Trace gases, submicron particle size distributions, and bulk filterable halogen content were measured on Ross Island, Antarctica, in austral spring 2007. During several surface level, partial ozone depletion events, enhanced submicron particle concentrations, and changes in filterable halogens were observed. These events were characterized by ozone depletions of 5–15 ppbv for durations between 6 and 48 h and associated with threefold-to-fourfold increases in submicron particle mass (PM1.0) over backgrounds of approximately 100 ng m−3. Peak particle number densities were centered on a mode at 500–600 nm in diameter, which is consistent with wintertime sea-salt aerosol size distributions. Filterable chloride also increased during these events, consistent with aerosol being of oceanic origin. Ozone depletion and particle enhancement events were accompanied by increasing temperatures and winds, suggesting that halogen-containing aerosol is generated from windblown snow and brine from the snow pack or sea ice near the ice edge.
 During the spring, ozone concentrations in the coastal Antarctic boundary layer dramatically decrease for periods ranging from hours to weeks [Jones et al., 2006]. Observed in both polar regions, these ozone depletion events (ODEs) are thought to be a naturally occurring phenomenon and the result of catalytic reactions of halogens released to the atmosphere from sea salt [Simpson et al., 2007b]. Due to the similarity of timescales for transport and the chemical transformations necessary to maintain enhanced amounts of reactive halogens, there is still significant uncertainty as to the mechanisms for liberation and transport of sea-salt-derived reactive halogens.
 Initial investigations have suggested that brine, first-year sea ice, or frost flowers may be the major sources of reactive halogens [Kaleschke et al., 2004; Sander et al., 2006]. Satellite observations have shown a correlation between areas of increased reactive bromine (i.e., BrO) and regions conducive to the formation of frost flowers [Kaleschke et al., 2004]. However, ozone depletion has also been observed over ice shelves, at significant distances from sea ice or open water. This suggests a non-oceanic source of reactive halogens or the transport of nonreactive forms over long distances before activation into ozone-destroying forms [Jones et al., 2006].
 By concentrating bromine over their windblown surface area, brine pools or frost flowers may be direct sources of gas phase bromine [Rankin and Wolff, 2002]. Conversely, recent evidence suggests that frost flowers themselves may not be a direct source of gas phase halogens [Simpson et al., 2005] and that enhanced BrO is better correlated with first-year sea ice than with frost flowers [Simpson et al., 2007a]. The depletion of sodium sulfate from aerosol hints that frost flowers and brine may still play a role as a source of bromine-enriched sea-salt aerosol, either through fracturing [Rankin et al., 2000] or by contaminating windblown snow with concentrated brine [Roscoe et al., 2011]. Snow on sea ice is another possible source for sea-salt aerosols [Yang et al., 2008], and measurements in the Antarctic have shown such snow to be rich in sea-salt-derived ions [Aristarain and Delmas, 2002].
 While the circumstantial evidence increasingly points to sea-salt aerosols as the transport mechanism for halogens from sea ice, snow pack, or frost flowers, there is little observational evidence for a direct correlation between halogenated aerosol and ozone depletion, particularly in the Antarctic. Early measurements from the Arctic have shown an anticorrelation between filterable bromine and boundary layer ozone [Barrie, 1986; Langendorfer et al., 1999], strongly suggesting that aerosols are a factor in ozone depletion events. A number of studies done in the Antarctic have characterized aerosols and the halogen content in the aerosol fraction using filters, impactors, and optical methods [Hall and Wolff, 1998; Hara et al., 2004; von Glasow and Crutzen, 2004; Fattori et al., 2005; Tomasi et al., 2007]. These measurements show significant halogen content in aerosols both at coastal and inland Antarctic sites, confirming the possibility that sea-salt aerosol may play a significant role in Antarctic ODEs. Unfortunately, none of these aerosol measurements coincided with observations of ozone and gave only limited information about aerosol size distributions. Size distribution is likely to be important in establishing a relationship between aerosols and ozone depletion as both heterogeneous ozone destruction and the release of gas phase halogens may be dependent not only on total aerosol mass but on the available aerosol surface area.
 This study, which combined fast-response and high-resolution observations of ozone and submicron particles during the onset of ozone depletion events near-coastal Antarctica, was undertaken to determine the role of long-range transport of oceanic sources of halogens in ODEs. Enhancements of submicron aerosol with high salt content, important for sustaining high abundances of BrO through subsequent heterogeneous reactions on enhanced particulate surface areas, were typically observed in conjunction with ozone loss.
 Measurements of ozone mixing ratio, submicron particle size distributions, and bulk filterable chloride, bromide, sulfate, and iodide were made during the austral spring of 2007 at the Cosmic Ray (COSRAY) observatory, located near McMurdo Station on Ross Island, Antarctica (77.85°S, 166.73°E). Measurements were made nearly continuously from 27 August to 27 October 2007 (Figure 1) with filter measurements taken on a periodic basis. During this time period the measurement site was approximately 50 km from the sea ice edge, and the terrain between the site and the ice edge was predominantly snow-covered first-year and multiyear sea ice. Particle size distributions were measured with an Ultra High Sensitivity Aerosol Spectrometer (UHSAS), ozone was measured with a Thermo Environmental Instruments Model 49C, and filter samples were returned to Colorado for anion analysis using ion chromatography and inductively coupled plasma mass spectrometry. This study was undertaken as a measurement of opportunity, and future more comprehensive studies are planned to further explore this phenomenon. Further details of the measurement methods and limitations are described in the supporting information.
 The ozone time series for the duration of the campaign is shown in Figure 1 (top). The overall trend for ozone during the measurement period was a gradual decrease from approximately 36 to 26 ppbv punctuated by four distinct longer-duration (> 12 h) partial ODEs.
 Both total submicron aerosol concentrations and particle size distributions were highly variable throughout the course of the campaign, with many distinct particle events. Average background particle mass (PM1.0) from 1 to 9 September, prior to occurrence of ODEs, was 70 ng m−3, with a maximum of 150 ng m−3 (in hourly averaged data). Over these dates, a relatively stable aerosol loading was observed despite a wide variety of weather conditions. In contrast, during the onset of the first ODE (18 September 12:00—all times are Universal Time (UT)), sustained PM1.0 mass concentrations exceeded 400 ng m−3 for a period of 18 h. The highest hourly PM1.0 mass loading of 570 ng m−3 (9 times above typical background values) was observed at 13:00 on 18 September during the onset of an ODE. For the campaign as a whole, there is a correlation between PM1.0 enhancements and ozone depletion, with an r2 value of 0.58 and statistical significance at the greater than 95% confidence level. During the four partial ODEs highlighted in Figure 1, the increase in particle volume was concentrated in a mode with optical diameters between 200 and 1000 nm, with a peak at approximately 600 nm. Furthermore, a general increase in particle volume was observed from 14 October onward, which is again driven by an increase in a 200–1000 nm particle mode. This baseline increase is accompanied by a corresponding decrease in the baseline ozone mixing ratio.
 Particulate chloride mass loading ranged from 12 to 328 ng m−3, with a mean concentration of 85 ng m−3. This is of a similar magnitude to total PM1.0 mass during filter collection periods. Given that the dominant source of chloride is sea-salt-derived NaCl, it is likely that the sea-salt mass including the cation (Na+) is greater than the total PM1.0 mass, which indicates that a significant portion of the measured chloride mass is contained in particles > 1 µm. However, the variability in the ratio between calculated PM1.0 mass and chloride mass, particularly during ODEs (for example, a PM1.0:Cl− ratio of 1.3 during the ODE around 19 September compared to a ratio of 2.2 during the ODE on 20 October), would indicate that behavior of the submicron aerosol is largely decoupled from the behavior of the supermicron sea-salt aerosol.
 The highest chloride concentrations were observed in early September (Figure 1, bottom). These concentrations are on average lower than, yet still within the range of, the coarse fraction (d > 2 µm) measurements of Fattori et al.  from Terra Nova Bay (hereafter, TNB). They are also somewhat lower than the mean annual chloride concentration (272 ng m−3) measured in total aerosol at Halley Bay (HB) by Rankin and Wolff . The lower chloride concentrations may be explained by COSRAY's greater distance from the ice edge during this measurement campaign or the timing of the measurements earlier in the winter/spring season. The measured chloride concentrations are significantly larger than the fine-fraction chloride concentrations from TNB; however, the TNB measurements were taken further into the summer season. Conversely, year-round measurements from HB show that a majority of sea-salt ionic mass (both Cl− and Na+) occurs in the submicron aerosol during the winter and spring, suggesting that two different sea-salt aerosol regimes may exist, depending on season and distance from the ice edge [Rankin and Wolff, 2003]. Furthermore, measurements at Dumont d'Urville also show an increase in the submicron fraction of sea-salt aerosol during the winter months [Jourdain and Legrand, 2002]. Over the duration of the campaign, the ratio of chloride to total Sub-Micron Volume (SMV) decreases. This could be to the result of a shift in the composition of the particles (e.g, a decrease in the relative amount of soluble chlorine, in the case of internal mixtures) or of a shift in the partitioning of chloride between submicron and supermicron particles. This behavior is consistent with the transition from the winter/spring sea-salt aerosol submicron size mode toward a summertime supermicron size mode, as was observed at HB by Rankin and Wolff .
 Sulfate concentrations during the measurement period are consistent with those seen at HB for the same period, with a mean loading of 76 ng m−3. Sulfate is weakly correlated with PM1.0, except during ODEs. For example, the ODE on 18 September coincides with an approximately fourfold increase in PM1.0 and Cl− over background concentrations, relative to a 1.5-fold increase in SO42− over background concentrations. The relative decrease in the SO42− to PM1.0 and Cl− during ODEs would be consistent with the source of aerosols shifting toward a sea-ice/snow pack origin, which would be expected to be depleted in sulfate, as opposed to direct oceanic emissions. While we are not able to calculate non-sea-salt sulfate concentrations, it is likely that there is a significant non-sea-salt contribution to the sulfate loading from biogenic sources and direct oceanic emissions. Finally, due to the bulk nature of the particle composition measurements, we cannot differentiate between internal and external aerosol mixtures. For halogen ratios, it is safe to assume that these mixtures are internal mixtures as both Br− and Cl− originate from the same source. However, sulfate can be present in both sea-salt aerosol and in secondary particles nucleated from gas phase dimethyl sulfide, which could produce external mixture of sulfate and halogen aerosol and thus bias halogen to sulfate ratios for determining particle source.
 Total bromine ranged from the detection limit (< 2 ng m−3) to 31 ng m−3 with a mean of 3.2 ng m−3. The ratio of Cl− to Br− in the aerosol averaged 50:1, which is significantly lower than the seawater mass ratio of 290:1, even when the large uncertainty in the bromide concentration is considered. This indicates a depletion in Cl− or an enhancement in Br− relative to seawater ratios. This could imply that the source of the measured aerosols is snow on sea ice or frost flowers which have been exposed to brine at the snow/ice interface and then have been cooled below −21°C leading to precipitation of Cl− and the relative enhancement of Br− [Kalnajs and Avallone, 2006; Obbard et al., 2009]. It is also possible that chlorine has been liberated to the gas phase, leading to a chloride deficit in the aerosol phase, as was seen by Hara et al.  .
2.1 Features of Ozone Depletion Events
 Between 12:00 on 13 October and 00:00 on 15 October there is a single longer-duration ODE with the lowest ozone mixing ratio observed during the entire measurement period. Ozone and aerosol data from this event are shown in Figure 2. The increase in the submicron aerosol volume and surface area density is driven almost entirely by the coarse fraction, defined here as 200–1000 nm. The depletion is accompanied by rising temperatures, with the peak temperature of −12°C occurring simultaneously with the minimum ozone mixing ratio. Two hours after the peak in aerosol mass, temperatures begin to fall rapidly, returning to below −25°C at 18:00 on 14 October, accompanied by a rapid decrease in aerosol number to pre-ozone depletion amounts by 20:00. Accompanying this drop in temperature, the wind direction switches from a primarily easterly to primarily southerly direction. The ODE is also observed at the Arrival Heights site (2 km North and 200 m above COSRAY), in agreement with COSRAY to within 1 ppbv, signifying a wider geographic extent to the event and indicating that the larger-scale depletion is not caused by local emissions.
 Unfortunately, there were no aerosol composition measurements during this period. The pattern of events described in this case study is typical of several other smaller ODEs observed later in the measurement campaign (for example, events on 20 and 23 October) which will not be discussed in detail here, other than to note that this was not an isolated event (although it was the most severe of this type of event).
 The 13 to 15 October event appears to be a preceded sudden change in air mass, accompanied by a rapid ozone depletion event (~1.8 ppbv h−1) during a period of warming. The recovery from the ODE is even more rapid (~2.5 ppbv h−1) and is coincident with a colder air mass moving in. A possible explanation for this sequence of events is that the north/northeasterly winds before the ODE brought in warmer air from the sea ice edge east of COSRAY near Cape Crozier. This air mass brought with it enhanced aerosol concentrations and is depleted of ozone. The maximum ozone depletion occurs simultaneously with the peak in aerosol volume, and the shape of the ozone depletion mirrors the shape of the coarse particle volume (Figure 2). The peak in ozone depletion also occurs at approximately midnight local time. While there is 24 h illumination at this point in the spring season, nighttime solar fluxes are extremely small (solar zenith angle (SZA) = 94.2), so it is unlikely that a depletion of this magnitude could be sustained through local in situ photochemistry. The recovery of ozone and reduction in aerosol volume are accompanied by a wind direction change and dramatic temperature drop, suggesting the arrival of a new air mass from the Antarctic continent, which would be depleted in sea-salt aerosol relative to coastal air masses. While this is the only example discussed in detail here, the general meteorological and chemical features of the four depletion events (Figure 1) follow a similar pattern.
 The first simultaneous measurements of ozone and high-resolution submicron particle size distributions near McMurdo that are presented here reveal a strong and robust correlation between enhanced aerosol concentration and ozone depletion events. Sea-salt aerosols have been postulated as a transport mechanism for halogens from sea ice/seawater source regions to locations further inland, as well as being a direct source for gas phase bromine [Yang et al., 2008]. Measurements by Domine et al.  suggest that windblown aerosol is a source of halogens in the Arctic snow pack, and by extension, that this may be a source of bromine that contributes to ODEs. Friess et al.  present measurements showing increased BrO, depleted ozone, and increased light scattering that is suggestive of a correlation between enhanced aerosol concentrations and ozone depletion. There are many more studies of BrO and ozone depletion that have suggested that aerosols may either be involved in polar ODEs directly, through heterogeneous ozone destruction or release of gas phase bromine, or indirectly, through deposition of halogens on the snow pack [e.g., Simpson et al., 2007b, and references therein]. These studies, and many more, build a strong case for the involvement of sea-salt aerosol in ODEs, but few present any direct observational evidence, particularly in the Antarctic. Recent modeling studies have also indicated that sea-salt aerosol, or other halogen-containing aerosol, is a likely [Yang et al., 2008; Saiz-Lopez et al., 2008] or even necessary [Piot and von Glasow, 2008] component to bromine-catalyzed ODEs. The measurements presented here provide strong evidence for a link between aerosol abundances and ozone depletion in the Antarctic boundary layer.
 Determining the origin of aerosols associated with ODEs is important to understanding their composition and how they could interact with ozone. Unfortunately, due to the complex terrain and wind field in the Ross Island region, the single measurement site and the lack of meteorological observations, it was not possible to calculate reliable local back trajectories for the periods of ozone depletion using standard tools (e.g., Hybrid Single-Particle Lagrangian Integrated Trajectory). However, the meteorology associated with the observations summarized here, as well as during several other ODEs not discussed in detail, reveals a consistent pattern, with warming that is either preceded or accompanied by strong winds and indicative of a change in air mass.
 Increased temperatures accompanying ODEs near McMurdo indicate that the air masses originated from the ocean or coastal regions farther north, and not from the colder continental interior. Strong winds during these episodes would also entrain blowing snow aerosols from the snow pack or sea ice, probably a necessary condition for mobilizing active halogens. Modeling work by Yang et al.  suggests that snow on sea ice may be a more significant source of sea-salt aerosol than the open ocean. This is also consistent with observations of the depletion of sodium and sulfate in aerosol during winter at Halley Bay [Rankin and Wolff, 2003]. The study by Yang et al.  also indicates a critical wind speed of about 7 m s−1 for lofting of snow from the surface. Modeling of frost flowers as a source of halogenated aerosol similarly indicates that halogen particle production is likely to increase significantly at wind speeds over 7 m s−1 [Piot and von Glasow, 2008]. At Syowa Station, which is near the Antarctic coast like McMurdo, Hara et al.  found a correlation between sea-salt concentrations in particles > 1 µm and low air pressure, strong winds, and higher temperatures. This may explain the increase in filterable chloride associated with the ODE around 18 September (Figure 1) in this study; the Ultra High Sensitivity Aerosol Spectrometer (UHSAS), with a Mie scattering threshold of 1 µm, will undersize larger wind-lofted particles and snow crystals.
 In order for sea-salt-derived halogens to deplete ozone, it is necessary for reactive bromine to be released from the aerosol phase to the gas phase, or for heterogeneous ozone depletion to occur on the aerosol particles. In either case, submicron aerosols provide a greater specific surface area per mass than supermicron aerosols for ozone depletion or bromine release. Submicron aerosols also have a longer atmospheric residence time than larger particles due to slower saltation [Yang et al., 2008], which could also explains the shift to submicron sea salt-containing aerosol in the winter months when distances from the open ocean are longer. Moreover, Fattori et al.  compared the alkalinity of the supermicron and submicron fractions at TNB and found that the supermicron fraction is alkaline whereas the submicron fraction is generally acidic. A commonly proposed mechanism for the release of gas phase bromine is the reaction of gas phase nonradical species with aqueous bromide as suggested by Fan and Jacob . Since this mechanism is highly dependent on available H+, it should be favored in the submicron aerosol, even with reduced halogen content [Fickert et al., 1999]. Combined with the relative contribution to surface area per mass of submicron particles, this may imply that submicron particles are a more potent source of reactive halogens to the atmosphere than supermicron particles.
 Ozone depletion events in the early spring at McMurdo Station are accompanied by increases in the number of submicron particles. The near-tenfold increase in particle volume during ozone depletions is largely driven by an increase in larger submicron particles with diameters centered around 500–600 nm. This increase in submicron particles is also accompanied by an increase in filterable chloride and sulfate masses. These particles are attributed to a sea-salt-derived source; meteorology and bulk particle composition suggests that source is the sea ice or ice edge in the vicinity of Ross Island.
 This material is based upon work supported by the National Science Foundation under Grant No. 1043266.
 The Editor thanks Xin Yang and an anonymous reviewer for their assistance in evaluating this paper.