under a Creative Commons License. Atmospheric Chemistry and Physics Discussions

DOAS formaldehyde absorption cross sections T. Brauers et al.


Introduction
Stable isotope studies have been used for a long time to constrain fluxes and processes taking place at the surface of the Earth.Mass-dependent processes such as kinetics or thermodynamic equilibria lead to a quasi-linear relationship between oxygen isotope ratios in oxygen bearing compounds: δ 17 O 0.52δ 18 O, with the isotopic content reported as enrichments with respect to a standard reference:  (Thiemens, 1999).Due to a mass-independent formation process experimentally discovered by Thiemens and Heidenreich (1983), the oxygen isotopic composition in atmospheric O 3 does not fit the terrestrial fractionation line produced by mass-dependent processes (Mauersberger et al., 2001).This anomaly, termed by several authors as the Mass Independent Fractionation (MIF) contribution, is quantified as a deviation from the mass-dependent fractionation line often using the following linear expression: ∆ 17 O=δ 17 O−0.52×δ 18O.Tropospheric O 3 exhibits large ∆ 17 O values (up to 35‰ as measured by Krankowsky et al., 1995, andJohnston andThiemens, 1997, and calculated by Lyons, 2001) and there is considerable evidence that this anomaly is transmitted through chemical reactions in the atmosphere to other oxygen bearing compounds (Thiemens, 1999).Michalski et al. (2003) showed that atmospheric models could reproduce the temporal evolution of the anomaly measured in particulate nitrate oxygen in a polluted marine boundary layer.Here we present the coupled evolution of ∆ 17 O(NO − 3 ) and O 3 at Alert, Nunavut (82 • 30 N, 62 • 19 W) during the polar sunrise period in springtime 2004.At Alert, like everywhere else in the coastal Arctic, surface O 3 is subject to severe depletion events (ODE) in springtime (Bottenheim et al., 1986), during which the ozone mixing ratio can decrease from around 40 parts-per-billion in volume (nmol mol −1 ), its mean background level, to values as low as 1 nmol mol −1 and below, and remain at this level for several days (Bottenheim et al., 2002).Ozone depletion is due to catalytic cycles involving ocean-originating halogen oxides and radicals such as BrO (H önninger and Platt, 2002) but the understanding of the processes governing the release of such compounds from the surface is still a subject of debate (Domin é and Shepson, 2002).
To the best of our knowledge, our measurements are the first measurements of ∆ 17 O in atmospheric nitrate in coastal arctic, which therefore enhances the global representation of this variable.Permanent sunshine, sea-ice proximity and the associated halogen chemistry responsible for ODEs make the interpretation of our results a chal- EGU lenge for existing theories about isotopic anomalies in atmospheric nitrate and about ODE-related chemical mechanisms.Our analysis demonstrates the power of this isotopic tool to assess the magnitude of the reaction pathways leading to the formation of nitrate in various environments.

Measurement site and sample collection
Measurements and sample collection were performed at the Global Atmospheric Watch (GAW) observatory and at the Special Studies Trailer (SST) which are located on a plateau, 190 m above sea level and 6 km to the South-South West of Canadian Forces Station Alert, the main base at the coast of Northern Ellesmere Island.
Surface O 3 was measured at the SST site with a UV absorption instrument (model TEI 49-C, Thermo Environmental Instruments, Inc., Franklin MA, USA), calibrated against standards traceable to the National Institute of Standards and Technology (NIST).This instrument was terminated on 4 May.In order to analyze ozone measurements during the entire period of isotopic measurements, the gap between 4 and 18 May is filled by using the raw (uncalibrated) O 3 GAW dataset, and recalibrating using a linear regression against the SST ozone dataset over the overlapping period of measurements.The time resolution for ozone measurements is 30 min.
Samples were collected between 29 March 2004 and 18 May 2004, on cellulose acetate filters (Whatman 41) on a semi-weekly basis using a high volume aerosol sampler operated by Environment Canada since 1980 (see e.g.Sirois and Barrie, 1999, for details on sampling procedures).Each of the 14 samples was collected over a period of 3 or 4 days.First, ionic species trapped on the filter were dissolved in 40 ml of ultra-pure water (Millipore).For each of the 14 aqueous samples, an aliquot (100 µl) was taken out for the ion-chromatography analysis, and the rest of the sample was available for the isotopic analysis.
Inorganic anions (Cl − , Br − , NO − 3 and SO 2− 4 ) and cations (Na + , Ca 2+ , Mg 2+ , K + and NH + 4 ) were measured using the ion-chromatography system described by Jaffrezo et al. (1998).Four field blanks were performed to assess the contribution of possible contaminations induced by the manipulation of filters and by the filters themselves, as Whatman 41 filters have been subject to questions regarding their initial loading in some trace elements (Watts et al., 1987).Field blank samples were obtained by loading clean filters in the sampler, keeping them there for 5 min and packing them exactly like the actual samples.The contamination was estimated by dividing the amount of species collected on the filter by the typical volume pumped through regular filters (ca. 10 000 m 3 ).
For the samples presented here, the blank contribution, integrating all possible sources of contaminations, represents less than 4% for all the measured cations.The contribution of contamination to the total measured NO − 3 , SO 2−

and Br
− is less than 1%.Only Cl − measurements are affected by a 10% blank effect.It must also be noted that the isotopic measurements undertaken here are not sensitive to non optimal aerosol collection efficiencies (though estimated by Sirois and Barrie, 1999, to be >95%, based on previous work by Watts et al., 1987).We therefore mainly use the chromatographic measurements to estimate the relative composition of the aerosol, although the absolute concentration may be affected by a 10% bias.
Existing literature on the collection of atmospheric inorganic nitrate by means of high volume sampling on a variety of media show that none is devoid of sampling artefact (e.g.Schaap et al., 2002).

EGU
nitric acid) and gaseous nitric acid (HNO 3 ), the sum of which being referred to as total inorganic nitrate (TIN, Bottenheim et al., 1993).However, it is worth keeping in mind that the measurements of ∆ 17 O(NO − 3 ) based on filter collection are not sensitive to the chemical form of nitrate (either gaseous HNO 3 or particulate NO − 3 ) because any fractionation occurring at the interface between the gas phase and the particulate phase obeys mass-dependent relationships, without affecting the ∆ 17 O value.In what follows, we present and discuss ∆ 17 O(NO − 3 ) measurements, which represent the integrated isotopic anomaly of oxygen in TIN.

Isotopic analysis
The isotope analysis was performed using the method developed by Michalski et al. (2002).Chloride and sulphate were precipitated out of the samples, using BaCl 2 for SO 2− 4 and subsequent removal of chloride using an Ag-form ion-exchange resin (AG 50W-X8 200-400 Mesh size, Bio-Rad, Hercules CA, USA).Nitrate was then isolated by Dionex ion chromatography and converted to silver nitrate using a ion exchange column (Dionex AMMS III, 4 mm) where the usual regenerant was replaced by Ag 2 SO 4 (2.5 mmol.l−1 ).The isotopic composition of oxygen in AgNO 3 was determined by thermal decomposition at 550  Thiemens, 1999;Michalski et al., 2003;Alexander et al., 2004).We are aware that alternative expressions have been proposed for the mathematical expression of ∆ 17 O (see for instance Thiemens, 1999;Miller, 2002;Kaiser et al., 2004), the best of which being the non approximated and standard-independent expression proposed by Miller (2002): This expression unambiguously defines ∆ 17 O as the deviation from the massdependent fractionation curve in term of a power law relationship between isotope ratios (Miller, 2002).Since we focus on nitrate, characterized by large and strongly variable heavy oxygen enrichments and anomalies, the use of this definition is not as crucial as for slightly anomalous species (e.g.N 2 O), for which the historical linear definition (Thiemens, 1999) may be responsible for inaccuracies and interpretation errors, as pointed out by Kaiser et al. (2004).For instance, using the linear expression for our nitrate samples (see data in Table 1) results in an overestimation of ∆ 17 O of (0.74±0.07)‰ with respect to the non-linear expression (Miller, 2002), which lies within the range of the stated uncertainty (see above We present the data acquired in Alert, Nunavut, between 29 March 2004 and 18 May 2004.They mainly consist of meteorological observations (temperature, wind speed and direction), ozone mixing ratios, aerosol inorganic composition and the triple isotopic composition of oxygen in inorganic nitrate.

Meteorology
During the course of the measurement campaign, sunlight was permanent.The mean daily temperature ranged between −30 and −10 • C, gradually increasing over the measurement campaign.Cumulative precipitation (snow) was 22 mm during this period, with only one major (>8 mm) snowfall occurring on 12 April.Weather conditions were most of the time calm and clear.

Ozone
The ozone mixing ratio was highly variable during this period, and its time series features several drastic drops, commonly referred to as ozone depletion events (ODEs) that are ubiquitously observed in coastal polar regions (see also high resolution time series on Fig. 3) (Tarasick and Bottenheim, 2002).To quantify the link between the ozone mixing ratio and the composition of aerosols, the O 3 mixing ratio is averaged over each sampling period (i.e.3-4 days).

Particulate bromide and BrO
The Fig. 1 shows the coupled evolution of the ozone mixing ratio and the particulate bromide load.Particulate bromide (Br − ) ranged between 2 and 20 ng m −3 , and therefore represents a negligible fraction of the anionic load.However, elevated Br − contents (>10 ng m −3 ) were associated with ozone depleted air masses, which is consistent with the observations first made by Barrie et al. (1988)  EGU mainly originates from wet deposition of active bromine species such as Br and BrO, a covariation of BrO and Br − is expected.However, as evidenced by Lehrer et al. (1997) and Evans et al. (2003), inferring BrO mixing ratio from Br − is only possible for limited period of times.
Unfortunately no continuous record of BrO is currently available for the spring 2004 period at Alert, but on some occurrences the mixing ratio of BrO could be derived from Multi-Axis Differential Optical Absorption Spectroscopy (MAX-DOAS) measurements carried out at Alert or in its vicinity.Morin et al. (2005) have shown that on 22-23 April 2004, BrO ranged between 1-3 pmol mol −1 ("normal" ozone) and 10 pmol mol −1 (strong ODE).This is inline with measurements presented by H önninger et al. ( 2004), carried out in the vicinity of Alert in spring 2004.Overall, there seems to be evidence that, even during the strongest ODEs, BrO never reached values significantly above 10 pmol mol −1 in spring 2004.

Origin of air masses
Following the approach of Bottenheim et al. (2006) 1 and previous trajectory studies, 6days backtrajectories (Stohl et al., 1995;Stohl and Seibert, 1998;Stohl, 1998)  EGU trajectories are considered.Each backtrajectory represented in Fig. 2 is a mean of all the backtrajectories starting from Alert at 500 m altitude every 6 h during the collection period (about 3-4 days) of each sample.They provide an indication of the mean origin and history of air masses contributing to a given aerosol sample.Backtrajectories starting at 1 km altitude (not shown here) were also considered but the overall features were similar.Each mean trajectory (corresponding to an aerosol sample) is color-coded in Fig. 2 according to its particulate bromide content, which is indicative of the activity of the heterogeneous bromine chemistry responsible for ODEs.
The conclusions than can be drawn from the analysis of the Fig. 2 are in-line with other studies discussing the origin of ozone depleted air masses (e.g.Hopper et al.,     1998; Morin et al., 2005; Bottenheim et al., 2006 1 , and references therein).In most cases, drastic changes in the ozone mixing ratios observed at Alert at polar sunrise can be related to the movement of air masses.During the course of our measurement campaign, most air masses depleted in ozone were advected at low altitude above the Arctic Ocean before reaching the coastal sites.

Inorganic chemical composition of the aerosols
Our main focus being the analysis of the isotopic composition in TIN, we only present here a summary of our measurements of inorganic chemical composition of aerosols.Extensive reviews of the composition of inorganic aerosols observed at Alert can be found elsewhere (e.g.Sirois and Barrie, 1999).
The relative composition of aerosol in terms of inorganic anions was found to be rather invariant with time during April and May, but the total amount varied greatly.NO − 3 +SO 2− 4 +Cl − ranges between 145 and 1900 ng m −3 , in which Cl − , NO − 3 and SO 2− 4 represent (7±6)%, (23±8)% and (69±8)% of the molar composition respectively.As the chloride composition is affected by the strongest blank effect, and that halide chemistry is involved in ozone depletion, we prefer to use Na + for the estimation of the marine contribution to the aerosol loading.Using a molar sea salt ratio of 0.0603 (Holland, 1978), the calculated non-sea salt sulphate component (nss−SO

EGU
for more than 99% of the sulphate concentration.Aerosols collected during this study represent a mixture of sulphuric acid droplets and sea salt particles, on which inorganic nitrate is adsorbed (Sirois and Barrie, 1999).We see no noticeable correlation between the concentration of inorganic nitrate and the origin of air masses (data not shown).Based on nitrate concentrations only, it is virtually impossible to estimate the different sources of particulate nitrate, as alkylnitrate, peroxyalkylnitrates (PANs) and other organic nitrate dominate the budget of nitrate in the Arctic.Thus, the variability in particulate nitrate concentrations does not provide any insight in the budget of nitrate (Sirois and Barrie, 1999).

Isotopic composition of oxygen in atmospheric inorganic nitrate
We report elevated 18 O enrichments, with δ 18 O values ranging from 78 to 92‰.
3 ) ranges between 29 and 34‰ (see Table 1).The large temporal variability of ∆ 17 O(NO − 3 ) during this period is shown in Fig. 3. High isotopic enrichments and anomalies are to be expected in atmospheric nitrate: in terms of range, our measurements are consistent with previous measurements of δ 18  3 )/‰= 0.15 O 3 /(nmol mol −1 ) + 28.6, with R 2 =0.70 (n=12), showing that a strong link exists between these two variables (see Fig. 4).

Discussion
In this section, we first review the current knowledge regarding the origin of the isotopic anomaly in atmospheric nitrate and adapt it to the conditions experienced in the arctic springtime.
Based on this analysis, we then derive from known chemical mechanisms a relationship between ∆ 17 O in inorganic nitrate and the relative magnitude of the processes involved in its formation.We test this relationship using measured or estimated variables during the course of the measurement campaign, using a simple Monte-Carlo approach.
4.1 Overview of the present theory about the origin of ∆ 17 O to atmospheric nitrate Based on field measurements in coastal California and modeling, the large ∆ 17 O measured in atmospheric nitrate has been convincingly attributed by Michalski et al. (2003) to a transfer of the ozone oxygen isotope anomaly to atmospheric nitrate.In these conditions, it is well established that oxygen anomalies observed in stable compounds such as particulate nitrate are due to chemical interactions between their precursors (NO x ) and ozone.Our observations somewhat support this theory by bringing to light a co-evolution between the O 3 mixing ratio and the isotopic composition of oxygen in aerosol nitrate during ODEs, showing that O 3 mixing ratio and nitrate isotopes are intimately coupled.However, in the arctic lower troposphere in springtime BrO plays an important role an interacts greatly with ozone and NO x (Platt and H önninger, 2003).

EGU
The traditional view on ∆ 17 O(NO − 3 ) has been extensively described in Michalski et al. (2003) and we briefly summarize the main aspects here.The origin of the isotopic anomaly of oxygen in nitrate can be accounted for considering a two-steps chemical oxidation mechanism for nitrogen oxides: -NO x are primarily recycled through the following photochemical cycle (Finlayson-Pitts and Pitts, 2000): where RO 2 represents all peroxy and alkyl-peroxy radicals.Without taking into account BrO chemistry, at steady state, the anomaly in NO 2 only depends on the ratio between the rate of oxidation of NO by O 3 and RO 2 (R öckmann et al., 2001;Michalski et al., 2003).
-NO 2 is converted into atmospheric HNO 3 and particulate NO − 3 through one of the following main reaction pathways (Finlayson-Pitts and Pitts, 2000): Each of these reaction pathways (R5, R6 followed by R7, and R6 followed by R8) involves NO 2 , and also O 3 for the pathways (R6+R8) and (R6+R7), and therefore The first step acts upon time scales of minutes (lifetime of NO 2 with respect to photolysis) and the second one acts upon time scales of hours to days.It is therefore possible to fully decouple them when modeling the origin of the oxygen anomaly in particulate nitrate (Michalski et al., 2003).

Bromine oxides and ozone depletion events
It is now well established that bromine is the key element in the processes responsible for tropospheric ozone depletion in the Arctic (Bottenheim et al., 2002).Bromine originates from sea-salt, but the exact location where bromide is activated into the radical form, which destroys ozone, is still a matter of debate (Domin é and Shepson, 2002).Sea-salt aerosols (Fan and Jacob, 1992), but also the seasonal snowpack (Tang and McConnell, 1996;Sander et al., 1997;Michalowski et al., 2000) should be the predominant sources of bromine in the atmosphere.
The catalytic cycles of Br/BrO in both the gas phase and the condensed phase are responsible for the so-called "bromine explosion" (Wennberg, 1999) which results in a fast and efficient removal of ozone from the arctic boundary layer.In these conditions, significantly high levels of BrO were measured (H önninger and Platt, 2002) EGU gaseous phase and on wetted surfaces: In what follows, we neglect Cl chemistry, since Br is by far the most active halogen during ODEs (H önninger and Platt, 2002).However, taking into account these species would be crucial for stratospheric studies where chlorine chemistry dominates over bromine chemistry.

Oxidation of NO in the Arctic
We now examine the competition between the three major NO oxidants identified in the arctic troposphere in springtime (O 3 , RO 2 and BrO), in terms of kinetics and the subsequent influence on the isotopic composition of oxygen in NO 2 .

Reaction (R3
According to the analysis of R öckmann et al. ( 2001) and Michalski et al. (2003), in "normal" conditions (i.e.not influenced by bromine induced ozone depleting chemistry) this is the reaction responsible for the transmission of anomalous oxygen from ozone to NO 2 .Therefore the rate of transfer for ∆ 17 O between O 3 and NO 2 is one of the most critical parameter for our analysis.Since little is known regarding both the chemical mechanism for this reaction and the intramolecular isotope distribution in ozone, contradictory parameterizations have been published so far in various modeling studies: for some authors (see e.g.Lyons, 2001;R öckmann et al., 2001), only the terminal atom of ozone reacts with NO to yield NO 2 , although there is mechanistic evidence that the central atom of ozone also reacts with NO (van den Ende et al., 1982).In addition, the intramolecular distribution of oxygen isotopes in ozone is very likely to differ greatly from a stochastically expected distribution: this is the case for 18 O (Janssen, 2005)

EGU
we assume that 17 O behaves in a similar manner (Lyons, 2001).The combination of these two sources of uncertainties has made it impossible to consistently account for the transfer of anomaly resulting from Reaction (R3).Since it may be difficult to separate these two effects (even in controlled experiments), we introduce an integrated ∆ 17 O transfer coefficient (τ) for this reaction, such that at photochemical steady-state, when only considering Reaction (R3) and the subsequent photolysis of NO 2 : Assuming that the overall transmission of anomalous oxygen from ozone to NO 2 proceeds as if the bulk composition of ozone was more or less equally transferred by each of its atoms, which is equivalent to τ=1, the parameterization of Michalski et al. (2003) has proven so far to be the most realistic in terms of comparison between modeled and measured ∆ 17 O(NO − 3 ).Using an atmospheric chamber, Savarino et al. (2006) 2 are investigating this reaction to experimentally derive a measurement of τ in controlled conditions.So far none of these measurements contradict the simplified approach of Michalski et al. (2003).In addition, in no case the ∆ 17 O of the O-atom transferred to the resulting NO 2 is measured higher than the initial bulk ozone oxygen anomaly.Given the fact that the terminal atoms in ozone are more anomalously enriched in heavy atoms (Janssen, 2005), this is an additional evidence that the alternative hypothesis made by Lyons (2001) is not correctly describing the transfer of ∆ 17 O from ozone to NO 2 .
For the analysis of the results presented here, we take τ=0.9−1, which lies within the range derived from Savarino's experiments so far, and is the most consistent with the existing literature.

Reactions (R4): RO
In this reaction, R represents H or any organic radical.As peroxy radicals are derived from atmospheric O 2 , they do not bear a significant isotopic anomaly, so we take Introduction

Conclusions References
Tables Figures

Back Close
Full Screen / Esc

Printer-friendly Version
Interactive Discussion

Evaluation of ∆ 17 O(BrO)
In the arctic troposphere after polar sunrise, BrO is mostly produced by reaction with ozone, and plays an important role in the ozone destruction catalytic cycle (Platt and H önninger, 2003): Similarly to the NO+O 3 reaction and the associated photochemical cycle, which is conceptually similar to some extent to the bromine oxide ozone destruction catalytic cycle, it is expected a steady state isotopic equilibrium between BrO and O 3 to be reached within a few minutes.Since the position of O atoms in the ozone molecule determines their load of isotopic anomaly (Janssen, 2005), the steady state ∆ 17 O(BrO) value is expected to depend both on the branching ratio between central/terminal O atom transfer to BrO and the intramolecular isotopic composition of ozone.Theoretical considerations based on kinetic measurements (Toohey et al., 1988) and molecular crossed beam studies (Zhang et al., 1997) show that Br + O 3 is a direct reaction, and that the Br atom most likely attacks a terminal ozone atom.Lyons (2001) calculated that terminal atoms in ozone carry an anomaly between 70 and 80‰.More recent work by Janssen (2005) shows that the intramolecular distribution of heavy isotopes in ozone is temperature dependent, favouring higher enrichments for terminal atoms at higher temperature, so the calculation by Lyons (2001) should be seen as an upper limit.An experimental confirmation of this prediction is strongly needed, but this calculation seems to indicate that BrO could feature the highest isotopic anomaly in a chemically active molecule in the troposphere.Given the increasing influence given to BrO in EGU chemical mechanisms implemented in atmospheric models (e.g.Platt and H önninger, 2003;von Glasow et al., 2004;Lary, 2005), this potentially unique isotopic signature could represent a tool of choice to test these hypothesis.

Reaction (R9): BrO
Since BrO is present in the arctic boundary layer after polar sunrise (Hausmann and Platt, 1994;H önninger and Platt, 2002), NO oxidation by BrO must be taken into account for the assessment of the origin of the oxygen isotopic anomaly in NO 2 .When comparing the kinetic rates of oxidation of NO by BrO and O 3 (Atkinson et al., March 2005 version) in the context of the arctic boundary layer with temperatures ranging between 233 and 273 K, one realizes that 1 pmol mol −1 BrO has the same power as 2-4 nmol mol −1 O 3 in oxidising NO.In addition, as pointed out previously, the high isotopic anomaly in BrO makes it even more active as a source of anomalous O atom in NO 2 .

Derivation of the equation for the anomaly in NO 2
We calculate steady-state abundance ratios for a given NO 2 isotopologue using the kinetic equations derived from the photochemical cycling of NO x and ozone, disturbed by peroxy radicals and bromine oxides.We neglect species containing more than one heavy O atom ( 17 O or 18 O) because of their extremely low abundances.In addition, in the kinetic equations for NO and NO 2 , NO x sink terms such as the formation of HNO 3 are negligible compared to the NO−NO 2 interconversion terms.Indeed, the NO and NO 2 chemical lifetime (about a min) is much shorter than NO x lifetime (about a day

EGU
of the isotopomers of a given species X.

Conclusions References
Tables Figures

Back Close
Full Screen / Esc

Printer-friendly Version
Interactive Discussion

EGU
This further reduces to: The latter term is equal to zero when τ=1, and ranges between 2-3‰ in the conditions presented here (see data in Table 1) and τ>0.95.However it must be noted that if τ was to vary significantly from 1, this would alter Eq. ( 5) and require a specific analytical treatment.For the development of this expression we neglect the deviations which would result from τ values differing from 1, given the analytical uncertainties and the lack of knowledge regarding the transmission of anomaly from O 3 to NO 2 .Additionally, as discussed above, ∆ 17 O(RO 2 ) 0. We therefore rewrite an approximated form for this equation: In the extreme case where peroxy radicals disturbance of the photochemical cycling between NO x and ozone is negligible and BrO does not play a significant role, we find indeed that at photochemical equilibrium ∆ 17 O(NO 2 )=τ ∆ 17 O(O 3 ), with τ 0.9−1.
In addition, this equation is compatible with the formalism adopted by Michalski et al. (2003), which did not include BrO.A generalisation of this approach is possible: noting X i a population of oxygen bearing NO oxidants, with τ i their ∆ 17 O transfer coefficient (function of the intramolecular distribution of oxygen isotopes in X i and of the mecha-

Nitrate formation
As discussed before, because sunlight was permanent at the sampling site and in the Arctic basin during the whole measurement campaign, any processes involving nighttime radicals such as NO 3 is considered insignificant (the lifetime of NO 3 with respect to photolysis is only a few seconds).As a result, the pathways (R6+R8), and (R6+R7) can be discounted.In these conditions, two processes arise to account for the formation of significant levels of inorganic nitrate in the lower troposphere: NO 2 either reacts with OH to form gas-phase nitric acid, or reacts with BrO to form BrONO 2 .Bromine nitrate subsequently undergoes hydrolysis on ice surfaces and yields particulate nitrate (Hanson et al., 1996), or sustains nitric acid formation through a gas-phase equilibrium with HOBr and HNO 3 (Sander et al., 1999).The role of BrONO 2 has been investigated since the early stages of the research effort on ODEs and is an important Br recycling mechanism (Fan and Jacob, 1992).However, its concomitant impact on nitrate formation has not drawn a similar attention until recent years.In a modeling effort dealing mainly with gas-phase mechanisms, Calvert and Lindberg (2003) showed that BrONO 2 would be the most abundant N-containing species in the arctic boundary layer.For Lary (2005), up to 35% of HNO 3 in the free troposphere could be formed through the BrONO 2 hydrolysis.
Based on temperature dependent kinetic rates (Atkinson et al., March 2005 version) and heterogeneous loss rates from Ridley and Orlando (2003) and Michalowski et al. (2000), we compared the speed of formation of nitrate through BrONO 2 hydrolysis and reaction of NO 2 with OH.We find that with only 1 pmol mol −1 BrO and 10 6 cm −3 OH, 40% of inorganic nitrate is produced through the BrONO 2 hydrolysis channel.This Introduction

Conclusions References
Tables Figures

Back Close
Full For the gas phase OH + NO 2 reaction, it is clear that in the resulting nitrate two O atoms carry an isotopic signature corresponding to the parent NO 2 , and one with the isotopic signature of OH (Michalski et al., 2003), so that For the reaction involving BrONO 2 , it is necessary to address the origin of the third atom in the resulting nitrate, since it could carry the isotopic signature of BrO or that of water depending on the hydrolysis mechanism.By means of independent techniques, Gane et al. (2001) and McNamara and Hillier (2001)  Finally, combining these two competing mechanisms and the numerical expression for ∆ 17 O(NO 2 ) yields: with r the proportion of nitrate produced through BrONO 2 hydrolysis.This relationship allows for the calculation of ∆ 17 O(NO − 3 ) produced in conditions influenced by the local chemical composition of the lower atmosphere.Long range transport does not have any influence on the isotopic budget of nitrate: NO y are mainly transported as PAN and locally decompose to yield NO 2 .In turn, it quickly conforms to the isotopic composition of ambient NO 2 .If particulate nitrate was to be transported to the Arctic, it is likely that it would undergo dry or wet deposition on the sea surface or onto the seasonal snowpack.NO x subsequently emitted from the snowpack (Ridley et al., 2000;Beine et al., 2002) would also quickly gain a local isotopic signature.Finally, if particulate nitrate was to be transported as it to the Arctic without any modification of its isotopic composition, this would not explain the large anomalies we measured because air masses originating from the south would most probably carry an isotopic signature comparable to the values reported by Michalski et al. (2003), i.e. between 23 and 25‰ for this time of the year.We consequently assume that all of the inorganic nitrate we analyzed carries a local signature, and that our measurements only cannot address the question of the origin of nitrate in the Arctic.

Estimation of variables of Eq. (18) and implications
To quantitatively interpret isotope data on inorganic nitrate during ODEs, we derived from existing chemical reactions a relationship between ∆ 17 O(NO − 3 ) and variables that Introduction

Conclusions References
Tables Figures

Back Close
Full Screen / Esc Printer-friendly Version

Interactive Discussion
EGU are likely to describe the isotope kinetics in the arctic lower troposphere (see Eq. 18).We now inspect the terms in this equation and propose a set of parameters which consistently account for the observed co-evolution of ∆ 17 O(NO − 3 ) and the mixing ratio of ozone during ODEs.
Except for ∆ 17 O(NO − 3 ), the only measured variable in this equation is the mixing ratio of ozone.Other variables are estimated as follows: We use the widely accepted tropospheric value for ∆ 17 O(O 3 ) of 35‰ (Thiemens, 1999;R öckmann et al., 2001;Lyons, 2001;Michalski et al., 2003Michalski et al., , 2005;;McCabe et al., 2006).However, no systematic measurement of ∆ 17 O(O 3 ) in the troposphere has ever been undertaken.To the best of our knowledge, only two studies (Johnston and Thiemens, 1997;Krankowsky et al., 1995) attempted to describe this variable in the troposphere based on field measurements.Johnston and Thiemens (1997) show highly variable measurements, with ∆ 17 O(O 3 ) ranging between 22 and 35‰ depending on the environmental conditions at the sampling site, and especially the NO x load of the atmosphere.According to these authors, "the greater the degree of photochemistry, the more mass-dependently fractionated the ozone" i.e. the lower the ∆ 17 O(O 3 ), because it is controlled by the balance between ozone formation (mass-independent) and its decomposition (mass-dependent).Lyons (2001) modeled the vertical distribution of ∆ 17 O(O 3 ) in a standard atmosphere, using approximated kinetic data and the pressuretemperature dependencies described by Morton et al. (1990), and showed that at the surface ∆ 17 O(O 3 )=35‰.We are aware that in the Arctic this value may not be valid, but until a systematic study of the variability of ∆ 17 O(O 3 ) in the troposphere has been achieved we use the most trusted value as a starting point.To the best of our knowledge no surface peroxy radical measurements have ever been undertaken at Alert in early spring and very little RO x −HO x data is available in the arctic boundary layer during ODEs.Nonetheless, we calculate the evolution of RO 2 ratio during ODEs based on the field data acquired in the arctic boundary layer during the campaign TOPSE 2000 (Tropospheric Ozone Production about the Spring Equinox experiment, see Atlas et al., 2003).Evans et al. (2003) summarized the boundary layer data acquired during TOPSE by binning mean measurements of species according to the ozone levels.According to Cantrell et al. (2003), the [HO 2 ]/[RO 2 ] ratio was found to be more or less constant (0.75) during this whole campaign so that [HO 2 ]=0.75 [RO 2 ].
The dominant alkyl-peroxy radical is CH 3 O 2 (Cantrell et al., 2003), so for the purpose of this study we take In what follows, we therefore calculate k RO 2 +NO = 0.75k HO 2 +NO +0.25k CH 3 O 2 +NO so that RO 2 is treated as a virtual species integrating the combined kinetic rates of its two main components according to their relative abundances.Evans et al. (2003) showed that during ODEs, when the ozone mixing ratios drops from about 40 nmol mol −1 to 5 nmol mol −1 on average (consistent with our averaging approach), RO 2 concentrations decrease from (2.2±1.2) 10 8 cm −3 to (1.3±0.6) 10 8 cm −3 , corresponding roughly to a factor 2 division.At the same time, during ODEs, CH 2 O was found to drop by about a factor 4. This is consistent with the simplified RO 2 chemical budget described in Evans et al. (2003) where the main source of RO 2 is the photolysis of species such as CH 2 O, and the main sink is the self reaction RO 2 +RO 2 (mainly HO 2 +HO 2 ).Thus, at steady state, the RO 2 concentrations are expected to be proportional to the square root of the concentration of CH 2 O, which appears to be mostly the case during TOPSE 2000.CH 2 O mainly originates from the snowpack (Sumner and Shepson, 1999) and its drop during ODEs seems to be driven by reactions with halogen atoms, according to model studies (Evans et al., 2003)
Since in no case the average ozone mixing ratio during the sampling time of an aerosol sample was lower than 5 nmol mol −1 , we do not consider here the bin corresponding to the lowermost ozone mixing ratio.From these data, we observe that RO 2 and O 3 are positively correlated.From these data, to estimate the concentration of RO 2 as a function of the ozone mixing ratio at Alert during our measurement campaign, we applied the following relationship: RO 2 /10 8 cm −3 =(0.09O 3 /nmol mol −1 +1.31) 1/2 , and calculated the corresponding [RO 2 ] † accordingly, using kinetic rates from Atkinson et al. (March 2005 version).

[BrO] †
In springtime 2004, some BrO measurements were carried out at Alert: Morin et al. (2005) show that on the frozen ocean a few km off the coast north of Alert, in ODE conditions instantaneous BrO levels did not exceed 10 pmol mol −1 , and that in normal conditions BrO was in the 1-3 pmol mol −1 range.This is quite similar to the observations made in spring 2003 by H önninger and Bottenheim (2003) at Alert.In addition, the detailed study carried out by H önninger and Platt (2002) shows that BrO reaches significant levels (above 5 pmol mol −1 ) only when ozone drops below 5 nmol mol −1 : in no case this situation happened during the entire course of an aerosol sampling period, since the minimum averaged ozone mixing ratio is 5.2 nmol mol −1 .
To check the consistency of our approach and its predictive capacity, we implemented a simple Monte-Carlo method to calculate the following remaining variables: r, τ, ∆ 17 O(BrO) and [BrO] † .We assume that they are constant during the measurement campaign, the idea being to derive mean values for ∆ 17 O(BrO) and [BrO] † that would be representative of ODE conditions.In these simulations, r was allowed to vary Introduction

Conclusions References
Tables Figures

Back Close
Full Screen / Esc Printer-friendly Version

Interactive Discussion
EGU between 0 and 1, τ between 0.9 and 1, ∆ 17 O(BrO) between 20 and 80‰, and [BrO] † corresponding to a 1-10 pmol mol −1 BrO range.For each variable, 10 values linearly distributed in the stated interval were considered, and the chosen values correspond to a minimization of the quantity: for the i = 1...12 data points analyzed for this study, and ∆ 17 O(NO − 3 ) calculated,i using Eq. ( 18) and the variables described above.
With this method, we find r=0.9,τ=1, [BrO] † corresponding to a 1.9 pmol mol [BrO] † , the value we obtain (1.9 pmol mol −1 BrO mixing ratio) is perfectly in-line with the data presented by H önninger and Platt (2002) and Morin et al. (2005) for situations where the ozone mixing ratio remains over 5 nmol mol −1 (which, as discussed previously, was on average the case for all the situations experienced during the measurement campaign).
Figure 6 shows a direct comparison of the measured and calculated ∆ 17 O(NO − 3 ) values, using Eq. ( 18) and variables as determined above.
3 ) with ozone mixing ratio (see above), and on simulated data the slope is 0.10, with very comparable regression coefficient (R 2 =0.70 for measured values, and R 2 =0.74 for simulated data).This shows that the broad picture is well captured by our interpretation, even if improvements are highly desirable for a better quantification of the processes at play.

Conclusions
Using present knowledge regarding the transmission of the ozone oxygen isotopic anomaly to particulate nitrate through photochemical reactions between NO x and ozone, and the mechanisms responsible for the conversion of NO 2 to inorganic nitrate in the polar atmosphere after polar sunrise, we demonstrated in this paper that its isotopic composition records a footprint of the tropospheric ozone depletion events which occur in this environment, based on atmospheric inorganic nitrate samples collected at Alert during 7 weeks in spring 2004.Including Arctic-specific processes in the analytical scheme first proposed by Michalski et al. (2003), we show that most of the nitrate is produced through hydrolysis of BrONO 2 after the BrO + NO 2 reaction.This complements the widely accepted crucial role of this process in recycling BrO and acting to destroy ozone in the arctic boundary layer in springtime (Fan and Jacob, 1992).We also showed that, in the conditions experienced during our measurement campaign, the expected increase in BrO mixing ratio during strong depletion events did not happen significantly, since the correlation between ∆ 17 O(NO − 3 ) and the mixing ratio of ozone during the sampling interval was not changed at low ozone mixing ratio.This extended theoretical framework predicts that, in situations of sustained strong ODE during the whole sampling interval of a filter sample (i.e.O 3 <5 nmol mol −1 ), the Introduction

Conclusions References
Tables Figures

Back Close
Full Screen / Esc Printer-friendly Version

Interactive Discussion
EGU accompanying high BrO levels could result in higher ∆ 17 O(NO − 3 ) values, since elevated BrO levels would then efficiently transfer their anomaly to NO 2 , hence to nitrate.Indeed, we found evidence for such a situation in samples collected at Barrow, Alaska in March 2005 (Morin et al., 2006 3 ), where the linear correlation between ∆ 17 O(NO − 3 ) and the ozone mixing ratio reversed for low ozone mixing ratios.
In addition to the analysis of this specific situation, we generalized the isotopic massbalance approach proposed by Michalski et al. (2003) to a situation where more than one anomalous oxygen bearing NO oxidant acts in the system.Our formalism is compatible with previous studies and it demonstrates the urgent need for better characterizations of ∆ 17 O values for active species (namely, tropospheric ozone and BrO in various environments) and anomaly transfer coefficients (τ values) for reactions with NO.
Finally, our results can be interpreted in term of the reaction of the atmospheric reactor to a change in its oxidative power: during ODEs, the ozone mole fraction drops are associated with a strong decrease in the oxidative power of the atmosphere, as O 3 is a prominent oxidant of the atmosphere and chemically linked to HO x chemistry.It has been suggested, in light of the first findings on the origin of the oxygen anomaly in nitrate, that the oxygen isotopic content of nitrate recorded in polar ice cores could be used as a proxy for past oxidative power, crucially needed to take into account chemical feedbacks on the Earth climate.More work is required to test this idea on a broader dataset, and to use these empirical observations to develop a model that would account for these links in a predictable manner.In addition, post-depositionnal effects within the firn and in the ice have to be studied carefully in order to assess past ∆ 17 O(NO Full  (March-April 2004).Each backtrajectory is a mean of all the backtrajectories starting from Alert every 6 h during the collection period (about 3-4 days) of each aerosol sample (for a total of 12 samples).This provides an indication of the origin and history of air masses contributing to a given aerosol sample.The filled circles represent the position of the air masses every 2 days.The circles and lines are color-coded according to the particle bromide concentration measured in each sample (in ng m −3 ).Note that most samples with enhanced bromide concentration are associated with trajectories originating from the siberian side of the Arctic basin or to rather motionless air masses, sitting over the Arctic Ocean.Such areas are believed to be key locations for the bromine heterogeneous chemistry involved in Arctic surface ozone depletion events.Introduction

Conclusions References
Tables Figures

Back Close
Full Note that these channels are suppressed in the case of this study, since sunlight was permanent during the sampling campaign.Major reactions involved in anomalous oxygen transfer (for day-light reactions only) are highlighted using thicker arrow.Simplified processes for the release and the cycling of BrO are included in this figure (modified after H önninger and Bottenheim, 2003).We neglect Cl-bearing species for clarity.Homogeneous and heterogeneous reactions are not specified in this scheme, since some may proceed both pathways (e.g.BrONO 2 + H 2 O → HONO 2 + HOBr).This has no implication on the isotopic budget presented here since gas-to-particle conversions obey mass-dependent relationships.Only significant reaction products are kept in this figure.Introduction

Conclusions References
Tables Figures

Back Close
Full O and ∆ 17 O, such as those carried out by Michalski et al. (2003) (La Jolla, California: δ 18 O(NO − 3 )=50-89‰ and ∆ 17 O(NO − 3 )=20-31‰ during a year-round sampling campaign).High δ 18 O(NO −3 ) values were also reported in snow at high latitudes in the northern hemisphere.For example,Hastings et al. (2004) reported δ 18 O(NO − 3 ) values ranging between 65 and 80‰ in samples collected at Summit, Greenland.Heaton et al. (2004) also found high δ 18 O(NO − 3 ) values (60-85‰) in samples collected at Ny Ålesund, Svalbard.Finally, Alexander et al. (2004) measured ∆ 17 O(NO − 3 ) in the Site A ice core in Greenland, with values ranging between 27 and 29‰.In polluted boundary layer conditions, Michalski et al. (2003) have measured ∆ 17 O(NO − 3 )=23-25‰ in March-May.Our measurements for the same period of the year are much higher and highly variable, with variations as large as 5‰ from one sample to the other.Most of these variations are associated to the occurrence of which ∆ 17 O(NO − 3 ) drops to values as low as 29‰.A linear regression between ∆ 17 O(NO − 3 ) and the average ozone mixing ratio over the collection period of each aerosol sample gives: ∆ 17 O(NO − to over 80% in conditions more representative of the arctic boundary layer in spring time, with 2-5 pmol mol −1 BrO (H önninger and Platt,

)Fig. 1 .Fig. 2 .
Fig. 1.Time series of the ozone mixing ratio (diamonds, black line) and particulate bromide concentration (circles, green dashed line).The error bars associated to the ozone mixing ratio represent its variability (1σ) within each averaging period.The concentration of particulate bromide carries an uncertainty of about 10%, not shown for clarity.

Fig. 3 .Fig. 4 .Fig. 5 .
Fig. 3. Coupled evolution of surface O 3 mixing ratio (plain line) and ∆ 17 O(NO − 3 ) (•).Error bars represent the uncertainty due to the mass-spectrometer internal standard error and the correction of the blank effect.

)Fig. 6 .
Fig. 6.Coupled evolution of surface O 3 mixing ratio (plain line), measured ∆ 17 O(NO − 3 ) (•) and calculated ∆ 17 O(NO − 3 ) (squares).Errors bars on measured values represent the uncertainty due to the mass-spectrometer internal standard error and the correction of the blank effect.The calculated isotopic anomalies are based on the analysis summarized in Eq. (18) and variables measured or estimated (see text for details).
sample /R SMOW −1) × 1000 where R is the 17 O/ 16 O or 18 O/ 16 O ratio in the sample or the Standard Mean Ocean Water (SMOW) taken as a reference uncertainty are correlated to the size of the O 2 samples.Because of a too low concentration of nitrate in two samples, the thermal decomposition yielded a too small amount of O 2 to be reliably analyzed.Therefore we present here the data obtained on 12 samples in terms of their isotopic composition.
• C in a vacuum line and analysis on a isotope ratio massspectrometer (Dual Inlet IRMS, Finnigan MAT 253).All the analytical steps were simultaneously performed on nitrate standards (International Atomic Energy Agency USGS 34 and USGS 35, with ∆ 17 O=0‰ for USGS 34 and 21.1‰ for USGS35, B öhlke et al.,     2003).Standards were only used for the correction of biases related to a blank spectrometer (accounting for about ±0.3‰ for a detector voltage of 500 mV in average) and from the correction of the blank effect (accounting for about ±1.0‰).Both of Introduction at Alert.Since particulate bromide Introduction transfer of anomalous oxygen to HNO 3 in different proportions.It is possible to calculate the resulting ∆ 17 O(HNO 3 ) for each channel (seeMichalski  et al., 2003, for details).
, with peak values as high as 30 pmol mol −1 during a strong ODE near Alert in spring 2000.
-Formation of nitrate through the hydrolysis of bromine nitrate (BrONO 2 ) in both Introduction , and in lack of experimental data regarding the intramolecular distribution of 17 O in ozone Introduction where [X] denotes the sum of atmospheric concentrations Introduction reached the conclusion that in the hydrolysis process the O atom initially bound to Br would bind the N atom in NO 2 to form nitrate, incorporating the isotopic signature of BrO.Thus in nitrate produced Introduction . Overall, during ODEs, RO 2 concentrations drop by a factor much lower than the reduction factor Introduction BrO mixing ratio and ∆17 O(BrO)=44‰.The r value is in excellent agreement with the previously mentioned kinetics calculations showing that in the conditions of the arctic boundary layer in springtime most of the nitrate should be produced through the BrO + NO 2 reaction, followed by the hydrolysis of BrONO 2 .τ=1 is fully consistent with current knowledge regarding the NO + O 3 reaction (see above).∆ well reproduced by our simulated values, as well as the range of variation.Our analysis would be greatly improved if continuous BrO measurements were available during the course of this measurement campaign, however even if we do not Introduction

Table 1 .
Geophys.Res.Lett., 23, 2633-2636, 1996.6268Tarasick,D.andBottenheim,J.: Surface ozone depletion episodes in the Arctic and Antarctic from historical ozonesonde records, Atmos.Chem.Phys., 2, 197-205, 2002.6262Thiemens,M.: Mass-Independent Isotope Effects in Planetary Atmospheres and the Early Solar System,Science, 283, 341-345, 1999.6257,6261, 6279Introduction Summary of the data used in this study.Date ON represents the time (UTC) when the sample started to be collected.Less than 5 min separate each sampling period (data not shown), so the end of sampling correspond to the beginning of the next sample.All data represent averages over each sampling period.∆