Enhancement of the volcanogenic “ bromine explosion ” via reactive nitrogen chemistry ( K ı̄ lauea volcano , Hawai ’ i )

G. G. Salerno, C. Oppenheimer, V. I. Tsanev, A. J. Sutton, T. J. Roberts, and T. Elias Istituto Nazionale di Geofisica e Vulcanologia, sezione di Catania, Piazza Roma, 2, Catania, 95123, Italy University of Cambridge, Department of Geography, Downing Place, Cambridge, CB2 3EN, UK Le Studium®, Institute for Advanced Studies, Orleans and Tours, France Institut des Sciences de la Terre d’Orléans, 1a rue de la Férollerie, Orléans, cedex 2, 45071, France University of Cambridge, Department of Chemistry, Lensfield Road, Cambridge, CB2 1EW, UK United States Geological Survey, Hawaiian Volcano Observatory, P.O.-Box 51, Hawaii National Park, HI 96718, USA


Conclusions References
Tables Figures

Introduction
Volcanoes release significant quantities of reactive gases and particles into the atmosphere, with fluxes and relative abundances that tend to reflect levels of volcanic activity, tectonic setting and extent of magmatic-hydrothermal interaction (e.g., Carroll and Holloway, 1994;Mather et al., 2003;Oppenheimer, 2003).Volcanism has played Introduction

Conclusions References
Tables Figures

Back Close
Full Screen / Esc Printer-friendly Version Interactive Discussion Ernst, 2005), and continues to play a significant role in the composition and radiative transfer of the atmosphere (e.g., Kelly et al., 1997;Grainger and Highwood, 2003).
The study of volcanic emissions is thus essential not only for modelling eruptive activity and for hazard evaluation but also to evaluate fully their impacts on the Earth system.With the recent discovery of reactive halogens, notably BrO, in volcanic emissions from both passive (e.g., Bobrowski et al., 2003) and explosive (e.g., Theys et al., 2009) volcanic degassing, the role of volcanism in atmospheric ozone abundance, acidity and oxidising capacity has become the focus of interest in the wider context of the roles of halogens in the troposphere.Chlorine and fluorine species are among the most abundant volatile species in volcanic emissions following water vapour, CO 2 and SO 2 (e.g., Symonds et al., 1994).Heavy halogens, although present in lesser quantities, nevertheless are expected to partition strongly into magmatic gas phases (Bureau et al., 2003), and assume special significance on account of their important atmospheric chemistry.Typically, halogens are more abundant in gases released by convergent-plate-sited volcanoes, due to their associated water and halogen-rich magmas compared with divergent-plate-sited and hot-spot volcanoes (e.g., Hilton et al., 2002;Oppenheimer, 2003).Despite the noteworthy influence of halogens in magmatic processes (Hedenquist and Lowenstern, 1994;Aiuppa, 2008a) and atmosphere chemistry (e.g., Von Glassow et al., 2009), their role in volcano dynamics remains poorly constrained (Aiuppa et al., 2008b).Owing to the low abundance of heavy halogens in magma, more is known of their behaviour in convergent-plate magmas (e.g., Carol and Holloway, 1994;Muranatsu and Wedepohl, 1998).The inventories of bromine content in magma (e.g., Wedepohl, 1974;Wehrmann, 2005), in volcanic gas (e.g., Sugiura et al., 1963;Snyder et al., 2002), and Br-oxide emission rates (e.g, Bobrowski et al., 2007b), remain weakly constrained (Pyle and Mather, 2009).
Thanks to the proliferation of low cost ultraviolet spectrometers and associated analytical techniques such as Differential Optical Absorption Spectroscopy (DOAS) in the last decade (e.g., Platt et al., 1994;Plane and Saiz-Lopez, 2006) several volcanogenic BrO measurements have been carried out.In particular, since the first detection of BrO Introduction

Conclusions References
Tables Figures

Back Close
Full by UV DOAS (at Soufri ère Hills Volcano, Montserrat; Bobrowski et al., 2003), numerous efforts have been made to advance the knowledge of the origin and contribution of this reactive halogen in volcanic plume chemistry.Several studies have argued that BrO is not a primary magmatic species, but that it forms in volcanic plumes via the so-called "bromine explosion" (Wennberg, 1999), i.e., by heterogeneous, photochemical processing originating with oxidation of a small quantity of magmatic HBr (e.g., Oppenheimer et al., 2006;Bobrowski et al., 2007a;Roberts et al., 2009).Recent observations at Masaya volcano where BrO was observed only during daylight hours provide strong evidence for the proposed chemical schemes (Kern et al., 2008).
Here we report and discuss the first observations of BrO emission from a hotspot volcano, namely Kīlauea (Hawai'i, USA).et al., 2003), whereas Kīlauea summit activity began on 12 March 2008 (Orr et al., 2008;Patrick et al., 2008;Poland et al., 2008).The Summit and East Rift activity and degassing have been the subject of many studies, which demonstrate that much of the magmatic supply of CO 2 is released at the summit (e.g., Gerlach and Graeber, 1985).UV spectra were acquired within 1-1.5 km of both the Pu'u' Ō' ō crater on 3 September 2007 and Halema'uma'u crater on 13 May 2008 (Fig. 1).Surveys were performed between 09:00 and 16:00 LT (Local Time), dictated largely by the prevailing weather conditions and wind direction.SO 2 , BrO and NO 2 Slant Column Amounts (SCA) were retrieved from successively collected passive UV skylight spectra recorded by an Ocean Optics USB2000 spectrometer with spectral range 295-375 nm and ∼0.44 nm (full-width-at-half-maximum) nominal resolution.The spectrometer was housed in a thermo-stabilised enclosure (held at 15 • C) and coupled by a fibre optic bundle to a single lens telescope (full plane field of view of 8 mrad).The azimuthal viewing directions were north-west and east in 2007 and 2008, respectively; the elevation angle of the field of view was ∼30 • .Each spectrum was recorded with an overall time step of ∼20 s by exposing the detector for ∼200 ms and co-adding up to 100 successive individual spectra (so as to maximise the signal-to-noise-ratio whilst avoiding saturation in the wavelength region used in the retrieval).In each survey 3-4 datasets were recorded, each consisting of ∼80 spectra collected over a ∼20 min period.
The cooling and thermo-stabilisation reduced the associated dark current and avoided temperature drifts that are known to affect the performance of the detector and optical bench.For a few datasets in both surveys, the thermo-stabilised instrument was not used, in which case we collected Clear Sky Reference (CSR, out of plume), dark and offset spectra, routinely at the start and the end of each data set sampling to account for instrumental drift.The CSR spectra were recorded by pointing the telescope away from the plume with the same elevation used to observe the plume.
Each recorded spectrum was evaluated following standard DOAS methodology (e.g., Noxon, 1975;Perner et al., 1979;Platt and Stutz, 2008) using the WinDOAS V2.10 software package (Fayt and Roozendael, 2001).The Ring spectrum (e.g., Solomon 1987;Fish and Jones, 1995) was calculated from the CSR following the approach of Chance et al. (1998).Both laboratory spectra of SO 2 , BrO, O 3 , and NO 2 (293, 298, 293 and 223 K, respectively;Wahner et al., 1998;Bogumil et al., 2003) and the Ring spectrum were convolved to the spectrometer's resolution.In the fitting process, a first order linear offset was also included to account for possible instrumental and/or atmospheric stray light, residual dark current and for different (time variant) Rayleigh/Mie contributions in the recorded spectra.Each spectrum was shifted and stretched until Introduction

Conclusions References
Tables Figures

Back Close
Full Screen / Esc Printer-friendly Version Interactive Discussion the best fit was achieved.The optimal fitting windows for SO 2 (305.0-316.0nm), and BrO and NO 2 (345.0-358.0nm) were selected to avoid interferences between gases during the retrieval, and yielded near random structures with minimal standard deviation in the residuals.Additionally, BrO and NO 2 were retrieved for a wider spectral window (330-370 nm) to check for artefacts in the retrieved column amounts.Figure 2 shows an example of the BrO and NO 2 fit over the 345.0-358.0nm spectral range for a spectrum collected at Halema'uma'u crater on 13 May 2008.

Results
In Table 1, we compile the results obtained for the Pu'u' Ō' ō and Halema'uma'u crater measurements together with the prevailing environmental and plume characteristics.The Pu'u' Ō' ō spectra yield a mean SO 2 abundance around 4 times lower than those measured at the summit (∼3×10 18 molec cm −2 ) and did not reveal BrO or NO 2 .Conversely, at Halema'uma'u crater, BrO was successfully detected (mean of 3×10 15 molec cm −2 ) as well as negative (differential) amounts of NO 2 (mean of −6×10 16 molec cm −2 ).Note that negative abundances signify lower NO 2 relative to the background.Figure 3 shows SO 2 versus BrO SCA measured at Halema'uma'u crater in 2008.The derived SO 2 /BrO plume molar ratio is 1600 mol mol −1 (R 2 :0.84) equivalent to a S/Br mass ratio of 645.
Based on weekly SO 2 flux measurements of the Halema'uma'u plume (Elias et al., 2010), this suggests a BrO emission rate of ∼8 g s −1 .Note that the linear regression of the SO 2 versus BrO data does not go through the origin.We interpret this to suggest the presence of BrO in the CSR spectrum, possibly hinting at differential rates of chemical processing of SO 2 and BrO in the local atmosphere.Taking this into account as well as the likelihood that other reactive Br species are present in the plume, our Introduction

Conclusions References
Tables Figures

Back Close
Full Taking the mean measured BrO and SO 2 SCA and estimated optical path length through the plume of 600 m, the corresponding mean in-plume BrO and SO 2 mixing ratios were ∼2.3 and 2140 ppbv, respectively.
Nitrogen oxides play significant roles in reactive halogen chemistry -for instance, high levels of NO x increase production of BrONO 2 , which accelerates the bromine explosion, (Oppenheimer et al., 2006;Roberts et al., 2009).Elevated NO x levels (in excess of 100 ppbv) have also been associated with the active lavas at Kīlauea (Huebert et al., 1999;Witham, 2005), thus we co-retrieved NO 2 in the UV spectra with BrO to explore for possible effects.We found NO 2 strongly negatively correlated with both SO 2 (R 2 =0.87;Fig. 4a) and BrO (R 2 =0.84;Fig. 4b).As discussed in the next section, we believe that this striking finding reflects local depletion of NO x as a result of reactive bromine chemistry that acts to promote BrO formation.

Discussion
The first question to address is why BrO was detected at the Halema'uma'u vent but not at Pu'u' Ō' ō.The most likely explanation is that the optical density of the plume gas species was around 4 times higher in the case of the observations at the summit crater, thus making it easier to detect associated levels of BrO.Given that previous studies have not revealed any evidence for significant fractionation of halogens between the summit and rift zone magmatic degassing (see data compilation in Symonds et al., 1994), we consider that BrO is very likely present in the (daytime) Pu'u' Ō' ō plume but not in sufficient abundance to detect readily with our instrumental set up (consider that the maximum SCA of SO 2 detected at Pu'u' Ō' ō was 3×10 18 molec cm −2 (Table 1) and consult Fig. 3, and it is readily apparent that conclusive detection of BrO at Pu'u' Ō' ō would have been challenging for our spectrometers (the detection limit of BrO is ∼3×10 14 molec cm −2 ).Introduction

Conclusions References
Tables Figures

Back Close
Full Screen / Esc Printer-friendly Version

Interactive Discussion
Other factors can contribute to the detection of reactive halogen species, including sensitivity and selectivity of the spectrometer and the spectroscopy, as well as diurnal, seasonal and meteorological factors (e.g., RH, temperature, actinic flux, and cloudiness, which can affect the quality of the CSR; see Bobrowksi and Platt, 2007;Kern et al., 2008;Von Glasow et al., 2009).In addition, the section of the plume transected (margins versus core) and its young age (i.e., distance from source and wind speed less than 2 min) can significantly affect the observed SO 2 /BrO ratio (Bobrowski et al., 2007b(Bobrowski et al., , 2009;;Oppenheimer et al., 2006;Vogel et al., 2008), as well as the overall source flux, which affects plume-air mixing (Bani et al., 2009).Our observations pertain to plume ages of ∼2-8 min during which time the Br explosion is underway; thus, observed SO 2 /BrO ratios may be quite sensitive to the exact plume age (at Mt.Etna, Oppenheimer et al., 2006, observed BrO formation within 2-3 min of emission from source).Variability in air entrainment into the plume likely accounts for some of the observed scatter in the plot of SO 2 versus BrO seen in Fig. 3.
Note that the plumes at both Summit and East Rift were occasionally ashy and dense and the admixture of water droplets and aerosols resulted in high plume opacity.
If we assume that the measured SO 2 /BrO ratio characterises the near source emission for eruptive vents on Kīlauea, then we can take the long-term SO 2 flux observed for the volcano (0.62 Tg yr −1 ; Sutton et al., 2001;Elias andSutton, 2002, 2007) and the S/Br mass ratio to calculate a nominal minimum production of ∼15 g s −1 of reactive Br (equivalent to ∼480 Mg yr −1 ).This is within the range of estimates of other volcanoes, including Soufri ère Hills Volcano and Mt.Etna (Bobrowski et al., 2003;Oppenheimer et al., 2006).Gerlach (2004) predicted that although non-arc basaltic magmas at 1100-1200 A further key issue to address regarding the observations is the anticorrelation between BrO (and SO 2 ) and NO 2 abundance in the retrievals of the differential spectra.The pattern strongly suggests a deficit of NO 2 within the plume compared to an elevated background.The mean "missing" NO 2 in the plume is ∼6×10 16 molec cm 2 (Table 1).This is equivalent to a mixing ratio in the ambient atmosphere at Kīlauea summit of around 50 ppbv.This is well in excess of the typical NO x mixing ratio in the Hawaiian free troposphere (e.g., <1 ppbv measured at the Mauna Loa Observatory; Carroll et al., 1992) but NO x levels of 10 s to a few 100 s of ppbv have been reported in the immediate vicinity of Kīlauea associated with thermal fixation of atmospheric nitrogen at the high temperature surfaces of active lava (Hubert et al., 1999).Witham (2005) reported NO 2 mixing ratios measured along the Chain of Craters Road in the Hawaii National Park that average ∼150 ppbv; even background sites had NO 2 levels of around 24 ppbv.If these observations are indicative then there appears to be plenty of NO x in the local atmosphere at Kīlauea to account for our observed deficit in the young volcanic plumes.
Given the parallel observations of BrO in the plume, we speculate that the differential spectra indicating NO 2 loss in the plume reflect rapid titration of NO 2 out of entrained ambient air via the following chemical schemes (Oppenheimer et al., 2006;Roberts et al., 2009): NO 2 and BrO combine to form BrONO 2 which reacts with HBr in aerosol via a multistage process to form HNO 3 and Br 2 .Photolysis of Br 2 (once released to the gas phase) generates Br radicals that react with ozone to form BrO, with an overall doubling of BrO in this autocatalytic cycle (see Roberts et al., 2009).In this way, the presence of Introduction

Conclusions References
Tables Figures

Back Close
Full NO x promotes the bromine explosion and may represent a relevant process enhancing reactive bromine production at Kīlauea and other volcanoes where high temperature lava surfaces or high temperature gas jets generate NO x (Huebert et al., 1999;Mather et al., 2004;Oppenheimer et al., 2005;Martin et al., 2006Martin et al., , 2009)).Note that dissociation, or photolysis of BrONO 2 can also occur (regenerating BrO and NO 2 , or Br and NO 3 ), but that the reaction of BrONO 2 on aerosol is sufficiently rapid under the acidic, high-aerosol loading of volcanic plumes to promote the bromine explosion.The presence of NO x also alters the distribution of reactive bromine species between different forms, for example increasing BrONO 2 at the expense of BrO.Thus, despite promoting the overall bromine explosion, volcanogenic NO x can reduce BrO abundance making it harder to detect by UV spectroscopy.
Because of the path-integrated nature of the DOAS measurements it is difficult to confirm this hypothesis with the data available to us but the evidence for complex chemical pathways involving halogens, NO x , NO y and ozone in plumes is mounting (e.g., Oppenheimer et al., 2010).Careful in situ measurements, combined with model studies, are required to resolve the details.Model simulations (Roberts et al., 2009) of volcanic plume halogen chemistry have predicted in-plume NO 2 depletion and formation of nitric acid (HNO 3 ) via the BrONO 2 reaction scheme, supporting our observations.The combined observations and model simulations thus provide an explanation for elevated concentrations of HNO 3 that were reported at other volcanoes, including Villarrica, Masaya, Etna L áscar and Erebus (Mather et al., 2004;Oppenheimer et al., 2010), and cannot be explained by high-temperature chemistry alone (Martin et al., 2006).These findings highlight the importance of NO x for BrO-chemistry in volcanic plumes, and identify NO x to HNO 3 conversion as a consequence of volcanic plume halogen chemistry, in addition to known impacts such as ozone depletion.Introduction

Conclusions References
Tables Figures

Back Close
Full Screen / Esc Printer-friendly Version Interactive Discussion

Concluding remarks
In 2007 and 2008, we carried out UV spectroscopy measurements of the volcanic plume emitted by Pu'u' Ō' ō crater and by the new magmatic vent that opened in March 2008 within the Kīlauea Summit caldera.While BrO was not conclusively detected at Pu'u' Ō' ō, we observed BrO slant column amounts of 10 14 to 10 15 molec cm −2 and BrO/SO 2 molar ratio of 1600 mol mol −1 for the Summit plume.We observed apparent depletion of NO 2 (∼6×10 16 molec cm −2 ), equivalent to ∼50 ppbv of NO 2 loss within the plume, with amounts strongly correlated with SO 2 and BrO abundance suggesting an active role of NO x in the plume chemistry.Specifically, the negative NO 2 amounts might result from its reaction with BrO, enhancing the "bromine explosion" and converting NO x into nitric acid (which has also been detected close to source; Witham, 2005).
Based on a mean optical path through the plume of 600 m, we estimate a characteristic BrO mixing ratio of ∼2.3 ppbv, roughly double that observed in other tropospheric volcanic plumes.Based on the observed BrO/SO 2 ratio and the long term SO 2 emission rate from Kīlauea, we estimate that the volcano emits on the order of 480 Mg yr −1 of reactive bromine, suggesting that it might play a significant role in local ozone depletion and NO x to HNO 3 conversion.The observation of BrO in emissions from a relatively halogen-poor magma confirms predictions for Kīlauea based on thermodynamic arguments (Gerlach, 2004) and extends the database on the global volcanogenic reactive bromine source to include a hotspot volcano for the first time.
Screen / Esc Printer-friendly Version Interactive Discussion estimated BrO emission rate for the summit plume at Kīlauea underestimates the total reactive bromine flux.

Fig. 2 .
Fig. 2.Example of fitting process for a spectrum with high volcanic gas content.The spectrum was collected in the plume emitted from the new magmatic vent in Halema'uma'u crater on 13 May 2008.The measured differential optical density (grey curves) is fitted by the Ring, NO 2 , O 3 , and BrO differential laboratory spectra (black curves).The residual after fitting shows random modulation, validating the quality of the fit.Systematic structures at ∼345.5 and ∼348.0 nm are due to instrumental noise (hot pixels).A second order polynomial was applied to subtract low frequency contributions to the atmospheric attenuation due to Rayleigh and Mie scattering.The high differential ozone SCA results from the difference in the optical paths through the whole atmosphere of the CSR compared with the throughthe-plume spectra.

Figure 4 .
Figure 4. Scatter plots showing SCA of (a) SO 2 versus NO 2 , and (b) NO 2 versus BrO.The retrieved NO 2 SCA at Halema`uma`u, is strongly anti-correlated with both SO 2 and BrO (R 2 = 0.87 and 0.84, respectively).This close relationship between the three species, suggests active involvement of NO x in reactive bromine formation.The solid lines indicate the linear regression fitted to the data, and

Figure 4 .
Figure 4. Scatter plots showing SCA of (a) SO 2 versus NO 2 , and (b) NO 2 versus BrO.The retrieved NO 2 SCA at Halema`uma`u, is strongly anti-correlated with both SO 2 and BrO (R 2 = 0.87 and 0.84, respectively).This close relationship between the three species, suggests active involvement of NO x in reactive bromine formation.The solid lines indicate the linear regression fitted to the data, and the dotted and dashed lines represent prediction and 95% confidence limits of the regression,

Fig. 4 .
Fig. 4. Scatter plots showing SCA of (a) SO 2 versus NO 2 , and (b) NO 2 versus BrO.The retrieved NO 2 SCA at Halema'uma'u, is strongly anti-correlated with both SO 2 and BrO (R 2 =0.87 and 0.84, respectively).This close relationship between the three species, suggests active involvement of NO x in reactive bromine formation.The solid lines indicate the linear regression fitted to the data, and the dotted and dashed lines represent prediction and 95% confidence limits of the regression, respectively.

2 Observation sites and methods
These were obtained by ground-based UV spectroscopic observations carried out in 2007 and 2008 for plumes emitted by both the Pu'u' Ō' ō crater (on the East Rift Zone) and by a new magmatic vent that opened in March 2008 within Halema'uma'u crater.We discuss the reactive bromine observations in light of striking results concerning the co-retrieved abundances of NO 2 .

Table 1 .
GGS thanks the Dipartimento di Protezione Civile della Regione Sicilia for support via the project "Sviluppo di sistemi di monitoraggio".Finally, both GGS and CO acknowledge additional support from the NOVAC (Network for Observation of Volcanic and Atmospheric Change) project of the EU Sixth Framework Programme.Figures Summary of BrO, SO 2 , and NO 2 retrievals for spectra collected at Pu'u' Ō' ō and Halema'uma'u on 3 September 2007 and 13 May 2008, respectively.Geographic location, timing of data collection, and environmental and chemical parameters are given.See Fig.1for location map.