200-year ice core bromine reconstruction at Dome C (Antarctica): observational and modelling results

. Bromine enrichment (Br enr ) has been proposed as an ice core proxy for past sea-ice reconstruction. Understanding the processes that inﬂuence bromine preservation in the ice is crucial to achieve a reliable interpretation of ice core signals and to potentially relate them to past sea-ice variability. Here, we present a 210 years bromine record that sheds light on the main processes controlling bromine preservation in the snow and ice at Dome C, East Antarctic plateau. Using observations alongside a modelling approach, we demonstrate that the bromine signal is preserved at Dome C and it is not affected by the strong variations in ultraviolet radiation reaching the Antarctic plateau due to the stratospheric ozone hole. Based on this, we investigate whether the Dome C Br enr record can be used as an effective tracer of past Antarctic sea ice. Due to the limited time window covered by satellite measurements and the low sea-ice variability observed during the last 30 years in East Antarctica, we cannot fully validate Br enr as an effective proxy for past sea-ice reconstructions at Dome C.


Introduction
Halogens play an important role in the chemistry and oxidising capacity of the Earth's atmosphere: they take part in new particle formation processes, promote mercury oxidation, influence the budget of HO x and NO x radicals, and cause ozone depletion through efficient catalytic cycles (Saiz-Lopez and Von Glasow, 2012;Simpson et al., 2007). Volcanic eruptions (Gutmann et al., 2018) and the ocean (Parrella et al., 2012;Prados-Roman et al., 2015) represent the main natural sources of halogens to the atmosphere, releasing significant amounts of bromine (Br) and iodine (I) (Cuevas et al., 2018;Carpenter et al., 2013). In this work, we focus on bromine, which has been shown to dominate halogen emissions and chemistry in the polar atmosphere through the so-called "bromine explosion events" (Pratt et al., 2013;Platt and Lehrer, 1997). These are heterogeneous autocatalytic photochemical reactions, which were first described in the Arctic boundary layer and that cause the bromineinduced ozone depletion events (Fan and Jacob, 1992;Vogt et al., 1996;Foster et al., 2001;Wennberg, 1999;Barrie et al., 1988;Kreher et al., 1997). These autocatalytic mul-F. Burgay et al.: 200-year ice core bromine reconstruction at Dome C (Antarctica) tiphase chain reactions require both acidic conditions and sunlight to produce an exponential increase in atmospheric bromine concentration, mainly as gaseous BrO, Br 2 , and HOBr (Schönhardt et al., 2012;Zhao et al., 2016;Nghiem et al., 2012). In the polar regions, the most favourable substrate (i.e. with a large bromide content) to produce such bromine explosion and ozone depletion events during springtime is the sea salt aerosol derived from surface blowing snow deposited over first-year sea ice (FYSI), which is characterised by acidic conditions, higher Br − / Cl − ratio (Pratt et al., 2013), and higher salinity than the snow deposited over multi-year sea ice (MYSI) . Direct observations from two winter cruises in the Weddell Sea (Antarctica) and model simulation showed that significant bromine losses take place in the aerosol phase, indicating that sea salt aerosol debromination from salty blowing snow over sea ice represents a relevant source of gas-phase inorganic bromine to the troposphere Parrella et al., 2012;Yang et al., 2005). Note that most inorganic bromine gasses present in the atmosphere are highly water-soluble and they suffer wet and dry deposition over the ice sheets (Legrand et al., 2021;Parrella et al., 2012;Fernandez et al., 2019).
Especially during the "bromine explosion events", bromine, that shares the same sources as sodium (i.e. sea salt), is significantly enriched compared to sodium (Na) in the FYSI surfaces, exceeding the bromine-to-sodium mass ratio of seawater (Millero et al., 2008). Being promoted by the presence of FYSI compared to MYSI, this bromine enrichment (Br enr , Eq. 1) has been proposed as a potential tracer for the reconstruction of past FYSI conditions (Spolaor et al., 2013b(Spolaor et al., , 2016Maffezzoli et al., 2019;Vallelonga et al., 2017). Qualitatively, higher Br enr values were linked to larger FYSI extent (Spolaor et al., 2013b). However, many unknowns, mainly related to the source, transport, and preservation of bromine within the snowpack, still remain (Maffezzoli et al., 2019). For example, it has been suggested that the anthropogenic-induced acidity increase of the snow deposited over the sea-ice surface can enhance the sea salt debromination rates enhancing the release of reactive bromine from sea salt aerosols into the atmosphere (Maselli et al., 2017;Sander et al., 2003). Further, bromine can also be re-emitted from the snowpack after deposition and prior to burial. However, the results obtained from previous studies are contradictory and site-specific (Mcconnell et al., 2017;Legrand et al., 2016;Dibb et al., 2010;Spolaor et al., 2019). In Greenland, Dibb et al. (2010) showed that bromine photoactivation was present during spring/summer and highlighted an efficient Br chemical cycling above the snow. In the Svalbard Archipelago, a high-temporal resolution study designed to investigate the potential photoemission of bromine from the snowpack  did not highlight any bromine diurnal cycle, suggesting its preservation in snow. In Antarctica, through the investigation of Na and Br fluxes against snow accumulation rate, Mc-Connell et al. (2017) found that bromine re-emission from the Antarctic snowpack is inversely dependent on the accumulation rate. It was shown that the bromine loss from the snowpack was higher (65 %) at sites with the lowest accumulation rate (50 kg m −2 yr −1 ), and it decreased to 11 % at sites with high annual accumulation rate (300 kg m −2 yr −1 ). Based on these observations, virtually all bromine deposited at Dome C (Antarctica), where the annual snow accumulation is ≈ 25-28 kg m −2 yr −1 , would have been re-emitted to the atmosphere prior to burial (Mcconnell et al., 2017;Maffezzoli et al., 2019). These conclusions contrast with previous observations performed at Dome C that reported no significant bromine re-emission from the snowpack (Legrand et al., 2016). An additional process that can affect bromine preservation within the snowpack has been identified in coincidence with the 17.7 ka Mt. Takahe volcanic eruption, when the combination of an increased surface ultraviolet (UV) radiation, due to stratospheric ozone depletion, and high acidity conditions was associated with a decrease in ice bromine concentration (Mcconnell et al., 2017). To our knowledge, there are no investigations that focused on the effects of the modern UV radiation changes reaching the Antarctic plateau, due to the ozone hole formation, on bromine preservation in snow.
To unravel the physicochemical processes that can influence bromine preservation in the snow, we investigated the main pathways that can induce its emission from the snowpack to the atmosphere. Bromine is mainly present in the Antarctic snowpack as bromide (Spolaor et al., 2013a) and it can be oxidised by OH radicals (George and Anastasio, 2007;Abbatt et al., 2010) to form evaporable gaseous bromine. The main q OH source within the snowpack is the photolysis of hydrogen peroxide (Chu and Anastasio, 2005), nitrate (Chu and Anastasio, 2003;Abbatt et al., 2010), and nitrite (Chu and Anastasio, 2007). Understanding the relevance of each of these photochemical pathways in explaining the preservation of bromide in the Antarctic snowpack before and after the onset of the modern-ozone hole is then crucial for a reliable interpretation of the bromine enrichment profile observed in ice core records.
In this study, we present the first bromine record retrieved from a shallow firn core collected at Dome C, Antarctica, covering the period 1800-2012. Through the evaluation of the bromine profile, it is possible to provide new evidence about the role of the enhanced solar UV radiation due to the onset of the modern Antarctic ozone hole (1975) on bromine preservation in the snowpack. An extended evaluation of the role of q OH precursors and their relevance at Dome C is also performed. Lastly, an assessment of the suitability of Br enr as a potential past sea-ice tracer at Dome C is also addressed by combining air-mass transport reanalysis and sea-ice extent satellite observations. The results presented in this paper open new perspectives on future long-term bromine studies from ice cores retrieved from low accumulation areas, aimed at forecasting future deep core drillings at Dome C, such as those planned for the Beyond EPICA project.  Spolaor et al. (2021), with age uncertainties ranging from 1 year at surface to 5 years at the bottom of the core. Dome C is a suitable Antarctic site for performing photochemical studies related to the preservation of reactive elements and halides within the snowpack Cairns et al., 2021;Song et al., 2018;Spolaor et al., , 2021. This location presents a low and rather constant accumulation rate (25.3 ± 1.3 kg m −2 yr −1 from 1816-1998, 28.3 ± 1.3 kg m −2 yr −1 from 1965-1998, 28 kg m −2 yr −1 from 2004-2011) (Frezzotti et al., 2005(Frezzotti et al., , 2013Genthon et al., 2016), and it is located about 1000 km away from shorelines, thus not being directly affected by local coastal emissions. Matching these criteria is essential for the evaluation of the effects of modern stratospheric ozone loss due to long-lived ozone depleting substances in the potential bromine release from the snowpack. The shallow ice core was collected using a hand drill (3 in. diameter); the sections were sealed in plastic containers and shipped to the Institute of Polar Sciences of the National Research Council (ISP-CNR) in Venice (Italy). The ice core sections were subsequently sampled at 5(±1) cm resolution (corresponding to approx. 1 year) using a ceramic knife, rinsed with Ultra Pure Water (UPW, Elga Lab, UK) after each use. Only the central part of the core was collected into 50 mL pre-cleaned polyethylene (PE) vials for subsequent analyses, while the outer 2 cm was removed by scraping with a ceramic knife. The core samples were processed in a class 1000 inorganic clean room under a class 100 laminar-flow bench. Samples were kept at −20 • C and under dark conditions until the analysis to avoid any possible photolysis reaction. Since we were only interested in the water-soluble bromine and sodium fractions and to avoid potential halogen loss (Flores et al., 2020), the analyses were conducted on melted and not acidified samples by inductively coupled plasma sector field mass spectrometry (ICP-SFMS; see Sect. 2.2). The sodium record considered in this study had some gaps from 1989-1997 (n = 12, corresponding to 5 % of the total amount of samples) that were filled with Na concentration data retrieved from two snow pits collected in 2013 and 2017, as reported in Spolaor et al. (2021).

Instrumental analysis and cleaning procedure
Total sodium and bromine concentrations were determined by ICP-SFMS following Spolaor et al. (2016). Each analytical run started and ended with an Ultra-Pure Water (UPW) cleaning session of 3 min to ensure a stable background level throughout the analysis. The external standards that were used to calibrate the analytes were prepared by diluting a 1000 ppm stock IC (ion chromatography) standard solution (TraceCERT ® purity grade, Sigma-Aldrich, MO, USA). The standard concentrations ranged between 1 and 200 ng g −1 for sodium and 0.05 and 0.200 ng g −1 for bromine. As for the samples, the standards were not acidified. Precision and accuracy of the measurements were determined through the multiple reading of selected ice samples and external standards, respectively. The relative standard deviation (RSD %) was low for all the analytes, ranging between 3-4 % for sodium and 5-7 % for bromine, while accuracy, expressed as the ratio between the observed and the true values, was 105 % for sodium and 92 % for bromine. The instrumental limit of detection (LoD), calculated as 3 times the standard deviation of the blank (n = 10), was 1 ng g −1 for Na and 0.05 ng g −1 for Br. Bromine and sodium concentration values were used to calculate bromine enrichment (Br enr ) as follows: where Br and Na are the concentrations obtained from the Dome C record, and 0.0062 reflects the bromine-to-sodium mass ratio in seawater (Spolaor et al., 2013b;Millero et al., 2008). All the plastic material used for sample storage and analysis was washed 5 times using UPW and filled with UPW for 1 week. Then, it was rinsed again 5 times with UPW and dried under class-100 laminar flow hood before use.

Back-trajectories calculation, satellite observations, and statistical analysis
Backward air mass trajectories that reach the Dome C site (75 • 05 59 S, 123 • 19 56 E) were calculated to identify the most likely ocean and sea-ice areas that release bromine species to the atmosphere and that can be transported to the interior of Antarctica. Back-trajectories were obtained from the Hybrid Single-Particle Lagrangian Integrated Trajectory (HYSPLIT) model (Stein et al., 2015) using European Centre for Medium-range Weather Forecasts (ECMWF) ERA5 meteorological reanalysis (Hersbach et al., 2020). ERA5 is available on 37 pressure levels with a regular spatial grid of 0.25 • × 0.25 • at hourly temporal sampling. However, due to the huge amount of trajectories needed for this study, for computation requirements, we considered ERA5 parameters on a spatial grid of 0.5 • × 0.5 • every 3 h and 24 pressure levels (Becagli et al., 2022). Five-day backward trajectories were calculated every 3 h at 1000, 2000, and 3000 m above Dome C model terrain height for the period 1979-2018. Each back trajectory was then projected on the seaice concentration field (SIC) and the 10 m wind field (still in the ECMWF ERA5 reanalysis), associating each value along the trajectory path with the nearest SIC and wind speed val-394 F. Burgay et al.: 200-year ice core bromine reconstruction at Dome C (Antarctica) ues. The main paths of air masses reaching Dome C were highlighted dividing the Southern Hemisphere in a regular 1 • × 1 • mesh and counting the total number of backtrajectories points at 5 d, falling in each grid cell (i.e. the hours spent by the air mass in each grid cell). Since bromine species are emitted in the marine boundary layer (MBL) (Sander et al., 2003), source bromine areas were evaluated, selecting only the trajectory where the air mass paths lie within the MBL and over sea ice and counting the total number of resulting points, where conditions are fulfilled, in each of the 1 • × 1 • grid cells. The height of MBL was set equal to the 900 hPa isosurface (Lewis et al., 2004) and a value of SIC >15 % was considered in order to simulate the presence of the sea-ice cover (Becagli et al., 2022). The sea-ice concentrations used in this work are derived from passivemicrowave radiometers on NASA's satellites. The data are publicly available at the NASA Earth Science portal (https: //earth.gsfc.nasa.gov/, last access: 15 March 2022) and at the National Snow and Ice Data Center portal (http://nsidc.org, last access: 15 March 2022). The sea-ice extents (in km 2 ) are calculated as the hemispheric total as well as five regions in the Southern Ocean (Indian Ocean, Western Pacific, Ross Sea, Bellingshausen and Amundsen seas, and the Weddell Sea).
The correlations computed among the different variables of this study (Na, Br, Br enr , sea-ice extent data, Southern Annular Mode (SAM) index) were performed using a 3 years moving average. This choice takes into account both the dating error of the core (≈ 3 years) and the effects of wind erosion on the age distribution of surface snow that spans over more than a year (Picard et al., 2019). To identify any abrupt change point in the records, the findchangepts() Matlab (Mathworks) function was used.

CAM-Chem model set-up
The wavelength-dependent solar ultraviolet (UV) radiation reaching the Antarctic plateau surface at Dome-C during the 1950-2010 period was computed using the Community Earth System Model (CESM) (Tilmes et al., 2016). The setup of the atmospheric component of the model (CAM-Chem, version 4) was identical to the one used in previous studies addressing the evolution of iodine ice core records in the Arctic (Cuevas et al., 2018) and Antarctica , and considers prescribed sea surface temperatures and sea-ice distributions following the CCMI-REFC1 recommendation (Eyring et al., 2013). The CAM-Chem VSL setup includes geographically distributed and seasonally-dependent natural oceanic emissions of five bromocarbons (VSL Br = CHBr 3 , CH 2 Br 2 , CH 2 BrCl, CHBrCl 2 , CHBr 2 Cl) and four iodocarbons (VSL I = CH 3 I, CH 2 ICl, CH 2 IBr, CH 2 I 2 ), whose oceanic flux is assumed to remain constant during the whole modelling period (Ordóñez et al., 2012). The chemical scheme includes the additional inorganic chlorine and bromine contribution arising from the so-called sea salt dehalogenation recycling occurring in the marine boundary layer and the free troposphere (Fernandez et al., 2014;Fernandez et al., 2021). The model was configured in free-running mode, with 26 vertical levels expanding from the Earth's surface to approximately 40 km (3.5 hPa in the upper stratosphere), and with a spatial resolution of 1.9 • latitude by 2.5 • longitude. The CAM-Chem REFC1 configuration used here provides a reasonable representation of the evolution of the size and depth of the ozone hole, presenting an excellent agreement with satellite ozone observations during the modelled period (Fernandez et al., 2017;. Here, we compute the photolysis rate constant (J-value) of different q OH precursors involved in the preservation of bromine in the snowpack, following the same approach used in Spolaor et al. (2021). The molar absorptivities, quantum yields, and species concentrations for hydrogen peroxide, nitrate, and nitrite at Dome-C are summarised in Table 1. The modelled actinic flux reaching the Antarctic surface includes 100 bins expanding from 121 to 750 nm, with a spectral resolution ranging from less than 1 nm in the UV to 50 nm in the visible edge. In particular, the model bandwidth within the 280-400 nm spectral range considered in this work possesses a mean resolution of 4 nm. Thus, the CAM-Chem surface actinic flux for each wavelength grid was linearly interpolated into a 1 nm bandwidth, and the mean J-value during the whole sunlit period (i.e. from 1 September of a given year to 28 February of next year) was computed offline at the closest grid box to Dome C (74.84 • S; 122.5 • E; model mean altitude of 3300 m a.s.l.). The complete sunlit period (spring + summer) was selected because even when the largest changes in surface actinic flux associated with the ozone hole formation are observed during spring, the UV radiation intensity reaching the Antarctic plateau maximises during the summer .

Results and discussion
3.1 Sodium, bromine, and bromine enrichment profiles from the Dome C shallow core Air masses arriving at Dome C originate from a vast area that extends over the eastern Antarctic Ocean, Ross Sea, and in minimal percentage from West Antarctica. Back trajectory analyses ( Fig. 1)  Concentration at Dome C (Chu and Anastasio, 2007, 2005 H 2 O 2 0.7 2 ng g −1 (Frey et al., 2006)  at Dome C (Caiazzo et al., 2021). Sodium concentrations along the entire record spanned from 12 to 117 ng g −1 , with an average value of 40 ± 13 ng g −1 . Na profile (Fig. 2) shows an increase in concentration from 1800-1850, followed by a long-term decreasing trend until 1994. Over the last 18 years of the record, higher sodium concentrations were recorded, suggesting an enhanced transport towards Dome C. Bromine concentration at Dome C ranges from below the LoD (0.05 ng g −1 ) to 0.41 ng g −1 , with an average value of 0.10 ± 0.05 ng g −1 along the entire record. A significant bromine increase was detected since 2004 when Br concentration increased from 0.10 ± 0.05 ng g −1 (pre-2004) to 0.23 ± 0.09 ng g −1 (post-2004). The absence of an abrupt change of the bromine signal at the onset of the ozone hole (1975) indicates that Br is preserved in the snowpack independently on the incoming UV-radiation (see Sect. 3.2). Sodium and bromine did not show any significant correlation (r = 0.06, p-value = 0.20) along the entire record, suggesting different deposition velocities during transport from the coast, with sodium being deposited faster than bromine. Indeed, in the polar atmosphere, sodium is present in the aerosol phase and it is mainly affected by wet deposition processes. It was observed that its concentration decreases significantly with distance from the coast, reaching a rather constant deposition rate at approximately 400 km inland . Contrarily, bromine exists in both the aerosol and in the gas phase, its atmospheric lifetime is also driven by dry deposition processes, it can experience heterogeneous chemical recycling during transport, and its concentration gradually decreases from 100-1000 km inland (Simpson et al., 2005;Vallelonga et al., 2021).
Br enr values ranged between 0.07 and 1.6 with an average value of 0.4 ± 0.3 along the entire record. Contrarily to sodium and bromine, the enrichment did not show any significant increase in the recent part of the record. On the contrary, a significant regime change was detected in 1825, when the Br enr mean value changed from 0.7 ± 0.3 (1800-1825) to 0.4 ± 0.3 (1825-2012). A two-sample t-test strengthened the significant difference between the two periods (p-value <0.001). In general, Br enr values were mainly below 1 (i.e. bromine is depleted relative to sodium), which is expected at remote locations like Dome C, since the Br to Na ratio depends on the relative transport times of sea salt aerosol and gaseous bromine compounds in the atmosphere (Spolaor et al., 2013b;Simpson et al., 2007;Vallelonga et al., 2021). Thus, our findings agree with the synthesis of Vallelonga et al. (2021), which show low Br enr values near the coast, a gradual increase at 300-600 km inland and a following decrease to values lower than 1 for all sites located more than 800 km from the coast. The few Br enr values higher than 1 may indicate larger FYSI surface from the source areas (Spolaor et al., 2016) or different bromine partitioning between the aerosol and gas phase that depends on the aerosol size and, consequently, on atmospheric resident times (Legrand et al., 2016;Maffezzoli et al., 2019;Vallelonga et al., 2021). In addition, changes in background atmospheric q OH and NO x might have had an impact on the gas-phase bromine partitioning between reservoir and reactive species that might have led to a faster bromine deposition, since species like BrONO 2 and HOBr have larger and more efficient deposition velocities than reactive species like BrO and Br (Saiz-Lopez and Fernandez, 2016;Fernandez et al., 2019).
In Antarctica, few other long-term bromine records exist and they were all collected from coastal sites (Spolaor et al., 2013b;Vallelonga et al., 2017). In contrast to Dome C, these records are more directly influenced by local marine contributions. The shorter atmospheric transport time from the source to the deposition location is reflected by the higher Br enr values observed in these cores (Vallelonga et al., 2017).

Bromine preservation in the snowpack at Dome C
When reaching the snowpack, UV radiation can rapidly break weak chemical bonds and, due to its high energy, can promote photochemical reactions, especially in the UV-A (320-400 nm) and UV-B (290-320 nm) wavelength bands (Grannas et al., 2007). Due to its low accumulation rate, Dome C is the perfect location for performing UV photolysis studies on chemical species occurring at the ice and snow surface (Frey et al., 2009;. The stratospheric ozone layer depletion observed since 1975 has caused an increase in the incoming solar UV radiation over Antarctica at λ<300 nm, enhancing, for example, the photochemical iodide oxidation and its subsequent release from the snowpack . As highlighted in Sect. 3.1, no significant changes either in bromine concentration or in bromine enrichment have been detected at Dome C since 1975, suggesting that, contrarily to iodine , the ≈ 10-fold enhanced UV radiation reaching the Antarctic plateau has not altered bromine preservation within the snowpack (Fig. 2). Moreover, laboratory and chamber experiments showed enhanced photochemical oxidation and subsequent release of I 2(g) from artificial snow/ice and the snowpack through the formation of a critical I-O 2 complex having an absorption band centred at 290 nm (Kim et al., 2016). At present, there is no evidence and/or available literature describing a similar Br-O 2 complex, and/or any other brominated intermediate product, that leads then to the release of Br (2)(g) . In fact, the main inorganic route for bromide oxidation requires radical oxidants (e.g. q OH) to drive the redox production of hypobromous acid (BrOH) (Artiglia et al., 2017). This oxidised species can then combine with other reduced halide ions to form molecular halogen compounds that are released into the gas phase (Reactions R1-R4) (George and Anastasio, 2007) as follows: Over ice and snow substrates, hydroxyl radicals ( q OH) can be produced by the photolysis of hydrogen peroxide (H 2 O 2 ), nitrate (NO − 3 ), and nitrite (NO − 2 ) (Chu and Anastasio, 2005;Abbatt et al., 2010;Chu and Anastasio, 2007). Between 290-340 nm wavelengths, H 2 O 2 has a wavelength-dependent molar absorptivity that is 2.5-7.1 times lower than that for NO − 3 . Nevertheless, H 2 O 2 has a ≈ 160 times greater quantum yield for q OH production ( Table 1) that is insensitive to ionic strength, pH, and wavelength (Chu and Anastasio, 2005). Therefore, for a given concentration, H 2 O 2 is a much more effective source of q OH than nitrate. To our knowledge, the only H 2 O 2 concentration value available at Dome C is 2 ng g −1 , derived from a sample collected at 3.5 m depth (Frey et al., 2006). This low value, compared to other locations, is consistent with semi-empirical models that predict a complete hydrogen peroxide loss when the accumulation rate is below ≈ 70 kg m −2 yr −1 and the annual mean temperature is −50 • (Frey et al., 2006). Considering that Dome C has an annual mean accumulation of ≈ 25-28 kg m −2 yr −1 (Genthon et al., 2016) and an annual mean temperature between −54 and −50 • (Genthon et al., 2021), we assume that the majority of the deposited or in situ-produced H 2 O 2 is rapidly lost to the atmosphere.
Alternatively, NO 3 photolysis ( The q OH radicals, formed by nitrate photolysis can produce Br 2(g) , following Reactions (R1)-(R4). The typical snowpack nitrate profile at Dome C ranges between 22 and 147 ng g −1 (Caiazzo et al., 2021;Spolaor et al., 2021) and shows an exponential decay in concentration with depth driven by nitrate UV photolysis and recycling (Winton et al., 2020;Röthlisberger et al., 2000;. Due to its higher concentration compared with hydrogen peroxide, nitrate may represent a relevant q OH source at Dome C despite its lower quantum yield for q OH production. The nitrate UV photolysis, followed by q OH formation and Br 2(g) emission, has been reported under laboratory conditions with a significant dependency on the ice pH, with the largest Br 2(g) emissions observed at low pH (George and Anastasio, 2007).
Another source of OH radicals is nitrite (Minero et al., 2007), which can be produced by the dissociation of nitrous acid (HONO) in the condensed phase (Reaction R9), or by the direct formation from the minor channel (quantum yield of approximately 0.0011) of nitrate photolysis (Reaction R5) (Dubowski et al., 2002). Nitrite displays two major absorption bands peaking at 300 and 354 nm and, through its photolysis, it can produce q OH (Reactions R10-R11) as follows: The q OH quantum yield from nitrite photodissociation depends both on the wavelength (increases with decreasing wavelength) and on temperature (decreases with decreasing temperature). In addition, nitrite has a ≈ 2-fold higher molar absorptivity than nitrate between 280 and 300 nm (Chu and Anastasio, 2007) and its q OH quantum yield in ice is equal to 0.020 (at 240 K, λ = 300), which is 6-fold higher than the one calculated for nitrate (Table 1). Unfortunately, there are no direct measurements of nitrite at Dome C, meaning that its concentration needs to be estimated. Following the approach used by Chu and Anastasio (2007), we assumed that at Dome C the nitrite concentration is similar to the one calculated at the South Pole, which is 0.092 ng g −1 . This assumption is based on: (a) similar superficial NO 3 concentrations recorded both at the South Pole (99 ng g −1 ) and at Dome C (90-147 ng g −1 ); (b) the use of NO − 3 photolysis as the main source for nitrite in the snow (Chu and Anastasio, 2007); and (c) the similar total UV radiation intensities and seasonality reaching both locations.
To evaluate the relevance of these processes on the OH radical production in the snow grains, and consequently their role in promoting Br 2(g) emission from the snowpack, we modelled the hydrogen peroxide, nitrate, and nitrite photoactivation before  and after the ozone hole formation (post-1975) at Dome C, following the wavelengthdependent CAM-Chem model actinic flux output and the methodology described in Spolaor et al. (2021). Both H 2 O 2 and NO − 3 exhibited a small, but significant, enhancement on their surface photolysis (J-value) after the onset of the ozone hole (≈20 % increase) due to the higher actinic flux reaching the surface at λ<300 nm, where most of the H 2 O 2 absorption occurs, and to a limited extent also NO − 3 (Figs. 3, 4). In contrast, NO − 2 does not show a significant trend on their J-values because both absorption bands maximise at longer wavelengths, within a spectral region that is not directly affected by the formation of the ozone hole (Figs. 3, 4). It is important to notice that the normalised photolysis ratio between the ozone hole period and the pre-ozone hole period strongly depends on the wavelength range considered to compute the J-value integration. For example, the ratio between the ozone hole and the pre-ozone hole periods for H 2 O 2 ranges from a minimum value of 1.2 (280-390 nm) to a maximum value larger than 5 (280-300 nm) (Fig. S1). Equivalent results are obtained for the J-nitrate and J-nitrite enhancements (Fig. S1), although for these species, with the strongest absorption at longer wavelengths, the upper bandwidth limit used to perform the integration should not be located at values below 310-320 nm, which results in an up to 3-fold increase on the normalised ratio during the preozone hole period. For these reasons, and based on the ob-398 F. Burgay et al.: 200-year ice core bromine reconstruction at Dome C (Antarctica) served molar absorptivities of each species shown in Fig. 3a, our best estimate of the normalised photolysis ratio shown in Fig. 4c (Fig. 4b). Those values are calculated by multiplying the mean photolysis rate constant for q OH formation during the ozone hole (pre-ozone hole) period (i.e. 6.4 × 10 −8 (5.9 × 10 −8 ) s −1 for NO − 3 , 5.9 × 10 −6 , (5.36 × 10 −7 ) s −1 for H 2 O 2 , and 6.2 × 10 −5 (6.2 × 10 −6 ) s −1 for NO − 2 ) by the estimated or real snow grain concentration (110 ng g −1 or 1.77 × 10 −6 M for NO − 3 , 2.2 ng g −1 , or 6.47 × 10 −8 M for H 2 O 2 , and 0.092 ng g −1 or 2 × 10 −9 M for NO − 2 ). Our results are different from those computed at Neumayer station (Chu and Anastasio, 2007), where the dominant contributor to q OH production was H 2 O 2 (2.3 × 10 −11 M s −1 ), followed by NO − 3 (3.9 × 10 −13 M s −1 ) and NO − 2 (1.8 × 10 −13 M s −1 ). Further, the q OH production rate at Dome C for H 2 O 2 is 2 orders of magnitude lower than at Neumayer station, while it is similar for NO − 3 . The contribution of nitrite at Dome C is 1 order of magnitude lower than that computed at Neumayer station, where nitrite has been already considered as an insignificant source of q OH because of its very low estimated concentration (Chu and Anastasio, 2007). Overall, we can conclude that the contribution of H 2 O 2 , NO − 3 , and NO − 2 in forming OH radicals is low at Dome C both before and after the ozone hole period, and their change in photolysis is unlikely to have affected bromine preservation within the snowpack. This is in agreement with previous empirical observations (Legrand et al., 2016). We then propose that bromine release into the atmosphere can be favoured only in those locations where high snow acidity (e.g. in correspondence to a volcanic horizon or at coastal sites where biogenic marine emissions can enhance snow acidification) and high concentration of q OH precursors (e.g. H 2 O 2 ) are found.

The role of volcanic eruptions in bromine preservation at Dome C
Volcanic eruptions are a significant halogen source with the emission of large amounts of HCl, HF, and HBr (Pyle and Mather, 2009). In particular, BrO formation through heterogeneous photochemical reactions was detected in a volcanic plume where local O 3 destruction occurred (Von Glasow et al., 2009). However, the role of volcanic eruptions in affecting bromine concentration in ice and snow has been poorly addressed. Studies performed in the European Alps (Legrand et al., 2021) and in the West Antarctic Ice Sheet Divide (Mcconnell et al., 2017), showed opposite results, with recorded bromine increase and depletion in coincidence with volcanic events, respectively. The Dome C shallow core presented in this work covers a period characterised by at least seven inter-hemispherical volcanic eruptions that were identified in other snow pits and deep cores, using both nssSO 2− 4 and Fe(II) as volcanic proxies (Castellano et al., 2005;Burgay et al., 2021;Gautier et al., 2016): Pinatubo/Cerro Hudson (1991, VEI = 6), Agung (1963, VEI = 5), Krakatau (1886, VEI = 6), Cosiguina (1835, VEI = 5), Tambora (1815 = 7), and an unknown eruption (UE) 1809 (1809, VEI ≥ 5). VEI stands for Volcanic Explosivity Index, a commonly used quantity to define the magnitude of a volcanic eruption (Newhall and Self, 1982). Its values range from 0 (Hawaiian eruption) to 8 (ultra-Plinian eruption). In this record, we did not detect any clear fingerprint either as bromine increase or as depletion compared to the adjacent periods, suggesting a negligible role of volcanic eruptions in affecting the Br snow chemistry in the inland Antarctic plateau (Fig. S2). Nevertheless, we highlight that, due to low snow accumulation and to strong wind erosion, not all the volcanic eruptions listed above might be present in our record. Indeed, a previous investigation that compared the sulfate signal from five ice cores drilled 1 m apart from each other at Dome C showed a bulk probability of 30 % of missing volcanic events when a single core is used as the site reference (Gautier et al., 2016). Among the volcanic events embraced by our record, only Krakatau, Cosiguina, and UE 1809 were observed in all the previously mentioned five replicate cores, giving us confidence that for these eruptions the volcanic fingerprint is present also in our record. However, we acknowledge that a proxy-based volcanic reconstruction is missing for our core, and considering the strong spatial variability observed at Dome C, further and more specific studies are needed to investigate the impact of large inter-hemispherical volcanic eruptions on the preservation of bromine in the snowpack.
3.4 Can Br enr at Dome C be used as proxy for past sea-ice extent?
Having presented evidence to demonstrate the preservation of bromine in the snowpack at Dome C and the absence of a link between our Br signal and the formation of the ozone hole, we now investigate the suitability of Dome C Br enr as a proxy for past sea-ice variability. Previous studies supported the use of Br enr in reconstructing past Antarctic sea-ice extent (Vallelonga et al., 2017;Spolaor et al., 2013b). However, these ice core records were retrieved at coastal sites close to local source areas, where Br enr values where enriched with respect to sea water mass ratio. To the contrary, due to its position, Dome C receives atmospheric signals from a vast area of the East Antarctic sector (Fig. 1), which extends from the Indian Ocean to the Ross Sea, potentially giving a reconstruction of past-sea-ice extent over a broader region.  (1975) represents the beginning of the ozone hole period. The horizontal coloured lines represent the average magnitude during the preozone hole period .
Our 200-year ice core record shows that Br enr has an average value of 0.4 ± 0.3, meaning that it is typically depleted at Dome C. This reflects the differences in Na and Br depositions as a function of the distance from the coast, with resulting Br enr values lower than 1 recorded at sites located more than 800 km far from the coast . Further, due to the low snow accumulation at this location and to the low concentrations of bromine, Br enr values can be influenced by surface snow removal by wind erosion, changes in meteorological patterns, and changes in wind field. For these reasons, and with the current state of knowledge, the presented bromine record should be interpreted with caution.
To understand the driving patterns of the Dome C record and its suitability to reconstruct past-sea-ice variability, we compare the Dome C Br enr record with the annual averaged Southern Annular Mode (SAM) Marshall index (Marshall, 2003), satellite observations of FYSI extent from the source areas over the period 1979-2012, and with the Law Dome methanesulfonic acid (MSA LD ) profile (Curran et al., 2003) ( Table 2). SAM describes the poleward/equatorward movement of the westerly winds that circle Antarctica. When these winds, known as southern westerly winds (SWWs), contract towards Antarctica, the SAM is in its positive phase, and vice versa in its negative phase. The strength of wind patterns likely influences the amount of sea salt aerosols deposited at Dome C (Crosta et al., 2021), as indicated by the positive correlation of Br and Na with SAM index (0.41 and 0.61, pvalue <0.01, respectively). This is in agreement with recent findings that highlight a prominent northward flow during the SAM negative polarity at Dome C (Kino et al., 2021). We did not observe any correlation between the SAM index and Br enr values.
We find that for the past decades, Br enr at Dome C is mainly influenced by Br deposition, given the positive and significant correlation of Br enr with total Br (r = 0.76, pvalue ≤ 0.01) and the negative correlation with Na (r = −0.37, p-value ≤ 0.01) over the entire record (Table 2). Since gas-phase bromine is emitted in enhanced concentrations (with respect to sea water ratio) from sea salt aerosol derived from surface blowing snow deposited over FYSI, the Br enr signal at Dome C is likely to be mainly controlled by emissions and recycling from seasonal sea ice at the Antarctic coast rather than long-range air mass transport of sea salt aerosols (Spolaor et al., 2013b).
To test this hypothesis, we compared our record with FYSI extent data (Parkinson and Cavalieri, 2012) during the satellite era  over the main source areas defined through back-trajectory analysis (Fig. 1). As previously stated, the large majority of the back trajectories that reaches Dome C over the period 1979-2018 originated from the WP sector (see Sect. 3.1, Fig. 1). However, we found significant but weak, correlations between Br enr and FYSI only with the IO sector (11 % of the back-trajectory points satisfying bromine loading condition) and the RS (21 %) with r = 0.35 (p-value <0.1) and r = 0.3 (p-value <0.1), respectively. In contrast, the closer WP sector does not show any significant correlation with Br enr (Table 2). Given the main source areas from back-trajectory analysis are located in the WP (45 %, Fig. 1), we further investigate this sector by considering the MSA record retrieved from the Law Dome ice core (hereafter MSA LD ), located at a coastal site facing the WP (Curran et al., 2003). MSA LD shows a positive and significant correlation with the past-sea-ice extent in the WP sector (r = 0.89, p-value ≤ 0.01), but it does not correlate with Dome C Br enr , strengthening the idea that Dome C is influenced by a broader source area than Law Dome. Based on these findings, a possible interpretation is that the IO and RS seasonal sea ice might have a stronger influence on the Dome C Br enr profile than WP, due to their 211 % and 157 % average larger FYSI extent than the one recorded in the WP (Fig. S3), leading to a larger emission of reactive bromine into the atmosphere. Overall, we found a weak, but significant, correlation between the Br enr record and sea-ice extent in East Antarctica (WP + RS + IO) (r = 0.35, p-value ≤ 0.1). Nevertheless, we need to consider that overall sea-ice extent in East Antarctica has not undergone significant changes over the last 3 decades, with an interannual variability of ∼ 20 %. Moreover, taking into account the observed Br enr depletion at Dome C and the difficulties in capturing relatively small sea-ice variabilities due to snow remobilisation, and changes in meteorological patterns and in wind fields , we hypothesise that sea-ice extent variability observed over the last decades has not been large enough to cause a significant variability in the Br enr signal at Dome C. However, it cannot be ruled out that when longer periods which extend further back in the past are considered (e.g. glacial/interglacial transitions), Br enr variations could be used as a qualitative tracer (i.e. to identify transitions between large and small FYSI extent) for FYSI variability in East Antarctica.

Conclusions
In this paper we presented the first long-term ice core record of bromine and bromine enrichment from Dome C (Antarctica). Based on observations and modelling results, we propose that bromine is effectively preserved within the Antarctic plateau snowpack regardless of the intensity of the incoming UV-radiation. Furthermore, we find that the change in surface UV-radiation due to ozone hole formation does not affect the contribution of H 2 O 2 , NO − 3 and NO − 2 to the production of OH radicals and consequently the dominant OHdriven bromide oxidation channel remains slow. We suggest that neither of these photochemical mechanisms are likely to take place at Dome C, mostly due to the low concentration of H 2 O 2 and NO − 2 as well as the low q OH quantum yield from the NO − 3 photolysis. Due to the variety of chemical reactions that can influence bromine preservation within the snowpack, we suggest the inclusion of site-specific studies to assess to what extent bromine is preserved at different specific locations, i.e. through the analysis of q OH precursors (H 2 O 2 , NO − 3 and NO − 2 ). Finally, we provided preliminary insights on the effects of volcanic eruptions on the preservation of bromine in snow at Dome C, where we did not observe any influence on the Br record. However, due to the specific and peculiar environmental conditions at Dome C and to the lack of a proxy-based volcanic eruption reconstruction for this record, our findings are not conclusive and we encourage specific investigations on this topic. We also inquired whether Br enr at Dome C can be used as a sea-ice proxy for the East Antarctic sector. Despite finding weak -but significant -correlations with the Indian Ocean and Ross Sea sectors (which are the ones presenting the largest FYSI extents) it is difficult to validate Br enr as an effective proxy for past sea-ice reconstructions in East Antarctica; this is primarily due to low sea-ice variability observed during the last 30 years. Future investigations at Dome C need to focus on glacial/interglacial transitions to assess whether Br enr at Dome C is somehow related to largescale variations of sea-ice extensions.
Author contributions. FB: conceptualisation, data curation, formal analysis, investigation, methodology, visualisation, and writing (original draft preparation). RPF: data curation, formal analysis, investigation, methodology, software, visualisation, and writing (original draft preparation and review and editing). DS: data curation, formal analysis, methodology, software, visualisation, and writing (original draft preparation and review and editing). CT: data curation, methodology, and writing (review and editing). CSBB: writing (review and editing). RHR: writing (review and editing). CS: software and writing (review and editing). VC: software and writing (review and editing). CB: funding acquisition, supervision, and writing (review and editing). ASL: investigation and writing (review and editing). AS: conceptualisation, investigation, resources, supervision, and writing (review and editing).

402
F. Burgay et al.: 200-year ice core bromine reconstruction at Dome C (Antarctica) Financial support. This research has been supported by the Horizon 2020 (Beyond EPICA; grant no. 815384), by the Programma Nazionale per la Ricerca in Antartide (PNRA; project no. PNRA16_00295), and by the bilateral international exchange award Royal Society (UK)-CNR, titled "Antarctic sea-ice history: developing robust ice core proxies" (grant no. IEC/R2/202110), awarded to Rachael H. Rhodes and Andrea Spolaor.
Review statement. This paper was edited by Florence Colleoni and reviewed by two anonymous referees.