Reactivity of BrNO2 and ClNO2 with solid alkali salt substrates

Article abstract of DOI:10.1021/jp982000f

The reactions of BrNO₂ and ClNO₂ with bromide have been studied in a Knudsen-cell reactor in order to better understand the elementary reactions of these potentially important tropospheric species in the presence of solid sea-salt aerosol. BrNO₂ efficiently reacts with bromide to produce molecular bromine (γ > 0.3). To better understand the fate of BrNO₂ in the Knudsen-cell reactor, we have developed a BrNO₂ source based on the reaction of molecular bromine with solid KNO₂. We have determined that the lifetime of BrNO₂ under our experimental conditions is on the order of 10 s and is limited by heterogeneous decomposition reactions. The interaction of BrNO₂ with KBr, KCl, and KNO₂ was studied using the external BrNO₂ source. ClNO₂ can be converted to BrNO₂ in the presence of solid bromide, characterized by an uptake probability of γ = 1.3 × 10^-4, in good agreement with results obtained for several substrate presentations such as salt powder, grain, and single crystals. We compared these results with previous work and briefly discuss atmospheric implications.

Full text of DOI:10.1021/jp982000f

7470
J. Phys. Chem. A 1998, 102, 7470-7479
Reactivity of BrNO2 and ClNO2 with Solid Alkali Salt Substrates
†
Fran c¸ ois Caloz, Sabine Seisel, Frederick F. Fenter, and Michel J. Rossi*
Laboratoire de Pollution Atmosph e´ rique et Sol (LPAS), Swiss Federal Institute of Technology (EPFL),
1
015 Lausanne, Switzerland
ReceiVed: April 27, 1998; In Final Form: June 23, 1998
The reactions of BrNO
better understand the elementary reactions of these potentially important tropospheric species in the presence
of solid sea-salt aerosol. BrNO efficiently reacts with bromide to produce molecular bromine (γ > 0.3).
To better understand the fate of BrNO in the Knudsen-cell reactor, we have developed a BrNO source
based on the reaction of molecular bromine with solid KNO . We have determined that the lifetime of BrNO
under our experimental conditions is on the order of 10 s and is limited by heterogeneous decomposition
reactions. The interaction of BrNO with KBr, KCl, and KNO was studied using the external BrNO source.
ClNO can be converted to BrNO in the presence of solid bromide, characterized by an uptake probability
2 2
and ClNO with bromide have been studied in a Knudsen-cell reactor in order to
2
2
2
2
2
2
2
2
2
2
-
4
of γ ) 1.3 × 10 , in good agreement with results obtained for several substrate presentations such as salt
powder, grain, and single crystals. We compared these results with previous work and briefly discuss
atmospheric implications.
Introduction
BrNO + KBr f Br + KNO
2
(3)
2
2
It has been well established that N2O5 can react with salt to
produce volatile nitryl halides of the type X-NO2 1 in what
appears to be a displacement reaction:
0
-1
The enthalpy of reaction ∆Hr 298 ) -16.2 kJ mol of
-7
reaction 3 has been estimated using the value for the heat of
formation of BrNO2 (71.1 ( 7.5 kJ mole ) given by Wine
-1
9
and co-workers.
N O + NaCl f ClNO + NaNO
(1)
(2)
2
5
2
3
In addition, we will report our results on the reaction of
ClNO2 with bromide. ClNO2 is not reactive toward chloride.
Therefore the reaction with bromide is of interest not only
because of its atmospheric importance, but also because it
represents a second pathway for BrNO2 generation inside our
Knudsen-cell reactor:
N O + KBr f BrNO + KNO
3
2
5
2
The presence of these compounds may have an important
impact on the oxidizing potential of the Earth’s troposphere, as
they represent activated halogen compounds which may release
a halogen atom upon photolysis. Although bromide is a small
ClNO + KBr f BrNO + KCl
(4)
2
2
-
-
fraction of sea salt ([Cl ]/ [Br ] is approximately 600/1), there
is growing evidence that catalytic reactions are responsible for
the importance of atmospheric bromine chemistry. Because of
the potential role of reactions 1 and 2 in the atmosphere, we
have investigated the fate of the product species BrNO2 and
ClNO2 in the presence of salt surfaces using a low-pressure
Knudsen reactor which is part of a flowing gas experiment.
0
-1
Reaction 4 (∆Hr 298 ) 15.6 kJ mol ) may represent a
competitive pathway to ClNO2 photolysis and may be efficient
at volatilizing bromine during the night. Moreover, hydrolysis
of BrNO2 according to reaction 5 may represent an additional
source of HONO in the marine atmosphere.
BrNO + H O f HONO + HOBr
(5)
2
2
An important goal of this kinetic study is to characterize the
behavior of the key intermediate, BrNO2, under our typical
experimental conditions in light of earlier work carried out on
For this study, we have developed an alternatiVe BrNO2
source which is essential for testing the various hypotheses
concerning the reactivity of BrNO2 under our experimental
conditions. The source is based upon reaction 6:
1
,8
this reaction system in previous laboratory studies.
In
particular, we have shown in previous work that the only
observed stable gas-phase product of reaction 2 under our
conditions is molecular bromine. This result is in disagreement
with work performed in other laboratories, where direct evidence
for the presence of BrNO2 has been obtained.8 We postulated
that BrNO2 is formed as a primary product, which however
remained undetectable under our experimental conditions due
to its reactivity with bromide:
Br + KNO f BrNO + KBr
(6)
2
2
2
Experimental Details
For this study, we used a Teflon-coated Knudsen reactor,
1
0,11
recently described in detail.
Briefly, the reactor is a two-
chamber reactor operated in the molecular flow regime (see
Table 1), which allows one to isolate the reactive surface and
to perform reaction ON/reaction OFF experiments. Because
*
Author to whom correspondence should be addressed.
Current address: Jet Propulsion Laboratory, Chemical Kinetics and
†
-
3
Photochemistry Group, 4800 Oak Grove Dr., Pasadena, CA 91109.
of the low partial pressure on the order of 10 mbar or less,
S1089-5639(98)02000-3 CCC: $15.00 © 1998 American Chemical Society
Published on Web 08/25/1998

Reactivity of BrNO2 and ClNO2
J. Phys. Chem. A, Vol. 102, No. 38, 1998 7471
TABLE 1: Knudsen Cell Parameters
Using the observation that the reaction yield of HCl with NaNO2
is 90% to result in HONO, the loss of the HCl MS signal may
be quantitatively related to HONO (m/e 47).
cell parameter
value
3
volume (V)
estimated surface area (total)
1830 cm
1300 cm
19.6 cm
2
In this work we have synthesized ClNO2 according to the
12
2
procedure proposed by Ganske and co-workers. Briefly, pure
HCl gas is passed through a mixture of fuming nitric and sulfuric
acid, and the evolving gases are collected in a liquid nitrogen
cooled trap. The obtained product mixture is then twice distilled
to eliminate molecular chlorine and nitric acid. The purity of
the synthesized ClNO2 has been checked by MS. No Cl2 (m/e
70) and no HNO3 (m/e 63) have been observed.
surface area (A
S
, sample)
(1-1000) × 10 cm-3 a
10
number density range
surface collision frequency (Z
)b
38.7 × (T/M)
0.02 × (T/M)
1/2
s
s
-1
1
A
S
1
/2 -1
escape rate constant for the 1 mm
c
aperture (kesc
)
0.22 × (T/M) s-1
1
/2
escape rate constant for the 4 mm
c
aperture
0.80 × (T/M) s-1
1/2
escape rate constant for the 8 mm
c
aperture
To differentiate between the various potential sinks of BrNO2
in our reactor, both gas-phase and heterogeneous, it is necessary
to rely on a source of BrNO2 other than the in-situ reactions 2
and 4. In this work we developed a BrNO2 source appropriate
for a low-pressure reactor, based on reaction 6. The design of
the source is schematically presented in Figure 1, where we
show a second Knudsen cell mounted upstream to the reactor
and containing solid granular KNO2. Molecular bromine is
passed through the source reactor in order to produce BrNO2
via reaction 6. Two parameters are important for the source
design: (1) The escape rate constant of the source Knudsen
1.03 × (T/M) s-1
1/2
escape rate constant for the 9 mm
c
aperture
1.77 × (T/M) s-1
1/2
escape rate constant for the 14 mm
c
aperture
a
i
i
Calculated using the relation F ) Vkesc[M], where F is the flow of
molecules, V is the reactor volume, and [M] the number density.
Calculated for a sample surface of 19.6 cm . Determined directly
b
2 c
by experiment.
gas-wall collisions are favored over gas-gas collisions in the
Knudsen reactor, making it well-suited for the study of
heterogeneous processes. The different kinetic parameters
which are necessary in order to use a Knudsen cell may be
determined using simple gas kinetic expressions and are
summarized in Table 2. The modulated effusive molecular
beam leaving the Knudsen cell is analyzed by a quadrupole mass
spectrometer (MS). The mass spectra of all observed com-
pounds are listed in Table 3. In addition, the experiment was
-1
cell has to be as large as possible (>5 s ) in order to minimize
the residence time of BrNO2 in the source reactor, and (2) the
orifice of the source reactor has to be as small as possible in
order to avoid back-diffusion of BrNO2 from the main reactor;
i.e., the volume of the source reactor has to be small compared
to that of the main reactor. The chosen source reactor geometry
was a cylinder of 6 cm height and 4 cm diameter with an escape
orifice diameter of 8 mm. According to the equations presented
in Table 2, and using a value for the uptake coefficient for
reaction 6 of 0.32 (see below), the source is able to convert
1
1
equipped with laser-induced fluorescence detection. In the
present work this technique was used to unambiguously detect
NO2 after excitation at 403 nm and broad band detection to the
red of 500 nm using a photomultiplier protected by a cutoff
filter. The signal acquisition was performed using a boxcar
integrator (delay 9 µs, width 10 ns, average 30 pulses) resulting
9
0% of the molecular bromine at a residence time for BrNO2
in the source reactor of less than 1 s. In addition to BrNO2,
the major emitted gases were Br2 and NO2 in a ratio of
approximately 1:10.
To take into consideration diffusion13 of the gas molecules
9
3
in a detection limit of 1 × 10 molecule/cm .
The gas-phase reactants are introduced into the Knudsen cell
either through a glass capillary inlet or via a pulsed solenoid
valve allowing the introduction of millisecond pulses. Depend-
ent on the method of introducing the test gas, two different types
of experiments referred to as steady-state and pulsed-ValVe
experiments are routinely performed. Steady-state experiments
are performed by introducing into the reactor a constant flow
of molecules. By analysis of the change of the MS signal levels
of the corresponding compounds upon opening and closing the
sample chamber, a value for the net uptake coefficient γ may
be calculated. Pulsed-valve, thus real-time, experiments have
been performed by introducing a pulse of the test gas into the
reactor. A reference pulse is introduced while the reactive
surface is still isolated from the reactor volume. The rate
constant for effusive loss kesc is determined by simple fitting of
an exponential decay function to the experimental MS signal
trace in the absence of reaction. Repeating the same process
with the plunger lifted, thus with the sample exposed, a reactive
pulse is obtained. The observed single-exponential decay in
the presence of a reactive surface is characterized by a new
rate constant, kdec, defined by kdec ) kreac + kesc.
into the bulk of the salt under our experimental conditions we
have used different types of surfaces: bulk salt powders, sieved
grain substrates having grain diameters between 300 and 400
µm, spray deposited surfaces, and single-crystal optical salt flats,
the two latter without internal surface. Before each experiment,
the fresh salt surface has been held under vacuum for several
hours in order to desorb the adsorbed surface water. Occasion-
ally, the salt samples were heated under vacuum to 500 K to
completely eliminate adsorbed water. Single-crystal optical flats
have been treated in two different ways, referred to as polished
and depolished. Polished surfaces are prepared by gliding a
sheet of wet optical paper over the surface and then allowing it
to dry. The salt flats are depolished by gentle rubbing using
fine-grained sandpaper with careful subsequent elimination of
the powder generated at the surface. All salts are commercially
available: Fluka NaCl p.a., KNO2 p.a., KBr p.a., NaNO3
MicroSelect, KCl MicroSelect. The various kinds of surface
1,14
preparation have been described in detail in previous work.
Results
The gas densities were determined in mass flow calibrations.
The flow rate into the Knudsen reactor was measured by
recording the pressure change as a function of time in a
calibrated volume behind the capillary while monitoring the
corresponding MS signal. The flow of molecules may then be
related to the concentration of the gas molecules in the reactor.
The HONO calibration has been performed in situ by reacting
a known flow of HCl with solid NaNO2 in the Knudsen cell.
To study the reactivity of BrNO2 toward various salts we
had to develop an appropriate BrNO2 source. As already
described in Experimental Details, the source is based on the
reaction of Br2 with KNO2 (reaction 6). We begin this section
with a presentation of our results of a study of reaction 6 because
these data provide the needed information on the BrNO2 lifetime
and reactivity characterizing our source.

7472 J. Phys. Chem. A, Vol. 102, No. 38, 1998
Caloz et al.
TABLE 2: Relevant Equations
number
equation
) (8RT/πM)0.5(1/4V)
note
Gas-wall collision frequency (molecule cm-2). V is the volume of the reactor (cm ).
-1
3
1
2
3
Z
1
-
1
2
kesc ) Z
uni ) kesc (S
1
A
H
H
Escape rate constant (s ). A is the escape orifice surface area (cm ).
-
1
k
0
/S
R
- 1)
First-order rate constant for the uptake (s ). S
0 R
and S refer to the MS signals measured
before and during reaction.
4
5
γ ) kuni/(Z
1
A
S
)
V
Uptake coefficient. A
Vapor pressure determined in the steady-state experiment described in the text.
S
refers to the geometrical area of the sample surface.
i
P ) RTF /kesc
TABLE 3: Mass Spectral Data
parent
peak
a
fragmenta fragmenta fragmenta fragmenta
species
BrNO
ClNO
Br
NO
2
93, 95 (1)
49 (1)
160 (100) 79, 81 (17)
46 (50) 30 (100)
46 (100)
46 (100)
30 (80) 79, 81 (10)
30 (50) 35 (10)
2
2
2
HONO 47 (12) 30 (100)
a
Ion mass with peak intensity in parentheses, given as percent of
the most intense peak. The numbers in italics indicate the mass
fragments which were used to monitor the species by MS.
Figure 2. Steady-state experiment on the reaction of Br
2
+ KNO
flow of 3.8
10 molecules s . During this experiment we monitored the MS
2
(
×
reaction 6) performed using the 14 mm orifice with a Br
2
14
-1
+
+
signal at m/e 93 (BrN ) and at m/e 160 (Br
2
).
-
1
molecules s ] was allowed to react with KNO2. These
experiments showed a large uptake rate of Br2 (detected at m/e
+
1
(
60) and the appearance of new signals at m/e 30 (NO ), 46
+
+
NO2 ), and 93 and 95 (BrN ), as shown in Figures 2 and 3.
The product mass spectrum indicates the presence of BrNO2.
When the gas-phase residence time in the Knudsen reactor is
increased by reducing the exit-orifice diameter from 14 mm to
1
mm, the ratio m/e 30 to 46 increases (see Figure 3). This
suggests that secondary products are formed which have a strong
+
contribution at m/e 30 (NO ). Possible candidates are BrNO,
HONO, NO2, and NO. BrNO and HONO both have strong
parent-ion signals at m/e 109, 111 and at m/e 47, respectively,
and thus can be unambiguously monitored by MS. In ancillary
experiments on reaction 6, we investigated the formation of these
2
Figure 1. Schematic drawing of the external source of BrNO . The
source is a small Knudsen cell containing nitrite and mounted directly
on the main reactor. Molecular bromine is introduced through a capillary
or a needle valve.
possible secondary products which may contribute to mass 30
+
(
NO ). BrNO was not detected (m/e 109 and 111), but we did
observe slow HONO production (m/e 47), as shown in Figure
We attribute this to the hydrolysis reaction of BrNO2
TABLE 4: Results on Reaction Br
2
+ KNO
2
4
.
substrate
expt type
uptake coeff
no. of expts
(reaction 5).
powder
powder
spray sample
spray sample
steady-state
pulsed-valve
steady-state
pulsed-valve
0.32 ( 0.03
0.28 ( 0.05
0.36 ( 0.05
0.3 ( 0.1
10
2
2
The yield of HONO is 10% with respect to Br2 taken up on
samples that were dried under vacuum without heating. Thus,
when we take into account reaction 6 under our experimental
conditions, only 10% of the BrNO2 undergoes hydrolysis.
HOBr, however, was not detected because it undergoes a fast
reaction with KBr according to reaction 7:
1
av 0.32 ( 0.05
Br2 + KNO2: A Source Reaction for BrNO2. Kinetics of
Reaction 6. Reaction 6 has been measured to be fast with an
uptake coefficient of γ ) 0.32 ( 0.05. This value has been
found to be independent of reactant density and of sample
surface presentation (see Table 4). We therefore conclude that
reaction 6 is a first-order process. The independence of the
results on the surface presentation is in agreement with the
HOBr + KBr f Br + KOH
(7)
2
Reaction 7 has recently been studied in our laboratory and
shown to be fast (γ > 0.1), and molecular bromine was detected
1
5
as a product. HONO disappeared from the product mass
spectrum of reaction 6 when the salt sample was dried under
vacuum in the Knudsen reactor by heating it to 500 K using a
high-temperature sample support.
1
3
model of surface diffusion proposed by Keyser and co-workers
which does not predict any significant changes in the observed
uptake coefficient for the case of large uptake rates. This case
may be compared to the interaction of ClONO2 with salt surfaces
that has been studied previously.14
Product Analysis. Initial experiments were conducted using
the Knudsen-cell reactor in the steady-state configuration, in
We concluded that the product mass spectrum of reaction 6
was consistent with the presence of large amounts of NO2. To
simplify the analysis of the NO2 yield, it was necessary to
conduct experiments using LIF detection of NO2. These LIF
experiments were performed at an excitation wavelength of 403
1
4
which a constant flow of molecular bromine [(1-10) × 10

Reactivity of BrNO2 and ClNO2
J. Phys. Chem. A, Vol. 102, No. 38, 1998 7473
Figure 4. The steady-state experiment on the reaction of Br
2
+ KNO
2
(
reaction 6) presented in this figure has been performed using a Br
2
14
-1
flow of 4.5 × 10 molecules s and the 1 mm orifice. HONO and
Br are detected at m/e 47 and m/e 160, respectively, during the reaction.
2
Figure 5. Typical signals obtained during a combined LIF/MS
experiment on the reaction of Br + KNO (reaction 6). This experiment
was performed using the 4 mm orifice at a Br
2
2
1
4
2
flow rate of 5.3 × 10
-1
molecules s . The sample chamber was opened after 220 s and closed
after 250 s. The LIF signal is displayed here as a negative-going signal.
During this experiment we have measured a NO
molecular bromine taken up.
2
yield of 0.89 per
Figure 3. This figure displays two steady-state experiments on the
reaction of Br
2
+ KNO
2
(reaction 6) performed using two different
TABLE 5: Results of LIF Experiments of the Reaction Br2
14
orifice sizes and a molecular bromine flow rate of 4.5 × 10 molecules
+ KNO2
-
1
s
2
. The residence time of BrNO in the reactor is varied by about 2
orifice [mm]
k
esc [s-1]
t
res [s] NO
2
yielda m/e 46b
m/e 30b
orders of magnitude. Note the variations in the ratio of the recorded
signals at mass m/e 46 and m/e 30 which decreases with decreasing
orifice size.
14
9
8
3.1
1.7
1.33
0.33
0.19
0.11
0.03
0.32 25 ( 10%
1
1
1
1
1
0.85 ( 0.05
0.85 ( 0.05
0.85 ( 0.05
1.0 ( 0.3
0.6
0.8
3
40 ( 10%
50 ( 20%
90 ( 20%
95 ( 20%
4
nm, and an example is shown in Figure 5. The results are
summarized in Table 5. The NO2 yield with respect to the
consumption of Br2 increases with increasing residence time
and reaches 200% when using the smallest escape orifice
3
5.3
1.0 ( 0.3
2
1
.3
9.1 130 ( 30%
33 200 ( 50%
a
lost through reaction. b Normalized residual MS
2 2
NO yield per Br
(residence time of approximately 25 s). To interpret these
signal after subtraction of the NO contribution. The values for the
2
results, we propose the following reaction mechanism:
two smallest orifices are not given because the residual is too small.
Hinshelwood mechanism displayed in reaction 8a.
Br + KNO f BrNO (ads) + KBr
(6)
(8)
2
2
2
BrNO (ads) f 1/2 Br + NO
2
2
2
2 Br(ads) f Br2
(8a)
1
/2 Br + KNO f KBr + NO
The observed mass spectrum for BrNO2, after correction for
the presence of NO2, does not change significantly as a function
of residence time (see Table 5), in support of this mechanism.
The radical product, NO2, originates from the decomposition
of BrNO2. As discussed below, this decomposition (reaction
8) takes place on some uncoated parts of the internal surfaces
of the reactor and has only a vanishingly small homogeneous
gas-phase contribution. This step is strongly favored because
the gas-wall interaction is predominant under the low-pressure
2
2
2
We note that reaction 8 may not correspond to an elementary
reaction. The most probable mechanism involves a fast
heterogeneous recombination of atomic bromine yielding Br2
(
reaction 8a). We did not observe free bromine atoms under
the same conditions where we have observed Br in the gas phase
from the reaction of NO3 free radical with KBr.26 Therefore,
we think that the recombination of Br follows a Langmuir-

7474 J. Phys. Chem. A, Vol. 102, No. 38, 1998
Caloz et al.
MS intensities at m/e 30 and 46 have been determined in good
accuracy from experiments displayed in the first three entries
of Table 5. In ancillary experiments using the external BrNO2
source, the complete MS spectrum has been determined to be
4
6 (100%), 30 (80%), 79 (10%), and 93, 95 (1%); here the
relative intensities are given in parentheses and are found to be
in agreement with the results presented above.
In conclusion, the uptake of Br2 on KNO2 is fast and produces
BrNO2 with a lifetime of 10 s in our reactor. Therefore the
reaction is suitable as a source of BrNO2.
Reactivity of BrNO2. Once we had established an appropri-
ate BrNO2 source, we studied its reactivity on various salts.
One disadvantage of this source is that an exact calibration of
the MS signal of BrNO2 is not available. Therefore the study
is limited to the measurement of uptake coefficients and the
identification of products.
Figure 6. Display of the fractional yield of BrNO
2
as a function of
in the Knudsen
the calculated gas-phase residence time of BrNO
reactor. The solid line is a single-exponential fit.
2
BrNO2 + NaNO3. To obtain information on the heteroge-
neous decomposition of BrNO2 on salt surfaces (reaction 8),
we performed uptake experiments of BrNO2 on NaNO3. NaNO3
is a non-halogen-containing salt. Thus, we are able to dif-
ferentiate between the heterogeneous decomposition according
to reaction 8 and the reaction with salt according to reaction 3.
Steady-state experiments on the uptake of BrNO2 on NaNO3
powder have been performed in which BrNO2 was monitored
at m/e 95 using the external BrNO2 source. We observed a
weak interaction of BrNO2 resulting in an uptake coefficient of
conditions. According to the above scheme, reaction 8 produces
molecular bromine and NO2; thus, for each mole of Br2
consumed in the reaction with KNO2, 1 mol of NO2 and 0.5
mol of Br2 are produced, resulting in a 200% yield of NO2 with
respect to loss of Br2.
Lifetime of BrNO2 in the Knudsen Cell. Detailed analysis
of the LIF experimental results on reaction 6 allows the
determination of the lifetime of BrNO2 in our reactor. At steady
state, the flow of BrNO2 molecules from the reactor is given
by the consumption of the molecular bromine flow and by the
production of nitrogen dioxide, according to
-3
γ ) 5 × 10 . No changes of either Br2 or NO2 have been
observed in the gas phase, indicating that the interaction is
nonreactive and that decomposition of BrNO2 does not take
place on NaNO3 under the prevailing experimental conditions.
This result asserts that products observed during the uptake of
BrNO2 on halogen-containing salt are due to the reaction of
BrNO2 on the alkali halide salt sample.
o
i
o
o
F (BrNO ) ) [F (Br ) - F (Br )] - 0.5 F (NO ) (E1)
2
2
2
2
where the superscripts o and i represent the outgoing and
incoming gas flows. The LIF signals, which correspond to a
density measurement, have to be converted into equivalent flows
BrNO2 + KBr. We have shown in ref 1 that BrNO2, the
primary product of reaction 2, remained undetectable under our
experimental conditions. The only observed stable gas-phase
products were molecular bromine and a small amount of HONO
which was attributed to a secondary reaction (reaction 5). To
explain these experimental results we postulated that BrNO2 is
highly reactive toward bromide (reaction 3). Using the external
BrNO2 source presented above, we were able to test this
hypothesis. Steady-state experiments of the uptake of BrNO2
on KBr powder substrates indeed obtained a fast rate of uptake
of BrNO2 on KBr with an uptake coefficient of γ g 0.3. The
only detectable gas-phase product was molecular bromine, in
agreement with reaction 3. No NO2 formation has been
observed using sensitive LIF detection, indicating that the
reaction of BrNO2 with bromide (reaction 3) is faster than its
heterogeneous decomposition (reaction 8).
o
for eq E1, using the relation F (NO2) ) [NO2]kesc(NO2)V. In
Figure 6 we have plotted the fractional yield, f, of BrNO2 against
residence time, whose variation was obtained by changing the
escape orifice:
o
F (BrNO )
2
f )
(E2)
i
o
F (Br ) - F (Br )
2
2
The solid line shown in the figure is an exponential fit that
corresponds to a BrNO2 lifetime of 10 s within our Knudsen-
cell reactor. Additional experiments on the lifetime of BrNO2
in our reactor have been performed using the external BrNO2
+
source. A comparison of the MS signals at m/e 95 (BrN ) at
different residence times resulted in the same lifetime as
determined above, supporting the heterogeneous decomposition
mechanism proposed above.
BrNO + KNO . The assumption that two molecules of NO
2
2
2
In Table 5 we present the residual MS signal after subtraction
of the NO2 contribution for both m/e 46 and m/e 30, determined
from a calibration of the LIF signal for NO2. We point out
that this residual MS signal corresponds to the sum of all
possible reaction products except NO2. This spectrum remains
constant for different orifice sizes and therefore for different
residence times, which is evidence for the presence of only one
important product in the mixture, namely BrNO2. The slight
increase in the relative signal at m/e 30 at long residence times
might be due to an additional contribution to this mass peak by
HONO, but the uncertainties of the measurements and of the
subtraction procedure (large numbers leading to a small residual)
are large for long residence times. We conclude that the relative
MS intensities for BrNO2 resulting in a ratio of 0.85 for the
are produced for each Br molecule lost according to the sum
2
of reactions 6 and 8 is the basis for the lifetime analysis carried
out above. It is therefore necessary to investigate the potential
NO formation by the secondary reaction between BrNO and
2
2
nitrite (reaction 9), which results in the same net reaction and
relative yields of NO per Br lost as the net of reactions 6 and
2
2
8 in order to assert that BrNO in fact undergoes heterogeneous
2
decomposition (reaction 8) releasing NO2:
Br + KNO f BrNO + KBr
(6)
(9)
2
2
2
BrNO + KNO f 2 NO + KBr
2
2
2
Br + 2 KNO f 2 KBr + 2 NO
2
2
2

Reactivity of BrNO2 and ClNO2
J. Phys. Chem. A, Vol. 102, No. 38, 1998 7475
We have performed steady-state experiments on the uptake
of BrNO2 on KNO2 powder using the external BrNO2 source.
Since Br2 is always emitted from the source and readily reacts
with KNO2 according to reaction 6, these experiments are
difficult to interpret. In cases where higher Br2 concentrations
were present in the reactor an increase in the MS signal at m/e
TABLE 6: Results of LIF Experiments of the Reaction
ClNO
2
+ KNO
2
(Reaction 11)
a
orifice [mm]
τ (ClNO ) [s]
2
γ (ClNO2)
NO yield
2
-2
1
4
8
4
1
0.3
0.7
2.4
26
1.8 × 10
25%
37%
46%
-2
1.7 × 10
-
-
2
3
1.5 × 10
5.4 × 10
194%
93 has been observed, indicating formation of BrNO2 according
to reaction 6. At the same time, the formation of NO2 has also
been observed using LIF detection. However, at small levels
a
NO
2
yield per ClNO
2
lost through reaction.
i
14
TABLE 7: Br and NO Yields of Reactions 2 and 4
of Br2 (F < 10 molecules/s) emitted from the BrNO2 source,
we observed a weak interaction of BrNO2 with KNO2 ac-
companied by a decrease in the LIF signal of NO2, indicating
a decrease in the rate of formation of NO2 according to reaction
2
2
yielda
yieldb
reaction
+ KBr (2)
ClNO
expt type
Br
2
NO
2
N
2
O
5
steady-state
steady-state
pulsed-valve
0.35 ( 0.05
0.61 ( 0.22
0.5 ( 0.2
2
+ KBr (4)
0.2 ( 0.1
8
. Therefore, the uptake of BrNO2 on KNO2 is nonreactive,
ClNO + KBr (4)
2
similar to the uptake on NaNO3, and no NO2 is formed. We
conclude that reaction 9 is too slow under our experimental
conditions so that the observed NO2 results from the heteroge-
neous decomposition shown in reaction 8.
a
Determined per loss of reactant molecule (see text). b Given for
the 1 mm orifice.
Reactivity of ClNO . N O reacts with chloride according
2
2
5
BrNO2 + KCl. The interaction between BrNO2 and KCl may
produce two different products:
to reaction 1, producing ClNO which is stable with respect to
2
NaCl and therefore easily observable under our experimental
1
conditions. However, ClNO2 does react with bromide accord-
BrNO + KCl f ClNO + KBr
(10a)
2
2
ing to reaction 4. This reaction may have implications for
atmospheric chemistry and in addition represents a second option
for BrNO2 production in the laboratory.
ClNO2 + KNO2. Steady-state experiments of the interaction
of ClNO2 with KNO2 powder substrates have been performed
using MS detection at m/e 49 for ClNO2 and LIF detection for
BrNO + KCl f BrCl + KNO2
(10b)
2
In steady-state experiments the initial uptake coefficient of
BrNO2 on KCl powder substrates monitored by MS at m/e 95
-
2
was measured as γ ) 5 × 10 . The BrNO2 rate of uptake
decreased with exposure time. A second exposure on the same
salt sample, again of a length of approximately 1 min, already
reduced the uptake coefficient by a factor of 2. No chlorine-
containing gas-phase product was observed. The expected
products of reaction 10a and 10b do not interact with KCl, an
observation which has been verified in an ancillary experiment.
NO2. An average initial uptake coefficient of γ ) (1.7 ( 0.2)
-2
×
10 has been measured. No HONO formation (m/e 47) has
been observed indicating the absence of HNO3 which could have
reacted with KNO2 and released HONO. We conclude that
ClNO2 did not undergo hydrolysis under the present experi-
mental conditions. The unique product detected in the gas phase
was NO2. The yield of NO2 with respect to ClNO2 consumed
has been measured at 200% at the longest ClNO2 residence time
used (26 s). At decreasing ClNO2 residence times, the NO2
yield gradually decreased. The experimental results are sum-
marized in Table 6 and reveal the occurrence of reaction 11:
8
This result is in disagreement with the work of Frenzel et al.
who observed the formation of ClNO2 during the uptake of
BrNO2 on chloride solutions. The only gas-phase product we
were able to detect was molecular bromine followed by MS at
mass m/e 160. No formation of HONO has been observed
which excluded the hydrolysis of BrNO2 (reaction 5) followed
by reaction 7 as a sink for BrNO2. Since the BrNO2 source
does not allow one to establish a mass balance, it is unclear
whether molecular bromine is formed by decomposition of
BrNO2 according to reaction 8 or by fast reaction of either
ClNO + KNO f KCl + 2 NO
2
(11)
2
2
ClNO2 + KBr: Product Analysis. The product analysis of
steady-state experiments showed Br2 as the only detectable
product of reaction 4 with MS intensities at m/e 79, 81 and
158, 160 and 162 in agreement with the Br2 mass spectrum.
The mean Br2 yield per molecule of ClNO2 lost by uptake has
been determined to be 0.55 ( 0.2 (see Table 7). An example
of a steady-state experiment performed to measure the yield of
Br2 per loss of ClNO2 is shown in Figure 7. During reference
experiments in which KBr was exposed to a flow of Br2, an
interaction between Br2 and KBr was measured which possibly
1
4
BrNO2 or BrCl with impurities of bromide in the KCl sample
reaction 3). However, we have some indirect evidence for the
(
occurrence of a secondary reaction: (i) BrNO2 does not
decompose on NaNO3, and therefore efficient heterogeneous
decomposition on the salt may be excluded; (ii) during the
uptake of BrNO2 on KCl, no formation of NO2 following the
heterogeneous decomposition (reaction 8) has been observed
using LIF. This suggests that BrNO2 reacts faster than it
heterogeneously decomposes on the KCl salt surface according
to reaction 8; (iii) the KCl sample contains 0.05% bromide as
an impurity. Experiments on NaCl with a bromide content of
1
5
may explain a yield of Br2 less than unity.
In real-time experiments we were able to detect MS signals
+
at m/e 160, 79, and a weak contribution at m/e 93 and 95 (BrN )
0
.005% resulted in an initial uptake coefficient of γ0 ) 2.2 ×
as product signals, along with the time-dependent disappearance
-
2
1
0 , which is significantly smaller than the value obtained on
of ClNO2 monitored at m/e 46. We did not detect the fragment
+
the KCl sample. On the other hand, when the bromide content
is increased to 2% the uptake of BrNO2 increases to a initial
value of γ ) 0.2. At the same time a higher yield of Br2 has
been obtained. These facts, namely the correlation of both the
initial uptake coefficient and the Br2 yield with the bromide
content of the salt, the saturation behavior of the uptake, and
the absence of NO2 formation let us conclude that BrNO2 is
reacting only with the bromide impurity of the salt and not with
chloride itself.
at m/e 97 corresponding to BrO . We were also able to rule
out BrCl as a product of the reaction of ClNO2 with bromide
under our conditions. A typical pulsed-valve experiment is
displayed in Figure 8. Data analysis of pulsed-valve experi-
ments allow the determination of the uptake coefficient by fitting
the reactant pulse to an exponential decay and extracting the
decay constant. The yield is evaluated by integrating the
calibrated product pulse at mass m/e 160 and comparing this
value to the difference between the calibrated integrated

7476 J. Phys. Chem. A, Vol. 102, No. 38, 1998
Caloz et al.
and 4. This reaction scheme led to a decay of BrNO2, monitored
at m/e 93, which was faster than experimentally observed in
Figure 8. Therefore, we proposed a modification to the above
reaction mechanism. Two reasons may be at the origin of this
discrepancy: (1) BrNO2 may remain adsorbed on the surface
after its production, and its appearance in the gas phase is
therefore delayed. This corresponds to the following mecha-
nism:
ClNO (g) + KBr(s) f BrNO (ads) + KCl(s) (12)
2
2
BrNO (ads) f BrNO (g)
(13)
(14)
2
2
BrNO (ads) + KBr(s) f Br (g) + KNO (s)
2
2
2
(
2) The reactive uptake of ClNO2 is a two-step process,
Figure 7. Steady-state experiment on reaction of ClNO
reaction 4) using the 1 mm orifice and a continuous ClNO -flow rate
of 2 × 10 molecules s . The sample (6 g of KBr powder) exposition
begins at 50 s and ends at 900 s. The delay in the Br signal rise (50
s < t < 250 s) is due to the filling time of the Knudsen cell. During
the reaction, Br was detected with a yield of 0.6 with respect to loss
of ClNO , as integrated between the dotted lines.
2
+ KBr
according to reactions 15 to (17):
(
2
1
5
-1
ClNO (g) f ClNO (ads)
(15)
(16)
(17)
2
2
2
2
ClNO (ads) + KBr(s) f BrNO (g) + KCl(s)
2 2
2
BrNO (g) + KBr(s) f Br (g) + KNO (s)
2
2
2
We are not able to distinguish between the two proposed
mechanisms using the experimental results of the pulsed-valve
experiments. However, it seems unlikely that the ClNO2 uptake
is a complex process, passing through an adsorbed state. The
molecule diffuses into the internal void of the grain samples
according to observation and theory (see below) which indicates
that the lifetime of ClNO2(ads) must be small relative to 1/kesc.
This hypothesis is supported by surface residence time experi-
ments of ClNO2 on KBr where no measurable residence time
1
6
16
has been observed. In addition, Koch et al. determined a
residence time of τ ) 0.7 ms of ClONO2 on NaCl which they
explain by the interaction of the positively polarized Cl in
-
ClONO2 with Cl of the salt. Since bromine in BrNO2 is also
positively polarized, the adsorption of BrNO2 on KBr resulting
in a long-lived surface complex before reaction 14 may be
possible. Therefore, the mechanism consisting of reactions 12-
1
4 is not only consistent with residence time measurements16
but may also explain the less than 100% yield of Br2 observed
in the gas-phase (Table 7) and the nonreactive uptake of BrNO2
on KCl (see above). Consequently, the solid lines presented in
Figure 8 have been obtained by using the mechanism involving
the adsorption of BrNO2 on the KBr powder corresponding to
reactions 12-14. The rate-determining step in the mechanism
given by reactions 12-14 is the desorption of BrNO2 (reaction
13) with a rate constant k13 ) 3.1 s corresponding to a
residence time on the salt of τ ) 0.3 s.
Kinetics of the Reaction. We have investigated the reactivity
Figure 8. Pulsed-valve experiment of ClNO
performed using the 14 mm orifice. From this experiment we are able
to measure the uptake coefficient by directly fitting the reactive ClNO
pulse to an exponential decay (γ ) 0.01). The Br yield per loss of
ClNO is determined to be 0.61 by comparing the integrals of the
reactant and product signals. BrNO is detected at mass m/e 93. The
2
on KBr (5 g of powder)
2
2
2
2
-
1
solid lines are fits obtained using a two-parameter model (see text).
nonreactive (sample absent) and reactive (sample present)
reactant pulses monitored at mass m/e 46, corresponding to the
amount of ClNO2 lost per pulse.
of the ClNO /KBr system as a function of reactant density and
2
as a function of the surface presentation of the salt. We did
We attribute the MS signal intensity at m/e 93 and 95 to
BrNO2. The product spectrum of this system is therefore
consistent with the following mechanism:
not detect any systematic variation of the initial uptake as a
function of the gas-phase density when the sample was presented
-
2
as a powder. The uptake coefficient, γ ) (1.9 ( 0.6) × 10 ,
measured on powders remained constant over the density range
ClNO + KBr f BrNO + KCl
(4)
(3)
10
13
-3
2
2
of 10 to 10 molecules cm . Pulsed-valve experiments
performed on powder surfaces also showed a similar uptake
BrNO + KBr f Br + KNO
-2
2
2
2
coefficient of γ ) (1.5 ( 0.6) × 10 . The density indepen-
dence of the measured initial uptake coefficient on one type of
surface confirms that the process follows first-order kinetics.
To study the dependence of reaction 4 on the presentation of
the reactive surface, we have performed uptake experiments on
various types of substrate (powder, grain, spray-deposited salt
In additional experiments we confirmed that reaction 3 is fast
and produces molecular bromine (see above).
In our first attempts to fit the pulsed-valve data for Figure 8,
we used the simple two-step mechanism given by reactions 3

Reactivity of BrNO2 and ClNO2
J. Phys. Chem. A, Vol. 102, No. 38, 1998 7477
TABLE 8: Results Obtained for the Reaction of ClNO
KBr as a Function of Surface Presentation
2
+
detection. Indeed, NO2 was observed, but the yield of NO2
per ClNO2 consumed was small, not exceeding 20% for the
longest residence times, as summarized in Table 7. The
appearance of NO2 may be due to heterogeneous decomposition
of BrNO2 (reaction 8). In addition, in the fast secondary reaction
of BrNO2 with KBr (reaction 3), KNO2 is formed, which may
react with ClNO2 according to reaction 11. The uptake
coefficient of ClNO2 on KNO2 powder was found to be γ ) 2
number of
measured uptake corrected value
type of surface
spray
window-polished
window-depolished
powder
experiments
coefficient
of γa
-
-
-
-
4
5
4
2
2
1
1
6
19
(1.0 ( 0.5) × 10
(1.5 ( 1.0) × 10
(1.0 ( 0.5) × 10
(1.9 ( 0.6) × 10
-4
-4
1.0 × 10
1.3 × 10
-
4
grain (350 µm)
(1-100) × 10
-
2
×
10 (see above), thus of the same magnitude as the uptake
a
Corrections obtained by using the surface diffusion model described
coefficient of ClNO2 on KBr powder. Therefore it seems
possible that at long residence times, when a sufficient amount
of KNO2 has been built up, ClNO2 undergoes competitive
reactions with KBr and KNO2, both leading to a certain amount
of NO2 according to reactions 8 and 11, respectively. On the
basis of our experiments, we may not distinguish one reaction
channel from another. We therefore state that the observed NO2
may be due to both reactions, namely, heterogeneous decom-
position of BrNO2 (reaction 8) and the reaction of ClNO2 on
KNO2 (reaction 11) which has been formed in reaction 3.
in ref 1.
Discussion
We have presented results on the reactivity of ClNO2 and
BrNO2 toward various solid alkali salts. The reactions presented
above have been studied in other laboratories using different
5
,8,17
techniques
which have been applied at higher pressures
using aqueous solutions as a reactive surface.
1
7
George and co-workers have shown that ClNO2 reacts
-
efficiently with halogen anions, notably with I ; although their
study was carried out involving the solution phase, the observed
reactivity is analogous to our observations for reaction 4.
Frenzel and co-workers have studied the same reactions as
presented here on aqueous solutions using a wetted-wall flow
Figure 9. Results of monodisperse grain experiments (grain size 0.35
mm) using different orifice sizes. The dashed line is a fit obtained using
1
3
the diffusion model proposed by Keyser and co-workers using the
-
4
value for the true uptake coefficient of 1.3 × 10
.
8
tube experiment. In general, our results agree well with the
surfaces, and polished and depolished single-crystal optical
flats). These experiments, summarized in Table 8, showed a
strong variation of the measured uptake coefficient as a function
of the surface presentation. In light of these results, we
performed experiments using monodisperse grain samples of
different mass in order to determine if the surface diffusion
model of Keyser and co-workers13 is applicable to the present
case. Figure 9 displays the results obtained from these
experiments. The dashed line is the best fit obtained using the
findings of Frenzel et al. in that Br underwent a fast reaction
2
with KNO to form BrNO , and ClNO reacted readily with
2
2
2
-
Br solutions to produce BrNO . Hydrolysis of ClNO and
2
2
BrNO was found to be slow. In agreement with our study,
2
BrCl was not observed as a product of reaction 10. During the
reaction of ClNO with KNO , slow release of NO into the
2
2
2
gas phase has been observed.
However, there are two main discrepancies between the
8
results of Frenzel et al. and the ones presented in this work.
-
4
surface diffusion model corresponding to γtrue ) 1.3 × 10 .
We note that the predicted value for the true uptake coefficient
in the monodisperse grain experiments presented in Figure 9
They report that BrNO2 has a lifetime that exceeds 1 h under
18
their conditions. This is in apparent contradiction to our results
which show a lifetime of BrNO in our reactor on the order of
2
-
4
(
γtrue ) 1.3 × 10 ) is in excellent agreement with the value
10 s. The heterogeneous decomposition (reaction 8) is surpris-
measured on spray-deposited KBr samples and on depolished
optical KBr flats. Powders are substrates composed of a great
number of grain layers; for this type of sample the correction
factor corresponds to the limit of a large number of layers and
fortunately, in this limit, the correction factor becomes inde-
pendent of the grain diameter. The diffusion model applied in
this manner predicts a value of γtrue which is in excellent
agreement with the results obtained on other KBr samples,
namely, γtrue ) 1.0 × 10 , as shown in Table 8. The uptake
coefficient of reaction 4 measured on polished KBr flats has
been determined to be 1 order of magnitude smaller than the
true uptake coefficient (γtrue) determined for all the other
surfaces. We obtained a similar discrepancy for the case of
N2O5 interacting with salt which was presented in previous
work.1 This discrepancy may perhaps be attributed to a change
in crystallinity of the surface.
ingly rapid, which indicates that the process may take place on
some uncoated parts of the Teflon-coated walls of our low-
pressure reactor by virtue of its large surface area (Table 1). A
semiquantitative RRK calculation using the dissociation energy
determined by Kreutter et al. indicates that at the temperature
and total pressures used in this study, the unimolecular
decomposition of the BrNO molecule should still be far in the
2
falloff region, predicting a much longer lifetime with respect
-
4
9
to only gas-phase decomposition. This highlights a funda-
mental limitation of the Knudsen-cell technique: when working
with compounds that are only marginally stable in the presence
of surfaces, the loss to the walls may become a dominating effect
because of the efficiency of the gas-wall interaction.
1
9
Recent calculations by Lee show the existence of three
different BrNO isomers, trans-BrONO, cis-BrONO, and BrNO .
2
2
The most stable isomer, BrNO , has a calculated room-
2
-1
We have also carried out steady-state experiments on the
interaction of ClNO2 with bromide using NO2-selective LIF
temperature bond dissociation energy of 94.1 kJ mole , whereas
cis-BrONO has a bond dissociation energy of 67.4 kJ mole .
-
1

7478 J. Phys. Chem. A, Vol. 102, No. 38, 1998
Caloz et al.
TABLE 9: Summary of Reactions Studied in This Work
reaction
γ
∆Hr0298a
∆Hr0298b
ClNO
ClNO
BrNO
BrNO
2
+ KCl f Cl
+ KBr f BrCl + KNO
+ KCl f ClNO + KBr (10a)
+ KCl f BrCl + KNO (10b)
+ KBr f BrNO + KCl (4)
+ KBr f Br + KNO (3)
+ KNO f KCl + 2NO
+ KNO f KBr + 2NO
f 0.5 Br + NO (8)
2
+ KNO
2
no uptake
no uptake
no uptake
no uptake
54.3 (50.1)
26.0 (19.1)
4.9 (7.6)
30.9 (26.7)
-4.9 (-7.6)
4.3 (-2.6)
-13.1 (-8.9)
-8.2 (-1.3)
-2.0
2
2
2
2
-15.6 (-12.9)
10.4 (6.2)
15.6 (12.9)
2
2
-
4
4
ClNO
BrNO
ClNO
2
2
1.3 × 10
2
2
2
>0.3
-16.2 (-23.1)
-
2
2
2
(11)
(9)
3.0 × 10
BrNO
BrNO
2
2
2
no uptake
-28.7 (-21.8)
-22.5
2
2
2
a
Standard heats of reaction in kJ/mol for salt in the pure crystalline state and in aqueous solution at infinite dilution (values in brackets), using
0
b
0
∆
H
f
2 f 2
(BrNO ) ) 50.6 kJ/mol from ref 19. Same as footnote a, using ∆H (BrNO ) ) 71.1 kJ/mol from ref 8.
Therefore, one additional possibility to explain the difference
in the lifetime of BrNO2 measured in our and in Zetzsch’s
laboratory may be due to the production of a different isomer
as a primary product. Reactions 2, 4, and 6 may generate the
less stable cis-BrONO as the initial product, which decomposes
rapidly on our reactor walls, but isomerizes to BrNO2 under
high-pressure conditions used by Frenzel et al.8 However, this
kJ/mol which is hardly of significance. In addition, an
uncertainty in the calculated value of ∆Hf (BrNO2) may
invalidate the preference for the standard state of salts in their
aqueous solution.
0
The difference in the reactivity of the two molecules, BrNO2
and ClNO2, may be rationalized by considering the difference
in the oxidation state of the halogen atom, Br being in the
oxidation state (+1) whereas Cl is formally in the (-1) state.
Experimental evidence from this study supporting this hypoth-
esis includes (1) the relative instability of BrNO2 in the presence
of bromide, and (2) the observed hydrolysis products: ClNO2
is known to be the mixed anhydride of nitric and hydrochloric
+
hypothesis seems unlikely because only BrN fragments but
+
no BrO fragments, the latter indicating the presence of BrONO,
were observed in our product spectrum of reaction 6 not
withstanding the fact that isomerization of an unstable form of
BrNO2 may be even faster on a solid substrate.
2
0
A second difference between our results and the results of
acid, whereas BrNO2 hydrolyzes to HOBr and HONO.
HONO has been unambiguously observed in reaction 6 as well
8
Frenzel et al. is the reactivity of BrNO2 toward chloride. They
1
found that reaction 4 is reversible in solution and that the
equilibrium between ClNO2 and BrNO2 is established. This is
in disagreement with our results where reaction between BrNO2
and NaCl as well as on KCl has not been observed. The lack
of reactivity is in agreement with the endothermicity of the
elementary reaction 10a when the calculated standard heat of
formation for BrNO2 calculated by Lee19 is used (Table 9).
However, Frenzel et al. conceded that the interconversion
reaction may not be a direct halogen exchange but may proceed
via a complex reaction sequence involving Br2 and BrCl.
In summary, we have studied the heterogeneous reactions of
ClNO2 and BrNO2 with KCl, KBr, and KNO2 and have found
as reaction 2. The fact that hydrolysis has not been observed
in the reaction of BrNO2 with KCl (NaCl) (reaction 10) and on
NaNO3 may be due to HONO yields below our detection limit.
As stated above, the hydrolysis of BrNO2 is a slow process,
only 10% of the BrNO2 is undergoing hydrolysis on KNO2.
Considering the hygroscopic nature of KNO2, the hydrolysis
of BrNO2 may be faster on KNO2 compared to NaCl and
NaNO3.
Atmospheric Implications. Reactions 1 and 2 generate
photolyzable halogen-containing species which will ultimately
lead to the release of active halogen into the atmosphere. The
study shows that if ClNO2 is produced in the presence of
bromide, it can be converted to BrNO2, even if sea-salt aerosol
exists in a solid, albeit humid state. During the day ClNO2 is
them to be in agreement with the standard heats of reaction
0
∆
Hr 298 as displayed in Table 9, as expected. Column three of
Table 9 displays the thermochemistry using the calculated
mainly removed by photolysis; the photolysis rate constant kp
standard heat of formation of BrNO219 [∆Hf (BrNO2)] whereas
0
) 1.6 × 10
-4 -1 21,22
s
is several orders of magnitude higher
9
column four uses the value reported by Kreutter et al. which
than the heterogeneous removal rate constant. However, at
night, heterogeneous reactions may represent an efficient sink
for ClNO2. Any BrNO2 that is released as a result of a
heterogeneous reaction represents a source of photolyzable
bromine. The flux of photolyzable bromine from the sea-salt
aerosol may even be further enhanced if BrNO2 reacts with
available bromide to form Br2. Atmospheric modeling studies
are currently under way to assess the impact of these hetero-
geneous halogen sources on the tropospheric chemistry within
the marine boundary layer.
probably addresses the species BrONO rather than BrNO2 as
discussed by Frenzel et al.8 Owing to our inability to observe
0
reaction 10a, we support the lower value of ∆Hf (BrNO2)
calculated by Lee in agreement with the conclusions reached
by Frenzel et al. This is displayed in Table 9 where the lower
0
value of ∆Hf (BrNO2) leads to a positive heat of reaction as
0
opposed to a negative one when the higher value for ∆Hf -
(BrNO2) is used. The identical argument also applies to reaction
4
which is the inverse of reaction 10a and which has been
observed to occur under our conditions. The values for the heats
of reaction depend somewhat on whether one uses the standard
heats of formation for the salts in their pure crystalline state or
as aqueous solutions at infinite dilution. We prefer at this point
the solution values because they make reaction 3 slightly
exothermic in relation to the values for the salts in their solid
crystalline state. The solution values correspond more closely
to the concept of the quasi-liquid state of the interface even
though one has to exercize caution as the liquid layer is certainly
highly concentrated. However, when one chooses to use
standard heats of formation of concentrated aqueous salt
Because we carry out our experiments at very low total
pressures, thus at low partial pressures of water vapor, it is
necessary to comment on the role of humidity. In fact, because
the kinetics of water exchange with salt substrates at low
2
3
pressures is inefficient, our substrates may retain significant
quantities of adsorbed water. Recent experiments performed
by Finlayson-Pitts and co-workers have shown that adsorbed
water plays a critical role in relation to interfacial chemistry on
2
4,25
salt surfaces.
These studies point toward the existence of a
so-called quasi-liquid layer of concentrated salt solution, and
they suggest that the observed heterogeneous reactions take
place on or within a thin film of solution even on nominally
0
solutions, the values for ∆Hr 298 change by approximately 1

Reactivity of BrNO2 and ClNO2
J. Phys. Chem. A, Vol. 102, No. 38, 1998 7479
dry salt substrates. We have evidence that this layer is preserved
for many days even under our low-pressure conditions. For
example, when we heat a “dry” spray-deposited salt film in our
Knudsen reactor to 700 K, the integrated water flux from the
(4) Laux, J. M.; Hemminger, J. C.; Finlayson-Pitts, B. J. Geophys. Res.
Lett. 1994, 21, 1623.
(
5) Behnke, W.; Scheer, V.; Zetzsch, C. J. Aerosol Sci. 1993, S24,
S115.
(6) George, Ch.; Ponche, J. L.; Mirabel, P.; Behnke, W.; Scheer, V.;
Zetzsch, C. J. Phys. Chem. 1994, 98, 8780.
sample corresponds to tens of formal monolayers of surface-
adsorbed water.23 We never observed any influence of the
(7) Behnke, W.; George, Ch.; Scheer, V.; Zetzsch, C. J. Geophys. Res.
1
997, 102, 3795.
8) Frenzel, A.; Scheer, V.; Sikorski, R.; George, Ch.; Behnke, W.;
Zetzsch, C. J. Phys. Chem. A 1998, 102, 1329.
9) Kreutter, K. D.; Nicovich, J. M.; Wine, P. H. J. Phys. Chem. 1991,
5, 4020.
amount of water adsorbed on the salt surface on our initial
uptake coefficients. However, we observed the effect of
adsorbed water on the product formation: BrNO2 undergoes
hydrolysis on salt, forming HONO, and upon drying the salt,
the HONO yield may be reduced. Thus it seems that water is
not the limiting reagent and is abundantly available to form
localized regions of concentrated salt solutions upon or within
which interfacial reactions may take place.
(
(
9
3
1
1
5
(
10) Fenter, F. F.; Caloz, F.; Rossi, M. J. ReV. Sci. Instrum. 1997, 68,
180.
(11) Caloz, F.; Fenter, F. F.; Tabor, K. D.; Rossi, M. J. ReV. Sci. Instrum.
997, 68, 3172.
12) Ganske, J. A.; Berko, H. N.; Finlayson-Pitts, B. J. J. Geophys. Res.
992, 97 (D7), 7651.
13) Keyser, L. F.; Moore, S. B.; Leu, M.-T. J. Phys. Chem. 1991, 95,
496.
14) Caloz, F.; Fenter, F. F.; Rossi, M. J. J. Phys. Chem. 1996, 100,
7494.
(15) Mochida, M.; Akimoto, H.; van den Bergh, H.; Rossi, M. J. J. Phys.
Chem. 1998, 102, 4819.
(
(
Note Added in Proof. S. Fickert, F. Helleis, J. Adams, G.
K. Moortgat, and J. N. Crowley in their paper “Reactive Uptake
of ClNO2 on Aqueous Bromide Solutions”, submitted to this
Journal, reach similar conclusions resulting from the interaction
of ClNO2 with aqueous solutions of bromide.
(
(16) Koch, T.; Rossi, M. J. J. Phys. Chem. 1998, submitted.
(
17) George, C.; Behnke, W.; Scheer, V.; Zetzsch, C.; Magi, L.; Ponche,
Acknowledgment. This work was funded by a contract from
the Office F e´ d e´ ral de l’Education et de la Science (OFES) as
part of the SALT project carried out in the framework of the
EU Environment and Climate Program. We thank Professor H.
van den Bergh for fruitful discussions as well as for his lively
interest and support.
J. L.; Mirabel, P. Geophys. Res. Lett. 1995, 22, 1505.
(18) Scheffler, D.; Grothe, H.; Willner, H.; Frenzel, A.; Zetzsch, C.
Inorg. Chem. 1997, 36 (6), 335.
(19) Lee, T. J. J. Phys. Chem. 1996, 100, 19847.
(20) Hollemann A. F.; Wiberg, E. Lehrbuch der Anorganischen Chemie;
de Gruyter: Berlin, 1985; p. 605.
(21) Atkinson, R., et al. J. Phys. Chem. Ref. Data 1992, 21 (6).
(
22) Finlayson-Pitts, B. J.; Pitts, J. N. Atmospheric Chemistry; John
Wiley & Sons: New York, 1986.
23) Caloz, F. Thesis No. 1628, Swiss Federal Institute of Technology
References and Notes
(
(1) Fenter, F. F.; Caloz, F.; Rossi, M. J. J. Phys. Chem. 1996, 100,
(EPFL), 1997.
1
008.
(24) Beichert, P.; Finlayson-Pitts, B. J. J. Phys. Chem. 1996, 100, 15218.
(25) Allen, H. C.; Laux, J. M.; Vogt, R.; Finlayson-Pitts, B. J.;
Hemminger, J. C. J. Phys. Chem. 1996, 100, 6371.
(26) Seisel, S.; Caloz, F.; Fenter, F. F.; van den Bergh, H.; Rossi, M. J.
Geophys. Res. Lett. 1997, 24 (22), 2757.
(2) Livingston, F. E.; Finlayson-Pitts, B. J. Geophys. Res. Lett. 1991,
1
8, 17.
(3) Leu, M. T.; Timonen, R. S.; Keyser, L. F.; Yung, Y. L. J. Phys.
Chem. 1995, 99, (9), 13203.