Svalbard is an archipelago in the Arctic Ocean consisting of Spitsbergen, Bear Island, and Hopen, which together cover about 62 000 km2. Ny-Ålesund, whose geographical coordinates are 78∘55′ N, 11∘56′ E, is situated on the western coast of Spitsbergen and is an international center of scientific research and environmental monitoring in the central Arctic.

Aerosol samples were collected from 3 to 23 August 2012, using an individual particle sampler at the Chinese Arctic Yellow River Station, Ny-Ålesund (Fig. S1 in the Supplement). We use copper TEM grids coated with carbon film (carbon type-B, 300-mesh copper, Tianld Co., China) to collect aerosols by a single-stage cascade impactor with a 0.5 mm diameter jet nozzle at a flow rate of 1.0 L min-1. The collection efficiency of the impactor is 50 % for particles with an aerodynamic diameter of 0.3 µm and almost 100 % at 0.5 µm if the density of particles is 2 g cm-3. The sampling duration of each sample varied from 20 min to 2 h, depending on the aerosol dispersion on the film that was estimated by optical microscopy after the sampling. Then we placed the grids in sealed, dry, plastic capsules to prevent contamination. Finally, the samples were stored in a desiccator at 25 ∘C and 20 ± 3 % relative humidity (RH) until analysis (Li et al., 2011a).

According to the sampling time and sample quality, 23 aerosol samples were selected for analysis with TEM. During the sampling periods, temperatures were 1.6–7.3 ∘C; RH, 56–94 %; air pressure, 997.0–1020.7 hPa; and sampling wind speeds, 0–8.9 m s-1. Detailed sampling information can be found in Table S1.

Individual particle samples were examined by JEOL JEM-2100 transmission electron microscopy operated at 200 kV with energy-dispersive X-ray spectrometry (TEM/EDX) and scanning electron microscopy (SEM; Philips XL30) operated at a 20 kV accelerating voltage and a 80 µA filament. EDX spectra were examined within a maximum time of 30 s to minimize potential beam damage and efficiently collect particle elemental composition. TEM grids were made of copper (Cu) and covered by a carbon-reinforced substrate, so Cu had to be excluded from the quantitative analyses of the particles while residual C content was detected and overestimated in EDX spectra of individual particles. Elemental composition, morphology, size distribution, and mixing state have been studied by the TEM/EDX. Both submicron and supermicron particles occur on the TEM grids: coarser particles occur near the center, and finer particles are on the periphery (Li and Shao, 2009). To be more representative, 4–5 areas from the center to periphery of the sampling spot were chosen to analyze all the particles. Altogether we analyzed 1577 aerosol particles to understand the details of their mixing states, both internal and external. We used iTEM software to analyze the TEM images and obtained area, circularity, perimeter, and equivalent circle diameter of individual aerosol particles.

Some typical aerosol particles were analyzed for elemental mapping with a JEOL JEM-2100F TEM with scanning TEM (STEM) operation mode. Elemental mappings were collected in the annular dark-field imaging mode, with the electron beam focused on a corresponding spot of the sample but then scanned over this area in a raster. Elemental mapping has become an important tool in individual particle analysis in recent years (Ault et al., 2012; Li et al., 2013a, b), because it clearly displays the distribution of the detectable elements within each particle.

Individual aerosol particles were analyzed using a nanoscale secondary ion mass spectrometer (nanoSIMS) 50L (CAMECA Instruments, Geneviers, France), an ultra-high vacuum technique for surface and thin-film analysis at the Institute of Geology and Geophysics, Chinese Academy of Sciences. In this study, 12C-, 16O-, 12C14N-, 14N16O2-, 32S-, 35Cl-, and 23Na16O- ions in individual particles were obtained when the Cs+ primary ion beam caused the ionization of atoms within the particles. Furthermore, ion intensity mappings of individual particles with nanometer resolution can show the distribution of different ions. 12C- represents the organic matter in individual particles that excludes the contribution from the carbon substrate. Laboratory-generated NaNO3 particles on silicon substrate analyzed by nanoSIMS show 14N16O2- and 23Na16O- are their markers (Fig. S3).

Atomic force microscopy (AFM) with a tapping mode analyzed aerosol particles under ambient conditions. AFM, a digital NanoscopeIIIa Instrument, can detect the three-dimensional morphology of particles. The AFM settings contain imaging forces between 1 and 1.5 nN, scanning rates between 0.5 and 0.8 Hz, and scanning range sizes at 10 µm with a resolution of 512 pixels per length. After the AFM analysis, composition of the same particles was confirmed by TEM, with twelve SSA particles analyzed by this method. The NanoScope analysis software can automatically obtain bearing area (A) and bearing volume (V) of each analyzed particle according to the following formula: A=πr2=π×d22=πd24→d=4Aπ,V=43πr3=43×πD38=πD36→D=6Vπ3, where d is the equivalent circle diameter and D the equivalent volume diameter.

Additionally, we know the relation between d and D is D=0.7487×d, as shown in Fig. S2. As a result, the equivalent circle diameter (d) of individual aerosol particles measured from the iTEM software can be further converted into equivalent volume diameter (D) based on this relationship.

Based on their different morphology and composition, aerosol particles were divided into four major categories: sea salt (Fig. 1A-a), S-rich (OC coating) (Fig. 1A-b), Ca/Mg–S/N/Cl (Fig. 1A-c) and mineral (Fig. 1A-d). TEM observations indicated that SSA were most abundant from 100 nm to 10 µm in the summer Arctic atmosphere. Figure 1B shows that the relative abundance of SSA in the samples was about 72 % of the total aerosols. This result is consistent with the dominant SSA in different seasons reported for the Arctic atmosphere (Hara et al., 2003; Geng et al., 2010).

Based on their morphology and composition, we can further identify three types of SSA: fresh SSA, partially aged SSA, and fully aged SSA.

To summarize three types of SSA, we make one Na–Cl–O triangular diagram. Figure 5 gives general information about Cl depletion among the fresh, partially aged, and fully aged SSA. Three particle types in the triangular diagram display an interesting distribution: the fresh SSA around the NaCl, the partially aged SSA in the center of triangular (including partial NaCl), the fully aged SSA around NaNO3, and Na2SO4 (chloride full depletion).

Three-day (72 h) back trajectories of air masses were generated using the HYSPLIT model at the Chinese Arctic Yellow River Station during 3–23 August 2012, at an altitude of 500 m above sea level (Fig. 6). Most air masses originate in the Arctic Ocean, and are restricted to this vast marine region during the sampling periods. Figure 6 shows that two groups of back trajectories exhibit different ageing degrees of SSA: one group originates from the central Arctic Ocean and the other one from North America and Greenland. Fractions of the fresh, partially aged, and fully aged SSA are 43.07, 44.53, and 12.41 % in the first group and 18.72, 47.45, 33.83 % in the second group, respectively (Fig. 6). As a result, air masses from North America and Greenland brought large amounts of aged SSA into the Arctic area in summertime.

Figure 7 shows size distributions of the fresh, partially aged, and fully aged SSA. The peaks of fresh and partially aged SSA are 0.68 and 1.30 µm, respectively, and the fully aged ones display a broad size range from 0.9 to 2.0 µm (Fig. 7). Our results show that the average particle sizes gradually increase from fresh SSA, partially aged SSA, to fully aged SSA. Therefore, dry and wet deposition of aerosol particles could be changed following the degree of the particles ageing.

TEM observations classified the SSA into fresh, partially aged, and fully aged SSA in the Arctic atmosphere. Using individual particle analysis of the three types of SSA, it is shown that there is a major change in their internal structure and composition. The STEM further determined elemental mappings of Na, Mg, Ca, Cl, S, O, C, and N in the three kinds of SSA. Elemental mapping can clearly display the elemental distribution within each particle, which indicates the possible heterogeneous chemical reactions on their surfaces (Conny and Norris, 2011; Ault et al., 2012). Figure 8 shows that Cl content decreases and S, O, C and N contents increase from partially aged (particle B) to fully aged SSA (particle C). The Na–Cl–O triangular diagram further shows that Cl in SSA has been depleted and O content increases likely through additional chemical reactions (Fig. 5). Figure 9a shows 35Cl- in the NaCl core and minor 16O- and 32S- in the coating in fresh SSA. Figure 9b shows the absence of 35Cl- and the high intensity of 16O-, 14N16O2-, 32S-, and 23Na16O- in fully aged SSA. These results are completely consistent with the TEM and STEM observations in Figs. 2, 4, and 8. Based on the mappings of 32S- (sulfate), 14N16O2- (nitrate), and 23Na16O- in Fig. 9, the nanoSIMS analysis provides direct evidence of Na2SO4 and NaNO3 formation in fully aged SSA. As a result, sulfate and nitrate accumulated in SSA through atmospheric chemical reactions (e.g., Reactions R1–R4) in the Arctic area, which has been observed in polluted air by Laskin et al. (2012). These mapping results from STEM and nanoSIMS are consistent with the TEM/EDX results shown in Figs. 2–4 and with investigators who found that individual SSA in the Arctic air contained certain amounts of sulfate or nitrate by SEM/EDX and aerosol time-of-flight mass spectrometry (ATOFMS) (Hara et al., 2003; Geng et al., 2010; Sierau et al., 2014). However, we found that individual aged particles simultaneously contained sulfate and nitrate, suggesting individual SSA underwent heterogeneous reactions with different acidic gases during ageing processes in the Arctic atmosphere. A number of studies found that a large fraction of SSA internally mixed with sulfate and nitrate occurred from coastal to background marine air and from equatorial to high latitudes (Andreae et al., 1986; Pósfai et al., 1994; Middlebrook et al., 1998; Hara et al., 2003; Geng et al., 2010; Li et al., 2011b; Yao and Zhang, 2011; Laskin et al., 2012). Interestingly, we noticed that fully aged SSA (Fig. 8C) had an elevated carbon content, suggesting that organic matter might occur in the aged SSA (details in Sect. 4.2). Laskin et al. (2012) showed that SSA can react with secondary organic acids and result in formation of organic salts in the polluted air.

Na mappings show that Na is absent in the coating of fresh SSA (Fig. 8A) but is present in minor amounts in the coating of partially aged SSA (Fig. 8B), which is consistent with the EDX results as shown in Figs. 2–3. In contrast, Mg, S, and O are absent from the core of fresh SSA (Fig. 8A), but are present in certain amounts within the core of partially aged SSA (Fig. 8B). Therefore, we can deduce that MgCl2 coatings of the fresh SSA firstly reacted with acidic gases through chemical Reactions (R1) and (R2) as the fresh SSA were transformed into the partially aged SSA. The presence of MgCl2 coatings in the fresh SSA is particularly interesting because it is known to readily deliquesce at a considerably lower point (33 % RH at 298 K) than that of NaCl (75 %) (Wise et al., 2009). We conclude, therefore, that this MgCl2 coating consistently allowed an aqueous layer to coat the fresh SSA in the Arctic atmosphere, because the ambient RH (56–94 %) exceeded the deliquescence point. Despite the relatively low concentration of MgCl2 in SSA, the liquid surface significantly enhances heterogeneous reactions rates through the uptake of acidic gases (e.g., H2SO4/SO2, and HNO3/NOx) (Liu et al., 2007). Microscopic observations suggest that MgCl2 on fresh SSA likely initiated and promoted the heterogeneous reactions between SSA and acidic gases. To our knowledge, this is the first demonstration of the influence of minor MgCl2 in individual SSA in field studies, although this phenomenon has been reported in laboratory experiments (Zhao et al., 2006; Liu et al., 2007; Wise et al., 2009). In addition, MgCl2 appears to have greater importance in individual Arctic SSA than elsewhere because the Mg2+/ Na+ ratio in SSA increases following a temperature decrease (Hara et al., 2012). In the partially aged particles, the new coating containing MgSO4, Mg(NO3)2, and NaNO3 (Fig. 3) likely remained as a supersaturated liquid coating on the surface of the ambient SSA in the Arctic area (Zhao et al., 2006; Li et al., 2008; Woods et al., 2010). Therefore, the Mg salts in fresh and partially aged SSA are important surfactants to speed up particle ageing in the Arctic area.

In this study we found an abundance of fully aged SSA in this Arctic area (Figs. 4–6). Ault et al. (2013a) found that the Na+, Mg2+ and Ca2+ in individual SSA undergo a spatial redistribution after heterogeneous reaction with nitric acid in the laboratory. We noticed a similar phenomenon in which Na, Mg, and Ca mappings in particle C were different from particles A and B (Fig. 8). Na is enhanced at the surface; Mg and Ca tend to concentrate in the individual particle center. Figure 8C shows that carbon intensity becomes strong in fully aged SSA, similar to the C-rich coating in Fig. 4c–d. These results suggest that the aged SSA probably contain organic matter (details in Sect. 4.2). In addition, most of the fully aged SSA have a Na2SO4 core coated with NaNO3 (with minor Mg(NO3)2 and MgSO4) (Fig. 4). Hygroscopic experiments of individual SSA show that the surfaces of partially and fully aged SSA have earlier deliquescence than that of fresh SSA based on their hygroscopic growth on the substrate. The results indicate that the nitrate coatings likely influence particle surface deliquescence (Fig. S4). In the laboratory, the pure NaNO3 can take up water starting at 25 % RH and grow continuously with increasing RH (Lee and Hsu, 2000). Hu et al. (2010) also illustrated that NaNO3 did not exhibit an obvious deliquescence phenomenon in the hygroscopic experiment. As a result, some partially and fully aged Arctic SSA had liquid surfaces in ambient air (56–94 % RH) and this liquid amounted to certain amounts of water in these aerosol particles. Park et al. (2014) supplied the evidence of water in Arctic particles through the hygroscopicity and volatility tandem differential mobility analyzer. The hygroscopic growth of particles can cause 1.6 to 3.7 times more negative aerosol direct radiative effects (Sand et al., 2013) than particles without it.

Integrated observations of individual SSA through the TEM, STEM, and nanoSIMS provided direct evidence of the occurrence of sulfate and nitrate in aged SSA in the Arctic atmosphere. Laskin et al. (2012) showed that SSA effectively reacted with organic acids in polluted coastal air, leaving behind particles depleted in chloride and enriched in the corresponding organic salts. Moreover, substitution of Cl in SSA by weak organic acids was suggested in the laboratory by Ma et al. (2013). However, whether weak organic acids participated in the chloride depletion remains uncertain in the clean Arctic air.

NanoSIMS technology has been adopted to characterize organic matter in SSA through 12C- and 12C14N- mappings (Fig. 9). 12C- mapping of fresh SSA indicates extremely small amounts of organic matter in its coating (Fig. 9a). Tervahattu et al. (2002) found that organic films on SSA are common in fresh marine aerosols because of bursting bubbles from the spay of the waves. Quinn et al. (2015) also found that fresh SSA are internally mixed with organic matter in the northeastern Atlantic atmosphere using proton nuclear magnetic resonance (1HNMR). In comparison with fresh SSA in Fig. 9a, the 12C- intensity of the aged SSA in Fig. 9b was much enhanced. The only explanation is that more organic matter from the atmosphere was incorporated into the aged SSA. Similar phenomena in aged SSA have been observed by laser mass spectrometry at Cape Grim and the amounts of additional sulfate and organic matter in individual particles may be an indication of particles ageing (Middlebrook et al., 1998). Additionally, with increasing organic matter in SSA, their morphology and crystallization behavior changes (Ault et al., 2013b).

Interestingly, 12C- line scan in aged SSA further indicates that organic matter is mostly restricted to their surfaces (Fig. 9b). The near-edge X-ray absorption fine structure spectroscopy (NEXAFS) carbon K-edge spectrum of the organic phase reveals a dominant contribution from carboxylic acids on the surface of the aged SSA (Laskin et al., 2012). We further found organic matter internally mixed with nitrate on the surface for aged SSA. This finding supports the results from laboratory experiments and conforms to predictions of liquid–liquid phase separation theory (Ault et al., 2013a). As a result, we obtained one conceptual model based on these above findings summarizing the possible SSA ageing processes in the Arctic atmosphere (Fig. 10).

The nanoSIMS analyses of the aged SSA in the Arctic air show that their surface layers commonly contain organic matter. The organic layer may influence trace gas uptake, and subsequently impact trace gas budgets of O3 and NO2 in the Arctic (McNeill et al., 2006). This is the first demonstration of organic-coated SSA in clean Arctic air, which extends the application of heterogeneous reactions to aerosol particles (Fig. 10). Comparisons of fresh and aged SSA in Fig. 9 suggest that these organic coatings likely took part in the chloride depletion during particle ageing. The chloride depletion in the SSA induced by the presence of organic acids should be incorporated into the atmospheric chemistry models for clean marine air, in addition to the coastal urban regions reported by Laskin et al. (2012). Randles et al. (2004) reported that NaCl internally mixed with organic matter can reduce radiative cooling substantially compared to pure NaCl aerosol in the laboratory. In addition, how the organic layer influences hygroscopic and optical properties of the aged SSA in realistic atmospheres remains unknown. Our measurements make clear the need to understand SSA's complexity in Arctic air.

In summary, our results indicate that formation of sulfate, nitrate, and organic salts in the SSA can remove trace gases (e.g., organic acids, H2SO4/SO2, and HNO3/NOx) in the Arctic air, which were mostly emitted by various anthropogenic sources from middle latitudes (Law and Stohl, 2007; Chang et al., 2011). On the one hand, the results of these heterogeneous reactions could change the CCN or IN abilities, optical properties, and further dry and wet deposition of SSA. On the other hand, these surface reactions of SSA could affect the photochemical reactivity in the Arctic air because they release gaseous halogen species (De Haan et al., 1999; Knipping et al., 2000; von Glasow, 2008; Osthoff et al., 2008; Thornton et al., 2010; Phillips et al., 2012).