Release of gas-phase halogens by photolytic generation of OH in frozen halide-nitrate solutions: An active halogen formation mechanism?

Article abstract of DOI:10.1021/jp102072t

To better define the mechanisms by which condensed-phase halides may be oxidized to form gas-phase halogens under polar conditions, experiments have been conducted whereby frozen solutions containing chloride (1 M), bromide (1.6 × 10^-3 to 5 × 10^-2 M), iodide (<1 × 10^-5 M), and nitrate (0.01 to 1 M) have been illuminated by ultraviolet light in a continually flushed cell. Gas-phase products are quantified using chemical ionization mass spectrometry, and experiments were conducted at both 248 and 263 K. Br₂ was the dominant product, along with smaller yields of IBr and trace BrCl and I₂. The Br₂ yields were largely independent of the Br^-Cl^- ratio of the frozen solution, down to seawater composition. However, the yields of halogens were strongly dependent on the levels of NO₃^- and acidity in solution, consistent with a mechanism whereby NO₃^- photolysis yields OH that oxidizes the condensed-phase halides. In support, we observed the formation of gas-phase NO₂, formed simultaneously with OH. Gas-phase HONO was also observed, suggesting that halide oxidation by HONO in the condensed phase may also occur to some degree. By measuring the production rate of condensed-phase OH, using benzoic acid as a radical trap, we determine that the molar yield of Br₂ formation relative to OH generation is 0.6, consistent with each OH being involved in halide oxidation. These studies suggest that gas-phase halogen formation should occur simultaneously with NO_x release from frozen sea ice and snow surfaces that contain sufficient halides and deposited nitrate.

Full text of DOI:10.1021/jp102072t

J. Phys. Chem. A 2010, 114, 6527–6533
6527
Release of Gas-Phase Halogens by Photolytic Generation of OH in Frozen Halide-Nitrate
Solutions: An Active Halogen Formation Mechanism?
J. Abbatt,*,† N. Oldridge,† A. Symington,‡ V. Chukalovskiy,† R.D. McWhinney,† S. Sjostedt,†
and R.A. Cox‡
Department of Chemistry, UniVersity of Toronto, 80 St. George St., Toronto, ON, M5S 3H6, and Department of
Chemistry, UniVersity of Cambridge, Lensfield Road, Cambridge, U.K., CB2 1EW
ReceiVed: March 8, 2010; ReVised Manuscript ReceiVed: May 5, 2010
To better define the mechanisms by which condensed-phase halides may be oxidized to form gas-phase halogens
under polar conditions, experiments have been conducted whereby frozen solutions containing chloride (1
M), bromide (1.6 × 10-3 to 5 × 10-2 M), iodide (<1 × 10-5 M), and nitrate (0.01 to 1 M) have been
illuminated by ultraviolet light in a continually flushed cell. Gas-phase products are quantified using chemical
ionization mass spectrometry, and experiments were conducted at both 248 and 263 K. Br2 was the dominant
product, along with smaller yields of IBr and trace BrCl and I2. The Br2 yields were largely independent of
the Br-/Cl- ratio of the frozen solution, down to seawater composition. However, the yields of halogens
were strongly dependent on the levels of NO3- and acidity in solution, consistent with a mechanism whereby
NO3- photolysis yields OH that oxidizes the condensed-phase halides. In support, we observed the formation
of gas-phase NO2, formed simultaneously with OH. Gas-phase HONO was also observed, suggesting that
halide oxidation by HONO in the condensed phase may also occur to some degree. By measuring the production
rate of condensed-phase OH, using benzoic acid as a radical trap, we determine that the molar yield of Br2
formation relative to OH generation is 0.6, consistent with each OH being involved in halide oxidation.
These studies suggest that gas-phase halogen formation should occur simultaneously with NOx release from
frozen sea ice and snow surfaces that contain sufficient halides and deposited nitrate.
NO-3 + hν f NO-2 + O
(2)
1. Introduction
It is now well recognized that the snowpack is a chemically
complex environment, consisting of a wide range of organic
species, either formed biogenically or deposited from the
atmosphere, and inorganic species, such as deposited nitrate and
halides. So, the discoveries that have been made of late that
indicate that this environment can be highly chemically active
are not too surprising.1 In particular, it is now known that the
snowpack can also act as a source of small volatile organic
compounds (VOCs) to the atmosphere,2 especially with the
increased illumination and higher temperatures associated with
daytime, and that photolytic NOx release occurs in the form of
both NO2 and HONO.3,4
The branching ratio between reactions 1 and 2 favors the
former channel, by which OH is generated.6,7 Anastasio and
co-workers have confirmed that in frozen solution substrates,
OH is indeed formed in a manner similar to that which prevails
in water.7 Also, NO2 and HONO have been observed from these
photoreactions occurring in ice.8,9 The HONO may form either
-
from protonation of NO2 or by secondary chemistry arising
from NO2.10
Once OH is generated, it will react with chemically reduced
species, such as halides and organics. In the case of halides, it
was shown many decades ago that the ultimate fate of
photolytically generated OH radicals in seawater is to prefer-
entially oxidize Br- over Cl-, even though the concentration
of Br- is close to three orders magnitude lower than that of
Cl-.11-13 Frinak and Abbatt confirmed that Br2 is indeed released
from seawater solutions when exposed to gas-phase OH radicals
that interact heterogeneously with the solution substrates.14 These
results were in line with those of Oum et al. who inferred that
Cl2 is formed when OH interacts with aqueous particles
containing NaCl.15 Finally, George and Anastasio demonstrated
that a gas-phase species containing bromine was formed when
nitrate photolysis proceeds in a bromide-containing aqueous
solution.16
The goal of this work is to examine whether the snowpack
also has the potential to be a source of gas-phase active halogens,
through photolytic processes. Given known chemistry, it is likely
that halide oxidation will occur in ice environments where both
nitrate and halides are present. In particular, photolysis of nitrate
is a source of condensed phase OH radicals, through reaction
1, where the O- species that is formed will readily abstract a
proton from either water or an acid:5
Putting these findings together, it is reasonable to hypothesize
that Br2 production will proceed if nitrate photolysis occurs in
frozen aqueous media. However, experimental confirmation is
necessary to show the mechanism is viable. For example, can
OH oxidation of halides occur sufficiently fast to give high
halogen yields, compared to potential radical-radical recom-
NO-3 + hν f NO2 + O-
(1)
* To whom correspondence should be addressed.
† University of Toronto.
‡ University of Cambridge.
10.1021/jp102072t  2010 American Chemical Society
Published on Web 05/19/2010

6528 J. Phys. Chem. A, Vol. 114, No. 23, 2010
Abbatt et al.
bination steps, such as the reverse of reaction 1 or the reaction
of OH with NO2? Are the halogens released from the ice
substrates efficiently, or are they subject to chemical loss,
perhaps via reaction with condensed-phase OH? Which halogens
are formed, and at what relative rates? Our first attempt to
address these general issues was presented in a paper where
we demonstrated the gas-phase production of halogens, prin-
cipally Br2 with smaller yields of BrCl and IBr, when gas-phase
OH radicals are exposed to frozen halide solutions and humidi-
fied salts.17 However, to our knowledge, this is the first attempt
to demonstrate that halogen production proceeds photolytically,
from frozen halide solutions that contain an OH precursor. We
also consider the possibility that halogen production may
proceed via formation of HONO, generated in Reaction 2. In
particular, HONO is known to oxidize bromide in acidic ice
substrates, giving rise to BrNO formation (although Br2 forma-
tion has not yet been observed as a product).18,19
Figure 1. Photolysis Cell. Products that enter the gas phase are swept
away to the CIMS when valves 1 + 3 are open.
Although now widely recognized, the overall motivation for
understanding the formation of active halogens, both bromine
containing but also chlorine and iodine as well, is that high levels
of bromine have been observed to be associated with depletion
of ozone and mercury at polar latitudes in the springtime.20,21
The mechanisms for loss of such species under an environment
of high gas-phase active bromine loading are relatively well
understood, but the source of the bromine remains unquantified
two decades after the initial ozone depletion events were
observed. In particular, a “bromine explosion” driven by the
autocatalytic release of Br2 after HOBr uptake by the snowpack
is likely important at high active bromine levels,22 but the
bromine activation process at low bromine loadings and the
process that drives initial bromine release remain poorly
quantified. Without knowledge of the most important bromine
release initiation mechanism(s), our models do not yet have
predictive abilities for these ozone and mercury depletion events.
So, to better understand the formation of halogens under polar
conditions, the experiments described below were designed with
three specific goals in mind. First, are gas-phase halogens formed
from frozen solutions containing halides and an OH photolytic
precursor? Second, how does the production rate of these
halogens vary with experimental parameters such as temperature,
halide concentration, nitrate concentration, and acidity? What
chemical mechanism is consistent with these findings? And,
finally, what is the yield of halogens relative to the OH
production rate? That is, what fraction of OH radicals formed
go on to oxidize halides? If halogens are formed in stoichio-
metric amounts exceeding the production rate of OH, does this
imply a role for HONO as an oxidant?
Figure 2. Simplified schematic of the overall experimental apparatus.
through the cell, that is, reaction mode. Each run starts in bypass
mode while the ice sample is being prepared, and the pressure
in the reaction cell is the same as the ambient pressure. Once
the cell is opened in reaction mode, the pressure inside the
reaction cell decreases to the operating conditions of the
experiment, 90 Torr.
Light from a 1000 W xenon arc lamp was passed through a
water filter to remove infrared, through an iris to control beam
width, and then reflected off a concave spherical mirror before
being focused onto the ice surface. Any photochemical reactions
would occur in the ice at or below the surface where light shone,
which covered a circular area approximately 1.5 cm wide. We
assume the area of ice illuminated is approximately 2 cm2. The
spectrum of such lamps is not dissimilar to that of the sun, but
it does extend to below 300 nm, the threshold for ground level
actinic radiation. However, we used an unfiltered lamp for these
experiments to maximize the photolysis rate of nitrate, which
absorbs strongly below 300 nm as well. Because nitrate
photolysis only acts as the OH source for this work, we feel
that using UV below 300 nm does not affect the environmental
relevance of the results that are presented.
2. Experimental Section
The reaction cell has been described previously9 and is
illustrated in Figure 1. It is a cylindrical glass vessel, 8 cm high
with a radius of 5 cm. It is surrounded by another glass vessel,
similar in shape but larger, through which coolant flows. There
are three openings in the reaction cell: a transparent, fused-
silica cover to allow for the addition of sample and ultraviolet
light, an inlet for gas to be flowed in, and an outlet to allow
gases to escape. These two openings allow gaseous nitrogen to
pass through the cell, from the bottom to the top, carrying with
it any gases that may evolve from the ice.
There are three on/off valves to control the flow of nitrogen
in and around the cell. When valve 2 (Figure 1) is open and
valves 1 + 3 are closed, nitrogen bypasses the reaction cell
entirely. This is referred to as bypass mode. In the opposite
scenario, 1 + 3 are open and 2 is closed, and nitrogen is routed
The overall experimental apparatus is shown in Figure 2. The
outflow of nitrogen gas (190 sccm) containing the products is
sent to an ion-molecule reaction flow tube of a chemical
ionization mass spectrometer (CIMS), where the products mix
with an ionized gas. A dilute mixture of SF6 in N2 (8.0 L/min)
is sent through a Po-210 static charge eliminator (NRD LLC,
Grand Island, NY, USA). Here, SF6 is ionized by electron
-
attachment to SF6-. SF6 was chosen as the ionizing reagent
within the ion-molecule flow tube because of its ability to
ionize many different gases, including NO2, Br2, and other

Gas-Phase Halogens by Photolytic Generation of OH
J. Phys. Chem. A, Vol. 114, No. 23, 2010 6529
-
halogens.23 The interaction between SF6 and gases in the
molecules/cm2/s. Note that by expressing the fluxes per unit
surface area we do not mean to imply that heterogeneous
chemistry is driving the release; it is simply a convenient manner
by which to express the results.
ion-molecule flow tube generates product ions that pass to the
ion optics region of a mass spectrometer system. A single linear
quadrupole is set to selectively pass particular ions, which are
detected with a Channeltron electron multiplier. The mass
spectrometer has been described previously.24
Glass bulbs of trace Br2 (from liquid Br2, min 99.5%, ACP
chemicals) in N2 and NO2 (from liquid N2O4, Specialty Gases
of America, Toledo, OH) in N2 were prepared in known mixing
ratios. They were used for quantitative calibration of the CIMS
by addition of known flows under the operating conditions of
the experiment.
Solutions were prepared by dissolving NaCl (ACP Chemicals,
99.0%), NaBr (Sigma-Aldrich, 99.0%), and NaNO3 (Fisher
Scientific, 99.2%) in filtered water (resistance >18 MΩ). The
NaCl concentration was 1.0 M for most experiments, NaBr
concentration varied from 1.6 to 50 mM, and NaNO3 varied in
concentration from 10 mM to 1 M. The NaCl contained an
iodide impurity, < 1 × 10-5 M NaI. Acidity of the solutions
was controlled by addition of dilute sulfuric acid (diluted from
96% assay, Fisher Scientific).
With the reaction cell in bypass mode, 7 mL of solution were
added to the cell. Solutions at 263 K would remain liquid until
the cell was changed to reaction mode. At this point, the pressure
inside the cell would decrease from ambient pressure to 90 Torr,
and evaporative cooling caused the solution to freeze from top
to bottom over a time of 1-3 s.
These calibrations allowed us to determine our detection
limits, calculated for S/N ) 1 where the noise is the standard
deviation of our background signal, that is, this is the amount
of signal that we assume we can distinguish from background
noise. Depending on the operating conditions (such as the age
of the radioactive source), our detection limit for NO2 varied
from 2 × 1011 to 2 × 1012 molecules/s, and for Br2, from 7 ×
109 to 8 × 1010 molecules/s.
Calibrations were not performed for BrCl, IBr, and I2. We
assume that the CIMS is equally sensitive to each of these as it
is to Br2, given that their gas-phase electron affinities are so
similar to each other and to Br2.25 That is, we are assuming
that each of these dihalogens has the same collision-controlled
-
ion-molecule rate constant with SF6 as Br2. Calibrations of
HONO were not performed, and so we do not report flux data
for this photoproduct. In addition, as described above, we are
not fully confident that mass spectral intensity at m/z 66 is all
due to HONO.
We believe that the variability of the results, which is typically
a factor of 2 but occasionally higher, is larger than systematic
uncertainties arising from the calibrations (which are on the
order of (20% or so). The experimental variability was hard
to pin down specifically, but we believe it is probably due the
nature of the freezing process and the associated effects on the
flux of light in the ice matrix, for example, via multiple
scattering. It may additionally be due to the variability associated
with aligning the light source into the cell for each experiment
and the variations in the lamp output during the few month
duration of the experiments.
For the quantification of hydroxyl radical generation, we used
the benzoic acid radical trap method as described by Chu and
Anastasio.7 Sodium benzoate (>99.5%, Fluka) is added to the
reaction vessel at a concentration of 10 mM. Using a simple
kinetics model of the aqueous phase experimental conditions,14
it can be shown that at this concentration OH will react
quantitatively with benzoic acid, producing p-hydroxybenzoic
acid. The yield is 0.063 ( 0.014.7 The p-hydroxybenzoic acid
product was quantified with an HPLC fitted with a 100 µL
injection loop, a CSC-inertsil 150A C-18 column (10 × 0.46
cm, 5 µm packing), and a UV-absorbance detector, monitoring
at 256 nm. The separation was accomplished by three-step
gradient elution at 1 mL/min flow rate during the first step and
1.8 mL/min flow during the second and the third steps. The
mobile phase consisted of 13% acetonitrile and 87% of water
acidified with 0.13% trifluoroacetic acid during the first 3.5 min
step. The mobile phase contained 35% ACN and 65% acidified
water during the second 3 min step. The mobile phase was
changed back to 13% ACN and 87% acidified water during the
third 5 min step. The gradient elution method gave clearly
resolved peaks of benzoic, p-hydroxybenzoic, and m-hydroxy-
benzoic acids. Calibrations of p-hydroxybenzoic acid (>99%,
Fluka) were performed daily.
Signals from the CIMS were allowed to stabilize, at which
time the photolysis lamp was turned on. The mirror and cell
position were adjusted manually until the illuminated spot on
the ice was as bright as possible and about 2 cm2 in area. This
is the area used in calculations of gas-phase product flux. After
the product signals had remained stable for some time, the lamp
was turned off and the evolution of the product signals was
monitored for a short time.
Up to 10 different masses were monitored, for 0.5 s each.
Thus, each ion was measured at a time resolution of no more
than 5 s. However, typically, fewer species were monitored
because of lack of signal at all mass-to-charge ratios. Ions were
-
monitored at m/z ) 146 (SF6 ), 124 (SF4O-, from reaction with
H2O), 46 (NO2-, from reaction with NO2), 62 (NO3-, probably
-
from reaction with HNO3), 160 (Br2 from reaction with Br2),
206 (IBr- from reaction with IBr), 254 (I2- from reaction with
-
I2), 116 (BrCl- from reaction with BrCl), and 70 (Cl2 from
reaction with Cl2). We also monitored m/z 66, which we attribute
to FHNO2 . This ion is known to arise from SF6- reacting with
-
HONO. However, there is also the potential for HF, formed by
SF6- reacting with H2O, to react with NO2- to form this cluster.
Calibrations were performed by metering a constant flux of
Br2 or NO2 from the gas manifold, of fixed volume and
temperature, connected to the Br2 and NO2 reservoirs. The
pressure drop per unit time from the manifold was converted
to the number of Br2 or NO2 molecules released per unit time.
This gas release generated a CIMS signal in counts per second
(cps) on top of the background signal, so we could determine
a sensitivity in units of cps per molecule/s. In all cases, we
-
normalized the observed signals and sensitivities to the SF6
3. Results and Discussion
reagent ion signal prevalent during the experiment. For each
experiment, the signal intensities that were stable with time were
averaged to calculate the signal generated while the lamp was
on. Using the calibrated sensitivity, we could then determine
the flux of molecules per second that evolved from the ice.
Dividing by the area that is illuminated (approximately 2 cm2)
gives the rate of halogen evolution from the ice, in units of
3.1. Initial Observations. Typical results are shown in Figure
3 for selected ions. In particular, significant production of IBr,
Br2, NO2, and potentially HONO is observed. There is a small
change in the reagent ion signal with the products being formed,
but we account for this, as mentioned above, by referencing all
signals to the intensity of the reagent ion, when fluxes were

6530 J. Phys. Chem. A, Vol. 114, No. 23, 2010
Abbatt et al.
Figure 3. Typical output from a photolysis experiment. The system is allowed to equilibrate until 100 min, at which point the lamp is turned on.
Signal intensities of Br2, NO2, IBr, and probably HONO all increase upon illumination, suggesting that they are products or byproduct of nitrate
photolysis. This output is from a frozen solution of 1 M NaCl, 1 M NaNO3, 0.05 M NaBr, pH 2.5, at 248 K.
3.2. Effect of Nitrate Concentration on Br2 Production.
To help evaluate the mechanism for the halogen release,
solutions of 1 M NaCl, 0.05 M NaBr, and pH 2.5 were prepared
with varying amounts of nitrate. At 248 K, solutions containing
more nitrate produced more Br2 (Figure 5), with the effect
saturating at high nitrate concentrations. In early experiments
at 263 K (data not shown), the same general trend of saturation
was also apparent.
The saturation effect in the Br2 production rate with increasing
nitrate concentrations is understood from the optical thickness
of the solutions in the ultraviolet. For the most concentrated
solutions (i.e., 1.0 M), essentially all the light (for example,
between 280 and 300 nm) is absorbed within the 0.4 cm depth
of the frozen ice substrate, as calculated using the Beer-Lambert
Law and the absorption coefficients from Chu and Anastasio.7
Thus, increasing the concentration of nitrate further will not
lead to additional photolysis in solution. However, for concen-
Figure 4. Br2 signal from a frozen solution of 1.0 M NaCl, 0.05 M
NaBr, pH ) 2.5. The highlighted time, from 37 to 51 min, indicates
when the lamp was on.
trations of 0.1 M and lower, the light attenuation is ap-
proximately linear as a function of depth, that is, not all the
light is getting absorbed. In the nonsaturated regime, the
increasing fluxes of Br2 will be arising through increased rates
of nitrate photolysis, and so also OH production. We note that
calculated. By contrast, no Br2 was seen to evolve from solutions
the shape of the bromine production curve as a function of
that did not contain nitrate (Figure 4). Note that there were small
nitrate concentration resembles the analogous curve for the
and variable backgrounds at some mass-to-charge ratios, notably
those for Br2, probably arising from the stickiness of this
molecule on the surfaces of sampling lines.
aqueous reaction.16
3.3. Effect of Bromide Concentration on Br2 Production.
With other variables held constant, varying the concentration
It is also worth noting that while Br2 and IBr are the dominant
of bromide by over a factor of 30 in the ice did not affect Br2
halogen products, I2 and BrCl were also detected in some
fluxes (Figure 6), within the limit of our uncertainties. Note
experiments, albeit at levels close to our detection limit. Fluxes
of all species from the ice are documented in Table 1.
that the lowest concentration in Figure 6 is 1.6 mM, just twice
the level of bromide present in seawater. This observation is
In Figure 3, it is seen that after the lamp was turned off, each
consistent with efficient scavenging of photolytically generated
product ion signal decreased rapidly but not to background. We
OH by bromide, that is, as demonstrated for seawater by
Zafiriou13 and by George and Anastasio.16 In our case, where
freezing will be concentrating the halide salts into brine pockets
or films, it is even more likely that the levels of bromide will
be sufficiently high that it becomes the prime scavenger for the
OH radical.
interpret this behavior as due to slow diffusion of photoproducts
from the ice/brine mixture. The time scale for the release of
these trapped products is more consistent with diffusion from
an aqueous substrate than from a pure ice matrix, with very
low diffusion coefficients.
Our observation of NO2 production is qualitatively cor-
roborated by previously published laboratory studies showing
that illuminated frozen nitrate solutions produce NO2.8,9
3.4. Effect of Temperature on Halogen Release. Br2 and
NOx production both show a clear temperature dependence. For
otherwise identical conditions, solutions frozen to 263 K

Gas-Phase Halogens by Photolytic Generation of OH
J. Phys. Chem. A, Vol. 114, No. 23, 2010 6531
TABLE 1: Experimental Data
fluxes, molecules/cm2/s/1011
experiment ID No.
[NaCl] (mol/L)
[NaBr] (mol/L)
[NaNO3] (mol/L)
pH
T (°C)
NO2
Br2
IBr
I2
BrCl
0.06
1
2
3
4
5
6
7
8
1.0
1.0
1.0
1.0
1.0
1.0
1.0
1.0
1.0
1.0
1.0
1.0
1.0
1.0
1.0
1.0
0.05
0.05
0.05
0.05
0.05
0.05
0.05
0.05
0.05
0.01
0.0016
0.0016
0.05
0.05
0.01
0.01
1.0
1.0
0.5
0.5
0.5
0.1
0.1
0.01
0.01
0.01
0.01
0.01
0.1
0.1
0.1
0.1
2.5
2.5
2.5
2.5
2.5
2.5
2.5
2.5
2.5
2.5
2.5
2.5
2.5
2.5
2.5
2.5
-25
-25
-25
-25
-25
-25
-25
-25
-25
-25
-25
-25
-10
-10
-10
-10
378
414
103
78
342
149
196
10
11
37
22
11
17.9
15.8
6.2
11.8
29.5
12.1
17.6
5.7
3.8
2.7
3.6
2.5
3.8
3.8
1.4
0.7
2.9
0.44
0.22
0.19
0.06
0.8
0.1
0.1
9
10
11
12
13
14
15
16
0.1
0.1
11.1
7.8
2.9
3.5
869
358
537
654
70.3
75.1
26.2
24.0
0.06
0.14
0.08
0.07
produced about 4 times more NO2 than solutions frozen at 248
K. Bromine production increased by about a factor of 5 over
the same temperature range. For example, for 1 M NaCl, 0.05
M NaBr, 0.1 M NaNO3, pH 2.5, the average Br2 fluxes were
1.5 × 1012 molecules/cm2/s at 248 K and 7.3 × 1012 molecules/
cm2/s at 263 K. IBr production also exhibited a strong
temperature dependence, but the signals were sufficiently close
to detection limit to inhibit defining the increase accurately.
Although higher temperatures may increase rate constants
between OH and a dissolved halide, the oxidation reactions are
already fast, and the change in the kinetics due to the small
temperature differences are likely to be small. Instead, other
factors probably control why higher temperatures may increase
the flux of NOx and halogens from the ice. Most importantly,
the quantum yield of OH from nitrate photolysis decreases with
temperature. Chu and Anastasio measured that the quantum yield
for OH production in ice decreases by a factor of 3 with a 29
K decrease in temperature.7 This decrease is similar to the
change in fluxes that we observe over a 15 K temperature
change.
Also, the nature of the frozen solution is changing in character
as temperature is lowered. Koop et al. showed that at 250 K,
some sodium chloride crystallizes out of seawater solutions as
NaCl·2H2O.26 The eutectic point of sodium nitrate is 255 K,
and it stands to reason that a portion of nitrate would have
crystallized out of solution between 263 and 248 K as well.
This salt crystallization occurs concurrently with the formation
of ice. Overall, the ionic strength of the liquid component of
the frozen solution will increase as water crystallizes into a solid.
George and Anastasio found that bromine release decreased with
increases in the ionic strength of the liquid from which it
evolved.16 Thus, if these reactions are taking place in highly
concentrated liquid pockets within the ice, as is likely, it follows
that Br2 production will be affected.
3.5. Effect of Acidity on Br2 Production. Both halogen and
NOx production decreased with decreasing acidity so that, at
248 K, the halogen fluxes from frozen pH 7 substrates did not
give signals above detection limits. Instead, we report the acidity
dependence at 263 K where, in early experiments, pH 7 solutions
produced clearly measurable quantities of halogens. Although
the CIMS was not calibrated absolutely at this time, in a relative
sense these experiments showed that the Br2 yields were between
1 and 2 orders of magnitude smaller when pH was raised from
Figure 5. At 248 K, increased Br2 production is observed as
concentration of nitrate in the ice increases. 1.0 M NaCl, 0.05 M NaBr,
pH ) 2.5. The uncertainties in this plot and later plots are standard
deviations used to indicate the variance within the data set; with only
2-3 points, they do not have statistical validity.
Figure 6. At 248 K, Br2 production does not vary with concentration
of bromide in the ice. 1.0 M NaCl; 0.01 M NaNO3; pH ) 2.5.

6532 J. Phys. Chem. A, Vol. 114, No. 23, 2010
Abbatt et al.
2 to 7 for solutions of 1.0 M NaCl, 0.05 M NaBr composition,
with 0.1-1.0 M NaNO3.
We note that both Frinak and Abbatt,14 using a gas-phase
source of OH, and George and Anastasio,16 using a photolytic
source, also observed greater Br2 production from acidic
solutions. This acidity dependence is consistent with known
aqueous chemistry, as described in more detail by Frinak and
Abbatt and by George and Anastasio.14,16 Briefly, it is known
that the reaction between dissolved OH and Br- (kII ) 1.1 ×
1010 M-1 s-1 12 is faster than the reaction between OH and Cl-
)
(kII ) 4.3 × 109 M-1 s-1).27 However, since chloride ions
outnumber bromide ions by a large ratio, the reaction of OH
with chloride should dominate:
OH + Cl- T HOCl-
(3)
(4)
HOCl- + H+ f H2O + Cl
Figure 7. Total OH radical generation from photolysis as a function
of time. These values come from the measured concentration of
p-hydroxy benzoic acid produced in 7 mL solutions, and adjusted for
the yield of this product in OH reaction.7 Solutions contained 1.0 M
NaCl, 0.1 M NaNO3, 0.01 M sodium benzoate, pH ) 2.5, 263 K.
HOCl- + Cl- f Cl-2 + OH-
Cl- + Cl f Cl-2
(5)
(6)
in our experiments, so that we could relate it to the amount of
Br2 observed. Total OH production can be measured by adding
an in situ OH scavenger, such as benzoic acid, to solutions28 or
ices7 that do not contain other OH scavengers, such as bromide.
Benzoic acid is susceptible to reaction with hydroxyl radicals,
forming p-hydroxybenzoic acid among other products. After
exposing an ice containing nitrate and benzoic acid to light, we
used high performance liquid chromatography (HPLC) to
quantify the amount of p-hydroxybenzoic acid formed.
Chu and Anastasio have used this method to measure the
quantum yields of OH from nitrate photolysis in ice at 263 K
and found that the temperature dependence of the quantum yield
in ice can be modeled by the same regression line as the values
in aqueous solution.7 This suggests that the photolysis is
occurring in a liquid portion of the ice, rather than in the ice
bulk itself.
Cl-2 + Br- f BrCl- + Cl-
BrCl- + Br- f Cl- + Br-2
Br2- + Br2- T Br-3 + Br-
Br-3 T Br2 + Br-
(7)
(8)
(9)
(10)
Note that the competition between 4 and 5 gives rise to the
acidity dependence, with reaction 5 being quite slow. The net
reaction becomes:
In our case, we exposed solutions (1.0 M NaCl, 0.1 M
NaNO3, 10 mM sodium benzoate, pH ) 2.5, 263 K) to light in
the same manner as for the other experiments described in this
paper. No bromide was added to these solutions, since it would
compete with benzoic acid as an OH scavenger. We found that
OH is produced linearly over time up to 30 min (Figure 7), at
a rate of 2.7 nmol/min ) 2.7 × 1013 molecules/s. Chu and
Anastasio also found this relationship to be linear. For a 2 cm2
illuminated area, this represents an OH production rate of 1.35
× 1013 molecules/cm2/s. This is significantly less than our
observed NO2 production rates for this ice (4 to 9 × 1013
molecules/cm2/s) indicating that not all O- species generated
in Reaction 1 form OH. However, this OH production rate is
quite similar to our observed Br2 production rates (8 × 1012
molecules/cm2/s), with about 0.6 Br2 molecules formed for each
OH produced. The overall uncertainty in this ratio is quite high,
on the order of a factor of 2 or so. Note that the stoichiometry
for the overall reaction of OH with Br- is for two OH radicals
to form one Br2 molecule (Reaction 11). So, our results confirm
that every OH radical generated is indeed involved in halogen
activation.
2OH- + 2Br- + 2H+ f Br2 + 2H2O
(11)
Note that the weak acidity dependence in the quantum yield
for OH production, where OH yields actually increase slightly
in going from pH 2 to pH 7,7 is thought to be dominated by the
chemistry just described.
It is also theoretically possible for Cl2- to self-react and form
molecular chlorine:
Cl-2 + Cl-2 f Cl2 + 2Cl-
(12)
However, chlorine production is not observed in cases where
sufficient quantities of bromide are available. For example,
Frinak and Abbatt did not observe Cl2 production from exposure
of gas-phase OH with solutions where chloride ions outnum-
bered bromide ions by 3 × 105 to 1.14 Similarly, we see no
evidence for Cl2 formation in our experiments.
Finally, we note that if some component of the halogen
formation proceeds via HONO formation, then enhanced
formation at low acidity would be consistent with HONO (pKa
) 3.3) being preferentially formed in Reaction 2.
4. Atmospheric Implications and Conclusions
3.6. Gas-Phase Br2 Yields Relative to OH Production. We
were interested to quantify the amount of OH that was forming
Our overall conclusion is that frozen solutions containing both
halides and nitrate produce gas-phase halogens when illuminated

Gas-Phase Halogens by Photolytic Generation of OH
J. Phys. Chem. A, Vol. 114, No. 23, 2010 6533
in the ultraviolet. For solutions with chloride concentrations
much higher than bromide that, in turn, was much more
prevalent than iodide, we find that Br2 is the dominant halogen
formed, with smaller but significant yields of IBr. Although
observed in some experiments, the yields of BrCl and I2 were
very small, and Cl2 was never observed.
from HNO3 deposition or deposition of acidic sulfate particles.
Likewise, the nitrate concentrations in the polar regions are
orders of magnitude lower than those we used, that is, they are
probably on the order of µM.33,34 However, we believe the
chemical mechanism should not be dependent on the nitrate
concentration, given that it is acting only as the source of the
OH radical. That being said, an important next step would be
to examine whether real-world samples of ice or snow do yield
gas-phase halogens upon ultraviolet illumination.
Given that frozen halide solutions are known to contain
concentrated brines down to low temperatures,26 we believe that
the halide oxidation occurs in such solutions. It is also known
that a small amount of solutes are soluble in ice29 and so, while
there is no need to invoke chemistry occurring in or on the ice
surfaces, we can not rule it out either. However, the observations
are all consistent with an aqueous phase mechanism. In
particular, we believe that OH is formed initially by nitrate
photolysis, given that we observe NO2 produced simultaneously.
The dependence of the halogen fluxes upon nitrate and the lack
of halogen formation in the absence of nitrate are in agreement
with nitrate being the OH precursor. Indeed, observations by
Chu and Anastasio have suggested that the photolysis of nitrate
is occurring in the liquid portion of ice matrices.7 Other evidence
in support of an OH-mediated reaction mechanism comes from
our earlier studies, where we observed that OH can heteroge-
neously oxidize frozen and aqueous halide solutions, forming
gas-phase molecular halogens.14,17 Also, George and Anastasio
have shown that OH formation drives gas-phase bromine release,
by using both nitrate and H2O2 as photolytic OH precursors.
Although we can not rule out HONO oxidation of halides
playing a role, we believe that OH is more important given that
the quantum yield for OH formation from nitrate photolysis is
substantially higher than that for HONO generation.6,7 In
agreement with this is the stoichiometric agreement between
the OH and Br2 production rates, suggesting that, to within
experimental uncertainties, we do not need to invoke a HONO
pathway.
The OH radical goes on to preferentially oxidize bromide,
even when it is present at close to seawater compositions. We
find that every OH radical is involved in halide oxidation,
through the radical trap experiments. The acidity dependence
is in agreement with a mechanism whereby the OH radical first
reacts with a chloride ion, then driving chemistry that ultimately
leads to bromine formation (Reactions 3-10). We note that the
persistent observation of IBr, and the observation of I2 in some
experiments, suggests that OH also has the ability to oxidize I-
in frozen solutions. Much less IBr and I2 were observed than
Br2, which is likely attributable to the high abundance of chloride
and bromide relative to iodide, which is only present as an
impurity. IBr formation has also been observed for the
heterogeneous reaction of OH with frozen halide solutions and
crystallized salts.17 We note that this chemistry is to be expected,
given that the rate constant for reaction between aqueous OH
and I- is similar to that for OH and Br-,30 that is, oxidation of
both species by OH is highly efficient.
Acknowledgment. The authors acknowledge the financial
support of NSERC and CFCAS. We thank Jamie Donaldson
for helping with the initial optical setup and for lending the
photolysis cell used in this work, and we thank Scott Mabury
for loan of the Xe lamp.
References and Notes
(1) Domine, F.; Shepson, P. B. Science 2002, 297, 1506.
(2) Grannas, A. M.; Shepson, P. B.; Guimbaud, C.; Sumner, A. L.;
Albert, M.; Simpson, W.; Domine, F.; Boudries, H.; Bottenheim, J.; Beine,
H. J.; Honrath, R.; Zhou, X. L. Atmos. EnViron. 2002, 36, 2733.
(3) Honrath, R. E.; Lu, Y.; Peterson, M. C.; Dibb, J. E.; Arsenault,
M. A.; Cullen, N. J.; Steffen, K. Atmos. EnViron. 2002, 36, 2629.
(4) Beine, H.; Colussi, A. J.; Amoroso, A.; Esposito, G.; Montagnoli,
M.; Hoffmann, M. R. EnV. Res. Lett. 2008, 3, 045005.
(5) Zepp, R. G.; Hoigne, J.; Bader, H. EnViron. Sci. Technol. 1987,
21, 443.
(6) Dubowski, Y.; Colussi, A. J.; Boxe, C.; Hoffmann, M. R. J. Phys.
Chem. A 2002, 106, 6967.
(7) Chu, L.; Anastasio, C. J. Phys. Chem. A 2003, 107, 9594.
(8) Honrath, R. E.; Guo, S.; Peterson, M. C.; Dziobak, M. P.; Dibb,
J. E.; Arsenault, M. A. J. Geophys. Res., [Atmos.] 2000, 105, 24183.
(9) Bartels-Rausch, T.; Donaldson, D. J. Atmos. Chem. Phys. Discuss.
2006, 6, 10713.
(10) Hellebust, S.; Roddis, T.; Sodeau, J. R. J. Phys. Chem. A 2007,
111, 1167.
(11) Matheson, M. S.; Mulac, W. A.; Weeks, J. L.; Rabani, J. J. Phys.
Chem. 1966, 70, 2092.
(12) Zehavi, D.; Rabani, J. J. Phys. Chem. 1972, 76, 312.
(13) Zafiriou, O. C. J. Geophys. Res. 1974, 79, 4491.
(14) Frinak, E. K.; Abbatt, J. P. D. J. Phys. Chem. A 2006, 110, 10456.
(15) Oum, K. W.; Lakin, M. J.; DeHaan, D. O.; Brauers, T.; Finlayson-
Pitts, B. J. Science 1998, 279, 74.
(16) George, I. J.; Anastasio, C. Atmos. EnViron. 2007, 41, 543.
(17) Sjostedt, S. J.; Abbatt, J. P. D. EnV. Res. Lett. 2008, 3, 045007.
(18) Seisel, S.; Rossi, M. J. Ber. Der Bunsen-Gesellschaft 1997, 101,
943.
(19) Diao, G. W.; Chu, L. T. J. of Phys. Chem. A 2005, 109, 1364.
(20) Barrie, L. A.; Bottenheim, J. W.; Schnell, R. C.; Crutzen, P. J.;
Rasmussen, R. A. Nature 1988, 334, 138.
(21) Schroeder, W. H.; Anlauf, K. G.; Barrie, L. A.; Lu, J. Y.; Steffen,
A.; Schneeberger, D. R.; Berg, T. Nature 1998, 394, 331.
(22) Vogt, R.; Crutzen, P. J.; Sander, R. Nature 1996, 383, 327.
(23) Huey, L. G.; Hanson, D. R.; Howard, C. J. J. Phys. Chem. 1995,
99, 5001.
(24) Thornton, J. A.; Braban, C. F.; Abbatt, J. P. D. Phys. Chem. Chem.
Phys. 2003, 5, 4593.
(25) NIST Chemistry WebBook, NIST Standard Reference Database No.
69; Linstrom, P. J.; Mallard, W. G., Eds.; 2010.
(26) Koop, T.; Kapilashrami, A.; Molina, L. T.; Molina, M. J. J.
Geophys. Res. 2000, 105, 26393.
(27) Jayson, G. G.; Parsons, B. J.; Swallow, A. J. J. Chem. Soc. -
Faraday Trans. I 1973, 1597.
(28) Zhou, X. L.; Mopper, K. Marine Chem. 1990, 30, 71.
(29) Hobbs, P. V. Ice Physics; Clarendon Press: Oxford, 1974.
(30) Buxton, G. V.; Greenstock, C. L.; Helman, W. P.; Ross, A. B. J.
Phys. Chem. Ref. Data 1988, 17, 513.
(31) Neubauer, J.; Heumann, K. G. Atmos. EnViron. 1988, 22, 537.
(32) Jaffe, D. A.; Zukowski, M. D. Atmos. EnViron. 1993, 27, 2935.
(33) Beine, H. J.; Domine, F.; Ianniello, A.; Nardino, M.; Allegrini, I.;
Teinila, K.; Hillamo, R. Atmos. Chem. Phys. 2003, 3, 335.
(34) France, J. L.; King, M. D.; Lee-Taylor, J. Atmos. EnViron. 2007,
41, 5502.
The relevance of the chemistry described in this paper will
be largely to sea ice to which nitrate has been deposited from
the atmosphere, probably in the form of HNO3.31,32 Or, the same
chemistry may occur in the snowpack to which marine aerosols
and nitrate have been deposited. Whereas the hydrogen ion
concentrations in these substrates will not be as high as those
used in most of our studies, we do note that we observed NOx
and halogen release from pH neutral solutions as well. The
acidity of sea ice and snowpack surfaces is not well-known,
especially with respect to the degree of acidification that arises
JP102072T