Characterizing the hygroscopicty of growing particles in the Canadian Arctic summer

. The impact of aerosols on clouds is a well-studied, although still poorly constrained, part of the atmospheric system. New particle formation (NPF) is thought to contribute 40–80% of the global cloud droplet number concentration, although it is extremely difﬁcult to observe an air mass from NPF to cloud formation. NPF and growth occurs frequently in the Canadian Arctic summer atmosphere, although only a few studies have characterized the source and properties of these aerosols. This study presents cloud condensation nuclei (CCN) concentrations measured on board the CCGS Amundsen in the eastern Canadian 5 Arctic Archipelago from 23 July to 23 August 2016 as part of the Network on Climate and Aerosols: Addressing Uncertainties in Remote Canadian Environments (NETCARE). The study was dominated by frequent ultraﬁne particle and/or growth events, and particles smaller than 100 nm dominated the size distribution for 92% of the study period. Using κ -K ohler theory and aerosol size distributions, the mean hygroscopicity parameter ( κ ) calculated for the entire study was 0.12 (0.06–0.12, 25th– 75th percentile), suggesting that the condensable vapours that led to particle growth were primarily non-hygroscopic, which 10 we infer to be organic. Based on past measurement and modelling studies from NETCARE and the Canadian Arctic, it seems likely that the source of these non-hygroscopic, organic, vapours is the ocean. Examining speciﬁc growth events suggests that the mode diameter ( D max ) had to exceed 40 nm before CCN concentrations at 0.99% SS started to increase, although a statistical analysis shows that CCN concentrations increased 13–274 cm − 3 during all ultraﬁne particle and/or growth times (total particle concentrations > 500 cm − 3 , D max < 100 nm) compared to Background times (total concentrations < 500 cm − 3 ) 15 at SS of 0.26–0.99%. This value increased to 25–425 cm − 3 if the growth times were limited to times when D max was also larger than 40 nm. These results support past results from NETCARE by showing that the frequently observed ultraﬁne particle and growth events are dominated by a highly non-hygroscopic fraction, which we interpret to be organic vapours originating from the ocean, and that these growing particles can increase the background CCN concentrations at SS as low as 0.26%, thus pointing to their potential contribution to cloud properties and thus climate through the radiation balance.

. Track of the NETCARE ship cruise (black and coloured points). Rectangles show the three legs and red triangles denote communities. The colour of the points show times used in calculations of Background (grey) and Growth > 40 nm (light blue). Additional times included in the UFP and Growth calculations are shown in magenta.

CCN data analysis
The first 30 s of CCN data at every SS were excluded because the chamber temperatures require time to stabilize when changing to a new SS. Only values with the temperature stabilized flag, as reported by the CCNC, were used in this analysis. Additionally, 125 the sample times that were previously identified to be contaminated by Collins et al. (2017) were excluded from the CCNC data.
It has been reported that the total uncertainty in the estimated CCN concentration above 100 cm −3 varies between 7 and 16% due to factors such as temperature, pressure, flow, etc. (Moore et al., 2011). One second CCN concentrations were matched to the 5 min SMPS sample times and the median concentrations calculated. The median was used to avoid the influence of outliers that may have remained after filtering and are, on average, only 1.2% lower than mean concentrations. Of the 2061 130 concurrent SMPS and CCN observations, 69 were excluded due to high variability (standard deviation to median ratio was greater than 0.5) or the concentrations exceeded 2000 cm −3 .

κ calculations
The hygroscopicity parameter (κ) was calculated using the aerosol size distribution, CCN concentrations and κ-K ohler theory (Petters and Kreidenweis, 2007): where A = (4M w )/(RT ρ w ), M w and ρ w are the molar mass and density of water, respectively, while R and T are the ideal gas constant and the absolute temperature, respectively. D a is the activation diameter and represents the diameter above which particles are large enough to activate as CCN at a saturation ratio of S c . D a was determined by integrating aerosol size distributions downwards from the largest size bin until the cumulative particle number concentration was equal to the corresponding 140 CCN concentration. The size bin at which this occurred was the D a . This method has been previously used in studies where it is not possible to size select the particles before they are counted by the CCNC (e.g. Collins et al., 2013;Gao et al., 2020).

Defining growth and background times
To determine the overall contribution of growing UFP on the CCN concentrations throughout this study, UFP and/or growth times were defined as times when the total number concentrations from the SMPS (N T ) > 500 cm −3 and D max ≤ 100 nm 145 (magenta and light blue points in Fig. 1), where D max is the statistical mode diameter in the size distribution (i.e. the diameter that had the maximum normalized concentration). An additional period when N T > 500 cm −3 and 40nm < D max < 100nm was defined as Growth > 40 nm (light blue points in Fig. 1), to examine CCN concentrations when the particles had grown larger than 40 nm but remained smaller than 100 nm. Both of these periods were compared to periods when minimal contributions from UFP and growth were expected. These Background periods were defined as times when N T ≤ 500 cm −3 (grey points in 150 Fig. 1). To retain clarity, periods when N T > 500 cm −3 and D max ≤ 40 nm (magenta points in Fig. 1) will be referred to as Growth < 40 nm in the legends, although these events do not always show growth and are not considered separately.
3 Results and discussion

General overview
The aerosol size distributions measured by the SMPS throughout the study and D max are shown in Fig. 2a, and the CCN 155 concentrations and N T are shown in Fig. 2b. The three legs are denoted by the shaded boxes above panel a. Summary statistics of CCN concentrations for the entire study and each leg are shown in Table 1. As seen in Fig. 2a, the study was characterized by frequent UFP and growth events, which caused N T to vary by three orders of magnitude. Collins et al. (2017) identified 14 UFP and growth events during this cruise which accounted for 41% of the sample times. However, particles smaller than 100 nm dominated the SMPS number size distribution for 92% of the study, suggesting that most of the aerosol particles observed 160 throughout the study had undergone secondary formation processes such as condensational growth. As such, even if particles were not actively growing at a given time, their CCN-activity and inferred chemical composition could still provide insight into the vapours that contributed to particle growth.
CCN concentrations at the higher SS usually varied according to N T , showing evidence that UFP and growth can lead to increased CCN concentrations. However, the median activation ratio (AR), defined as the CCN concentration / N T , was only 165 0.38 at the highest SS of 0.99%, suggesting that most of the particles were either too small and/or non-hygroscopic to activate.   (Table 2).
Over the entire study, the median and interquartile range of CCN concentrations increased with increasing SS (see Table 1), with a median CCN concentration of 29 cm −3 at SS of 0.17% compared to 98 and 228 cm −3 at 0.44% and 0.99%, respectively.

170
To put our observations into a global perspective, CCN and N T concentrations observed at a select number of polar and remote sites are shown in Table 2. This list is by no means exhaustive, but shows that overall, our CCN concentrations are generally higher than other Arctic observations at Ny Alesund (Jung et al., 2018) and the Central Arctic Ocean .
A notable exception is the aircraft observations over the Canadian Arctic Archipelago in 2008 which were influenced by biomass burning and transport of industrial emissions and resulted in higher aerosol concentrations (Lathem et al., 2013). Calculated κ values and summary statistics are also presented in Table 1. Overall, the values were very low, with medians for the entire study ranging from 0.07-0.11 over the six SS. Values were highest at the lowest SS (0.08-0.21), suggesting 180 that the larger particles were more hygroscopic, which is consistent with more processing in the larger aerosols or a different source resulting in a higher hygroscopicity, such as sea spray. The values also increased throughout the study, with the lowest values (0.04-0.08) corresponding to the first part of the study when the ship was in the warmer and more biologically-active waters of Baffin Bay where more UFP and growth events were observed, and the highest values (0.18-0.21) corresponding to

195
Based on the median calculated κ values from the entire study, we infer that >80% of the aerosol volume fraction was composed of a non-hygroscopic component, which we interpret as being organic. Previous modelling results have shown that the air sampled during our study was influenced by source regions within the Arctic circle which is mostly marine (Collins et al., 2017;Burkart et al., 2017), suggesting that these non-hygroscopic aerosols were influenced by the water. Other recent studies in the Canadian Arctic have inferred that VOCs are emitted from the ocean (Mungall et al., 2017), that the source . Together, these results suggest that the condensing material contributing to particle growth is non-hygroscopic, likely organic originating from marine biological sources.

Influence of particle growth on CCN
To further explore the potential role of UFP and growth on climate, CCN concentrations and D a were examined for specific events when UFP and growth were observed. Figure 3    obviously increased after 30 July 00:00 when D max exceeded 40 nm (light blue shading), and did not increase when bursts of smaller particles appear at 31 July 18:00 and 03 August 08:00 (magenta shading).
To determine the overall contribution of growing UFP on the CCN concentrations throughout this study, Fig. 5 Figure 6 shows the mean CCN concentrations at each SS for the full study period (solid black circles) and the three regions considered in the study. The lines in Fig. 6 represent the best fit of the empirical Twomey parameterization often used in models 240 to relate CCN to SS using:

Twomey parameterization
where C represents the CCN concentration at 1% SS and k is the power law exponent (Twomey, 1959). The parameters C and k are widely used in cloud microphysical models as they provide information about size and composition of the background aerosol concentration (Seinfeld and Pandis, 1997), although more recent parameterizations have adopted Eq. 2 to include more 245 information about the aerosol microphysical parameters (Khvorostyanov and Curry, 2006). The Twomey parameters calculated for our study are listed in Table 3. The C parameter for the full study period and the Baffin Bay and Nares strait regions are very similar (between 335 to 455 cm −3 ), whereas it was much lower (172 cm −3 ) for the Resolute Bay region, reflecting the overall lower N T during Leg 3. With the exception of the last leg, these values are all significantly higher than the 100-140 cm −3 determined for the Zeppelin station at Ny Alesund for July and August (Jung et al., 2018), likely due to the greater 250 aerosol concentrations caused by UFP and growth during our study resulting in more CCN at 1% SS. Similarly, the maximum k estimated at Ny Alesund was in July, with a value of 0.5606, which is lower than the values estimated for our study, especially the Baffin Bay and Nares Strait regions (1.2-1.4). This can also be attributed to the persistent UFP and/or growth events during our study since a greater number of small particles can activate as CCN at higher SS. In contrast, the Resolute Bay region, which had less UFP and growth events, had a much lower k of 0.81, showing that the CCN spectrum was less sensitive to 255 changes in SS.

Conclusions
This study reports CCN concentrations measured in the eastern portion of the Canadian Arctic Archipelago during the NET-CARE campaign onboard the CCGS Amundsen. The observations reported here took place from 23 Jul -23 Aug 2016 when UFP and/or growth events were highly prevalent, with particles smaller than 100 nm dominating the particle number size 260 distribution for 92% of the study. These UFP and/or growth events resulted in high particle concentrations which were also reflected in increased CCN concentrations, suggesting that the frequently-observed small, growing particles have the ability to contribute to CDNC and therefore the radiative budget in the Arctic. The mode diameter of the growing particles generally