Concentrations, Particle-Size Distributions, and Dry Deposition Fluxes of Aerosol Trace Elements over the Antarctic Peninsula

Size-segregated particulate air samples were collected during the austral summer of 2016-2017 at Palmer Station on the Anvers Island, west Antarctic Peninsula, to characterize trace elements in aerosols. Trace elements in aerosol samples, including Al, P, Ca, Ti, V, Mn, Ni, Cu, Zn, Ce, and Pb, were determined by total digestion and sector field inductively coupled plasma mass spectrometer (SF-ICP-MS). The results show that these elements are derived primarily from three sources: (1) 10 regional crustal emissions, (2) long-range transport, and (3) sea-salt aerosols. Elements dominated by a crustal source (Al, P, Ti, V, Mn, Ce) with EFcrust<10 were accumulated mostly in the coarse-mode particles (>1 μm) and peaked at 2.5–7.8 μm in diameter, reflecting the contributions of regional crustal sources. Non-crustal elements (Ca, Ni, Cu, Zn, Pb) showed EFcrust>10. Aerosol Pb was accumulated primarily in fine-mode particles, peaking at 0.078–0.25 μm, and likely was impacted by air masses from South America based on air-mass back trajectories. However, Ni, Cu, and Zn were not detectable in most size 15 fractions and didn’t present clear size patterns. Sea-salt elements (Ca, Na, K) showed single mode distribution and peaked at 2.5–7.8 μm. The estimated dry deposition fluxes of dust for the austral summer, based on the particle size distributions of Al measured at Palmer Station, ranged from 0.65 to 28 mg m-2 yr-1 with a mean of 5.5 mg m-2 yr-1. The estimated dry deposition fluxes of the target trace elements in this study were lower than most fluxes reported previously for coastal Antarctica and suggest that atmospheric input of trace elements through dry deposition processes may play a minor role in determining trace 20 element concentrations in surface seawater over the continental shelf of the west Antarctic Peninsula.


Sampling and sample treatment
Aerosol size-segregated samples were collected during austral summer from November 19, 2016 to January 30, 2017 at Palmer Station (64.77° S, 64.05° W, Figure 1), located on the southwestern coast of Anvers Island off the Antarctica Peninsula.
Sampling was conducted using a ten-stage Micro-Orifice Uniform Deposit ImpactorTM (MOUDI, MSP Corp., MN, USA) with 70 a 30 L min-1 flow rate. The 50% cut-off aerodynamic diameters of MOUDI are 0.056, 0.10, 0.18, 0.32, 0.56, 1.0, 1.8, 3.2, 5.6, 10, and 18 µm. In this study, size fractions ≤1.0 µm were summed to operationally define fine-mode particles and those ≥1.8 µm were summed to define coarse-mode particles, similar to previous studies which operationally divided aerosol particles into fine and coarse fractions using a cut-off size of 1.0-3.0 µm (Siefert et al., 1999;Chen and Siefert, 2004;Buck et al., 2010;Gao et al., 2019). The aerosol sampler was placed on a sampling platform which was ~300 m east from the station center and 75 ~3 m high above the ground (~20 m above sea-level) in "Palmer's backyard" (Gao et al., 2020). To avoid local contamination from the research station, a wind control system was set up to pause aerosol sampling when wind direction inside the sector ±60° from the direction of the station and wind speed <2 m s-1. Due to extremely low concentrations of aerosol trace elements over Antarctica, the sampling duration of each sample was approximately one week (Table 1).
After each sampling, the MOUDI sampler was carried back to the lab in the research station for sample filter changing and 80 sampler cleaning in a Class 100 cleanroom flow bench. Aerosol samples were stored frozen in pre-cleaned Petri dishes at -20℃ before analyses. A total of 8 sets size-segregated aerosol samples were collected on Teflon filters (1 µm pore size, 47 mm diameter, Pall Corp., NY, USA). A full set of blank filters (n = 11) was mounted on the sampler, carried to the sampling platform without running the sampler, and thus defined as field blanks. Meteorological conditions were recorded in situ by a weather station (Campbell Scientific, UT, USA) installed on the same platform (Table 1). 85

Trace elements in aerosols
Aerosol samples were analyzed for the concentrations of trace elements by an Element-1 sector field inductively coupled plasma mass spectrometer (SF-ICP-MS, Thermo-Finnigan, Bremen, Germany) at the Department of Marine and Coastal samples, at least two procedural blanks were processed in the same way as the samples to monitor for possible contamination.
During the ICP-MS analysis, duplicate injections of sample solutions were made every ten samples to check the instrument precision. The recovery of this analytical protocol was estimated by 7 separate digestions of the Standard Reference Material 100 (SRM) 1648a-urban particulate matter (National Institute of Standard and Technology, MD, USA) ( Table 2). The method limits of detection (LOD) were calculated as three times the standard deviation of 11 field blanks and a 200 m3 representative sampling volume ( Table 2). The medians of %blank in samples for detectable trace elements were calculated for quality control (Table 2). Elements with concentrations lower than the LOD, including Cr, Co, Cd and Sb, were measured but are not reported or discussed. Aerosol Fe concentrations and Fe solubility were measured in these samples and were reported in Gao et al. 105 (2020). Therefore, they are not included in this paper.

Ionic tracers in aerosols
The concentrations of water-soluble Na and K in aerosols were analyzed by ion chromatography (IC) (ICS-2000, Dionex, CA, USA) with an IonPac CS12A (2 × 250 mm2) analytical column at the Atmospheric Chemistry laboratory at Rutgers University.
The cations Na and K were used as tracers to estimate the portion of aerosols derived from seawater and biomass burning, 110 respectively. Sample processing for IC analysis was similar to the method used by Zhao and Gao (2008) and Xu et al. (2013).
Briefly, a quarter of the sample filter was transferred to a plastic test tube and leached with 5.0 mL Milli-Q water in an ultrasonic bath for 20 minutes at room temperature. Before being injected into the IC, the leachate was filtered through a PTFE syringe filter (0.45 µm pore size, VWR, PA, USA). The method LODs for Na and K based on 7 blanks and a 200 m3 representative sampling volume were 2 and 1 ng m-3, respectively. The precision of the analytical procedures based on seven 115 spiked samples was <±1%.

Enrichment factors
To achieve an initial estimate of the possible sources for trace elements, enrichment factors relative to upper continental crust (EFcrust) were calculated, using the equation: 120 where (Xi/XAl)sample is the mass concentration ratio of element i to the crustal reference element, Al, in aerosol samples, and (Xi/XAl)crust is the abundance ratio of element i to Al in the upper continental crust (Taylor and McLennan, 1995). The crustal reference element Al has been widely used to calculate crustal enrichment factors in the Southern Ocean and Antarctica (Zoller et al., 1974;Lowenthal et al., 2000;Xu and Gao, 2014). When the EFcrust is greater than 10, the element likely has additional 125 contributions from other sources (Weller et al., 2008). https://doi.org/10.5194/acp-2020-651 Preprint. Discussion started: 26 August 2020 c Author(s) 2020. CC BY 4.0 License.

Atmospheric dry deposition flux estimation
Dry deposition flux (Fd, mg m-2 yr-1) of each element in aerosols was calculated from the air concentration (Ce, ng m-3) and dry deposition velocity (Vd, cm s-1): where 0.315 is a unit conversion factor (Gao et al., 2013). The Vd for each trace element was computed by dry deposition rates (Vdi, cm s-1) and particle distribution ratios (Pdri, %) following the equation: The Pdri was derived from the concentrations of trace elements in different size fractions and Vdi was estimated using a combination model of Williams (1982) and Quinn and Ondov (1998). This model includes the effects of wind speed, air/water 135 temperature difference, sea surface roughness, spray formation in high wind speed and relative humidity. Meteorological parameters used for estimating dry deposition rates were measured in situ by the weather station with 1 min temporal resolution that was converted to 1 h averages. Sea surface temperature data was obtained from the Palmer Station Long-Term Ecological Research (LTER) study (https://oceaninformatics.ucsd.edu/datazoo/catalogs/pallter/datasets/28). Dry deposition rates of coarse-mode particles were dominated by gravitational settling, whereas the dry deposition rates of smaller particles were 140 controlled by environmental factors (Figure 2). The estimation of dry deposition flux carries substantial uncertainty due to the limited sample masses and the assumptions inherent to the Vdi estimation Gao et al., 2020). Dry deposition fluxes were calculated for the trace elements showing clear particle size distribution patterns, including Al, P, Ca, Ti, V, Mn, Ce, and Pb. For Ni, Cu, and Zn that didn't show clear size distributions, however, the ranges of their dry deposition fluxes were also estimated by applying their mean concentrations to the lowest (Pb, 0.11 ± 0.12 cm s-1) and highest (Al, 0.49 ± 0.28 145 cm s-1) dry deposition velocities. The dry deposition fluxes of dust were also estimated based on the concentrations and particle size distributions of Al in aerosols, assuming that Al accounted for 8% of dust mass (Taylor and McLennan, 1995).

Air mass back trajectories
To explore possible source regions of air masses affecting trace elements in aerosols collected at Palmer Station, the NOAA Hybrid Single Particle Lagrangian Integrated Trajectory Model (HYSPLIT) was used to calculate 72-hour air mass back 150 trajectories for each sampling duration (Rolph et al., 2017). In this study, the HYSPLIT model was driven by the meteorological data from the Global Data Assimilation System (GDAS) with a 0.5-degree resolution. Each air mass back trajectory was calculated at 3-hour intervals and started from one-half mixed boundary layer height. The back trajectories during each sampling period were used to calculate trajectory frequencies which were defined by the following equation:

Enrichment factors of trace elements
Crustal enrichment factors of trace elements in aerosols were calculated as the first step of source identification (Figure 3).
Two major EF groups were found, representing crustal and non-crustal elements as follows.

Crustal elements (P, Ti, V, Mn, Ce) 160
The values of EFcrust for Ti, V, Mn, and Ce in aerosol samples were less than 10, indicating that a crustal source is the dominant source for those elements ( Figure 3). As typical lithogenic elements (Boës et al., 2011), Ti and Mn in aerosols over oceanic regions are usually derived from natural dust emissions (Shelley et al., 2015;Marsay et al., 2018;Buck et al., 2019). Likewise, V and Ce in aerosols over the South Pole were reported dominated by crustal weathering (Zoller et al., 1974). In addition to crustal emissions, long-range transport may deliver some portion of these trace elements from remote sources to Antarctica 165 (Wagener et al., 2008). For example, aerosol Ce and Mn derived from anthropogenic emissions were thought to be contributed by additives in vehicle fuels (Fomba et al., 2013;Gantt et al., 2014), and V in aerosols was found associated with ship emissions due to the use of heavy oil fuel (Chen and Duce, 1983;Rahn and Lowenthal, 1984;Zhan et al., 2014). However, the EFcrust results from this study suggest that nearby fuel combustion did not cause significant enrichment of V in aerosols at Palmer Station. A similar phenomenon was observed at McMurdo Station where light-weight fuel oil was used that was not a 170 significant source of V (Lowenthal et al., 2000). We conclude that Ce, Mn, V, and Ti observed at Palmer Station were derived primarily from crustal sources.
The range of EFcrust for P was between 2 and 8, relatively higher than that of the other crustal elements. In Antarctic soils, P has been widely studied (Campbell and Claridge, 1987;Blecker et al., 2006;Prietzel et al., 2019). Around the Antarctic coast, including the northern end of the Antarctic Peninsula, high P inputs to the surface soil were found in seabird colonies 175 (Otero et al., 2018), and a high enrichment of P (EFcrust = 33) was reported previously at King George Island (Artaxo et al., 1990). The closest potential source, the penguin colony on Torgerson Island, is only about 1 km from Palmer Station. Given that regional wind-induced dust likely affects aerosol composition over the Antarctic Peninsula (Asmi et al., 2018;Gao et al., 2020), soil-derived P is likely to be emitted to the atmosphere. In addition, biogenic activities in Antarctica also produce abundant gaseous P, such as phosphine, through anaerobic microbial processes in soils and animal digestives, and phosphine 180 gas can be transformed to other low-volatile P-containing compounds in the atmosphere or soils (Zhu et al., 2006). A portion of P could derive from anthropogenic emissions, such as agricultural and industrial activities, through long-range transport as observed over the East China sea with EFcrust of 35 for P (Hsu et al., 2010). Primary biogenic aerosols, sea-salt aerosols, and volcanic emissions could also contribute P to the atmosphere, causing an elevated EFcrust for P (Zhao et al., 2015;Trabelsi et al., 2016).

Non-crustal elements (Ca, Ni, Cu, Zn, Pb)
The enrichment factors of atmospheric Ni, Cu, Zn, and Pb relative to the crustal element Al, were found to be greater than 10 in some samples, suggesting contributions from non-crustal sources during the corresponding sampling periods in this study ( Figure 3). High enrichments of these elements have commonly been found in regions where the aerosol composition is largely controlled by long-range transport, including the polar regions (Boutron and Lorius, 1979;Maenhaut et al., 1979;Shevchenko 190 et al., 2003;Xu and Gao, 2014;Zhan et al., 2014;Kadko et al., 2016). Aerosol Cu, Zn, and Pb are primarily associated with non-ferrous metal production (Pacyna and Pacyna, 2001;Shevchenko et al., 2003;Laing et al., 2014). Strong variations in the EFcrust of aerosol Cu, Zn, and Pb observed at the Dome C, Antarctica, over the 20th century, were attributed to volcanic activities (Boutron and Lorius, 1979). On the other hand, heavy oil combustion was found to be the major source of aerosol Ni and V, and V/Ni ratios are usually used to identify shipping emissions in Sweden (Isakson et al., 2001). Nevertheless, the 195 V/Ni measured at Palmer Station ranged from 0.01 to 0.2, much lower than the V/Ni = 3.2 ± 0.8 characteristic of the discharge from ship engines (Viana et al., 2009;Viana et al., 2014;Celo et al., 2015). Hence, despite the recent increase in tourist ship traffic, it looks that Palmer Station was barely impacted by ship emissions, which is consistent with the EFcrust of V. As a common crustal element, Ca accounts for about 3.5% of the weight of Earth's crust, while Ca is also a conservative major ion in seawater. The EFcrust of Ca in aerosol samples collected during this study varied from 11 to 48, indicating an enrichment 200 from other sources.

Concentrations of trace elements
The concentrations of Ca and Al, the two major elements measured in this study, were one to several orders of magnitude higher than other elements (Table 3). To place all of the results of this study into the context of past investigations, we provide a visual comparison of measured concentrations of aerosol trace elements over Antarctica (Figure 4). 205

Crustal elements (Al, P, Ti, V, Mn, Ce)
The concentrations of Al in aerosols varied from 1.2 to 7.9 ng m-3 with an average of 4.3 ng m-3 during the period of this study.
These concentrations are lower than the mean Al values of ~13 ng m-3 reported previously at King George Island (62.02° S, 58.21° W), the northern end of Antarctic Peninsula (Artaxo et al., 1992), but were slightly higher than the 2-year mean Al concentrations of 1.9 ng m-3 observed at King Sejong Station (62.13° S, 58.47° W) (Mishra et al., 2004), ~385 km northeast 210 of Palmer Station, the mean of 0.194 ng m-3 observed at Larson Ice Shelf to the southeast of Palmer Station (Dick et al., 1991), the 5-year summer mean of 1.3 ng m-3 in East Antarctica (Weller et al., 2008) and the summer average of 0.57 ng m-3 measured at the South Pole (Zoller et al., 1974;Maenhaut et al., 1979) (Figure 4b). All these results were much lower than the average Al concentrations of 180 ng m-3 and 250 ng m-3 observed on the two sites at McMurdo Station (Mazzera et al., 2001) , 1990). Further, the P values from this study are much lower than the P concentrations previously observed on the east coast of Asia (Cohen et al., 2004), the Australian coast (Maenhaut et al., 2000;Vanderzalm et al., 2003), and Europe (Virkkula et al., 1999) which have been heavily affected by biomass burning, industrial activities and other anthropogenic sources. Comparing global aerosol P concentrations, we find that P concentrations over the Antarctic Peninsula are in the same 235 range as those over the Central Pacific Ocean (Chen, 2004). Confirming that Palmer Station was little influenced by aerosols derived from biomass burning through long-range transport, the calculated non-sea-salt-K was indistinguishable from zero.
These results agree well with the air mass back trajectories, which indicate that most samples collected during this study were barely affected by South America (Figure 5). In addition, for most of the 72-hour air-mass back trajectories, the highest frequencies were found around northern Antarctic Peninsula, suggesting that aerosol crustal elements observed at Palmer 240 Station were impacted by sources in that region ( Figure 5).

Non-crustal elements (Ca, Ni, Cu, Zn, Pb)
The highest concentration of Ni observed during this study was up to 320 pg m-3 with an average of 75 pg m-3, while the lowest  (Figure 4d). In addition, Na, as a tracer of sea-salt, and K, as a tracer of biomass burning , were used to further evaluate the contribution of trace elements from other sources. The concentrations of Na and K showed average concentrations of 890 ± 310 and 28 ± 11 ng m-3, respectively. The 270 Na/K and Na/Ca ratios were 32 ± 3.5 and 31 ± 5.5 which are close to their average mass ratios, 27 and 26, in seawater (Millero, 2016). The results suggest that Ca was dominated by sea-salt aerosol and Palmer station was barely affected by the biomass burning.

Aerosol particle-size distributions of trace elements
The concentrations of trace elements in aerosols observed during this study varied as a function of particle size ( Figure 6). 275 Trace elements in aerosols can be classified into three groups based on their potential dominant sources, with each group showing unique size distribution pattern: (1) elements from the continents, due to crustal weathering and wind-induced resuspension, (2) elements from combustion or other anthropogenic sources, as a result of long-range transport, and (3) elements from the ocean, through bursting bubbles of seawater, by which droplets are ejected into the atmosphere.
The first group includes Al, P, Ti, V, Mn, and Ce, which were derived from crustal sources and accumulated in the coarse-280 mode particles. This group was dominated by particles with diameters larger than 2.5 µm (Figure 6), consistent with the coarsemode dominance of natural dust (Adebiyi and Kok, 2020). For Al, the highest mass concentrations in sample M1, M2, M4 and https://doi.org/10.5194/acp-2020-651 Preprint. Discussion started: 26 August 2020 c Author(s) 2020. CC BY 4.0 License.
M7 were found in the >18 µm fraction whereas in the other 4 sample sets, Al peaked at around 2.5-7.8 µm. The notably high concentrations of Al in coarse-mode particles found in M1, M2 and M4 may indicate stronger contributions from regional or local crustal materials. In most sample sets, P, Ti, V, Mn and Ce peaked at around 2.5 to 7.8 µm. Additionally, M2 and M4 285 showed bimodal distribution for V which had a small peak in fine-mode at around 0.14-0.44 µm. A similar distribution was found for Mn in sample M4. Such enrichments of V and Mn in fine-mode particles might hint at a minor contribution of aerosols from distinct sources, such as ship emissions (Lowenthal et al., 2000) or long-range transport of dust and biogenic aerosols (Artaxo et al., 1994;Weller et al., 2008).
The mass distributions of Ni, Cu, Zn, and Pb as the second group were different from those of the elements derived from 290 crustal sources ( Figure 6). Because of the low concentrations and the resulting high blank corrections, the particle size distribution of Ni, Cu, and Zn was not obtained. The mass distribution of Pb showed either a single mode or bimodal distributions in most samples with Pb being accumulated in the fine-mode particles, peaking at around 0.078-0.25 µm. The enrichment of Pb in fine particles is consistent with relatively long residence time and transport distances in the atmosphere (Seinfeld and Pandis, 1998); this was particularly evident in the size distribution of M4, which was impacted by air masses 295 derived from South America (Figure 5c). In addition, sample M4 showed trimodal distribution for Pb, exhibiting 3 peaks at around 0.078-0.25 µm, 0.78-1.4 µm, and 4.4-7.8 µm, suggesting that additional Pb was contributed by distinct sources. The fine-mode Pb exhibited the highest concentration in most samples. Unlike the mass distributions of Pb, other non-crustal elements, including Ni, Cu, and Zn, did not clearly accumulate on either fine or coarse-mode particles, which may indicate that the dominant source of these elements varied during the sampling period. 300 The mass distributions of sea-salt elements (Ca, Na and K) as the third group were dominated by coarse-mode particles with diameters 2.5-7.8 µm (Figure 6), consistent with sea salt aerosols observed elsewhere (Lewis et al., 2004;Zhao and Gao, 2008;Xu et al., 2013). In this study, the correlation between the total concentrations of Ca and Na was strongly positive (R2 ± 0.82, p-value < 0.01), suggesting a strong contribution of Ca from sea salt aerosols.

Atmospheric dry deposition fluxes 305
Aerosol particle size distributions have been used to estimate atmospheric dry deposition fluxes of aerosol Fe to the Southern Ocean and coastal Antarctica (Gao et al., 2013;. Similarly, the dry deposition fluxes of aerosol trace elements were estimated in this study (Table 4). Given that the sea-salt elements were the most abundant, Ca showed the highest dry deposition flux which varied from 0.59 to 6.6 mg m-2 yr-1 with an average of 1.2 mg m-2 yr-1. At the other extreme, due to low concentration and fine-mode particle dominance, aerosol Pb showed ~5000-fold lower dry deposition fluxes with a mean of 310 0.30 ± 0.23 μg m-2 yr-1. The rough estimates of the dry deposition fluxes of Ni, Cu, and Zn at Palmer Station are close to the low deposition fluxes found in the North Atlantic Ocean (Shelley et al., 2017) but considerably lower than the other measurements conducted in mid-and low-latitude regions (Arimoto et al., 2003;Buck et al., 2019). The estimated dry deposition fluxes of total continental dust at Palmer station ranged from 0.65 to 28 mg m-2 yr-1 with a mean of 5.5 mg m-2 yr-through precipitation scavenging accounted for about 40~60% of the total deposition in the coastal and open oceans in the mid and low latitude oceans (Gao et al., 2003), and similar wet deposition fraction was also found in the Southern Ocean and Coastal East Antarctica (Gao et al., 2013). Assuming this wet deposition fraction applies to the Antarctic Peninsula region, we approximate roughly a total dust flux of 10 mg m-2 yr-1. This value is at the lower end of the dust flux range, ~5 to ~50 mg m-2 yr-1, estimated from ice core measurements at James Ross Island on the Antarctic Peninsula over the 20th century (McConnell 320 et al., 2007), but it is higher than dust deposition flux at Dome C (0.2-0.6 mg m-2 yr-1) (Lambert et al., 2012), at Talos Dome (0.70 -7.24 mg m-2 yr-1) (Albani et al., 2012) and the model estimate for this region (1.8-3.7 mg m-2 yr-1) (Wagener et al., 2008). However, this result is far lower than the total dust deposition fluxes in the North and South Atlantic Ocean (minimum 270 and 150 mg m-2 yr-1, respectively) (Barraqueta et al., 2019), the South Pacific Ocean (230 mg m-2 yr-1) (Prospero, 1989), South Indian Ocean (240 mg m-2 yr-1) (Heimburger et al., 2012) and McMurdo Dry Valleys (490 mg m-2 yr-1) (Lancaster, 325 2002), while it is close to the estimated total dust deposition fluxes for the 63°S region (between 62°53'S and 63°53'S, along 170°6'W; 33 mg m-2 yr-1) (Measures and Vink, 2000). Accordingly, the other crustal elements (P, Ti, V, Mn, Ce) were proportional to the dust dry deposition flux and showed extremely low values.
To examine the potential importance of atmospheric dust input to the particulate elemental concentrations in surface waters of the west Antarctic Peninsula shelf region, the maximum possible suspended particulate Al concentrations contributed by 330 dust were estimated. We used the maximum Al dry deposition flux of 2300 µg m-2 yr-1 (Table 4), assumed that dry deposition accounted for 40% of total deposition and assumed no settling loss from the mixed layer of mean depth 24 m (Eveleth et al., 2017) over a 4-month summer season. By this calculation, the accumulated concentration of suspended particulate Al contributed by total atmospheric deposition over this period is 3 nmol kg-1. While suspended Al reaches 135 nmol kg-1 in coastal surface waters close to the peninsula, outer shelf and off-shelf surface waters generally have concentrations <5 nmol 335 kg-1 (Annett et al., 2017), suggesting that this upper limit estimated dust flux could account for a substantial portion of observed surface ocean lithogenic particle concentration. However, the mean Al flux is just 20% of this maximum, and settling loss from the mixed layer over the course of the summer is likely significant. Still, this rough estimate suggests that atmospheric dust deposition, possibly dominated by regional sources on the Antarctic continent itself, may contribute a significant fraction of suspended particulate concentrations of lithogenic elements in the surface waters of the outer shelf and the proximal pelagic 340 Southern Ocean.

Conclusions and Implications
Results from this study indicate that trace elements in aerosols over Palmer Station are derived primarily from crustal and marine sources. Crustal elements (Al, P, Ti, V, Mn, Ce) and sea-salt elements (Ca, Na, K) showed single mode distributions and were accumulated in coarse-mode particles, whereas aerosol Pb that was likely impacted by air masses from South America 345 showed high EFcrust and was accumulated in finer mode particles. Most of the samples collected during this study were impacted by air masses originating around or passing over Northern Antarctic Peninsula, suggesting the regional influence on the https://doi.org/10.5194/acp-2020-651 Preprint. Discussion started: 26 August 2020 c Author(s) 2020. CC BY 4.0 License.
concentrations of aerosol trace elements observed at Palmer Station. Although the Antarctic Peninsula has experienced significant changes in the past decade, greater exposure of regional continental dust sources through glacial ice loss and more frequent shipboard tourism do not seem to drive the concentrations of these trace elements over the west Antarctic Peninsula 350 to higher values than observed in other remote Antarctic sites. In addition, the total dust deposition flux (~10 mg m-2 yr-1) estimated from the Al concentrations in aerosols obtained from this study and the assumption on the wet deposition suggests that dust deposition plays only a minor role in determining the concentrations of trace elements in coastal seawater around the west Antarctic Peninsula but may be more important in offshore regions. As the role of wet deposition is unquantified at present and remains poorly constrained for this region, however, the total deposition fluxes of trace elements may be 355 underestimated, and the importance of the atmospheric deposition of trace elements to the adjacent seawater may need to be re-evaluated.
Data availability. The data used in this paper has been deposited to the U.S. Antarctic Program Data Center. The DOI will be issued and the data will be freely accessible shortly. 360 Author contributions. YG conceived the research. YG and SY prepared field sampling and collected aerosol samples. SF and YG digested samples. KB and RMS conducted sample analyses. SF wrote the 1st draft of the manuscript. YG and RMS edited the drafts. All authors contributed to writing and approved the submission.   ND= not determined in samples because LOD was not exceeded.
The LODs were estimated based on a 200 m3 representative sampling volume. %Blank is the median field blank/sample concentration×100%, and n is the number of detectable samples.