Heterogeneous kinetics of the uptake of ClONO2 on NaCl and KBr

Article abstract of DOI:10.1021/jp953099i

The uptake kinetics of ClONO₂ on NaCl (reaction 1) and on KBr (reaction 2) have been studied in a low-pressure, Teflon-coated Knudsen reactor at room temperature. The initial uptake coefficient for both reactions has been measured as 0.23 ± 0.06 and 0.35 ± 0.06 for reactions 1 and 2, respectively, and is independent of reactant density in the range 10¹⁰-10¹³ molecules cm^-3. The values of the uptake coefficients are independent of presentation of the salt substrates: identical results are obtained on powder, grains, single-crystal surfaces, and thin deposited salt layers. The only product of reaction 1 is Cl₂. Reaction 2 initially produces Br₂, followed by BrCl and Cl₂. In our proposed mechanism, BrCl is the product of reaction 2, and secondary reactions between BrCl and the KBr substrate yield Br₂ at short reaction times and Cl₂ under prolonged exposure.

Full text of DOI:10.1021/jp953099i

7494
J. Phys. Chem. 1996, 100, 7494-7501
Heterogeneous Kinetics of the Uptake of ClONO2 on NaCl and KBr
Franc¸ois Caloz, Frederick F. Fenter, and Michel J. Rossi*
Laboratoire de Pollution Atmosphe´rique et Sol (LPAS), Swiss Federal Institute of Technology (EPFL), 1015
Lausanne, Switzerland
ReceiVed: October 20, 1995; In Final Form: February 15, 1996X
The uptake kinetics of ClONO2 on NaCl (reaction 1) and on KBr (reaction 2) have been studied in a low-
pressure, Teflon-coated Knudsen reactor at room temperature. The initial uptake coefficient for both reactions
has been measured as 0.23 ( 0.06 and 0.35 ( 0.06 for reactions 1 and 2, respectively, and is independent of
reactant density in the range 1010-1013 molecules cm-3. The values of the uptake coefficients are independent
of presentation of the salt substrates: identical results are obtained on powder, grains, single-crystal surfaces,
and thin deposited salt layers. The only product of reaction 1 is Cl2. Reaction 2 initially produces Br2,
followed by BrCl and Cl2. In our proposed mechanism, BrCl is the product of reaction 2, and secondary
reactions between BrCl and the KBr substrate yield Br2 at short reaction times and Cl2 under prolonged
exposure.
Introduction
line with the more qualitative conclusions of earlier studies by
Finlayson-Pitts and co-workers.16,18
The heterogeneous reactions of nitrogen oxides with salt are
coming under increased scrutiny, with the goal being to
understand the role they play in atmospheric chemistry. In
particular, nitrogen oxides such as N2O5, HNO3, and ClONO2
have been found to react with salt to produce gaseous chlorine-
containing species. If these products are photolyzed in sunlight,
then the chlorine atoms thus produced may affect the oxidative
potential of the local atmosphere.1,2 To this end, many efforts
are currently underway, including field studies,3,4 laboratory
measurements,2,5-10 and atmospheric modeling.11-15 Here, we
consider the reactions of gaseous chlorine nitrate with salt, which
are characterized by a displacement of halide for nitrate in the
condensed phase with a release of the corresponding molecular
halogen:
Here, we present our kinetic results for reactions 1 and 2,
obtained using a low-pressure flow reactor (Knudsen cell). We
include a study of the effects of the “internal” surface area of
salt samples, presented as powder, grain, single crystals, or
spray-deposited films, discussed in the frame of a model
proposed by Keyser and co-workers5,6,19-22 and in light of
previous work.2,9 By systematically varying the presentation
of the salt substrate, the reactivity of ClONO2 is studied as a
function of surface morphology.
Experimental Section
The experiments are carried out in a low-pressure, Teflon-
coated Knudsen reactor described elsewhere.2 The apparatus
consists of a gas-handling system, a Knudsen cell reactor, and
a vacuum chamber that houses the mass spectrometer (see Figure
1). Gaseous samples are prepared in the vacuum line and
injected into the reactor. From the pressure drop in a calibrated
volume, the gas flow (molecules s-1) is directly determined
using the ideal gas relation, and the mass-spectrometer signal,
for all species of interest, is calibrated against the known flow
rate. Flow as small as 5 × 1014 molecules s-1 can be directly
measured, whereas smaller flow rates are found by extrapolation
of the calibration curve. The Knudsen cell is a two-chamber
reactor in which the exposure of the reactive substrate is
controlled with a plunger (cf. Figure 1). In the absence of
reaction, the gas flowing into the cell subsequently leaves via
a small orifice into the vacuum chamber. Because the cell is
operated in the molecular flow regime, the loss through the
aperture is a first-order process with respect to the gas-phase
density. We experimentally determine the value of the rate of
effusive loss, kesc, for each species and for each aperture size.
The parameters for the Knudsen cell are summarized in Table
1. The molecules that effuse out of the Knudsen cell enter the
upper volume of a differentially pumped vacuum chamber. The
effusive beam is modulated by a tuning fork chopper located
at the orifice separating the two chambers and is subsequently
sampled by a quadrupole mass spectrometer (Balzers QMA
421). Phase-sensitive detection of the chopped signal using a
lock-in amplifier (SRS 830) eliminates background sources. The
amplified mass-spectrometer signal is proportional to the flow
of molecules exiting the Knudsen cell.
ClONO2 + NaCl f Cl2 + NaNO3
ClONO2 + KBr f BrCl + KNO3
(1)
(2)
In the atmosphere, the halogen will photolyze to produce
reactive chlorine and bromine atoms. It has been proposed that
the chlorine atoms liberated indirectly by the reactions of
nitrogen oxides with sea salt aerosol may provide a source of
atomic chlorine in the marine troposphere.16 This source may
be competitive with the homogeneous gas-phase reaction
between the OH radical and HCl:
HCl + OH f H2O + Cl
(3)
Also, the chlorine and bromine atoms liberated by reactions 1
and 2 may play a role in the catalytic loss of ozone after volcanic
injection of salt in the stratosphere, as suggested by in situ
measurements17 and modeling.14
Keyser and co-workers performed the first quantitative study
on reaction 1 at two different temperatures using a flow reactor
that significantly deviated from cylindrical symmetry.5 They
found Cl2 to be the sole product, and they determined the uptake
coefficient to be γ ) (4.6 ( 3.0) × 10-3 after having applied
a correction factor for internal diffusion. These results are in
* Author to whom correspondence should be addressed.
X Abstract published in AdVance ACS Abstracts, April 1, 1996.
S0022-3654(95)03099-1 CCC: $12.00 © 1996 American Chemical Society

Kinetics of the Uptake of ClONO2 on NaCl and KBr
J. Phys. Chem., Vol. 100, No. 18, 1996 7495
Figure 2. Low-dose steady-state experiments carried out on NaCl
powder substrate. The solid line shows the ClONO2 signal at m/e 46
(NO2+) and the dots give the Cl2 signal at m/e 70 (Cl2+). A large
decrease in ClONO2 signal is observed when the reactive salt surface
is exposed, corresponding to an uptake coefficient of 0.35. Calibration
of the signals reveals the Cl2 yield to be 85 ( 20%.
Figure 1. Scheme of the experimental setup. The upper part represents
the gas-handling system, including a reservoir for the gaseous reactant
(G and L), a calibrated volume (Vc), and a pressure gauge (P) for gas-
flow calibrations. The Knudsen cell is mounted on a differentially
pumped vacuum chamber, evacuated by two turbomolecular pumps
T1 and T2. The flux of gaseous species is measured by a mass
spectrometer (QMA) located in the lower of the two vacuum chambers.
TABLE 1: Knudsen Cell Parameters
cell parameter
value
volume
estimated surface area (total)
surface area (sample)
1830 cm3
1300 cm2
19.6 cm2
number density range
surface collision frequencyb
escape rate constant
(1-1000) × 1010 cm-3 a
67.6 s-1
for the 8-mm aperturec
for the 9-mm aperturec
for the 14-mm aperturec
0.80 × (T/M)1/2 s-1
1.03 × (T/M)1/2 s-1
1.77 × (T/M)1/2 s-1
Figure 3. Large-dose, steady-state experiment with KBr powder.
ClONO2 flow rate is monitored at m/e 46, Br2 at m/e 160, BrCl at m/e
116, and Cl2 at m/e 70.
a Calculated using the relation Fi ) Vkesc [M], where Fi is the flow
of molecules, V the reactor volume, and [M] the number density.
b Calculated for reactive surface of 19.6 cm2. c Determined directly by
experiment for ClONO2.
Since the measurement of an uptake rate depends upon the
precise knowledge of the escape rate constant (kesc), we
determine directly by experiment the true value of kesc for each
orifice. The agreement between calculated kesc (from gas-kinetic
theory) and the experimentally determined value is found to be
in good agreement for diameters under 10 mm; the discrepancy
is typically on the order of 20%, i.e., within the uncertainty
given by such parameters as reactor volume and true orifice
diameter. For larger orifices, the measured value of kesc becomes
significantly smaller than that determined from the gas-kinetic
expression. We suppose that this effect is due to a perturbation
of the molecular velocity distribution.23 In order to eliminate
the uncertainty due to this effect, kesc is measured directly using
a series of gases (He, O2, N2, SF6, etc.) for each orifice. The
formulas given in Table 1 are valid for all the gases studied.
The experiments described here are based on two distinct
experimental procedures, referred to as steady-state or pulsed-
ValVe experiments. A typical steady-state experiment is carried
out by (a) first isolating the surface sample by lowering the
plunger and setting a flow of ClONO2 and (b) lifting the plunger
to expose the reactive sample to the gas. We measure the first-
order rate of uptake, kuni, in terms of the rate of effusive loss
from the cell (kesc) and the change in mass-spectrometer
signal:2,24
Figure 4. Low-dose, steady-state experiment with KBr powder. The
solid line shows the ClONO2 signal at m/e 46 (NO2+), and the dots
give the Br2 signal at m/e 160 (Br2+). A large decrease in ClONO2
signal is observed when the reactive salt surface is exposed, corre-
sponding to an uptake coefficient of 0.34. Calibration of the signals
reveals the Br2 yield to be 110 ( 25%.
where Si and Sf correspond to the signals generated by the
effusive beam before and during the interaction between the
gas and substrate. Typical signals of steady-state experiments
can be seen in Figures 2-4.
A pulsed-valve experiment is carried out by using a solenoid
valve, through which gaseous reactant is introduced in mil-
lisecond pulses. The “control pulse” is produced while the solid
sample is still isolated by the plunger. The “reactive pulse” is
obtained by repeating the same operation with the plunger lifted.
When the heterogeneous reaction is first order with respect to
the gas-phase density, the observed rate of decay, keff, is simply
the sum of kesc and kuni and can be determined by exponential
fitting of the obtained signals. Additional kinetic information
is gained by characterizing the time-dependent “product pulse”,
S
kuni ) kesc i - 1
(
)
Sf

7496 J. Phys. Chem., Vol. 100, No. 18, 1996
Caloz et al.
where ω is the collision frequency with the reactive surface of
19.6 cm2 (Table 1) and 15.9 cm2. The collision frequency is
calculated using gas-kinetic theory.2 In the case of the 14-mm
orifice, for which the observed kesc is significantly smaller than
the gas-kinetic value, we expect a similar effect for the sample
collision rate. The true ω could be smaller than the calculated
gas-kinetic value used to calculate the γ’s for the experiments
performed using the 14-mm orifice; applying a correction would
only result in larger values than those reported here.
An important aspect of this work is the use of Teflon and
halocarbon wax coatings to render the glass and stainless steel
surfaces of the reactor inert. We have two means of testing for
an interaction between ClONO2 and the Teflon-coated walls of
the reactor. The first is the control experiment in which we
leave the sample dish empty and introduce ClONO2 under
specific conditions of flow and pressure. When the sample dish
is exposed, any reactivity with the Teflon coating becomes
evident as a change in the spectrometer signal. This experiment
is carried out frequently to verify the quality of the Teflon
coating of the sample dish; from these experiments, we
determined that the sticking probability of ClONO2 on the coated
sample holder is less than 0.01, i.e., much too small to interfere
with the kinetic measurements. We can also test for a gas-
wall interaction by simply making sure that the value of kesc
for ClONO2 scales correctly according to the expressions given
in Table 1 for each orifice. This was always the case, indicating
that any interaction between ClONO2 and Teflon is unobservable
under our experimental conditions.
Figure 5. Pulsed-valve experiment performed on powder NaCl sample.
This experiment was conducted using the 14-mm orifice. (a) By
integration of the “control pulse” (sample isolated), we determine that
the pulse contains 4 × 1014 molecules. By exponential fitting of the
decay, the effusive rate of loss is measured directly. Exponential fitting
of the “reactive pulse” (sample exposed) gives the first-order rate of
reactive uptake. In panel b, the simultaneous m/e 70 signal (Cl2+) is
displayed. From the integrations of the “reactive pulse” and the product
pulse, we calculate a yield of 110 ( 20% for this experiment.
Preparation of Gaseous Samples. The ClONO2 is synthe-
sized in our laboratory by adding Cl2O to an excess of N2O5,
according to the procedure described by Molina and co-
workers.25 The mixture is then stored in a dry ice/methanol
bath and subjected to several heating-cooling cycles (between
-78 and -25 °C) to accelerate the rate of reaction. The
resulting mixture of ClONO2 and N2O5 is distilled to yield the
final product, a yellow-green liquid that can be stored for
several weeks at dry ice temperature. The major impurity of
ClONO2 is chlorine, which is eliminated before each experiment
with a freeze-pump-thaw cycle. By monitoring the calibrated
signal at m/e 70 (Cl2+), we determine the Cl2 impurity to be
less than 1%. In addition, we detect trace quantities (,1%) of
HNO3 and HCl in the chlorine nitrate. Liquid Cl2O is obtained
by condensing Cl2 on an excess of yellow mercuric oxide (HgO)
and allowing the mixture to react at a temperature of -78 °C
for at least 12 h.26 Dinitrogen pentoxide (N2O5) is synthesized
by mixing NO2 with an excess of ozone, both dried by passage
through a P2O5 trap, and condensing the product in a dry ice/
methanol bath.9 Gaseous BrCl is obtained by mixing a small
amount of Br2 to an excess of Cl2 and allowing the mixture to
reach equilibrium
Figure 6. Pulsed-valve experiment on KBr powder. This experiment
was conducted using the 14-mm orifice. (a) “Control pulse” and
“reactive pulse” obtained as in Figure 5. (b) Simultaneous product
pulses of Br2 at m/e 160 and of BrCl at m/e 116 (expanded by a factor
of 10) are shown. The solid lines through the reactive pulse and through
both product traces are obtained by a simple model including only
reactions 2 and 4.
Br2 + Cl2 S 2BrCl
K
298 ) 7.6327
obtained by monitoring a product mass peak during the
introduction of the reactant. The real-time product profiles thus
obtained are useful for understanding (1) the mechanism of the
reaction, (2) the extent to which the product itself interacts with
the reactive substrate, and (3) the overall mass balance of the
reaction. Typical pulsed-valve signals are displayed in Figures
5 and 6. The utility of these experiments is illustrated in the
Results section.
Our working mixtures typically contained 80% Cl2, 18% BrCl,
and 2% Br2.
Table 6 shows the principal peaks in the mass spectra of the
gaseous samples used in the course of this work.
Preparation of Solid Samples. For this study, we used
commercial NaCl (Fluka puriss. p.a.) and KBr (Fluka puriss.
p.a. ACS). Because of the recent debate concerning the role
of diffusion into the interstitial spaces of powder and grain
samples, we have studied the reactivity of a range of samples,
providing a large variation of the total exposed surface area.
Salts are known to be nonporous materials; other laboratories
have found that, for example, the BET surface area of salt grain
corresponds to the total geometrical surface area.5 In order to
The values of γ listed in Tables 2-5 are calculated for
experimentally determined kuni using the following expression:
kuni
γ )
ω

Kinetics of the Uptake of ClONO2 on NaCl and KBr
J. Phys. Chem., Vol. 100, No. 18, 1996 7497
TABLE 2: (a) Steady-State, (b) Pulsed-Valve, and (c) Yielda Experiments on NaClb
salt/type
mass, g flow, molecules/s dose, molecules orifice, mm kesc, s-1 sample diam, mm ω, s-1 kuni, s-1
γ
yield
Section a
NaCl/powder
NaCl/powder
NaCl/powder
NaCl/powder
NaCl/powder
NaCl/powder
NaCl/powder
NaCl/powder
NaCl/powder
NaCl/powder
NaCl/powder
NaCl/powder
NaCl/window
NaCl/window
NaCl/window
NaCl/window
NaCl/window
NaCl/grain
1
1
1
1.3
1.2
1
8.00E+13
2.50E+14
8.00E+14
8.50E+14
1.00E+14
2.20E+14
1.35E+15
1.00E+14
5.00E+14
1.00E+14
4.90E+14
1.40E+14
1.00E+15
5.00E+14
2.00E+13
3.00E+13
1.30E+14
7.00E+14
6.50E+14
7.00E+14
8.00E+14
6.00E+13
1.50E+14
1.20E+14
1.00E+13
9
9
9
9
8
14
14
14
14
14
14
14
9
9
9
9
14
9
9
9
9
9
1.8
1.8
1.8
1.8
1.4
3.0
3.0
3.0
3.0
3.0
3.0
3.0
1.8
1.8
1.8
1.8
3.0
1.8
1.8
1.8
1.8
1.8
1.8
3.0
3.1
50
50
50
50
50
50
50
50
50
50
50
50
45
45
45
45
45
50
50
50
50
50
50
50
50
67.7
67.7
67.7
67.7
67.7
67.7
67.7
67.7
67.7
67.7
67.7
67.7
54.9
54.9
54.9
54.9
54.9
67.7
67.7
67.7
67.7
67.7
67.7
67.7
67.7
16.2
16.2
17.1
17.1
12.5
24.0
20.2
25.5
11.4
19.8
16.5
21.6
9.4
0.24 0.63
0.24 0.81
0.25 1.01
0.25 1.05
0.18
0.35 0.64
0.30 0.92
0.38 0.72
0.17 1.02
0.29 1.10
0.24 1.02
0.32 1.02
0.17
0.24
0.21
0.25 0.40
0.24 0.41
0.14
0.15
0.16
0.15
0.21
1
1.2
1.2
1.1
1.1
1.2
13.3
11.3
13.7
12.9
9.5
1.3
3.2
5.4
12
2.5
2.6
0.004
0.005
NaCl/grain
NaCl/grain
NaCl/grain
NaCl/grain
NaCl/grain
NaCl/spray
NaCl/spray
9.8
10.8
10.4
14.4
12.1
17.6
20.3
9
14
14
0.18 0.69
0.26 1.00
0.30
Section b
NaCl/powder
NaCl/powder
NaCl/powder
NaCl/powder
1.3
1.5
1.5
1.2
4.00E+14
2.80E+14
3.00E+14
1.50E+14
14
14
8
3.2
3.0
1.4
3.0
50
50
50
50
67.7
67.7
67.7
67.7
14.0
18.0
12.6
16.3
0.21
0.27
0.19
0.24 1.00
14
Section c
NaCl/powder
NaCl/powder
NaCl/powder
NaCl/window
NaCl/powder
1
1
1
1.60E+15
6.40E+14
2.80E+15
4.80E+13
8.00E+14
9
9
9
9
14
0.97
1.05
1.04
0.63
1.08
1.3
a These experiments are designed to calculate the yield of reaction 1. Surfaces are partially saturated to make sure steady-state conditions are
established. b The notation E+n represents × 10+n in all tables.
TABLE 3: (a) Steady-State and (b) Pulsed-Valve Experiments on KBr
salt/type
mass, g flow, molecules/s dose, molecules orifice, mm kesc, s-1 sample diam, mm ω, s-1 kuni, s-1
γ
yield
KBr/powder
KBr/powder
KBr/powder
KBr/powder
KBr/powder
KBr/powder
KBr/powder
KBr/powder
KBr/window
KBr/window
KBr/grain
1
1
1
2
2
2
2
2
9.00E+14
5.70E+14
5.00E+14
5.50E+14
3.60E+14
2.80E+15
4.00E+14
1.00E+14
6.50E+13
1.00E+14
1.90E+15
2.00E+13
9
9
9
1.8
1.8
1.8
1.4
3.0
3.0
3.1
3.0
1.8
3.0
1.4
3.1
50
50
50
50
50
50
50
50
50
50
50
50
67.7
67.7
67.7
67.7
67.7
67.7
67.7
67.7
67.7
67.7
67.7
67.7
23.4
22.3
23.2
21.1
22.8
20.6
25.0
24.9
21.7
35.7
24.6
27.0
0.35 0.86
0.33 0.9
0.34 0.88
0.31 1.08
0.34 0.9
0.30
8
14
14
14
14
9
14
8
0.37
0.37 1.14
0.32 0.8
0.53 0.84
0.36
2
KBr/spray
0.005
14
0.40 1.1
Section b
KBr/powder
KBr/spray
KBr/powder
1.2
0.004
1.1
3.80E+14
1.10E+14
2.00E+14
14
14
14
3.1
3.0
3.1
50
50
50
67.7
67.7
67.7
15.8
21.7
21.7
0.24 0.8
0.32 0.7
0.32 1.05
test the influence of total external surface area on the reactivity
of our systems, we carry out experiments on powders, single-
crystal optical flats, submillimeter grain samples, and spray-
deposited samples. Powders are obtained by grinding salt in a
ball mill and then sieving to isolate a size fraction. Typically,
we collect powder with grain size between 35 and 100 µm. The
total external surface of these samples is large and can be
estimated based on electron-microscope images. When the
sieving procedure is carried out for salt grain (diameters in the
range of 350-500 µm) a monodisperse fraction is obtained with
a large and quantifiable total exposed surface area. This is the
TABLE 4: Experimental Results: NaCl + ClONO2
substrate
expt type
no. of expts
uptake γ
powder
powder
window
grain (350 µm)
spray deposited
steady state
pulsed valve
steady state
steady state
steady state
12
4
5
6
2
0.27 ( 0.07
0.23 ( 0.05
0.22 ( 0.04
0.16 ( 0.04
0.28 ( 0.1
0.23 ( 0.06 av
ideal material for testing the role of diffusion in the internal
void of the solid substrates.9

7498 J. Phys. Chem., Vol. 100, No. 18, 1996
Caloz et al.
TABLE 5: Experimental Results: KBr + ClONO2
substrate
expt type
no. of expts
uptake γ
powder
powder
window
grain (350 µm)
spray deposited
spray deposited
steady state
pulsed valve
steady state
steady state
steady state
pulsed valve
8
2
2
1
1
1
0.33 ( 0.03
0.28 ( 0.1
0.42 ( 0.12
0.36 ( 0.1
0.40 ( 0.1
0.32 ( 0.1
0.35 ( 0.06 av
TABLE 6: Mass-Spectral Data
species
peak 1a
peak 2a
peak 3a
sensitivityb
ClONO2
Cl2
Br2
46 (100)
70 (100)
160 (100)
116 (100)
51 (14)
35 (33)
79 (25)
35 (28)
35 (7)
1.255
0.8
2.3
Figure 7. Steady-state experiments performed on NaCl grain. Sieved
salt has the average dimension of 350 µm and is held in a 5-cm-diameter
dish. Varying the sample mass from 1.2 to 12 g changes the number
of grain layers from 1 to 10. The solid curve shows a fit to our data
with the model developed by Keyser and co-workers. The ClONO2/
NaCl reaction is independent of sample mass over the range of values
studied.
81 (25)
79 (15)
BrCl
2.5
a Ion mass with peak intensity in parentheses, given as percent of
base peak. b Sensitivity given in 1015 molecules/V, measured at the base
peak.
With the single-crystal and spray-deposited samples, we study
the reactivity in the absence of the interstitial void present in
the grains and powders. Spectroscopy-grade salt windows (Dr.
Karl Korth, φ ) 500 mm for KBr and φ ) 45 mm for NaCl)
are typically treated by gliding a sheet of wet optical paper
across the surface, followed by rinsing with methanol and drying
(corresponding to wet-etched surfaces). The surfaces are
alternatively prepared by rubbing with fine-grained sandpaper
and subsequent blowing off of the surface powder. Spray-coated
optical flats are obtained by vaporizing a saturated salt solution
on a heated glass support. Methanol is used as the solvent,
and we obtain surface coverage approaching 100%. By
depositing only 1-5 mg of salt, the obtained salt film is smooth
and nonporous and has a thickness of less than 2 µm. The
spray-deposited surfaces are characterized in more detail in
previous work.9
signals are calibrated to monitor ClONO2 and Cl2, respectively,
during the milliseconds following the injection of ClONO2 into
the reactor. In Figure 5, we show the results of a typical run.
The “control pulse” shows the effusive loss of the ClONO2 from
the cell when the chloride is still isolated behind the sample
plunger. The “reactive pulse” is actuated under exactly the same
experimental conditions but with the NaCl sample exposed to
the reactor volume. The loss of ClONO2 due to reaction 1 is
exponential, and a rate for kuni can be determined directly by a
least-squares fitting of the experimental trace. From these
experiments, we confirm our earlier conclusion that the uptake
is a first-order process with respect to loss of gaseous ClONO2.
The value of the uptake coefficient determined from these
experiments, γ ) 0.23 ( 0.05, is in excellent agreement with
the value determined by the steady-state experiments. In Figure
5, it can be seen that the Cl2 signal (m/e 70) rises rapidly and
then decays according to the rate of effusive loss of chlorine
from the reactor. This demonstrates that there is no interaction
between the Cl2 formed as a product and the NaCl substrate.
In an additional series of experiments, we systematically
varied the total exposed surface area of the salt substrates to
see if this parameter plays a role in the observed rate of ClONO2
uptake. This dependence is most easily studied in a quantitative
manner by varying the number of layers of a monodisperse salt
grain within the sample holder. As the mass of the sample
increases, so does the interstitial void, and if diffusion into this
void is fast, the rate of uptake will increase with the extent to
which the sample holder is filled. In Figure 7, we see that there
is no variation in the initial rate of uptake of ClONO2 as the
sample mass of the monodisperse grain is varied by a factor of
10. In the other extreme of surface presentation, the internal
void can be eliminated by studying the reaction with salt single
crystals and with spray-deposited surfaces. The results of all
the experiments are listed in Table 2 and summarized in Table
4, where it is seen that all of the chloride substrates yield the
same uptake probability, within the uncertainties of the experi-
mental measurements. In a later section, we briefly discuss the
implications of this finding for the application of correction
factors to heterogeneous reaction rate constants on “porous”
substrates.
In general, the samples are pumped overnight after introduc-
tion into the Knudsen cell to eliminate residual water.
Results and Discussion
A. ClONO2 + NaCl f NaNO3 + Cl2. We investigated
reaction 1 by a series of steady-state experiments on powder
samples. Although most of the experiments are conducted in
the limit of small flow rates (to avoid saturation of the uptake),
a systematic study of the uptake rate dependence on the ClONO2
flow rate was also carried out. For all the experiments described
here, chlorine nitrate is monitored at the m/e 46 base peak
(NO2+). When a chloride sample is exposed to the ClONO2
flow, the rapid loss of signal at m/e 46 with new intensity at
m/e 35, 37, 70, 72, and 74 is observed (cf. Figure 2). From
these experiments, we can draw the following conclusions: (a)
Cl2 is the only product of the reaction, and it is formed with a
110 ( 15% yield, where the estimated error is based on the
uncertainties associated with the ClONO2 and Cl2 flow calibra-
tions. (b) The rate of uptake is independent of the flow of
ClONO2, confirming a first-order rate law for the reaction. (c)
The uptake coefficient, determined as the average of all the
experiments, is γ ) 0.27 ( 0.06, where the estimated error is
the 1σ statistical limit of the mean. As with all our kinetic
results described here, we do not include an estimate of
systematic error in the reported uncertainties of the rate constants
and uptake coefficients due to parameters such as surface
roughness; these may represent a source of systematic uncer-
tainty but are difficult to assess for surface reactions that occur
with very high sticking probability as measured in this work.
Our conclusions concerning reaction 1 are confirmed by the
results of pulsed-valve experiments. In these, m/e 46 and 70
In addition, reaction 1 was studied as a function of gas density
by systematically varying the gas-flow and escape orifice for
each type of substrate. In Figure 8, we see that the results are
independent of the gas density and of the surface presentation,
thus confirming the first-order rate law for reactive uptake on
reaction 1.

Kinetics of the Uptake of ClONO2 on NaCl and KBr
J. Phys. Chem., Vol. 100, No. 18, 1996 7499
Figure 8. Summary of all the steady-state experiments as a function
of flow rate, orifice size, and NaCl sample presentation for reaction 1:
(a) powder substrate, 9-mm orifice; (b) powder substrate, 14-mm orifice;
(c) window substrate, 9-mm orifice; (d) window substrate, 14-mm
orifice; (e) grain substrates, 9-mm orifice; (f) spray-deposited substrates,
14-mm orifice.
Figure 9. Summary of all the steady-state experiments as a function
of flow rate, orifice size, and KBr sample presentation for reaction 2:
(a) powder substrate, 9-mm orifice; (b) powder substrate, 14-mm orifice;
(c) window substrate, 9-mm orifice; (d) window substrate, 14-mm
orifice; (e) grain substrates, 9-mm orifice; (f) spray-deposited substrates,
14-mm orifice.
ClONO2 (m/e 46), BrCl (m/e 116), and Br2 (m/e 160) are
monitored in real time (in sequential experiments). The decay
of the ClONO2 signal corresponds to an uptake probability of
0.30 ( 0.1, in agreement with the value we determined using
the steady-state technique. In these experiments, a short-lived
BrCl signal is observed. The first-order rate constant for the
observed experimental decay of BrCl (10-15 s-1) is much
B. ClONO2 + KBr f products. As with the chloride
reaction, we began this study with a series of steady-state
experiments on powder substrates. On the basis of our
observations of reaction 1, we expected BrCl to be the main or
even the sole product of reaction 2. When the bromide powder
is exposed to the gas flow, a strong uptake of ClONO2 (m/e
46) is observed, but initially, no BrCl is formed (monitored at
m/e 116, BrCl+). In fact, the only new intensity observed in
the product spectrum is found at m/e 79 and 81 (Br+) and m/e
158, 160, and 162 (Br2+). After prolonged exposure to the
reactant, the molecular bromine signal decreases, and BrCl and
Cl2 are produced in small quantities. A typical time-dependent
mass scan is displayed in Figure 3 at a flow rate of ClONO2 of
2.8 × 1015 molecules/s. If BrCl is formed as the direct product,
it is itself reactive with respect to the bromide to yield Br2 and
with chloride to give Cl2. As with reaction 1, we carried out
most of the experiments at low flow rates of less than 2 × 1014
molecules/s to avoid “saturation” of the substrate (Figure 4).
Under these conditions, we find the uptake coefficient to be γ
) 0.34 ( 0.03, with molecular bromine as the product with
100 ( 15% yield, where the uncertainty is given by the
imprecision of the relevant mass-flow calibrations.
greater than the rate of effusive loss from the reactor (kesc
)
2.9 s-1). The Br2 signal decay, in contrast, is well described
by the first-order effusion rate. These observations indicate that
the BrCl reacts with the substrate to form Br2 and that Br2 does
not interact with the bromide. Indeed, a kinetic simulation using
only reactions 2 and 4 is capable of reproducing the experimental
observations, as shown in Figure 6. The model gives the best
agreement with the experimental data if the BrCl yield for
reaction 2 is set to 1.0 ( 0.2.
All our observations are consistent with the mechanism in
which the ClONO2 reacts with bromide to produce BrCl in
100% yield, with a second reaction between BrCl and bromide
responsible for the Br2 formation. When enough chloride has
accumulated on the surface of the salt substrate, it reacts directly
with gaseous ClONO2 to form the small amount of Cl2 observed
at long reaction times:
In a series of ancillary steady-state experiments, using
authentic samples of BrCl and Cl2, we confirm that the reactions
of BrCl and Cl2 with bromide are fast:
ClONO2(g) + KBr(s) f BrCl(g) + KNO3(s)
BrCl(g) + KBr(s) f Br2(g) + KCl(s)
ClONO2(g) + KCl(s) f Cl2(g) + KNO3(s)
Cl2(g) + KBr(s) f BrCl(g) + KCl(s)
(2)
(4)
(6)
(5)
BrCl(g) + KBr(s) f Br2(g) + KCl(s)
Cl2(g) + KBr(s) f BrCl(g) + KCl(s)
(4)
(5)
Reactions 4 and 5 are characterized by γ > 0.1 and have unitary
yields for the given stoichiometries. In the steady-state experi-
ments described above (Figure 4), the absence of a BrCl signal
is not surprising given the rapid conversion to Br2 in the
presence of bromide.
In Figure 3, the formation of BrCl and Cl2 at long exposure
times is easily interpreted in terms of the mechanism detailed
above.
Additional experiments were carried out on spectroscopy-
grade bromide windows and on spray-deposited samples. As
with the study of reaction 1, the reactivity was also investigated
as a function of the ClONO2 flow rate. The results on reaction
2 are listed in Table 3 and summarized in Table 5; in Figure 9,
it can be seen that the rate of uptake remains independent of
these variables over the range of experimental conditions
studied.
We were able to gain additional insight into the kinetics of
reaction 2 by following the real-time evolution of the reaction
after the pulsed-valve injection of ClONO2 into the reactor.
Typical experimental traces are shown in Figure 6, in which
C. Surface Morphology Dependence. With solid sub-
strates, and in particular with powders and grains, it is necessary
to consider the extent to which the total exposed surface area
and morphology has an influence on the observed rate of uptake
of a gaseous species. Recently, Keyser and co-workers20 have
proposed a model, developed originally by Wheeler and
others,28,29 in which the overall rate of uptake can be separated
into diffusional and reactive components. Based on the proper-
ties of the substrate and assuming Knudsen diffusion into the
interstitial void of the grain or powder, correction factors are
calculated and applied to experimentally determined values of
the uptake coefficient to obtain the reaction probability. This

7500 J. Phys. Chem., Vol. 100, No. 18, 1996
Caloz et al.
strat. density,c
TABLE 7: Comparison with Similar Reactions
approach is strengthened by the direct experimental character-
ization of the appropriate substrate, for example, by scanning
electron microscopy.
species
γNaCl
JNaCl,b s-1
molecules cm-3
a
<10-7
<1.4 × 10-10
2.3 × 10-5
3.7 × 10-4
1.7 × 10-3
2 × 109
5 × 108
2 × 109
7 × 108
The need to apply these correction factors is not universally
accepted. Hanson and Ravishankara, for example, have shown
that the rate of uptake of N2O5 and ClONO2 on ice is
independent or only weakly dependent of the thickness of the
ice coating applied to the walls of their flow-tube reactor under
their experimental conditions.30,31 As pointed out by Keyser
et al., the reactions that they chose to study are too rapid to
offer a sensitive test of the applicability of the theory.22 From
our previous observations, we put forward the hypothesis that
certain “sticky” molecules, such as HNO3, do not diffuse into
the interstitial void of a bulk sample at the gas-kinetic Knudsen
rate, so the diffusional contribution to the overall reactivity is
smaller than that predicted by the theory proposed by Keyser
et al. For other species, notably N2O5, the agreement between
the predictions of the theory and the experimental observations
is remarkably good.
NO2
N2O5
HNO3
ClONO2
(5 ( 2) × 10-4
(2.8 ( 0.3) × 10-2
0.24 ( 0.06
a Values for NO2 taken from unpublished results. Alternatives values
for N2O5 (<1 × 10-4), ClONO2 ((4.6 ( 3.0) × 10-3), and HNO3 ((1.3
( 0.4) × 10-2) can be found in ref 6. b Taken from the estimation of
ref 13 for post El-Chichon stratospheric salt injection. c Densities at
28 km taken from ref 31.
reaction, this would have the effect of lowering the observed
decay rate relative to its true value.
E. Atmospheric Importance. Chlorine nitrate has been
predicted to be present in the marine troposphere in pptv
concentrations,11 where it can react with salt-containing marine
aerosols according to reactions 1 and 2. Furthermore, ClONO2
is known to be present in the stratosphere, where it can come
into contact with salt aerosol after volcanic eruptions.17 The
reactions studied here are potentially of consequence to atmo-
spheric chemistry because of the photolyzable molecular halides
produced in reactions 1 and 2. Chlorine atoms are known to
play a role in the chemistry of the troposphere, where they
initiate hydrocarbon oxidation with great efficiency. The role
played by stratospheric chlorine atoms in ozone destruction has
been well elucidated.
For the two reactions studied here, the correction factors
calculated by the theory are small because the reaction prob-
ability is large enough so that diffusion represents only a minor
perturbation to the rate of overall uptake. For this case, the
correction applied would be a function of the total surface
exposed due to roughness or particle size. For example, for a
surface covered by half-spheres, the correction factor would be
about a factor of 2, approximately corresponding to the statistical
variability of the uptake coefficient measured (Figures 8 and
9). The curve drawn in Figure 7 shows the prediction of the
theory as a function of sample mass for grain samples of average
size of 350 µm. Because the uptake probability is so large, no
variation in the overall uptake is expected over the range of
experimental conditions studied. This reaction therefore does
not provide a sensitive test of the theory proposed by Keyser
and co-workers. Within the uncertainty of the measurements,
all the experiments on all substrates yield the same rate constant
for the uptake; this is somewhat surprising given the large
variation of surfaces employed and may indicate that ClONO2
is a species that, like HNO3, should be categorized as sticky.
On the basis of these observations and on our previous
experience with the reactions of HNO3 and N2O5 with salt, we
prefer to report our experimental values without applying the
corresponding correction factor.
The impact of reactions 1 and 2 should be considered within
a set of possible atmospheric reactions involving salts and
nitrogen oxides. Taking the example of the NaCl reactions,
ClONO2(g) + NaCl(s) f Cl2(g) + NaNO3(s)
2NO2(g) + NaCl(s) f ClNO(g) + NaNO3(s)
(1)
(7)
N2O5(g) + NaCl(s) f ClNO2(g) + NaNO3(s) (8)
HNO3(g) + NaCl(s) f HCl(g) + NaNO3(s) (9)
All these reactions have in common the loss of a nitrogen oxide
from the gas phase and the release of a volatile chlorine-
containing compound. Reactions 1, 7, and 8 produce photo-
lyzable molecules which will directly release chlorine atoms in
the atmosphere and will participate in important atmospheric
oxidation cycles.
D. Comparison with Previous Work. The first study of
reactions 1 and 2, by Finlayson-Pitts and co-workers,16,18
provided only qualitative details of the reaction kinetics. Their
observations, including the shift in the product spectrum at long
reaction times, are in accord with our own.
To illustrate the potential role of reaction 1 in the atmosphere,
in Table 7 we list some nitrogen oxides, the values of the
corresponding uptake coefficients with NaCl, and some typical
atmospheric concentrations. In addition, we include estimates
for the “rate of loss” of each species after volcanic injection of
salt into the stratosphere; these values are obtained according
to the approximation described in Michelangeli et al.13 The
rate of loss of ClONO2 in the presence of volcanic salt is seen
to be competitive with the rate of loss by photolysis calculated
by Michelangeli et al. (Jphot ) 1.8 × 10-4 s-1). In the table, it
is clear that the very rapid reaction between ClONO2 and NaCl
might play an important role in the atmosphere wherever
particulate salt comes into contact with ClONO2. The nature
of this influence can only be properly assessed by a complete
atmospheric model.33
The only previous quantitatiVe study, by Timonen et al.,5
reports measurements of the uptake probability at two temper-
atures. The experiment was conducted using a flow tube
coupled to a quadrupole mass spectrometer. Although the
product analysis is in good accord with ours, their experimentally-
determined rates are a factor of 7-8 smaller than those we
observed. After correcting their data for the effects of diffusion
into the interstitial spaces of their granular sample, they conclude
that the true reactive uptake coefficient is about 50 times smaller
than the value we report here for the same reaction. The reason
for this discrepancy is not evident; one contributing factor may
be that they neglected radial gas-phase diffusion in their data
analysis. According to the authors calculations, the maximum
effect due to radial diffusion is on the order of 10% for their
experimental conditions; it has, however, been pointed out by
others32 that the effect is difficult to assess for noncylindrical
flow reactors and may in fact be much greater. For a fast
Because the uptake experiments were carried out at low
water-vapor concentration and at room temperature, the results
must be considered as an indication of the relative atmospheric
importance of reactions 1, 7, 8, and 9. Since salt is known not
to interact with water vapor until the relative humidity has

Kinetics of the Uptake of ClONO2 on NaCl and KBr
J. Phys. Chem., Vol. 100, No. 18, 1996 7501
(8) Behnke, W.; Zetzsch, C. J. Aerosol Sci. 1990, 21, s229.
(9) Fenter, F. F.; Caloz, F.; Rossi, M. J. J. Phys. Chem. 1996, 100,
1008.
(10) Behnke, W.; Kru¨ger, H.-U.; Scheer, V.; Zetzsch, C. Proceedings
of EUROTRAC Symposium, 1992; SPB Academic: The Hague, The
Netherlands, 1992; p 565.
approached or exceeded the deliquescence point, the relative
reactivities presented in Table 7 are valid for the stratosphere
and for the dry troposphere.
Conclusions
(11) Singh, H. B.; Kasting, J. F. J. Atmos. Chem. 1988, 7, 261.
(12) Parrish, D. D.; Hahn, C. J.; Williams, E. J.; Norton, R. B.;
Fehsenfeld, F. C.; Singh, H. B.; Shetter, J. D.; Gandrud, B. W.; Ridley, B.
A. J. Geophys. Res. 1993, 98, 14995.
(13) Michelangeli, D. V.; Allen, M.; Yung, Y. L. Geoph. Res. Lett. 1991,
18, 673.
(14) Michelangeli, D. V.; Allen, M.; Yung, Y. L. Geophys. Res. 1989,
94, 18429.
(15) Parrish, D. D.; Buhr, M. P. AdV. Chem. Ser. 1993, 243.
(16) Finlayson-Pitts, B. J.; Ezell, M. J.; Pitts, J. N., Jr. Nature 1989,
337, 241.
(17) Woods, D. C.; Chuan, R. L.; Rose, W. I. Science 1985, 230, 170.
(18) Berko, H. N.; McCaslin, P. C.; Finlayson-Pitts, B. J. J. Phys. Chem.
1991, 95, 6951.
(19) Chu, L. T.; Leu, M. T.; Keyser, L. F. J. Phys. Chem. 1993, 97,
12798.
(20) Keyser, L. F.; Leu, M. T. J. Coll. Int. Sci. 1993, 155, 137.
(21) Chu, L. T.; Leu, M. T.; Keyser, L. F. J. Phys. Chem. 1993, 97,
7779.
(22) Keyser, L. F.; Leu, M. T.; Moore, S. B. J. Phys. Chem. 1993, 97,
2800.
Our major conclusions can be summarized in the following
points: (1) ClONO2 is taken up with very large probability by
both chlorides and bromides The values for γ that we determine
are roughly a factor of 50 greater than those reported in the
only previous quantitative study of the reactions. The sole
product of the reaction with chloride is molecular chlorine,
obtained in 100% yield. Although, under certain experimental
conditions, the apparent product of the ClONO2 reaction with
bromide is Br2, pulsed-valve experiments reveal that BrCl is
formed as the primary product and that BrCl is itself extremely
reactive with respect to bromide. (2) There is no observable
effect of the surface presentation on the overall uptake kinetics.
Because of this, we do not apply the correction factor suggested
by the treatment proposed by Keyser and co-workers. (3) The
very efficient uptake of ClONO2 by chlorides and bromides
should be taken into account in chemical modeling of the
atmosphere wherever particulate salt is present.
(23) Motz, H.; Wise, H. J. J. Chem. Phys. 1960, 32, 1893.
(24) Tabor, K.; Gutzwiller, L.; Rossi, M. J. J. Phys. Chem. 1994, 98,
6172.
(25) Molina, J. A.; Spencer, J. E.; Johnson, S. N. J. Chem. Phys. Lett.
1977, 45, 158.
(26) Schack, C. J.; Lindahl, C. B. Inorg. Nucl. Chem. Lett. 1967, 3,
387.
(27) Stull, D. R.; Prophet, C. B. JANAF Thermochemical Tables,
Publication NSRDS-NBS 37, 2nd ed.; National Bureau of Standards:
Washington, DC, 1971.
Acknowledgment. Funding for this work was provided by
the Office Fe´de´ral de l’Education et de la Science (OFES) as
part of the HALIPP subproject of the EUROTRAC program
and by the Fonds National de la Recherche Scientifique (FNRS)
under Grant 20-37599.93. We thank Professor Hubert van den
Bergh for his lively interest and input.
References and Notes
(28) Wheeler, A. AdV. Catal. 1951, 3, 249.
(29) Aris, R. The Mathematical Theory of Diffusion and Reaction in
Permeable Catalysts; Clarendon: Oxford, 1975; Vol. I.
(30) Hanson, D. R.; Ravishankara, A. R. J. Phys. Chem. 1992, 96, 2682.
(31) Hanson, D. R.; Ravishankara, A. R. J. Phys. Chem. 1993, 97, 2802.
(32) Hanson, D. R.; Ravishankara, A. R. J. Phys. Chem. 1993, 97, 12309.
(33) DeMore, W. B.; Sander, S. P.; Golden, D. M.; Hampson, R. F.;
Kurylo, M. J.; Howard, C. J.; Ravishankara, A. R.; Kolb, C. E.; Molina,
M. J. Chemical Kinetics and Photochemical Data for Use in Stratospheric
Modeling, EValuation Number 11; JPL Publication 94-26; NASA: Wash-
ington, DC, 1994.
(1) Finlayson-Pitts, B. J.; Pitts, J. N. Atmospheric Chemistry; John
Wiley & Sons: New York, 1986.
(2) Fenter, F. F.; Caloz, F.; Rossi, M. J. J. Phys. Chem. 1994, 98, 9801.
(3) Harrison, R. M.; Msibi, M. I.; Kitto, A. M. N.; Yamulki, S. Atmos.
EnViron. 1994, 28, 1593.
(4) Platt, U.; Hausmann, M. Res. Chem. Intermed. 1994, 20, 557.
(5) Timonen, R. S.; Chu, L. T.; Leu, M. T.; Keyser, L. F. J. Phys.
Chem. 1994, 98, 9509.
(6) Leu, M.-T.; Timonen, R. S.; Keyser, L. F. Yung, Y. L. J. Phys.
Chem. 1995, 99, 13203.
(7) Behnke, W.; Kru¨ger, H.-U.; Scheer, V.; Zetzsch, C. Proc. CEC/
EUROTRAC 1991, 1.
JP953099I