Kinetics of ClONO2 reactive uptake on ice surfaces at temperatures of the upper troposphere

Article abstract of DOI:10.1021/jp053477b

The reactive uptake kinetics of ClONO₂ on pure and doped water-ice surfaces have been studied using a coated wall flow tube reactor coupled to an electron impact mass spectrometer. Experiments have been conducted on frozen film ice surfaces in the temperature range 208-228 K with P _ClONO2 ≤ 10^-6 Torr. The uptake coefficient (γy) of ClONO₂ on pure ice was time dependent with a maximum value of γ_max ～0.1. On HNO₃-doped ice at 218 K the γ_max was 0.02. HOCl formation was detected in both experiments. On HCl-doped ice, uptake was gas-phase diffusion limited (γ > 0.1) and gas-phase Cl₂ was formed. The uptake of HCl on ice continuously doped with HNO₃ was reversible such that there was no net uptake of HCl once the equilibrium surface coverage was established. The data were well described by a single site 2-species competitive Langmuir adsorption isotherm. The surface coverage of HCl on HNO₃-doped ice was an order of magnitude lower than on bare ice for a given temperature and P_HCl. ClONO₂ uptake on this HCl/HNO₃-doped ice was studied as a function of P_HClHci. γ_max was no longer gas-phase diffusion limited and was found to be linearly dependent on the surface concentration of HCl. Under conditions of low HCl surface concentration, hydrolysis of ClONO₂ and reaction with HCl were competing such that both Cl₂ and HOCl were formed. A numerical model was used to simulate the experimental results and to aid in the parametrization of ClONO₂ reactivity on cirrus ice clouds in the upper troposphere.

Full text of DOI:10.1021/jp053477b

9986
J. Phys. Chem. A 2005, 109, 9986-9996
Kinetics of ClONO2 Reactive Uptake on Ice Surfaces at Temperatures of the Upper
Troposphere
Miguel A. Fernandez,† Robert G. Hynes,‡ and Richard A. Cox*,†
Centre for Atmospheric Science, Department of Chemistry, UniVersity of Cambridge,
CB2 1EW, United Kingdom, and CSIRO Energy Technology, Lucas Heights Science and Technology Centre,
PMB 7, Bangor, NSW 2234, Australia
ReceiVed: June 27, 2005; In Final Form: August 16, 2005
The reactive uptake kinetics of ClONO2 on pure and doped water-ice surfaces have been studied using a
coated wall flow tube reactor coupled to an electron impact mass spectrometer. Experiments have been
conducted on frozen film ice surfaces in the temperature range 208-228 K with PClONO e 10-6 Torr. The
2
uptake coefficient (γ) of ClONO2 on pure ice was time dependent with a maximum value of γmax ∼ 0.1. On
HNO3-doped ice at 218 K the γmax was 0.02. HOCl formation was detected in both experiments. On HCl-
doped ice, uptake was gas-phase diffusion limited (γ > 0.1) and gas-phase Cl2 was formed. The uptake of
HCl on ice continuously doped with HNO3 was reversible such that there was no net uptake of HCl once the
equilibrium surface coverage was established. The data were well described by a single site 2-species
competitive Langmuir adsorption isotherm. The surface coverage of HCl on HNO3-doped ice was an order
of magnitude lower than on bare ice for a given temperature and PHCl. ClONO2 uptake on this HCl/HNO3-
doped ice was studied as a function of PHCl. γmax was no longer gas-phase diffusion limited and was found
to be linearly dependent on the surface concentration of HCl. Under conditions of low HCl surface
concentration, hydrolysis of ClONO2 and reaction with HCl were competing such that both Cl2 and HOCl
were formed. A numerical model was used to simulate the experimental results and to aid in the parametrization
of ClONO2 reactivity on cirrus ice clouds in the upper troposphere.
Introduction
between observed low ozone and chlorine activation on cir-
rus.13,14 Model calculations suggest that the elevated ClO
concentrations could significantly increase ozone loss in this
region.13
Chlorine activating reactions on ice surfaces have been a
major focus of laboratory experiments since it was suggested
they play an important role in the depletion of stratospheric
ozone during the Antarctic spring. Since then, laboratory
studies1-8 have revealed that the heterogeneous reactions R1
and R2 proceed very efficiently on the surface of polar
stratospheric clouds (PSC) at temperatures of the polar strato-
sphere (T < 200 K).
The model calculations carried out to date have relied on
relatively simple parametrizations of heterogeneous reaction
kinetics using uptake coefficients for ice surfaces recommended
by the NASA panel for data evaluation.15 The recommendation
gives single, temperature-independent, values of the rate
determining uptake coefficients, which are based mainly on
experiments conducted under conditions relevant for polar
stratospheric clouds consisting of water ice or solid nitric acid
trihydrate at T < 200 K.2-4,6,7,16 There is now evidence from
laboratory studies that at higher temperatures appropriate for
cirrus clouds, the reactivity of the ice surface is lower and the
uptake coefficients are more strongly temperature dependent.6,17-22
In addition, we have shown that when multiple species are
simultaneously present (i.e., HNO3 and HCl), there is competi-
tion for adsorption sites which can affect heterogeneous reaction
rates. It is therefore important to understand the dependence of
heterogeneous reaction rates on the surface coverage of the
reactants. This would be a useful tool in differentiating between
Langmuir-Hinshelwood and Eley-Rideal surface reaction
mechanisms and thus aid in the parametrization of rate constants
in models.
ClONO2 + H2O f HOCl + HNO3
ClONO2 + HCl f Cl2 + HNO3
(R1)
(R2)
There is now emerging evidence that these reactions can occur
on the surface of ice particles in cirrus clouds at the warmer
temperatures, typically 205-230 K, of the Upper Troposphere
(UT). The potential for cirrus clouds to play a role in chlorine
activation in the UT was first suggested by Borrmann9 and later
by Solomon et al.10 Raman Lidar measurements of the UT which
showed ozone minima in the presence of ice cloud layers11 may
be explained by heterogeneous chlorine activation on cirrus ice
clouds. Indeed, model calculations using kinetic data applicable
to the colder temperatures of the polar stratosphere have shown
that up to 50% activation of reservoir chlorine on cirrus may
be possible.10,12 Observations of enhanced ClO concentrations
during cirrus events in the tropopause region support the link
Parametrization of reaction rates for varying conditions of
temperature, humidity, and reactant partial pressures using
surface coadsorption models has been described by Tabazedah
and Turco23 and Carslaw and Peter.24 These models have been
used by these groups to examine the available laboratory data
for the reaction of HCl with HOCl and ClONO2 on ice and
* Corresponding author. E-mail: rac26@cam.ac.uk. Phone: (+44) 1223-
336-253. Fax: (+44) 1223-336-362.
† University of Cambridge.
‡ Lucas Heights Science and Technology Centre.
10.1021/jp053477b CCC: $30.25 © 2005 American Chemical Society
Published on Web 10/15/2005

Kinetics of ClONO2 Reactive Uptake on Ice Surfaces
J. Phys. Chem. A, Vol. 109, No. 44, 2005 9987
NAT within the framework of a coadsorption model. This
analysis revealed large discrepancies in the experimental data
but enabled a parametrization for the reaction rates for NAT at
T < 210 K and calculations of polar ozone loss.24 The
discrepancies preclude any meaningful extrapolation of reaction
rate parameters to cirrus cloud temperatures and knowledge of
the rate coefficients for chlorine activating reactions in the upper
troposphere remains highly uncertain.
To address this uncertainty we have measured the reaction
probabilities of ClONO2 on a variety of ice surfaces at
temperatures relevant to upper tropospheric cirrus clouds and
parametrized the reactivity in terms of the above-mentioned
coadsorption models.
completely react. Purification of ClONO2 samples was carried
out to remove the impurities, Cl2, and unreacted N2O5. This
was done by trap-to-trap distillation at 173 K (methanol and
liquid N2) versus 77 K (liquid N2) to remove the Cl2, and again
at 195 K (dry ice/acetone slush) versus 77 K to remove any
unreacted N2O5. The purity of ClONO2 samples used in
experiments was determined by UV-vis spectrometry and found
to be >95% pure (the main impurity was Cl2). Gas dilutions of
ClONO2 in high-purity helium were prepared prior to every
experiment and stored in a 10 L blackened Pyrex mixing bulb.
HCl gas mixtures were prepared in mixing bulbs by diluting
HCl (Messer Gases; >99.8% purity) in high-purity helium.
Gaseous HNO3 was prepared prior to each experiment by
reacting sulfuric acid (96 wt %) with HNO3 (67 wt %). The
gaseous HNO3 was cryo-trapped at 77 K and then subjected to
several freeze-pump-thaw cycles before some condensed
HNO3 was allowed to evaporate into a blackened Pyrex mixing
bulb and diluted with high-purity helium. The uncertainty in
the mixing ratios of the reactants in the storage bulbs was ∼25%.
Measurement Procedures. Experiments conducted on sur-
faces doped with HNO3 and/or HCl required the use of a 2 mm
i.d. Teflon tube located at the upstream end of the ice surface
such that a continuous flow of dopant could be delivered to the
ice surface. Under these conditions the entire ice surface was
continuously doped and the sliding injector was always down-
stream of the Teflon tube such that only doped ice was exposed
to the reactant gas stream from the injector. In this study, partial
pressures of HNO3 were maintained at 1 × 10-6 Torr such that
at the temperatures employed and water vapor partial pressures
present, the NAT stability region of the HNO3-H2O phase
diagram was avoided. Similarly, to avoid the hydrate region of
the HCl-H2O phase diagram, partial pressures of HCl employed
were below 2 × 10-6 Torr. Mass spectrometer signals were
calibrated for each of the reactant species (HCl, ClONO2, and
HNO3) before every experiment. In each case the mass
spectrometer signals were found to be linearly dependent on
the partial pressure of the reactant. HCl was monitored at m/z
36. Both ClONO2 and HNO3 were monitored at m/z 46. The
mass spectrometer response at m/z 46 was more sensitive to
ClONO2 by ∼35% so changes in the relative amounts of these
gases present together could be monitored but with reduced
sensitivity. To distinguish the initial ClONO2 signal from the
HNO3 signal at m/z 46, it was necessary to introduce the gases
in sequence and subtract one signal from the other. After
exposure of ClONO2 to the ice surface, the NO2+ signal contains
contributions from both the ClONO2 and any desorbed HNO3
produced in the surface reaction and provides only qualitative
indications of ClONO2 concentration changes. The products Cl2
(m/z 70) and HOCl (m/z 52) were also calibrated. HOCl was
calibrated indirectly by using the relative magnitude of the
change in ClONO2 and HOCl signals for ClONO2 + ice
experiments during the initial stages of the uptake where
subsaturation quantities of HNO3 were formed on the surface
Experimental Methods
The Coated Wall Flow Tube. Experiments were carried out
in a double-jacketed cylindrical flow tube reactor coupled to
an electron impact quadrupole mass spectrometer (Hiden
Analytical model HAL IV) described in detail previously.25 The
system has recently been reorientated from a vertical to a
horizontal configuration, which has allowed an improved
protocol for ice film preparation to be used. In addition, the
mass spectrometer has been upgraded to digital electronics with
pulse counting capabilities and simultaneous multispecies
monitoring. The detection limit for ClONO2, monitored at mass
+
46 (NO2 fragment), was ∼4 × 10-8 Torr for S/N )1.
Ice films were prepared by establishing a uniform film of
liquid water on the inner surface of a Pyrex sleeve. This was
then inserted into the flow tube reactor precooled to 258 K.
The Pyrex sleeve was rotated constantly during this procedure
to ensure that an even ice film was formed throughout the length
of the tube and to minimize large variations in film thickness.
The flow reactor was then cooled to the required temperature:
208, 218, or 228 K. The water used was high purity (Acros
Organics; >99.99% purity). Ice films were maintained during
the course of an experiment by the addition of a co-flow of
helium saturated with water. This was achieved by bubbling
100 sccm of helium through a reservoir of pure water at a fixed
temperature. The water-saturated flow of helium was then passed
through a Pyrex spiral to remove supersaturated droplets prior
to entering the flow tube. The ice film thickness was calculated
using the volume of water used (measured accurately with a
pipet and kept constant between experiments), the density of
ice, and the inner surface area of the Pyrex insert. This
calculation yields an ice film thickness of ∼100 µm.
Reactant gases were exposed to a known length of ice using
a Pyrex sliding injector positioned along the central axis of the
flow tube reactor. Flow meters were used to deliver a known
flow rate of reactant into the reactor. The helium carrier gas
was fixed at 400 sccm and controlled with a mass flow
controller. The total pressure inside the reactor was ∼1.70-
1.80 Torr with a total flow velocity of ∼1500 cm s-1
.
+
Preparation of Chemicals. ClONO2 was prepared by react-
ing N2O5 with Cl2O. N2O5 was prepared by reaction of excess
O3 with NO (MG Gases, 99.99%). Ozone was produced by
passing a flow of oxygen (BOC high purity) through a high-
voltage commercial ozonizer. White crystals of N2O5 were
collected in a coldfinger at dry ice temperatures. Cl2O was
prepared by passing pure Cl2 (MG Gases, 99.99%) over yellow
mercury(II) oxide (Sigma-Aldrich >99.0%) in a Pyrex u-tube.
Cl2O appeared as a cherry red liquid at dry ice temperatures.
The Cl2O sample was then purified by trap-to-trap distillation
to remove excess Cl2 and transferred to the N2O5 sample vial.
This was left overnight in a cryogenic acetone/dry ice bath to
which did not desorb and contaminate the NO2 signal.
Desorption of HNO3 has been observed from saturated ice
surfaces at 218 K but no net desorption was found at subsatu-
ration coverages (<3 × 1014 molecules cm-2).21,26
Results
HCl Uptake on HNO3-Doped Ice. In these experiments, we
extended our previously reported results21 for the uptake of HCl
on HNO3-doped ice to cover a wider range of PHCl at three
temperatures (208, 218, 228 K). Our previous results at 218 K
and PHCl ) (0.5-1.5) × 10-6 Torr showed that (1) HCl surface

9988 J. Phys. Chem. A, Vol. 109, No. 44, 2005
Fernandez et al.
Figure 1. HCl uptake ([HCl] ) 3 × 1010 molecules cm-3) on HNO3-
doped ice at 218 K. The HCl gas stream is exposed to 6.8 cm of ice
between t ) 100 and 300 s. The surface coverage of HCl after exposure
is ∼5.6 × 1013 molecules cm-2. The smooth black line is the model fit
to the experimental data.
coverage, θ, increased with PHCl and, compared to the uptake
on “bare” ice, is reduced by approximately an order of magni-
tude for a given PHCl in the above range; (2) the uptake was,
within experimental uncertainty, completely reversible such that
the HCl adsorbed on exposing the ice surface to the HCl gas
stream was equal to the HCl desorbed when the injector was
returned to the zero position and exposure of the HCl gas flow
to the surface ceased; and (3) the surface coverage dependence
at low PHCl can be adequately modeled by using a Langmuir
adsorption isotherm for single site adsorption.
Figure 1 shows a typical uptake profile of HCl on a surface
saturated with HNO3 at 218 K. The sliding injector was
withdrawn to expose the ice surface to the HCl gas stream at
100 s and returned to the zero position at 300 s. The amount of
HCl desorbed was equal to that adsorbed within experimental
uncertainty.
Parts a, b, and c of Figure 2 show adsorption isotherms for
HCl uptake on HNO3-doped ice surfaces at 208, 218, and 228
K, respectively, with PHCl ) (0.3-4) × 10-6 Torr. In each case
the entire ice surface was continually exposed to HNO3 (PHNO
3
∼ 1.0 × 10-6 Torr). The amounts of adsorbed HCl were
calculated by integrating the signal from the initial drop in signal
until the point where the signal recovers to a steady level. The
surface coverage was then calculated by using the geometric
surface area of the ice exposed, the total flow rate, the integrated
signal area, and a calibration for the HCl signal. At 218 and
228 K, the PHNO and PH O in this study were such that ice was
Figure 2. Adsorption isotherms for HCl uptake on HNO3-doped ice
at (a) 208, (b) 218, and (c) 228K. Solid lines are best fits to a Langmuir
single site model with competitive adsorption between HNO3 and HCl.
adsorption is given below in eq 1:
3
2
the thermodynamically favorable phase in the HNO3-H2O
phase diagram. However, at 208 K the experimental condi-
tions were in a region of the phase diagram close to the
supercooled liquid coexistence line within the NAT region. The
possibility of “surface melting” may lead to larger time-
dependent uptakes observed by both ourselves and other
groups.27 Nevertheless, the HCl adsorption was reversible under
these conditions. The solid lines in Figure 2 represent Langmuir
fits to the experimental data using a single site 2-species
competitive adsorption model. It was assumed that the maximum
surface coverage (θmax) of both HNO3 and HCl was 3 × 1014
molecules cm-2 at all temperatures. This assumption is sup-
ported by numerous studies17,28,29 which suggest that the θmax
of a number of different species on ice tends toward 3 × 1014
molecules cm-2. The Langmuir equation for competitive
HCl
[HCl]ads
[HCl]gK
eq
θHCl
)
)
(1)
1 + [HCl]gKHeqCl + [HNO3]gKHNO3
θmax
eq
where [HCl]ads is the concentration of adsorbed HCl molecules
(molecules cm-2), [HCl]g is the gas-phase concentration of HCl
(molecules cm-3), [HNO3]g is the gas-phase concentration of
HNO3 (molecules cm-3), and Keq is the Langmuir equilibrium
constant for HCl or HNO3 (cm3 molecule-1).
HNO3
eq
The data in Figure 2 were well represented by eq 1. K
values were taken from a separate publication,30 which describes
the application of a numerical model to fit the HNO3-ice
experimental data of Ullerstam et al.26 These Keq values are
considerably higher than our earlier determination21 by fitting

Kinetics of ClONO2 Reactive Uptake on Ice Surfaces
J. Phys. Chem. A, Vol. 109, No. 44, 2005 9989
TABLE 1: Equilibrium Constants for HCl and HNO3
temp/K
Keq(HCl)/cm3 molecule-1
Keq(HNO3)/cm3 molecule-1
208
218
228
(1.8 ( 0.1) × 10-10
(5.6 ( 0.4) × 10-11
(2.6 ( 0.2) × 10-11
(4.3 ( 0.4) × 10-10
(3.2 ( 0.2) × 10-10
(1.9 ( 0.02) × 10-10
HCl
eq
with a dissociative Langmuir model. K
was determined
from our experimental data by fitting eq 1 to the isotherms in
Figure 2. The Keq values for both HNO3 and HCl are presented
in Table 1. The Keq values were used together with the gas-
phase HCl concentration to calculate the HCl surface coverage
in ClONO2 + HCl experiments which follow.
ClONO2 Reactive Uptake on Bare and HNO3-Doped Ice
Surfaces. Figure 3 shows an uptake experiment for exposure
of a 10 cm long bare ice film to ClONO2 at 218 K (PClONO
)
2
6.2 × 10-7 Torr). The ice was exposed to the ClONO2 gas
+
stream between t ) 250 and 860 s. Initially, the NO2 signal
level dropped very sharply and was accompanied by an increase
Figure 3. Uptake of ClONO2 (PClONO2 ∼ 6.2 × 10-7 Torr) on a “bare”
ice surface at 218 K. Ice was exposed to ClONO2 at t ) 250 s and
in the HOCl+ (m/z 52) signal, which peaked after ∼60 s. The
+
+
NO2 signal then recovered to a value lower than the initial
exposure was stopped at t ) 860 s. NO2 is shown by empty circles
and HOCl+ is represented by the noisy black line. The NO2 signal
+
signal indicating that some ClONO2 was continuously lost by
reaction to form HOCl at the ice surface and this is reflected in
the HOCl+ signal. The time-dependent component of the HOCl
production (t ) 250-400 s) is attributed to the partial
deactivation of the ice surface by adsorbed HNO3, a product of
the surface reaction between ClONO2 and H2O. The nature of
the HNO3 surface deactivation is not known but it is likely that
the presence of strongly adsorbed HNO3 reduces the number
of available surface water molecules, thus reducing the number
of sites at which ClONO2 can react. In addition, HNO3 may be
blocking sites for ClONO2 adsorption prior to reaction with H2O.
Under conditions of continuous HOCl production (t ) 400-
860 s) a steady state supply of reactive sites maintains a constant
reactive uptake coefficient for ClONO2 and therefore a constant
rate of HOCl production. The sliding injector was returned to
has been offset by +1 × 1010 molecules cm-3 for clarity. The smooth
line fits to the data represent model simulations for ClONO2 and HOCl
gas-phase concentrations with Eley-Rideal (black line) and Langmuir-
Hinshelwood (red line) reaction mechanisms.
rate constant (k) for removal of the reactant molecule from the
gas phase was calculated by using these signal differences as
shown in eq 2. The uptake coefficients were then obtained by
using eq 3:
S(0)
ν
z
k ) ln
(2)
(3)
(
)
S(t)
2k′r
+
the zero position at t ) 860 s at which point the NO2 signal
γ )
ω
temporarily exceeded the initial signal level. This apparent
desorption signal can be attributed to HNO3 or ClONO2
desorbing from the surface. We have shown previously21 that
HNO3 desorption occurs from ice surfaces which have a
saturated surface coverage of HNO3 of ∼3 × 1014 molecules
cm-2 at 218 K. At t ) 400 s (Figure 3) sufficient ClONO2 loss
(i.e., ∼3 × 1014 molecules cm-2) had occurred to produce
enough HNO3 to saturate the surface. This corresponds to the
point at which the HOCl production rate becomes constant. It
where ν is the flow velocity (cm s-1), z is the length of ice
exposed (cm), S(0) is the signal before exposure to the ice surface,
and S(t) is the signal after time t. γmax is calculated by taking
S(t) to be the lowest signal value after exposure to the ice while
γss takes S(t) to be the signal during continuous uptake. k′ refers
to the value of k corrected for radial gas-phase diffusion,33 r is
the inner radius of the Pyrex sleeve (0.80 cm), and ω is the
average molecular velocity of ClONO2. The correction of the
first-order rate constant for radial diffusion requires knowledge
of the diffusion coefficient of ClONO2 inside the flow tube,
which was calculated by using the Chapman-Enskog formula-
tion.34 The pressure-independent diffusion coefficient of ClONO2
was calculated to be ∼110 cm2 s-1 at 218 K and 1.7 Torr.
+
follows that the NO2 signal in Figure 3 cannot be used as a
quantitative description of ClONO2 uptake once a saturated
surface coverage of HNO3 has been attained. A mass balance
of the apparent total ClONO2 lost to the surface (as given by
+
the NO2 signal) versus the HOCl produced revealed less
ClONO2 lost than HOCl produced. This observation is explained
) (0.3-1.2) × 10-6 Torr, γmax
, with values in the range 0.09
ClONO2
+
At 218 K and P
by the contribution of HNO3 desorption to the NO2 signal
ClONO2
decreased with increasing P
during the latter stages of the uptake. However, mass balance
was observed during the time-dependent portion of the HOCl
production (i.e., t ) 250-400 s). The HOCl is observed in the
gas phase immediately upon exposure of the ice surface to the
ClONO2 gas stream, consistent with the known weak adsorption
of HOCl on ice at temperatures above 155 K.31,32 The rate of
at 3 × 10-7 Torr to 0.025 at 1.2 × 10-6 Torr. Typical values
of γss ∼ 0.02 were calculated with no discernible dependence
on P
. These values are summarized in Table 2.
ClONO2
In a separate set of experiments, the ice surface was first
doped with 1 × 10-6 Torr of HNO3 until the ice surface was
completely saturated. Saturated HNO3 surface coverage was
+
formation of HOCl peaked when the rate of decline of NO2
was maximum and then fell to a constant rate. No other reaction
products were observed in the gas phase.
The maximum (γmax) and steady state (γss) uptake coefficients
for ClONO2 uptake on bare ice were calculated by using the
maximum drop in signal for γmax and for γss, the drop in signal
once a stable and continuous uptake had been established. The
∼2.4 × 1014 molecules cm-2 21
.
Figure 4 shows a typical
ClONO2 (PClONO ∼ 7.0 × 10-7 Torr) uptake experiment on
2
HNO3-doped ice at 218 K. The HOCl production rate reached
a constant value without passing through the maximum, which
was observed on an initially bare ice surface. Moreover, although
HOCl appeared promptly in the gas phase it did not reach a

9990 J. Phys. Chem. A, Vol. 109, No. 44, 2005
Fernandez et al.
TABLE 2: Summary of Uptake Coefficients for ClONO2
Reactive Uptake
reaction
temp/K PClONO /Torr
γ
ClONO2 + bare ice
218
218
218
218
218
218
218
218
218
228
3.3 ×210-7
4.4 × 10-7
5.2 × 10-7
5.7 × 10-7
6.2 × 10-7
6.6 × 10-7
1.1 × 10-6
1.3 × 10-6
7.0 × 10-7
6.1 × 10-7
0.090 ( 0.025
0.065 ( 0.018
0.082 ( 0.033
0.069 ( 0.024
0.051 ( 0.018
0.069 ( 0.014
0.039 ( 0.006
0.024 ( 0.006
0.019 ( 0.006
0.009 ( 0.004
ClONO2 +
HNO3-doped ice
reaction
temp/K
P
HCl/Torr
γ
ClONO2 +
218
228
218
218
218
218
218
218
218
218
228
228
228
228
228
228
228
1.0 × 10-6
1.0 × 10-6
1.0 × 10-7
1.9 × 10-7
2.0 × 10-7
3.8 × 10-7
4.0 × 10-7
4.1 × 10-7
6.0 × 10-7
6.1 × 10-7
4.4 × 10-7
5.0 × 10-7
6.6 × 10-7
7.6 × 10-7
8.8 × 10-7
1.0 × 10-6
1.1 × 10-6
>0.1
>0.1
0.023 ( 0.012
HCl-doped icea
ClONO2 + HCl on
HNO3-doped ice
0.037 ( 0.010
0.048 ( 0.018
0.058 ( 0.021
0.057 ( 0.020
0.049 ( 0.015
0.072 ( 0.021
0.078 ( 0.025
0.020 ( 0.0045
0.023 ( 0.0081
0.025 ( 0.010
0.027 ( 0.0071
0.036 ( 0.0084
0.033 ( 0.0050
0.0040 ( 0.0068
Figure 5. ClONO2 uptake on HCl-doped ice at 218 K. Empty circles
represent the NO2 signal and the noisy black line the HOCl+ signal.
+
The NO2 signal has been offset by +1 × 1010 molecules cm-3 for
+
clarity. Ice was exposed to the ClONO2 gas stream at t ) 300 s and
exposure stopped at t ) 600 s. The solid line fits to the data represent
model simulations for ClONO2 and HOCl gas-phase concentrations with
Eley-Rideal (black line) and Langmuir-Hinshelwood (red line) surface
reaction mechanisms.
+
time dependence from the NO2 signal. This is supported by
a PClONO ) 5.0 × 10-7 Torr.
the observation that the total HOCl produced was much larger
than the apparent ClONO2 lost. However, kinetic information
can be obtained from the product signal and hence a reactive
uptake coefficient for ClONO2 can be determined from the
magnitude of the HOCl+ signal. Using this method of analysis,
γmax for ClONO2 on HNO3-doped ice was 0.019 ( 0.006 at
218 K and 0.009 ( 0.004 at 228 K. At 218 K this compares
well with the steady-state uptake coefficient on bare ice at
2
PClONO ∼ 7 × 10-7 Torr of γss ) 0.02, when the surface was
2
likely to be almost saturated with HNO3 molecules formed from
the hydrolysis reaction. These uptake coefficients are sum-
marized in Table 2.
ClONO2 Reactive Uptake on HCl-Doped and HCl/HNO3-
Doped Ice. Experiments were conducted to investigate the
reactivity of ClONO2 with adsorbed HCl at 218 K. The ice
surface was continuously exposed to 1 × 10-6 Torr of HCl until
a saturated surface coverage of HCl was achieved. The surface
coverage of HCl for these conditions was ∼2.0 × 1014 molecules
cm-2 35
Figure 5 shows an uptake experiment for ClONO2 on
.
HCl-doped ice. HCl-doped ice was exposed to ClONO2 between
t ) 300 and 600 s. The smooth lines are numerical model fits
to the experimental data discussed later. Rapid uptake of
ClONO2 was observed on the surface with γmax > 0.1. The rate
of uptake was diffusion limited so only lower limits to the uptake
coefficient could be determined. Cl2 was instantaneously
detected in the gas phase upon exposure of ClONO2 to the
surface with no detectable gas-phase HOCl. The rate of Cl2
formation reached a maximum immediately after ClONO2
exposure and then settled to a constant steady production rate.
It is concluded that ClONO2 reacts very quickly on an HCl-
doped ice surface to form Cl2 with a 1:1 stoichiometry. The
slight fall off in the maximum uptake rate is probably due to
reduction of the HCl surface coverage resulting from the fast
reaction and HNO3 production. This is supported by the
observed simultaneous loss of gas-phase HCl to the surface.
A further set of experiments was carried out to probe the
ClONO2 reactivity to HCl-doped ice at lower surface concentra-
tions of HCl. To reduce the concentration of surface HCl and
Figure 4. Uptake of ClONO2 (PClONO ∼ 7.0 × 10-7 Torr) on HNO3-
doped ice at 218 K. HNO3-doped ice 2was exposed to ClONO2 at t )
+
140 s and stopped at t ) 370 s. NO2 is shown by empty circles and
HOCl+ is represented by the noisy black line. The NO2 signal has
+
been offset by +1 × 1010 molecules cm-3 for clarity. The solid line
fits to the data represent model simulations for ClONO2 and HOCl
gas-phase concentrations with Eley-Rideal (black line) and Langmuir-
Hinshelwood (red line) surface reaction mechanisms.
maximum production rate until ∼60 s after exposure. The fact
that the HOCl signal dropped rapidly when exposure stopped
suggests that HOCl is not significantly adsorbed on HNO3-doped
ice at this temperature. We propose that the delay (∼60 s) in
the HOCl signal is due to build up of sufficient ClONO2 on the
surface to obtain the maximum rate of surface reaction. Since
the surface was initially saturated with HNO3 molecules, any
HNO3 formed from ClONO2 hydrolysis could have desorbed
immediately upon exposure of ClONO2 to the surface. This
desorption precludes any quantitative analysis of the ClONO2

Kinetics of ClONO2 Reactive Uptake on Ice Surfaces
J. Phys. Chem. A, Vol. 109, No. 44, 2005 9991
in turn the rate of ClONO2 processing, the ice surface was
continually doped with HNO3 (PHNO3 ) 1.0 × 10-6 Torr). Once
a saturated surface coverage of HNO3 had been attained, HCl
was introduced (PHCl ) (1-10) × 10-7 Torr) to the gas stream
at the upstream end of the flow tube such that the entire surface
of ice was then doped with HCl. The surface coverage of HCl
on this HNO3-doped surface was calculated by using the
isotherms in Figure 2a-c. Both HNO3 and HCl gas flows were
kept constant throughout the entire duration of these experi-
ments. Once an equilibrium surface coverage of HCl had been
attained, typically after approximately 1 h, a ClONO2 flow was
established via the sliding injector. Once all signals were stable,
the injector was pulled back to expose a known length of doped
ice to the ClONO2 gas stream.
Parts a and b of Figure 6 show ClONO2 uptakes on HCl/
HNO3-doped ice under two scenarios: (a) PHCl > PClONO and
2
(b) PHCl < PClONO . The initial HCl surface concentration in
2
Figure 6a was 1.8 × 1013 molecules cm-2 (PHCl ) 6.1 × 10-7
Torr), and that in Figure 6b was 6.1 × 1012 molecules cm-2
(PHCl ) 2.1 × 10-7 Torr). In both cases PHCl, and hence HCl
+
coverage, declined following exposure to ClONO2. The NO2
signal decreased sharply in both cases when the sliding injector
was pulled back to expose the ice surface. The time dependence
in the NO2+ signal (in both figures) reflected both ClONO2 loss
and HNO3 desorption from the surface as previously discussed.
Since some HNO3 molecules may desorb, the initial drop in
the NO2+ signal should represent a lower limit for the maximum
uptake, although the effect will be mitigated by the release of
sites occupied by HCl, and may allow some HNO3 produced
to remain adsorbed.
The “true” ClONO2 signal in Figure 6a should mirror the
Cl2 production trace that was used to calculate γmax and eliminate
the uncertainty introduced by HNO3 desorption. It was found
+
that γmax values for ClONO2 calculated with use of the NO2
signal were very similar to those calculated from the Cl2+ signal
suggesting that initially, HNO3 remains adsorbed on the ice
surface. Following this maximum uptake, the rate of reaction
depends on a balance between the rate of HCl replenishment at
the surface, the ClONO2 flux to the surface, and the reaction
rate. Both HCl and ClONO2 compete for adsorption at the
surface. The slow decrease in the gas-phase HCl concentration
to a new constant value is consistent with the establishment of
a new surface equilibrium. Experiments have been carried out
at 218 and 228 K with PHCl ) (0.2-1.0) × 10-6 Torr.
Figure 6. ClONO2 uptake experiments on ice surfaces doped with
HNO3 and HCl at 218 K. (a) PHCl > PClONO with an HCl surface
concentration of 1.8 × 1013 molecules cm-2. C2lONO2 exposure to the
ice surface starts at t ) 150 s and stops at t ) 390 s. Under these
conditions Cl2 is the only product detected. (b) PHCl < PClONO2 with an
HCl surface concentration of 6.1 × 1012 molecules cm-2. ClONO2
exposure to the ice surface starts at t ) 150 s and stops at t ) 370 s.
Both Cl2 and HOCl are detected as products. The HCl+ and NO2
+
signals have been offset by +2 × 1010 molecules cm-3 and the Cl2
signal offset by +1 × 1010 molecules cm-3 for clarity. Solid lines in
both figures are model fits to the experimental data. The smooth line
Figure 6b (PHCl < PClONO ) shows somewhat different
2
behavior from Figure 6a with both Cl2 and HOCl liberated to
the gas phase. During the uptake, the Cl2 production rate goes
through a sharp maximum on exposure of ClONO2 before falling
to a constant value. During this time, HOCl production reaches
a maximum value and remains constant until the exposure of
the ice surface to the ClONO2 flow is stopped. Evidently at
this low HCl surface coverage and low PHCl there is a
competition between ClONO2 reacting with H2O and HCl
(reactions R1 and R2). The same behavior was observed at 228
+
fit to the NO2 experimental data represents the model simulation
of the ClONO2 gas-phase concentration only and not the com-
bined contributions of ClONO2 + HNO3. The hydrolysis reaction
was modeled with a Langmuir-Hinshelwood surface reaction mech-
anism while reaction with HCl was modeled with an Eley-Rideal
mechanism.
state complicates any quantitative analysis of these data. Taking
the HCl signal level during steady state uptake and predicting
the surface coverages from Figure 2 results in coverages which
are far too large to result in the smaller uptake coefficients
measured. It is suggested that the reactive processing at the
surface is sufficiently fast that uptake of gas-phase HCl becomes
rate limiting and the surface concentration of HCl is reduced
below the Langmuir equilibrium surface coverage. The HCl
surface replenishment rate is governed by competitive adsorption
of ClONO2, HCl, and HNO3 on the HNO3-doped ice. The γss
of HCl in the two uptake experiments in parts a and b of Figure
6 were calculated from the drop in the HCl signal and found to
K when PHCl < PClONO
.
2
Analysis of the data from the steady-state region of the uptake
experiments is complicated by the uncertainty in the surface
coverage of HCl, which changes from the initial coverage to a
steady-state value as the reaction proceeds. In this steady-state
region the HCl surface concentration depends on the reactive
processing rate and the replenishment rate from the gas phase.
Furthermore, under conditions where there is a competition
between reactions R1 and R2 (Figure 6b), uncertainty in the
surface concentration of both HCl and H2O during the steady

9992 J. Phys. Chem. A, Vol. 109, No. 44, 2005
Fernandez et al.
Figure 7. ClONO2 maximum uptake coefficient dependence on HCl
surface coverage on HNO3-doped ice at 218 K.
Figure 8. ClONO2 maximum uptake coefficient dependence on HCl
surface coverage on HNO3-doped ice at 228 K.
be 0.005 and 0.004, respectively. For comparison, γss of ClONO2
during this period were calculated from the Cl2 signal and, in
the case of Figure 6b, the Cl2 and HOCl signals. The values
were 0.05 and 0.01, respectively. Hence the uptake coefficient
of ClONO2 on the doped ice surface was apparently higher than
Reaction Mechanism and Modeling
Reaction Mechanisms. To parametrize the ClONO2 reactiv-
ity under different atmospherically relevant conditions it is
necessary to have knowledge of the reaction mechanism.
Although our kinetic data do not provide conclusive proof of
the mechanisms it is clear that the hydrolysis reaction and
reaction with HCl (reactions R1 and R2) occur via different
reaction mechanisms. This conclusion is based on the differences
in the HOCl and Cl2 product profiles.
that for uptake of HCl. When PClONO < PHCl, there was
2
sufficient gas-phase HCl to maintain a flux of HCl sufficient
to support a constant reaction rate and make Cl2 formation
dominant over the hydrolysis reaction. When PClONO > PHCl
,
2
the gas-phase HCl concentration was insufficient to maintain a
high enough flux to the surface to maintain reaction R2 and
surface adsorbed ClONO2 undergoes hydrolysis to form HOCl.
The continuous production of Cl2 may reflect HCl impinging
on a partially ClONO2-doped surface, which arises as a result
of the slower reaction rate of reaction R1 compared with reaction
R2. The availability of surface water molecules is likely to be
affected by the presence of adsorbed HNO3 and HCl on the
surface.36 Despite the uncertainties which preclude any quantita-
tive analysis of these data it is clear that at the lowest surface
On bare ice (Figure 3), the rate of HOCl production passes
through a maximum before reaching a steady-state value. The
point at which the rate of HOCl production becomes constant
coincides with the uptake of ∼3 × 1014 molecules cm-2 of
ClONO2, which also corresponds to the saturated surface
coverage of HNO3 (the byproduct of reaction R1). It is
concluded that adsorbed HNO3 reduces the availability of
surface H2O molecules for hydrolysis of ClONO2. The HOCl
profile in Figure 4 and the observation that γss (HOCl from
bare ice) = γmax (HOCl from HNO3-doped ice) supports this
hypothesis. It was also noted that maximum HOCl production
from bare and HNO3-doped ice did not occur instantaneously
and instead took ∼60 s from the ClONO2 exposure. Given the
high propensity for HOCl desorption from ice at these temper-
atures31,32 and the rapid initial loss of ClONO2 from the gas
phase (see Figure 4 at t ) 140 s), it is suggested that the delay
in HOCl desorption comes from the accumulation of a precursor
species, namely surface adsorbed ClONO2. In contrast, Figures
5 and 6 show that Cl2 production from ClONO2 uptake on ice,
which has a saturated surface coverage of HCl, occurs instan-
taneously within the time response of the experimental system
and on the same time scale as ClONO2 loss to the surface.
ClONO2 has to compete for sites with HCl (and HNO3) before
it can adsorb and it is likely that the maximum uptake coefficient
of ClONO2 will have a major contribution from the direct
reaction with surface HCl but also contain a minor contribution
from adsorption to H2O sites, which only yields any detectable
HOCl when the surface HCl concentration is low as in Figure
6b. The instant production of Cl2 is consistent with a direct
reaction mechanism (Eley-Rideal). It follows therefore that the
results of these experiments do not support the indirect Cl2
formation pathway for R2 postulated in the literature5 where
ClONO2 first reacts with surface H2O to produce surface HOCl
coverages of HCl and when PClONO > PHCl, there exists a
2
competition between reaction R1 and reaction R2 such that both
Cl2 and HOCl are liberated to the gas phase. This scenario may
be the most applicable to the atmosphere where low PHCl will
result in low HCl surface coverage, especially in the presence
of HNO3.
Figures 7 and 8 show the dependence of the γmax ClONO2
as a function of initial HCl surface coverage at 218 and 228 K,
respectively. These data was obtained under conditions where
PHCl > PClONO . The solid lines are least squares weighted fits
2
to the data points. The ClONO2 uptake coefficient was found
to vary linearly with HCl surface coverage corresponding to
the range PHCl ) (1-10) × 10-7 Torr and was found to be
independent of PClONO in the range (4-10) × 10-7 Torr
2
indicating that any underestimation of γmax due to HCl surface
depletion is minimal. The data were adequately fit by straight
lines with intercepts of γmax ) 0.020 ( 0.005 at 218 K and
γmax ) 0.008 ( 0.002 at 228 K. The reaction probabilities at
the intercept correspond to the uptake of ClONO2 on HNO3-
doped ice in the absence of HCl. These are in good agreement
with independent measurements of these reaction probabilities
of γmax ) 0.019 ( 0.006 at 218 K and γmax ) 0.009 ( 0.004
at 228 K.

Kinetics of ClONO2 Reactive Uptake on Ice Surfaces
J. Phys. Chem. A, Vol. 109, No. 44, 2005 9993
which reacts with adsorbed HCl to release Cl2 to the gas phase.
If this reactive pathway were occurring we would expect to see
a Cl2 formation profile from ClONO2 + HCl-doped ice
experiments (Figure 5) similar to that in Figure 3 (ClONO2 +
H2O) for hydrolysis where the HOCl formation rate passes
through a maximum during the initial stages of the uptake due
to coverage of surface sites by strongly adsorbed HNO3. Instead
we see immediate and constant Cl2 formation in Figure 5.
The conclusions concerning the mechanism are in agreement
with those presented by Oppliger et al.8 based on experiments
at lower temperatures (180-200 K) with a Knudsen cell. In
the following section, a numerical model is used to simulate
the experimental data and to explore further the assignment of
reaction mechanisms to reactions R1 and R2.
Model Calculations. To further investigate the kinetics of
the heterogeneous reactions occurring in this system we have
used a numerical model to simulate the flow and uptake kinetics
of trace gases on surfaces in coated wall flow tube experiments.
The model uses the FACSIMILE program37 and is described
in detail elsewhere.30 The main features include transport of
gases along the flow tube, interaction of trace species with the
surface, simple Langmuir-type uptake, a diffusive process into
the subsurface layers of the ice film, and reaction between
species at the surface. Here, we have used the model to simulate
our experimental data to test the proposed mechanisms for
ClONO2 reaction at the ice surface, and to derive rate constants
for use in atmospheric models.
uptake of both HNO3 (reaction R3) and HCl (reaction R4) to
the ice surface. The forward and reverse rates of reactions R3,
R4, and R7 are determined by using eqs 4 and 5
γω
4Sinit
k(n)f )
(cm3 molecule-1 s-1)
(4)
(5)
k(n)f
k(n)r )
(s-1)
Keq
where n ) 1, 2 (or 5 for ClONO2 eq (R7)), γ is the uptake
coefficient, ω is the molecular speed (cm s-1) of the adsorbing
molecule, Sinit is the initial concentration of surface sites (cm-2
)
included to reduce the surface concentrations to dimensionless
surface coverages, and Keq is the surface equilibrium constant
(cm3 molecule-1).
The model includes reversible diffusive transport of surface
molecules (HNO3 and HCl only) into a subsurface layer of
thickness FT shown in reactions R5 and R6. The rate of diffusion
is calculated by using eq 6 and is described in terms of the
diffusion coefficient in ice (D/cm2 s-1), diffusion depth param-
eter (film thickness FT/cm), and time (t/s). D was assumed to
be constant and FT variable for fitting: n ) 3 or 4 in eq 6.
1/2
1 D
FT π
1
k(n)f ) k(n)r )
(s-1)
(6)
1/2
( ) ( )
t
The reaction schemes are described below where S is a
surface site (number cm-2), k(n)f are the forward rate constants,
k(n)r are the reverse rate constants, ks1and ks2 are the rate
constants for the ClONO2 Eley-Rideal reaction with HCl
(reaction R8) and H2O (reaction R9), respectively, and kh1 and
kh2 are the rate constants for the ClONO2 Langmuir-Hinshel-
wood reaction with HCl (reaction R10) and H2O (reaction R11),
respectively. The details of each study will be discussed
separately.
The solid line in Figure 1 is a model fit to the experimental
data for HCl uptake on HNO3-doped ice at 218 K. The model
reproduces the main features and time dependence of the uptake.
The equilibrium constants for both HNO3 and HCl were
variables in this model and determined by the best fit of the
model output to the experimental data. The equilibrium constants
determined from the model fits were in good agreement with
those presented in Table 1 from analysis of the data in Figure
2. The Keq values given in Table 1 are approximately an order
of magnitude greater than those reported in our earlier experi-
mental work for HNO3 and HCl uptake on bare ice.21,35 These
experimentally determined values were derived from surface
coverage measurements over a range of relatively high reactant
Adsorption of HNO3, HCl, and ClONO2 on ice:
HNO3(g) + S {k1f,k1r} HNO3(ads)
HCl(g) + S {k2f,k2r} HCl(ads)
(R3)
(R4)
(R5)
(R6)
(R7)
partial pressures (PHNO /HCl > 1 × 10-7 Torr). The uncertainty
3
in the magnitude of Keq is thought to come from a combination
of the following factors: (i) At high HNO3 partial pressures
the desorption of HNO3, and to a lesser extent HCl, from
downstream surfaces within the flow tube have a significant
HNO3(ads)
{
k3f,k3r} HNO3(bulk) + S
HCl(ads)
{
k4f,k4r} HCl(bulk) + S
+
effect on the behavior of the NO2 signal in the initial stages
ClONO2(g) + S {k5f,k5r} ClONO2(ads)
of the adsorption, which results in an underestimation in the
total amount of HNO3 adsorbed. This effect translates into
erroneously low equilibrium constants from the surface coverage
data. (ii) The criteria used to determine the surface coverage
may over- or undercompensate for diffusion to subsurface layers
requiring a correction in Keq values reported. (iii) The value of
ClONO2 reactiVe uptake on ice:
Eley-Rideal
ClONO2(g) + HCl(ads) ks18 Cl2(g) + HNO3(ads) (R8)
θ
max in our earlier report was assumed to be 1 × 1015 molecules
cm-3 whereas it is now clear that the max surface coverage is
∼3 × 1014 molecules cm-2 for both HNO3 and HCl.
ClONO2(g) + H2O(surf) ks28 HOCl(g) + HNO3(ads) (R9)
ClONO2 Uptake on Bare and HNO3-Doped Ice. The
reactive uptake of ClONO2 on bare and HNO3-doped ice was
modeled by using both Langmuir-Hinshelwood (reactions R7
and R11) and Eley-Rideal (reaction R9) mechanisms. In the
Eley-Rideal mechanism, a gas-phase ClONO2 molecule is
assumed to react on collision at a surface site occupied by a
water molecule, with a maximum uptake coefficient (γ0) on a
clean ice surface. The rate of reaction R9 (ks2) was calculated
by using eq 4. In the Langmuir-Hinshelwood treatment, the
Langmuir-Hinshelwood
ClONO2(ads) + HCl(ads) kh18 Cl2(g) + HNO3(ads) + S (R10)
ClONO2(ads) + H2O(surf) kh28 HOCl(g) + HNO3(ads) + S (R11)
HCl Uptake on HNO3-Doped Ice. The model uses a single
site Langmuir adsorption mechanism to describe the competitive

9994 J. Phys. Chem. A, Vol. 109, No. 44, 2005
Fernandez et al.
TABLE 3: Summary of Parameters Used in Model
gas-phase ClONO2 first establishes a population of surface
adsorbed molecules which react at sites occupied by water
molecules.
Calculations of ClONO2 Uptake on HCl/HNO3-Doped Icea
PHCl > P
PHCl < P
ClONO2
ClONO2
The solid lines in Figure 3 represent model fits to experi-
mental data for ClONO2 uptake on bare ice with use of an Eley-
Rideal mechanism (black line) and a Langmuir-Hinshelwood
(red line) mechanism. The main difference between the two
models occurs during the initial stages of the uptake. The Eley-
Rideal simulation results in an instant maximum production of
HOCl before returning to a lower steady signal level. The
experimentally observed HOCl production reaches a maximum
more slowly and begins to fall only after a few tens of seconds
at this maximum production rate. This delay in the formation
of HOCl has been observed in previous studies of this
reaction.8,38 In this study, only the Langmuir-Hinshelwood
simulation reproduced the delay in the attainment of maximum
HOCl production although the time to maximum was still shorter
than observed. The model output for ClONO2 only matched
218 K
228 K
218 K
228 K
Keq(HCl)/cm3 molecule-1 2 × 10-11 2 × 10-11 2 × 10-11 2 × 10-11
Keq(HNO3)
1 × 10-10 1 × 10-10 1 × 10-10 1 × 10-10
4 × 10-11 4 × 10-11 4 × 10-11 4 × 10-11
Keq(ClONO2)
γmax(HNO3)
γmax(HCl)
0.1
0.1
0.1
0.08
0.9
0.1
0.04
0.2
0.08
0.1
0.08
0.1
γr(ClONO2 + HCl)
γh(ClONO2 + H2O)
kh/cm2 s-1 molecule-1
0.02
0.02
0.1
0.1
1 × 10-15 1 × 10-15 1 × 10-15 1 × 10-15
a Keq ) equilibrium constant, γr ) ClONO2 uptake coefficient for
reaction with adsorbed HCl, γh ) ClONO2 uptake coefficient for
reaction with adsorbed H2O, kh ) rate constant for surface reaction
between adsorbed ClONO2 and adsorbed H2O (shown as kh2 in reaction
R11).
> PClONO , Cl2 was the only product detected in the gas phase.
2
The model reproduces the time dependence and magnitude of
the Cl2 production on exposure of the ice film to ClONO2. The
model predicts the formation of a small amount (∼2 × 109
molecules cm-3) of HOCl, which was close to the detection
limit of our mass spectrometer. The HCl time dependence is
adequately reproduced but the magnitude of the HCl loss to
the surface is underestimated by the model. The ClONO2
+
the experimental NO2 signal in the first 150 s; subsequently
desorption of product HNO3 also contributed to the NO2+ signal.
A model simulation of desorbed HNO3 revealed that sufficient
HNO3 desorption had occurred to account for the difference
between the experimental NO2+ signal and model ClONO2 trace.
The solid lines in Figure 4 show similar model fits for
ClONO2 uptake on HNO3-doped ice: Eley-Rideal (black line)
and Langmuir-Hinshelwood (red line). Again the main differ-
ences between the two model outputs occur during the initial
stages of the exposure. In this case the Langmuir-Hinshelwood
model accurately reproduces the time dependence in the HOCl
production while the product formation in the Eley-Rideal
model is far too rapid. Under HNO3-doped ice conditions it is
anticipated that HNO3 desorption from the surface will provide
+
simulations do not reproduce the NO2 signals which are
contaminated by HNO3 desorption. The model simulations do,
however, predict desorption of sufficient HNO3 to account for
+
the difference between the ClONO2 model and NO2 signal.
When PHCl < PClONO , there exists a competition between
2
the two reactive channels (reactions R1 and R2) for ClONO2
such that both HOCl and Cl2 were liberated to the gas phase.
The instant production of Cl2 was well reproduced as was the
more “sluggish” HOCl production. The model also reproduced
+
a greater contribution to the NO2 signal than in the bare ice
the HCl+ signal. The difference between the experimental NO2
+
experiments since the surface is already close to saturation. This
is reflected in the model outputs which show a much greater
difference between the experimental NO2+ signal and the model
ClONO2 than in the bare ice simulations. The model simulation
of HNO3 showed that sufficient desorption had occurred to
account for the differences in the experimental NO2+ signal and
the model ClONO2.
signal and the modeled ClONO2 could be quantitatively
accounted for by HNO3 desorption.
The experimental observations of adsorption of strong acids
(HNO3 and HCl) have been well reproduced with a simple
numerical model incorporating surface adsorption followed by
diffusion into subsurface layers. The diffusion parameters used
in the model have been described previously.30 It emerged that
model Keq for HNO3, HCl, and ClONO2 were not strongly
ClONO2 Reactive Uptake on HCl-Doped and HCl/HNO3-
Doped Ice. Figure 5 shows model fits to the experimental data
for uptake of ClONO2 on HCl-doped ice. In this case the Eley-
Rideal model (black line) provided a better fit to the NO2+ and
temperature dependent between 218 and 228 K with values of
HNO3
eq
K
∼ 1 × 10-10 cm3 molecule-1, KHeqCl ∼ 2 × 10-11
+
cm3molecule-1, and KeCqlONO2 ∼ 4 × 10-11 cm3 molecule-1
.
Cl2 data, particularly between t ) 300 and 380 s. Departure
+
of the model fit from the NO2 signal after 380 s can be
This is somewhat unexpected since Keq for HNO3 and HCl has
been shown by experiment to have negative temperature
dependence as shown in Table 1.21,26 The model Keq are
somewhat lower than those resulting from the Langmuir fit to
explained by desorption of product HNO3.
A final set of model calculations were carried out to simulate
ClONO2 reactive uptake on HCl/HNO3-doped ice. HNO3 and
HCl doping were modeled as described by reactions R3-R6.
ClONO2 uptake was split into reversible adsorption (reaction
R7) and two reactive channels represented by reactions R8 and
R11, i.e., reaction with surface adsorbed water molecules
(Langmuir-Hinshelwood) and direct reaction with surface
adsorbed HCl molecules (Eley-Rideal). The assignment of
these reaction mechanisms was based on the model fits for
ClONO2 uptake on bare, HNO3-doped, and HCl-doped ice
described above. The rate constants of reactions R7 and R8 were
calculated with eqs 4 and 5. The value of kh2 (reaction R11)
was variable with a maximum value of 1 × 10-15 cm2 s-1
molecule-1 (kinetic limit for reaction on every 2D encounter).
the HCl uptake on HNO3-doped ice data shown in Figure 2.
There are no previously reported K
the value reported here. Under conditions where PHCl < PClONO
ClONO2
eq
for comparison with
,
2
the model uptake coefficients (see Table 3) for ClONO2 + HCl
were γr218K ) 0.9 and γr228K ) 0.2. These values are in
agreement with the uptake coefficients (γ0) at maximum HCl
surface coverage (3 × 1014 molecules cm-2) determined from
218
extrapolation of the data in Figures 7 and 8 which gave γ0
)
228
0.9 (+0.08, -0.3) and γ0 ) 0.3 ( 0.1
The modeling simulations support the assignment of a
Langmuir-Hinshelwood reaction mechanism to the hydrolysis
reaction (reaction R1) and an Eley-Rideal mechanism to the
reaction with HCl (reaction R2). Table 3 shows a summary of
the modeling parameters which best fit the experimental
observations for the reaction mechanism assignment given
The solid lines in parts a and b of Figure 6 show the model
outputs for ClONO2 uptake on HCl/HNO3-doped ice at 218 K
for PHCl > PClONO and PHCl < PClONO , respectively. When PHCl
2
2

Kinetics of ClONO2 Reactive Uptake on Ice Surfaces
J. Phys. Chem. A, Vol. 109, No. 44, 2005 9995
above. On the basis of the above analysis, we believe that a
parametrization based on the assignment of reaction mecha-
nisms, together with the values presented in Table 3, can be
used to determine the rates of ClONO2-ice interactions in the
Upper Troposphere.
ClONO2 Lifetime and Steady-State ClO Production. The
lifetime of ClONO2 with respect to reactions R1 and R2 can be
estimated by using the parametrization presented above to
calculate γ, atmospheric concentrations of the reactants, and
eqs 10 and 11 below:
Parametrization of ClONO2 Reactivity on Cirrus Clouds.
A parametrization for the rate coefficient of ClONO2 reaction
on ice is presented below for ClONO2 hydrolysis and reaction
with adsorbed HCl based on parameters derived from experi-
mental results and modeling presented in this work.
ClONO2 + H2O. Equation 7 shows the mathematical expres-
sion for γ in a Langmuir-Hinshlelwood surface reactivity
model24,39
γωS
4V
khet
)
(10)
(11)
1
khet
τhet
)
where khet is the pseudo-first-order rate constant (s-1) for
heterogeneous loss of ClONO2 to the ice surface in cirrus clouds,
γ is the uptake coefficient of ClONO2 calculated with eqs 9
and 7, ω is the mean molecular speed (cm s-1) of ClONO2,
S/V is the surface-to-volume ratio of the cirrus cloud (cm2
cm-3), and τhet is the atmospheric lifetime of ClONO2 with
respect to heterogeneous loss on cirrus ice particles.
ωxσx
1
γ
1
S0
)
+
(7)
4kh[H2O]sK
where S0 is the accommodation coefficient (approximated to
be ∼0.9 from maximum uptake coefficient determination for
ClONO2 + HCl), ωx is the mean thermal velocity of ClONO2
(cm s-1), σx is the surface area occupied by one adsorbed
molecule (cm2 molecule-1), K is the modified Langmuir constant
(K ) Keq/(1 + Keq[ClONO2]s)), Keq is the surface equilibrium
constant for ClONO2, [ClONO2]s is the gas-phase concentration
of ClONO2 in the immediate vicinity of the surface (molecules
cm-3), kh is the second-order surface rate constant for hydrolysis
(cm2 molecule-1 s-1), and [H2O]s is the surface concentration
of H2O (∼1 × 1015 molecules cm-2).40 Since HNO3 is
ubiquitous in the UT with a high propensity for adsorption to
ice, it is important to include the effect of adsorbed HNO3 on
the number of available surface water molecules. In the presence
of gas-phase HNO3, we assume [H2O]s ) (1 × 1015) -
[HNO3]ads (i.e., ignoring sites occupied by HCl and other gases).
[HNO3]ads can be calculated by using the classical Langmuir
equation for physical adsorption given below:
Assuming maximum reactant mixing ratios of 2 pptv ClONO2,
500 pptv HCl and 1000 pptv HNO3, we obtain γ218 ) 0.1,
R1
γ218R2 ) 0.03, γ228R1 ) 0.09, and γ228R2 ) 0.009. These uptake
coefficients have been used together with a typical surface-to-
volume ratio of background cirrus clouds (∼2 × 10-5 cm2
cm-3 41
to calculate the lifetime of ClONO2. The lifetime of
)
ClONO2 was calculated to be ∼32 min with respect to loss via
reaction R1 and ∼40 min for loss via reaction R2 in the
temperature range 218-228 K.
The photolysis rate of ClONO2 in the UT is approximately 5
× 10-5 s-1, which corresponds to a lifetime of ∼6 h.9,10 The
lifetime calculation suggests that reactive uptake on background
cirrus can dominate over photolysis as a loss process for
ClONO2 in the UT.
The maximum uptake coefficients for ClONO2 were used to
calculate the steady-state concentration of ClO by assuming that
reactions R12-R15 represent the dominant production and
destruction pathways.
HNO3
[HNO3]ads
3 × 1014
[HNO3]gK
eq
k1
)
(8)
HNO3
eq
ClONO2 + HCl
98 Cl2 + HNO3
(R12)
(R13)
1 + [HNO3]gK
Cl2 + hυ
9
J8 2Cl
where [HNO3]g is the gas-phase concentration of HNO3
(molecules cm-3) and KeHqNO3 is the equilibrium constant of
HNO3 on ice (cm3 molecule-1).
ClONO2 + HCl. The uptake coefficient for ClONO2 on ice
in the presence of adsorbed HCl is well described by an Eley-
Rideal reaction mechanism. γ can be calculated by using the
expression shown in eq 9:
k2
Cl + O3
9
8 ClO + O2
(R14)
(R15)
k3
ClO + NO2 + M
98 ClONO2 + M*
A steady-state analysis was performed assuming that reactions
R12 and R15 are the major rate-limiting processes. Equation
12 shows the resulting steady-state expression for [ClO]:
γ ) γ0θHCl
(9)
where γ0 is the uptake coefficient at maximum HCl surface
coverage. From an extrapolation of experimentally determined
γ values at low HCl surface coverage (in Figures 7 and 8) to a
k1[ClONO2][HCl]surf
[ClO]ss )
(12)
k3[NO2][M]
saturated surface coverage of 3 × 1014 molecules cm-2, γ0
218
) 0.9 (+0.08, -0.3) and γ0228 ) 0.3 ( 0.1. θHCl in eq 9 is the
fractional surface coverage of HCl and can be calculated by
using the classical Langmuir eq 8. In the presence of HNO3, a
competitive adsorption Langmuir equation (eq 1) describes the
surface coverage of HCl.
Equation 9 can be used to predict the maximum reactive
uptake coefficient for reaction R2 at any HCl surface coverage.
It follows that, assuming a Langmuir adsorption isotherm for
HCl on ice, γmax can be predicted for atmospherically relevant
The gas-phase reactant concentrations were as follows:
[ClONO2] ) 1 × 107 molecules cm-3, [NO2] ) 6 × 109
molecules cm-3, and [M] (air) ) 9 × 1018 molecules cm-3 at
218 K, 200 mbar (∼10 km). [HCl]surf is the fractional surface
concentration of HCl calculated by using the classical Langmuir
eq 8, Keq ) 2.0 × 10-11 cm3 molecule-1, and [HCl] ) 3 × 109
molecules cm-3. The rate constant k1 (s-1) was calculated by
using the expression shown below (eq 13). k3 ) 2 × 10-12 cm6
molecule-2 s-1 at 218 K (1.7 × 10-12 cm6 molecule-2 s-1 at
228 K):15
PHCl
.

9996 J. Phys. Chem. A, Vol. 109, No. 44, 2005
Fernandez et al.
(6) Lee, S. H.; Leard, D. C.; Zhang, R. Y.; Molina, L. T.; Molina, M.
J. Chem. Phys. Lett. 1999, 315, 7.
(7) Leu, M. T.; Moore, S. B.; Keyser, L. F. J. Phys. Chem. 1991, 95,
7763.
γ0ω
k1 )
S
(13)
4
(8) Oppliger, R.; Allanic, A.; Rossi, M. J. J. Phys. Chem. A 1997, 101,
1903.
(9) Borrmann, S.; Solomon, S.; Dye, J. E.; Luo, B. P. Geophys. Res.
Lett. 1996, 23, 2133.
(10) Solomon, S.; Borrmann, S.; Garcia, R. R.; Portmann, R.; Thomason,
L.; Poole, L. R.; Winker, D.; McCormick, M. P. J. Geophys. Res., Atmos.
1997, 102, 21411.
where γ0 is the maximum uptake coefficient (0.92 at 218 K
and 0.32 at 228 K), ω is the mean molecular speed of the gas
(cm s-1), and S is the surface area-to-volume ratio of background
cirrus.
This analysis resulted in [ClO]ss ) 1 × 107 molecules cm-3
at 218 K (∼1 pptv at 200 mbar) and [ClO]ss ) 4 × 106
molecules cm-3 at 228 K (∼0.5 pptv at 200 mbar). A similar
calculation with hydrolysis of ClONO2 and the subsequent
photolysis of the HOCl product as the major source of ClO
was performed. This calculation gave [ClO]ss ∼ 4 × 107
molecules cm-3 at both 218 and 228 K (∼4 pptv at 200 mbar).
Borrmann et al.13 report airborne observations during descents
of the ER-2 aircraft through a cirrus cloud event near the
midlatitude tropopause with ClO abundances of 2.7 pptv. At
high latitudes, enhanced ClO mixing ratios of 15-20 pptv were
observed near the tropopause.14 The mechanism for maintaining
the observed ClO concentrations is unclear and the effects of
chemistry and transport both remain possible. The steady-state
ClO calculations performed here suggest that chemical chlorine
activation can account for the observations at midlatitudes.
Given the relative abundances of surface HCl and H2O on the
cirrus particle surface, it is important to note that chlorine
activation via ClONO2 hydrolysis (reaction R1) is likely to be
more significant for ClO production than reaction with adsorbed
HCl (reaction R2). Both reactions R1 and R2 result in the
formation of HNO3, which removes NOx from the gas phase.
This gas-phase denitrification may increase [ClO]ss by reducing
the loss rate of ClO (reaction R14).
(11) Reichardt, J.; Ansmann, A.; Serwazi, M.; Weitkamp, C.; Michaelis,
W. Geophys. Res. Lett. 1996, 23, 1929.
(12) Meilinger, S. K.; Karcher, B.; von Kuhlmann, R.; Peter, T. Geophys.
Res. Lett. 2001, 28, 515.
(13) Borrmann, S.; Solomon, S.; Avallone, L.; Toohey, D.; Baumgardner,
D. Geophys. Res. Lett. 1997, 24, 2011.
(14) Thornton, B. F.; Toohey, D. W.; Avallone, L. M.; Harder, H.;
Martinez, M.; Simpas, J. B.; Brune, W. H.; Avery, M. A. J. Geophys. Res.,
Atmos. 2003, 108, art. no. 8333.
(15) 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 No. 12; Jet Propulsion Laboratory: Pasadena, CA,
1997.
(16) Leu, M.-T. Geophys. Res. Lett. 1988, 15, 17.
(17) Abbatt, J. P. D. Geophys. Res. Lett. 1997, 24, 1479.
(18) Chaix, L.; Allanic, A.; Rossi, M. J. J. Phys. Chem. A 2000, 104,
7268.
(19) Chu, L.; Chu, L. T. J. Phys. Chem. A 1999, 103, 8640.
(20) Chu, L.; Chu, L. T. J. Phys. Chem. A 1999, 103, 691.
(21) Hynes, R. G.; Fernandez, M. A.; Cox, R. A. J. Geophys. Res.,
Atmos. 2002, 107, 4797.
(22) Zondlo, M. A.; Barone, S. B.; Tolbert, M. A. Geophys. Res. Lett.
1997, 24, 1391.
(23) Tabazadeh, A.; Turco, R. P. J. Geophys. Res., Atmos. 1993, 98,
12727.
(24) Carslaw, K. S.; Peter, T. Geophys. Res. Lett. 1997, 24, 1743.
(25) Percival, C. J.; Mossinger, J. C.; Cox, R. A. Phys. Chem. Chem.
Phys. 1999, 1, 4565.
(26) Ullerstam, M.; Thornberry, T.; Abbatt, J. P. D. Faraday Discuss.
2005, 130, DOI: 10.1039/b417418f.
(27) Hudson, P. K.; Shilling, J. E.; Tolbert, M. A.; Toon, O. B. J. Phys.
Chem. A 2002, 106, 9874.
(28) Sokolov, O.; Abbatt, J. P. D. J. Phys. Chem. A 2002, 106, 775.
(29) Winkler, A. K.; Holmes, N. S.; Crowley, J. N. Phys. Chem. Chem.
Phys. 2002, 4, 5270.
(30) Cox, R. A.; Fernandez, M. A.; Ullerstam, M.; Abbatt, J. P. D. Phys.
Chem. Chem. Phys. 2005. Submitted for publication.
(31) Chu, L.; Diao, G. W.; Chu, L. T. J. Phys. Chem. A 2000, 104,
3150.
The results from this study together with the parametrization
presented and the values of the important parameters (γ, Keq)
in Table 3 need to be incorporated into atmospheric chemistry
models and compared with field observations of ClO to decouple
the contributions from chemical processing and transportation
and to better understand the role of heterogeneous chemistry in
chlorine activation by cirrus clouds.
(32) Horn, A. B.; Sodeau, J. R.; Roddis, T. B.; Williams, N. A. J. Chem.
Soc., Faraday Trans. 1998, 94, 1721.
(33) Brown, R. L. J. Res. Natl. Bur. Stand. (U.S.) 1978, 83, 1.
(34) Sherwood, R. C. R. a. T. C. The Properties of Gases and Liquids,
3rd ed.; McGraw-Hill: New York, 1958.
(35) Hynes, R. G.; Mossinger, J. C.; Cox, R. A. Geophys. Res. Lett.
2001, 28, 2827.
Acknowledgment. The authors would like to thank the E.U.
project CUT-ICE (EVK2-CT-1999-00005) for funding this
research. M.A.F. acknowledges the support of a N.E.R.C. Ph.D.
research studentship.
References and Notes
(36) Delval, C.; Fluckiger, B.; Rossi, M. J. Atmos. Chem. Phys. 2003,
3, 1131.
(37) Curtis, A. R.; Sweetenham, W. P. AERE Harwell Publication R
12805 1987.
(38) Hanson, D. R.; Ravishankara, A. R. J. Phys. Chem. 1992, 96, 2682.
(39) Ammann, M.; Poschl, U.; Rudich, Y. Phys. Chem. Chem. Phys.s
2003, 5, 351.
(40) Haynes, D. R.; Tro, N. J.; George, S. M. J. Phys. Chem. 1992, 96,
8502.
(1) Tolbert, M. A.; Rossi, M. J.; Malhotra, R.; Golden, D. M. Science
1987, 238, 1258.
(2) Molina, M. J.; Tso, T.-L.; Molina, L. T.; Wang, F. C.-Y., Science
1987, 238, 1253.
(3) Hanson, D. R.; Ravishankara, A. R. J. Geophys. Res., Atmos. 1991,
96, 5081.
(4) Abbatt, J. P. D.; Molina, M. J. J. Phys. Chem. 1992, 96, 7674.
(5) Chu, L. T.; Leu, M. T.; Keyser, L. F. J. Phys. Chem. 1993, 97,
12798.
(41) Tabazadeh, A.; Toon, O. B.; Jensen, E. J. Geophys. Res. Lett. 1999,
26, 2211.