPF event with fairly rapid growth into the Aitken range.

Airmass origin 310
As already discussed above in sections 3.2 and 3.3, most of the airmasses sampled at our sites had crossed the Pacific coast 1 -2 days before arriving at our sites, and contained little or no detectable anthropogenic pollution. Thus, similar to other temperate/boreal sites, e.g., Hyytiälä and Pallas, our NPF events occurred in clean airmasses, mostly originating from the west and northwest, which had low levels of pollution aerosols that would suppress nucleation by acting as condensation sink (CS) (Sogacheva et al., 2005;Dada et al., 2017). 315 There was, however, no evidence that a marine influence on the airmass enabled or facilitated NPF at the EL site. The airmasses arriving on 8-13 Jun had not had contact with the sea surface, as they had either remained for the last 48 hours in the continental boundary layer (8 Jun) or had descended from the free troposphere. In these airmasses, NPF was just as active as in the airmasses arriving on 4-7 and 14-15 Jun, which had come from the surface level over the Pacific and had traveled in the boundary layer for their entire 48-h history. 320 https://doi.org/10.5194/acp-2021-838 Preprint. Discussion started: 18 October 2021 c Author(s) 2021. CC BY 4.0 License. as has been shown previously by Kulmala et al. (2004). Temperature effects on the rate of HOM formation and nucleation are not likely to be important over the limited range of ambient temperatures during our study. Low temperatures decrease the rate of HOM formation (Frege et al., 2018), but increase nucleation probability due to lower volatility of the oxidation products, resulting in only a modest net change of the nucleation rates from organic precursors (Simon et al., 2020). While more rapid HOM production during the warmer daytime could lead a build-up of HOMs, followed by NPF at the colder 405 nighttime temperatures, this effect is likely to be very small for the diurnal temperature range (about 10 ºC) at our sites (Simon et al., 2020).
Other than the diurnal change of relative humidity, there were no systematic RH variations that would allow examination of the effect of RH on NPF events. Based on previous studies, no strong effects would be expected, anyway. Low RH has been shown to favor NPF (Hamed et al., 2011;Dada et al., 2017). However, the mechanism proposed by Hamed et al. 410 (2011) is based on reduced H2SO4 production due to lower OH at very high humidities (>80 %), which were almost never present during our study. Reduced H2SO4 production would also not affect our proposed NPF mechanism, which is based on pure organic nucleation. The increase of the CS due to hygroscopic growth is also not likely to be significant, since the particles at our BC sites are presumed to be mostly organic and thus are not expected to show strong hygroscopic growth over the range of RH prevailing at our sites. Laboratory studies by Bonn et al. (2002) suggested that water vapor 415 suppresses the formation of ELVOCs from monoterpenes. Note that this effect is a function of absolute, not relative humidity. Contrary to what would be expected from findings of Bonn et al. (2002), we actually observed higher Nnuc during the (warmer) period with higher water vapor mixing ratio (11 -15 Jun, 6.7±0.9 g kg -1 ) than during the (cooler) period with lower water vapor (4 -10 Jun, 4.4±0.8 g kg -1 ).
Ozone plays a critical role as a key oxidant leading to formation of ELVOCS from monoterpenes (Ehn et al., 2014). The 420 minimum O3 levels required to initiate NPF with monoterpenes in chamber studies were 10-19 ppb (Ortega et al., 2012).
While we have no on-site ozone measurements, the ozone data from nearby sites indicate that there were sufficient O3 concentrations to fulfill this requirement. The O3 data from the Williams Lake monitoring site (the closest site to EL) ranged between an average morning low of 13 ppb and a late afternoon high of 34 ppb (https://envistaweb.env.gov.bc.ca/, last accessed 20 Jul 2021) for our study period (Supplemental Fig. S3). Hourly data were not available from Quesnel, the 425 closest monitoring site to NR, but data for the daily maximum concentrations showed values similar to Williams Lake.
Overall, then, our results are most consistent with a mechanism where HOMs/ELVOCs are formed by ozonolysis and/or OH-initiated autoxidation of monoterpenes followed by pure organic nucleation. The high incidence of nighttime NPF is consistent with the findings of Sulo et al. (2021), who showed that HOM dimers have their maxima during the night, whereas the HOM monomers and H2SO4 exhibit daytime maxima. The highest concentrations of both monoterpenes and 430 O3 can be expected in the late afternoon, which may explain why we frequently observed the onset of events with particles >10 nm around midnight, which, given growth rates of 2-3 nm h -1 , would imply that the actual nucleation event began around sunset. Analogously, the nighttime nucleation events at Hyytiälä, typically occur around sunset and are driven by HOMs from monoterpene oxidation (Rose et al., 2018), but in contrast to our sites, at Hyytiälä the particles from nighttime NPF never grow beyond a few nanometers . Our observed growth rates, averaging 2.1 and 3.2 nm 435 h -1 for the two sites, are in good agreement with the median GRs attributable to monoterpene oxidation products (1.0 -3.5 nm h -1 ) measured at Pallas by Asmi et al. (2011).
The pronounced difference in NPF frequency between our two sites may be related to the difference in vegetation in the two areas. Whereas the landscape around EL is completely dominated by conifers, a large fraction of the area around NR has been deforested in recent years either by logging or wildfires. These cleared areas are covered by a mixture of herba-440 ceous vegetation, small conifers, and abundant aspens. Consequently, one would expect lower concentrations of monoterpenes coupled with high levels of isoprene emitted by the aspens. Suppression of NPF in an isoprene-dominated forest https://doi.org/10.5194/acp-2021-838 Preprint. Discussion started: 18 October 2021 c Author(s) 2021. CC BY 4.0 License. environment has been observed by Kanawade et al. (2011) and investigated in the laboratory by Kiendler-Scharr et al. (2009), who attributed it to OH scavenging by isoprene. This mechanism is not likely to be important at our sites, since it would only affect daytime nucleation, whereas at NR, nighttime nucleation is also less frequent than at EL. More likely 445 is the alternative mechanism proposed by Heinritzi et al. (2020), wherein isoprene reduces the yield of dimer HOMs with 19 or 20 C atoms (C20), while increasing the yield of the more volatile dimers with 14 or 15 C atoms (C15), thereby reducing the rate of new particle formation.

Summary and conclusions
We observed a high frequency of NPF events during four weeks of measurements in June 2019 at two pristine sites in the 450 temperate/boreal transition zone of British Columbia, Canada. At the Eagle Lake site, every day was an event day, and many days had multiple NPF events. At Nazko River, the NPF frequency was lower, yet still, 50% of days were event days. In contrast to most sites studied previously, NPF occurred as frequently during nighttime as in daytime, with 14 out of a total of 27 events taking place at night.
Airmass trajectory analysis showed that most of the sampled airmasses had arrived from the Pacific Ocean and traveled 455 over land for 1 -2 days before arriving at our sites. The terrestrial fetch area has an extremely low population density and no industrial activity, resulting in essentially pristine atmospheric conditions. While a marine contribution to NPF cannot be excluded at our sites due to the limited instrumentation available for our campaign, it is likely not of significance, given that there was no preference for daytime nucleation (as would be expected for HIO3 or H2SO4 as nucleating species) and that NPF events were seen just as frequently on days with no previous ocean contact. 460 The average condensation sink at the start of events was 0.0025±0.0011 at EL and 0.0012±0.0010 s -1 at NR, well within the range where NPF has been observed at other temperate/boreal sites (Asmi et al., 2011;Dada et al., 2017;. The particle growth rates, with averages of 2.1±1.3 and 3.2±1.6 nm h -1 at EL and NR, respectively, were also within the range typically observed during spring/summer at temperate/boreal sites in Eurasia and North America Nieminen et al., 2018). Because of the relatively high lower cutoff diameter of our instrumentation 465 (10 nm), we could not obtain the actual nucleation rate, J * . The formation rates for particles at 10 nm, J10, averaged 0.32±0.16 and 0.12±0.11 cm -3 s -1 at EL and NR, respectively, also comparable to the Scandinavian and Siberian temperate/boreal sites listed by Nieminen et al. (2018).
While the limited data available from our campaign does not allow us to draw firm conclusions about the nucleating species responsible for NPF at our sites, several lines of evidence point to pure organic nucleation as the dominant mech-470 anism. The strongest argument comes from the fact that nighttime NPF was as frequent as daytime NPF, analogous to other sites where organic nucleation has been shown to dominate, and ruling out photochemical production of H2SO4 as source of nucleating species at least during the night. The lack of anthropogenic sources of SO2 and the independence of NPF from marine influence on the sampled airmasses also argues against nucleation driven by H2SO4. Finally, the presence of abundant monoterpene sources in the fetch (to the point that they sometimes could be detected by their odor) and 475 the potential dependence of the NPF frequency on the presence of isoprene-emitting vegetation also support organic nucleation as the dominant mechanism.
Our results are consistent with the model-derived importance of pure organic NPF in remote regions (Gordon et al., 2017;Zhu and Penner, 2019), however, they raise important questions about the extent to which pristine NPF conditions still exist in the present-day atmosphere. At the vast majority of remote sites studied so far, NPF is exclusively a daytime 480 phenomenon, suggesting dominance of H2SO4 as the controlling species. Would these sites have been dominated by pure organic nucleation in pre-industrial times, with frequent nighttime NPF? And, why is NPF so infrequent at the remote https://doi.org/10.5194/acp-2021-838 Preprint. Discussion started: 18 October 2021 c Author(s) 2021. CC BY 4.0 License. subboreal ZOTTO site in central Siberia, which is surrounded by vast coniferous forest (Wiedensohler et al., 2019;Uusitalo et al., 2021)? Possibly, the large distance to anthropogenic sources of SO2 has allowed it to be fully converted to sulfate aerosol, removing the source of H2SO4 while providing a condensation sink that prevents organic nucleation. 485 Is there a sequence of regimes, where at truly pristine conditions, pure organic nucleation dominates, followed by H2SO4driven nucleation with organic-dominated growth in the presence of small amounts of anthropogenic SO2, again followed by a "nucleation valley of death" where the CS from pollution aerosols suppresses nucleation, and finally the highly polluted regime where there is so much SO2 that H2SO4-driven nucleation can overcome the suppression by the elevated

CS? 490
Our study shows that there is a need for in-depth investigations at pristine continental sites to address these questions. A full set of instrumentation to identify nucleating species and precursors is required to elucidate nucleation and growth mechanisms. Also, our measurements were limited to a relatively short period in late spring to early summer, when previous studies at other sites have shown that both NPF events and HOM dimer precursor concentrations have their seasonal maxima (Nieminen et al., 2018;Sulo et al., 2021). Future studies should include the development of a site for 495 continuous long-term observations to investigate the seasonal and interannual variability of NPF over the remote North American forest regions.