Kinetics of the O + ClO reaction

Article abstract of DOI:10.1021/jp0100351

The rate coefficient for the reaction O + ClO → Cl + O₂ (1) was measured over the range 227 to 362 K using a discharge flow tube coupled to a pulsed laser photolysis-atomic resonance fluorescence apparatus. The title reaction was studied under pseudo-first-order conditions in O atoms in an excess of ClO. ClO radicals were produced in the discharge flow tube by the reaction of Cl with either O₃ or Cl₂O and its concentration was measured in situ by UV absorption. O atoms were produced by pulsed laser photolysis of ClO at 308 nm and detected by atomic resonance fluorescence. The measurements yield k₁(T) = (3.0 ± 0.8) × 10^-11 exp((75 ± 40)/T) cm³ molecule^-1 s^-1, where the uncertainties are at the 95% confidence level and include estimated systematic errors. These results are compared with previous data and an expression that can be used for atmospheric purposes is suggested.

Full text of DOI:10.1021/jp0100351

5402
J. Phys. Chem. A 2001, 105, 5402-5409
Kinetics of the O + ClO Reaction†
Leah Goldfarb,‡,§ James B. Burkholder, and A. R. Ravishankara*,§
Aeronomy Laboratory, NOAA, 325 Broadway, Boulder, Colorado 80305-3328
ReceiVed: January 4, 2001; In Final Form: March 28, 2001
The rate coefficient for the reaction O + ClO f Cl + O2 (1) was measured over the range 227 to 362 K
using a discharge flow tube coupled to a pulsed laser photolysis-atomic resonance fluorescence apparatus.
The title reaction was studied under pseudo-first-order conditions in O atoms in an excess of ClO. ClO radicals
were produced in the discharge flow tube by the reaction of Cl with either O3 or Cl2O and its concentration
was measured in situ by UV absorption. O atoms were produced by pulsed laser photolysis of ClO at 308 nm
and detected by atomic resonance fluorescence. The measurements yield k1(T) ) (3.0 ( 0.8) × 10-11
exp((75 ( 40)/T) cm3 molecule-1 s-1, where the uncertainties are at the 95% confidence level and include
estimated systematic errors. These results are compared with previous data and an expression that can be
used for atmospheric purposes is suggested.
Introduction
Chlorine-catalyzed ozone destruction via the reaction mech-
anism:
O + ClO f Cl + O2
Cl + O3 f ClO + O2
(1)
(2)
(3)
Net: O + O3 f 2O2
plays an important role in determining the abundance of upper
stratospheric ozone and its trend over the past two decades.1
(In this paper, O atom refers to the ground-state oxygen atom,
O(3P)). Reaction 1 is the rate-limiting step in this catalytic cycle.
Therefore, the rate coefficient, k1(T), and its temperature
dependence, is crucial for calculating ozone levels and its trend.
Reaction 1 has been studied numerous times over the past 15
years.2-8 While there is relatively good agreement among the
values of k1(T) reported in the 1980s, systematic errors in
determining the ClO concentration could have affected these
investigations. Also, most of these studies used the discharge
flow tube method, which constrained the pressures to a few
Torr. It is now recognized that ClO can form a dimer at low
temperature and more information about the generation of ClO,
possible reaction of vibrationally excited ClO with Cl atoms,9-11
and the absorption cross section of ClO has become available.12
For stratospheric modeling, the currently recommended12 value
of k1(T) is 3.0 × 10-11 exp (70/T) cm3 molecule-1 s-1; k1(298)
) 3.8 × 10-11 cm3 molecule-1 s-1 and is derived from the
studies of the 1980s. Therefore, a careful study of reaction 1 is
worthwhile.
Figure 1. A schematic of the experimental apparatus used to measure
k1. Individual components are noted in the figure and explained in the
text.
to measure ClO concentration, O atoms are produced via pulsed
laser photolysis of ClO, and the temporal profile of O atoms
are measured via atomic resonance fluorescence. The rate
coefficient for reaction 1, k1(T), was measured between 227 and
362 K in 2.5 to 18 Torr of N2.
Experiments
Reaction 1 was studied using a technique that couples a flow
tube with a pulsed photolysis system.11 We have recently used
such a system to study the reactions of ClO with OH10 and IO11
under pseudo-first-order conditions in OH and IO; these radicals
were monitored via laser-induced fluorescence. In this study,
we used atomic resonance fluorescence detection of O atoms
in an excess of ClO for measuring k1. This combination of a
flow tube and a pulsed photolysis method is described in detail
elsewhere.10,11,13
The apparatus used to measure k1 is shown in Figure 1. There
are four major components in this apparatus: (1) a flow tube
where ClO radicals were generated, (2) an absorption cell where
the concentration of ClO was measured, (3) a D2 lamp (30 W)
and a 0.28-m spectrograph equipped with a diode array detector
We have reinvestigated reaction 1 using an apparatus where
a discharge flow tube is coupled to a diode array spectrometer
* Author to whom correspondence should be addressed at NOAA,
R/AL2, 325 Broadway, Boulder, CO 80305-3328. E-mail: ravi@al.noaa.gov.
† This work was part of a Ph.D. thesis submitted by L. Goldfarb to the
University of Colorado, Boulder, Colorado.
‡ Presently at CNRS/SA, BP N° 3, 91371 Verrie`res-le-Buisson Cedex,
France.
§ Also affiliated with the Department of Chemistry and Biochemistry,
University of Colorado, Boulder, CO 80309.
10.1021/jp0100351 CCC: $20.00 © 2001 American Chemical Society
Published on Web 05/12/2001

Kinetics of the O + ClO Reaction
J. Phys. Chem. A, Vol. 105, No. 22, 2001 5403
to measure the concentration of ClO (as well as that of O3, Cl2O,
and Cl2), and (4) an aluminum reactor, where O atoms were
created by 308 nm (XeCl excimer) pulsed laser photolysis of
ClO and detected by atomic resonance fluorescence. The
individual components have been used in our laboratory
extensively.11,14,15 Therefore, we will describe here only the parts
essential for an understanding of this study.
This apparatus differs from the previous flow tube-pulsed
photolysis system that was used in the studies of OH + ClO10
and IO + ClO11 reactions in a few ways. First, we used an
atomic resonance fluorescence cell for the detection of O atoms.
Second, the pulsed photolysis-resonance fluorescence detection
component was at the end, as opposed to the middle, of the
absorption cell; however, it was also a part of the absorption
path. Third, the absorption cell and reactor were at the same
temperature.
Figure 2. Top panel: Plots of the differential absorption versus
wavelength for the reference spectrum of ClO (solid line) and the one
measured in a kinetics experiment (dashed line). Bottom Panel:
Residual from the subtraction of the two spectra in the above panel,
with a suitable multiplication factor included. Clearly, there is no
residual attributable to ClO.
ClO Generation. ClO radicals were produced at 298 K via
the reaction of Cl atoms with O3 or Cl2O:
Cl + O3 f O2 + ClO
(2)
(4)
The concentration of ClO was measured using the following
procedure: (1) a spectrum with only N2 flowing through the
cell, I0, was recorded. Then, the needed amounts of Cl2 and O3
or Cl2O were added and another spectrum, I1, was recorded.
From these two spectra, the concentrations of Cl2 and O3 or
Cl2O were determined. (These concentrations were also deter-
mined from flow rate and pressure measurements and agreed
well with the spectral determinations.) The microwave discharge
was turned on to generate Cl atoms in the flow tube. The Cl
atom concentration was variable but always much greater than
that of ClO. The presence of excess Cl was evident from the
visible emission (to the experimenter) due to the chemilumi-
nescent Cl-recombination reaction. The excess Cl atoms reacted
with O3 or Cl2O to generate ClO in the flow tube and ClO
generation (and other possible reactions that affected its
concentration) was complete in this region, i.e., before the flow
entered the absorption cell. Then, another spectrum, I2, was
recorded. From the measured spectral differences between I1
and I2, the absorption due to ClO was obtained. The reference
spectrum of ClO was subtracted from the difference spectrum
(I2 - I1) until no structure attributable to ClO remained in the
difference spectrum. One such example is shown in Figure 2,
where the residual is clearly small and does not show the
signature for ClO. The required absorption that had to be
subtracted to get a residual of zero was converted to concentra-
or
Cl + Cl2O f Cl2 + ClO
where k2(298 K) ) 1.2 × 10-11 cm3 molecule-1 s-1 and
k4(298 K) ) 9.6 × 10-11 cm3 molecule-1 s-1 12
. Cl atoms were
produced in a microwave discharge of a Cl2/He mixture in a
sidearm of the discharge flow tube. The concentration of Cl
atoms was in large excess over the initial concentration of O3
or Cl2O such that reaction 2 or 4 was essentially complete
(>99%) in the flow tube. Thus, all the O3 or Cl2O was consumed
before the flow entered the absorption cell. The absence of O3
or Cl2O was verified by measuring the absorption spectra of
the gas mixture flowing through the absorption cell.
A mixture of O3 or Cl2O in N2 was added through a movable
injector (4 mm i.d.) centered in the flow tube. An additional
flow of N2 (∼16 STP cm3 s-1) was added to this flow to bring
the pressure in the flow tube to the desired value of 2.5 to 18
Torr. The amount of He from the Cl2/He mixture in the flow
tube was usually less than ∼20% of the total flow; at the lower
pressures (i.e., less than ∼7 Torr) it constituted a larger fraction.
The Cl2 concentration was ∼1 × 1016 molecule cm-3. The
reaction region of the discharge flow tube (for reactions 2 and
4) was 35 cm long (i.d. ) 3 cm). The linear gas flow velocity
in this region was roughly 400-700 cm s-1. The Cl atom
concentration in the flow tube was greater than ∼2 × 1014 atom
cm-3. Therefore, the time for conversion of O3 or Cl2O to ClO
in the flow tube was 50-85 ms, in which time reactions 2 and
4 essentially went to completion and consumed all O3 or Cl2O.
The mixture of ClO, Cl2, He, and N2 (or Cl2, O3/Cl2O, He, and
N2 when the microwave discharge was off) flowed into the
absorption cell, which was downstream of the flow tube.
Absorption Measurements. The concentrations of ClO, O3,
Cl2O, and Cl2 in the gas mixture were measured via UV
absorption. The output of a D2 lamp was passed through the
combination of the absorption cell and the reactor and focused
on to the entrance slit of a spectrograph. The 0.28 m spec-
trograph was equipped with a 1024 element diode array detector
operated at ∼230 K. The resolution of the spectrograph was
∼1 nm (with a 100 µm entrance slit) and it covered the
wavelength range of 220 to 370 nm. The total optical path length
was measured to be 40.4 cm. The entire optical path was
maintained at the temperature of the reactor.
tion by using the absorption at 265 nm and the known absorption
265
ClO
cross section at this wavelength16 (σ
) 5.29 × 10-18 cm2
molecule-1). The ClO spectrum at this wavelength is nonstruc-
tured and its cross section here is independent of resolution and
temperature.10 The ClO reference spectrum at the temperature
where k1 was being determined and needed for the spectral
subtraction was measured each day using the same setup, but
with a high ClO concentration ([ClO] ∼ 1 × 1014 molecule
cm-3). (Note that unlike our previous experiments involving
ClO reactions, we did not have to correct for the number density
gradients in the absorption cell.) The concentration of Cl2O or
O3 was changed to vary the ClO concentration while keeping
the Cl atom concentration nearly constant during the determi-
nation of a given value of k1.
As discussed in our previous publications, we could detect
ClO without interference from the other species that were
present. The vibrational band structure of ClO was used to
separate the absorption due to ClO from that due to other species
in the absorption cell. This technique of locking onto the
structure of ClO spectrum to determine the absorbance due to

5404 J. Phys. Chem. A, Vol. 105, No. 22, 2001
Goldfarb et al.
ClO and then using the continuum part of ClO absorption to
deduce the concentration has been successfully used in our
laboratory in the past.10,11 Also, as discussed in those papers,
we measure the column abundance of ClO in the absorption
cell. The ClO concentration was essentially the same over the
entire length of the absorption cell. Small losses of ClO and
their consequences to the concentration of ClO in the reactor
are discussed later.
signal processing are basically the same as those used in
previous works from this laboratory.
O Atom Generation. For measuring k1, O atoms were
produced by the pulsed photolysis of ClO at 308 nm (XeCl
excimer laser; 20 ns pulse width):
ClO hν(λ)308 nm)8 Cl + O
(5)
Once the concentration of ClO was measured, O atoms were
generated via photolysis and their temporal profile was mea-
sured, as discussed below. After the temporal profile measure-
ment, the concentration of ClO was again determined by
measuring spectra I2 and then I1. The measured O atom temporal
profiles were excluded from the determination of k1 if the ClO
concentrations measured before and after acquiring the temporal
profile differed by more than 5%.
The products of reaction 5 are assumed to be ground-state O
and Cl atoms because there is insufficient energy to produce
excited atoms. Also, for calculating the initial concentration of
O atoms, [O]0, we assume that the quantum yield for dissociation
is unity since ClO does not fluoresce and the bands are quite
broad. Note that the ratio of [ClO]/[O]0 would only be larger if
this quantum yield is not unity and does not affect the measured
value of k1. The photolysis laser fluence was varied over the
range 4 to 18 mJ pulse-1 cm-2. For a given series of
experiments, where the ClO concentration was varied to obtain
k1, the laser fluence was kept constant. Thus, the fraction of
ClO photolyzed remained the same and [O]0 increased with
increasing [ClO], i.e., [ClO]/[O]0 was constant. The fraction of
ClO photolyzed was usually between 0.2 and 0.8%, using a
ClO absorption cross section at 308 nm of 3.0 × 10-19 cm2
molecule-1. (The XeCl laser beam is not monochromatic and
has two peaks. The above cross section is the effective cross
section for the wavelength distribution of the energy in the laser
beam and the structured absorption of ClO reported by Trolier
et al.17) Thus, pseudo-first-order conditions were ensured. In a
few experiments, the laser fluence was decreased as the ClO
concentration increased to keep the [O]0 approximately constant.
The laser repetition rate was varied between 7 to10 Hz, thus
ensuring that the laser beam photolyzed a fresh gas mixture
with each pulse.
The photolysis beam was passed along the same path as the
D2 lamp beam. A movable mirror attached to a positive position
mount was placed between the D2 lamp and the absorption cell
to introduce the photolysis laser beam into the reactor. The
positive position mounts ensured that the photolysis laser beam
traversed the same volume each time. A laser power meter was
placed between the reactor and the spectrometer to measure the
laser fluence. After acquiring the temporal profile, the mirror
and the power meter were removed to pass the D2 lamp beam
through the absorption cell.
Reaction Cell and Atomic Resonance Fluorescence Detec-
tion System. The reaction cell for O atom detection was
specially designed to reduce scattered light and, thus, improve
the sensitivity for detection of O atoms. The cubic reaction cell
was constructed from anodized aluminum with an internal
volume of ∼250 cm3 and an internal cross section of ∼20 cm2.
The O atom resonance lamp and solar blind photomultiplier
tube (PMT) were mounted perpendicular to each other and
orthogonal to the direction of the gas flow. Light traps were
mounted opposite to the resonance lamp and PMT. The light
traps measured 3.75 cm by 3.75 cm and consisted of an
assembly of roughly 100 tightly stacked shiny stainless steel
razor blades. The knife-edges of the blades faced the lamp and
the PMT. This assembly appeared black because the knife-edges
scatter the light sideways. A 3-cm long evacuated collar with
two windows and a ring (internal radius of 0.5 cm) was located
between the resonance lamp and reaction cell. This collar
assembly, combined with a knife-edge machined into the
reaction cell, improved the collimation of the resonance lamp
radiation so that essentially all of it impinged on the light trap.
Also, baffles were machined into the cell to minimize scattered
light from the resonance lamp reaching the PMT. A CaF2
window mounted in front of the PMT prevented Lyman-R
radiation from reaching the detector.
The quantity of interest in calculating k1 is the ClO concen-
tration, [ClO]Rx, in the volume where O atoms are detected.
The second-order self-reactions,
ClO + ClO f Cl2 + O2
(6a)
f ClOO + Cl
f OClO + Cl
(6b)
(6c)
deplete ClO along the absorption cell and the length of the
reactor. The rate coefficient for these three reactions together
is k6 () k6a + k6b + k6c) ) 1.6 × 10-14 cm3 molecule-1 s-1 at
298 K. When the concentration of ClO is less than 5 × 1013
cm-3, the average concentration along the length of the
absorption cell, [ClO]avg, lies within 5% of [ClO]Rx, as discussed
in our previous papers.10,11 However, there is a difference
between the position of O atom detection used here and the
OH and IO detection region used in previous studies. In the
present experiments, the O atoms were detected nearer to the
end of the absorption cell (with respect to the flow direction),
while the OH and IO detection regions in previous studies were
in the middle of the absorption cell. In terms of the time
available for ClO loss, however, the O atom detection zone was
nearly at the midpoint between the entrance and exit of the
absorption measurement region. This is because the cross section
of the O atom detection cell was more than a factor of 4 larger
than that of the glass absorption cell (see Figure 1) and the flow
slowed substantially in the O atom detection region.
The O atom detection limit in this reactor was measured to
be better than 5 × 107 atom cm-3 for a 1s integration. This
limit was determined by creating a known concentration of O
atoms via photolysis of a known concentration of O3 in N2 with
a known fluence of 308 nm from an excimer laser. This
detection limit is at least a factor of 4 better than the detection
sensitivities we obtained previously in cylindrical glass reac-
tors.14 The improved sensitivity made it easier to study reaction
1, a radical-radical reaction, under pseudo-first order conditions
in O atoms while keeping the ClO radical concentrations
relatively low (i.e., minimizing secondary chemistry of the ClO
radicals) and in the presence of a large abundance of Cl2, which
absorbs the resonance lamp output and O atom fluorescence.
The atomic resonance fluorescence lamp, detection system, and
In an excess of Cl atoms, as in our present study, channel 6c
does not lead to a net loss of ClO because OClO is converted

Kinetics of the O + ClO Reaction
J. Phys. Chem. A, Vol. 105, No. 22, 2001 5405
back to ClO,
Cl + OClO f 2ClO
(7)
A small fraction (∼10%) of the ClOO generated in reaction 6b
is also converted to ClO via the reaction,
Cl + ClOO f 2ClO
(8a)
(8b)
Cl + ClOO f Cl2 + O2
Thus, some of the ClO loss via reaction 6 does not lead to a
net loss of ClO.
The wall loss of ClO in the glass flow tube is negligibly small.
However, we have not previously used an aluminum cell and,
thus, were unsure of the possible loss to this surface. Therefore,
we checked for the wall loss of ClO by changing the flow rate
of the gases through the reactor by a factor of ∼3.5 at 298 K,
∼2.5 at 231 K, and ∼4 at 360 K. The measured values of k1
were not affected as noted later.
When the concentration of ClO entering the absorption cell
was 1 × 1014 cm-3, its concentration is reduced by 20% by the
time the flow exited the reactor due to reaction 6. The measured
column average concentration was less than 10% lower than
the concentration in the O atom detection region. At lower ClO
concentrations, this difference is even smaller, as noted above.
It should be noted that the depletion of ClO along the length of
the apparatus was small and errors in k1 introduced because of
the correction to ClO concentration should be small. For
example, if the error in the correction was 50% when the net
ClO depletion was 20%, the maximum error in k1 would be
less than 10%.
Figure 3. A temporal profile of O atoms recorded at 255 K with [ClO]
) 6.3 × 1013 cm-3, yielding an initial O atom concentration of ∼1 ×
1011 cm-3. Data are shown every 20 µs and is the average of the adjacent
10 points. The data were collected with using 2 µs wide bins. The O
atom signals in most experiments were much larger and the uncertainties
in measured values of k1′ were smaller.
as described by Cady.18 The largest impurity in the Cl2O sample
was Cl2 (5%), which did not affect our measurements. The
fraction of OClO in our sample of Cl2O was below our detection
sensitivity, <0.13%. Even if OClO were present, it would be
converted to ClO via reaction with Cl. A mixture of Cl2 (10%,
electronic grade) in helium (99.98%) was used in the microwave
discharge to generate Cl.
The concentration of Cl2 in the reactor was determined from
measured pressure and flow rates (using calibrated electronic
mass flow meters). It agreed (within 5%) with the concentration
measured via UV absorptions using the diode array spectrom-
eter. In some experiments where N2O photolysis was used to
generate O atoms (see below), the concentration of N2O was
calculated from measured flow rates and pressure. Nitrogen was
the carrier gas and its concentration was also obtained from
measured flow rates and pressure.
At lower temperature, while reaction 6 slows down, ClO is
lost via an additional pathway,
ClO + ClO + M f Cl2O2 + M
(9)
For the residence times and pressure used here, the decomposi-
tion of Cl2O2 back to ClO can be neglected. At 250 K and 15
Torr, the difference between [ClO]avg and [ClO]Rx was ∼15%
for a residence time of 0.1 s. The residence time for a gas
mixture in the absorption/photolysis/reaction cell varied between
0.06 and 0.1 s. This difference was smaller at lower pressures
and shorter residence times. The calculated ClO concentrations
in the O atom detection region explicitly accounted for the loss
of ClO via reactions 6 and 9 at each given concentration of
ClO. We assumed that the possible reaction of O atoms with
Cl2O2 is slower than reaction 1 and did not correct k1 for this
possible loss process. The measured linear variation of k1′ with
[ClO] supports this assumption.
The temperature of the gas flowing through the absorption/
reaction cell was measured with two chromel-alumel thermo-
couples in direct contact with the gas flow. One thermocouple
was located in the Pyrex absorption cell and the other in the
reactor where the O atoms were detected. The thermocouples
were inserted into the gas flow for the temperature measurement
and withdrawn during the kinetic and absorption measurements.
The gas temperatures measured in the two locations agreed to
within 2 K under all our flow conditions. The temperature-
controlled absorption cell protruded into the reactor (See Figure
1) such that there was very little difference in temperature
between the absorption cell and the region where O atoms were
detected.
Results and Discussion
A representative O atom temporal profile measured in the
presence of excess ClO is shown in Figure 3 and, as expected,
it followed the equation for first-order kinetics:
ln([O]o/[O]t) ) (k1[ClO] + kd)t ) k1′t
(10)
where kd is the first-order rate coefficient for the loss of O atoms
in the absence of ClO. The O atom temporal profiles were
exponential for at least two lifetimes. A linear least-squares fit
of the measured ln([O]t) versus time to eq 10 yielded k1′. The
solid line in Figure 3 is the obtained fit. The temporal profiles
of O atoms were recorded for at least 5, and usually 6 or 7,
different ClO concentrations over the range (1 to 20) × 1013
molecule cm-3 to get one value of k1. A plot of k1′ vs [ClO]
yielded k1 as the slope; one such plot is shown in Figure 4 where
the line is a linear least-squares analysis of these data. The rate
coefficient k1 was measured at 10 different temperatures between
227 and 362 K. The obtained values, along with the experi-
mental conditions used to derive them, are given in Table 1.
The average value of the nine measurements at 298 K, weighted
according to the noted errors in Table 1, yields k1(298 K) of
(3.90 ( 0.24) × 10-11 cm3 molecule-1 s-1
.
Materials. Ozone was introduced to the flow tube by passing
N2 through a silica gel trap containing O3. Cl2O was prepared
The intercepts in the plots of k1′ vs [ClO] could not be
measured directly since we used ClO itself as the source of O

5406 J. Phys. Chem. A, Vol. 105, No. 22, 2001
Goldfarb et al.
TABLE 1: Summary of Experimental Conditions and Results for O + ClO Kinetic Study
flow
a
b
c
T
(K)
pressure
(Torr)
[ClO]corr
[O]0
velocity
laser fluence
k1
(1013 cm-3
)
(1011 cm-3
)
(cm s-1
)
(mJ cm-2 pulse-1
)
(10-11 cm3 molecule-1 s-1
)
362
362
343
323
317
298
298
298
298
298
298
298
298
298
6.7
5.7
5.5
10.5
8
10.5
10.5
6.7
11.5
11.8
22
8
16.7
18.3
3.7-11.7
4.0-12.4
2.5-10.6
0.4-11.2
3.3-11.0
3.2-13.6
2.0-25.0
2.5-6.9
2.3-6.6
1.8-6.0
3.1-15.0
4.0-13.0
1.3-7.5
0.8-8.0
4-11
2-5
480
98-150
490
340-525
460
280
504
500
200
240-300
240
150
560
23
11-9
8
5-8
7-12
20-17
22
31-2
18
4
7-9
5
2
5
3.74 (0.17
4.12 ( 0.11
3.80 ( 0.20
3.60 ( 0.13
4.00 ( 0.15d
4.14 ( 0.18
3.78 ( 0.17
3.50 ( 0.35
3.28 ( 0.47
3.97 ( 0.71
4.02 ( 0.24
3.43 ( 0.23
4.36 ( 0.28d
4.27 ( 0.24d
3.90 ( 0.24
4.12 ( 0.35
4.20 ( 0.20
4.12 ( 0.14
4.46 ( 0.21
4.1 ( 0.37
1-4
0.1-4
1-6
3-11
2-27
0.2-10
2-5
0.3-1.2
1-6
1-3
0.1-0.6
0.2-2
average
1-5
200
279
279
255
237
231
231
231
227
7.3
11.0
6.0
7.8
7.0
9.0
22.0
8.0
4.6-8.4
3.2-12.6
2.0-17.0
2.0-12.0
1.3-4.4
1.7-9.2
3.4-11.2
1.8-9.1
560
300
590
730
500
770
280
780
5-13
6
4-8
11-7
3
13
6
1-3
0.4-6
1-4
0.2-1
1-8
4.1 ( 0.28
1-2
1-3
4.32 ( 0.44
12-7
4.43 ( 0.13d
a This is the concentration of ClO in the region where O atoms were detected. It was calculated by correcting the measured column ClO abundance
for the loss of ClO in the absorption cell, as described in the text. b The concentrations of O atoms were calculated from the measured fluence and
ClO concentrations and the known absorption cross section of ClO at 308 nm assuming the quantum yield for O atom production to be unity.c The
quoted uncertainties are 1σ and were obtained from the fit of measured values of k1′ at various values of [ClO] to expression 10, as discussed in
the text. d Cl2O was used in place of O3 to generate ClO via its reaction with Cl atoms.
photolyzed at 193 nm via the sequence:
N2O + hν f O(1D) + NO
(12)
(13)
followed by
O(1D) + N2 f O(3P) + N2
We used flow rates that were similar to those used in measuring
k1 and a 193 nm excimer laser beam whose diameter was the
same as that of the 308 nm beam used to photolyze ClO. The
O atom loss rate constant due to pump out and reaction with
impurities was measured to be 100 to 200 s-1. This is similar
to the values measured previously in our laboratory. Addition
of ∼1 × 1016 molecule cm-3 of Cl2 increased the O atom loss
to 600 ( 200 s-1. Note that the photolysis of Cl2 at 193 nm is
very small due to its low absorption cross section (<1 × 10-21
cm2).20 Therefore, it is clear that at least half of the O atoms
are lost in the absence of ClO due to reaction with Cl2. The
uncertainties in the intercepts in the k1′ vs [ClO] were such that
the rate coefficient measured with Cl2 overlapped with the
intercepts within the combined uncertainties. In any case, even
if there was an error of 20-30% in the value kd, it did not
introduce a large uncertainty in k1 because the values of k1′
measured in the presence of ClO were many thousands per
second.
Figure 4. A plot of the measured values of k1′ vs [ClO] at 255 K. The
error limits of the individual values of k1′ are 2σ from the fit of the O
atom temporal profile to eq 10. The line is a linear least-squares fit to
the equation: k1′ ) k1 ×[ClO] + kd. The intercept is roughly consistent
with the loss of O atoms due to pump out and reaction with Cl2.
atoms. The fitted values were in the range 200 and 1000 s-1
.
The O atom temporal profiles measured with Cl2, O3, and N2
bath gas, the same gases as for making ClO but with the
microwave discharge off (thus no ClO), could not be used to
determine the intercept because the 308 nm radiation photolyzed
Cl2 and generated ClO. Therefore, the loss of O atoms in the
absence of ClO was not routinely measured. This loss is
determined by reaction with impurities, pump out from the
detection zone, and its reaction with Cl2:
Figure 5 shows the measured values of k1 in an Arrhenius
form. A fit of the ln(k1) vs 1/T to a straight line yielded the
following value for the temperature dependence of k1:
k1(T) ) (3.0 ( 0.4) × 10-11
×
O + Cl2 f ClO + Cl
(11)
exp((75 ( 40)/T) cm3 molecule-1 s-1
where k11(298) ) 2.9 × 10-14 cm3 molecule-1 s-1 19
.
The values
of the first-order rate coefficient for the removal of O atoms
via pump out and reactions with impurities in the diluent gas
were measured in a set of experiments where N2O was
where the uncertainty in E/R is the 2σ precision of the fit and
uncertainty in A is twice the standard deviation in A defined as
σA ) Aσln A
.

Kinetics of the O + ClO Reaction
J. Phys. Chem. A, Vol. 105, No. 22, 2001 5407
O3 and ClO in the reaction zone and related by the equation:
[O]O
σ3O08
[O3]
3
3
)
×
(14)
308
[O]ClO [ClO]
σ
ClO
With the O atom quantum yields for O3 photolysis at 308 nm
under our conditions (N2 carrier gas) and of ClO photolysis at
308 nm both being unity, we could estimate [ClO]Rx from the
measured signals and the [O3] along with the known absorption
cross sections. The estimated value of [ClO] was within 15%
of the value calculated from the measured ClO absorption using
the diode array spectrometer and there was no systematic
difference in the two values. This agreement serves as another
verification of our determination of the ClO concentration. Such
measurements could have been done for each temporal profile
used to measure k1. However, the fluctuations in the O atom
detection sensitivity precluded such routine measurements.
The above tests and checks give us confidence in our
determined values of [ClO]Rx. Therefore, our measured rate
coefficients are also uncertain by that amount. Including this
Figure 5. A plot of the measured values of k1 (on a logarithmic scale)
against the reciprocal of temperature. The line is a linear least-squares
fit of ln(k1) vs 1/T data. The values of k1 measured using different ClO
radical sources are shown as: circles (Cl + O3) and squares (Cl +
Cl2O). The value shown at 298 K, open circle, is the average of nine
measurements listed in Table 1 for this temperature.
Estimating the Uncertainty in k1. A number of experimental
parameters were varied in the determination of k1 to minimize
the influence of systematic errors. These parameters included:
(1) total pressure and flow velocity (varied a factor of 3), (2)
different ClO radical precursors (O3 and Cl2O), (3) altering the
[ClO] over an order of magnitude, (4) changing the [ClO]/[O]o
ratio by a factor of 4, and (5) varying the photolysis laser fluence
a factor of 3. Variation of these parameters did not lead to
significant changes in the value of k1 (Table 1).
estimated systematic uncertainty yields k1(298K)
)
(3.90 ( 0.63) × 10-11 cm3 molecule-1 s-1 and k1(T) )
(3.0 ( 0.7) × 10-11 exp[(75 ( 40)/T] cm3 molecule-1 s-1
.
One of the possible complications in measuring k1 is the
potential interference through the reaction of ClO with O2(1Σ),
which can be a product of reaction 1.
O + ClO f O2(1Σ) + Cl
(15)
The most significant systematic uncertainty in the determi-
nation of k1 lies in the value of [ClO]Rx used to calculate the
rate coefficient. The uncertainty in the measured gas tempera-
ture, (2 K, has only a small (<2%) affect on the ClO
concentration and measured rate coefficient. The accuracy of
[ClO]Rx is dependent on several factors each of which have
associated uncertainties: the ClO absorption cross section,
calculation of [ClO]Rx from [ClO]avg, and the accuracy of the
absorption measurements and spectral analysis. The use of the
structured part of ClO spectrum to determine ClO absorption
and extracting ClO concentration from the unstructured part of
the spectrum leads to its better determination. As seen in Figure
2, the maxima in the residual was about 6% of the absorption
due to ClO in the reactor. Therefore, we believe that we could
measure ClO concentration with a precision better than about
6%. These factors are discussed elsewhere.10,11 This spectral
analysis method removes the common sources of error due to
baseline stability and interference from nonstructured absorbers.
When we combine this precision with the uncertainty in the
absorption cross section of ClO in the unstructured part (∼265
nm), about 5%, and the uncertainty in calculating the concentra-
tion of ClO in the O atom detection zone, we estimate the
uncertainty in the ClO concentration in the O atom detection
zone to be ∼15% at the 95% confidence level.
The subsequent reaction of O2(1Σ) with ozone, if present, can
regenerate O atoms. However, in our study, we used excess Cl
and there was no possibility of O atom regeneration via the
reaction of O2(1Σ) with O3 because of the absence of ozone.
Another possible potential source of error was the loss of ClO
on the walls of the aluminum reactor. As shown in Table 1,
variation of the flow rate through the system by factors of
roughly 3 did not vary the measured value of k1. Given that the
cross section of the aluminum cell is large and the flow rates
were quite large, we should not expect a significant wall loss
rate.
Comparison with Previous Measurements of k1. Previous
measurements of k1 include mostly flow tube studies, with one
notable exception being that by Nicovich et al.,8 who used
pressures of ∼50 Torr in a pulsed photolysis system. Of the
flow tube studies, only Leu3 assumed a stoichiometric conver-
sion of O3 (or Cl2O) to ClO and did not determine the ClO
concentration by another method. Since the studies of Leu, it
has become clear that neither O3 nor Cl2O is stoichiometrically
converted to ClO in an excess of Cl atoms due to the reaction
of Cl atoms with vibrationally excited ClO.3,9,10,21 Recently, we
have shown the difference between the expected and measured
ClO in such a system to be substantial.10 Therefore, the values
of k1 reported by Leu may be suspect. However, it should be
noted that at the low concentrations of ClO used by Leu, the
stoichiometry for conversion of O3 to ClO is nearly 1-to-1and
the rate constant may not be very far off the true value.
Margitan4 measured the concentration of ClO via UV absorption
in the gas flow after the O atom temporal profiles were
measured. He used the 277.5 nm band to quantify the ClO
concentration. The cross section of this band, contrary to what
Margitan noted, is quite sensitive to the resolution of the
measurement. At the resolution of 0.3 to 0.5 nm (assuming it
to be full width at half-maximum) used by Margitan, we estimate
the ClO absorption at the peak to be ∼6 × 10-18 cm2 and
Separate measurements were performed to test our determi-
nation of [ClO]Rx. First, O atom signals were measured as a
function of laser fluence (varied between 4 and 23 mJ pulse-1
cm-2) to confirm that it varied linearly with O atom concentra-
tion. Then, [ClO]avg was measured by absorption and the initial
O atom signal from the photolysis of ClO, [O]ClO, was measured.
The microwave discharge was then turned off. The O3 concen-
tration was measured by absorption and the O atom signal from
O3 photolysis at 308 nm, [O]O , was measured. All other
3
conditions were held constant during this sequence. The ratio
of the measured signals is proportional to the concentration of

5408 J. Phys. Chem. A, Vol. 105, No. 22, 2001
Goldfarb et al.
TABLE 2: The Values of k1 at 298 K Reported in Previous Studies and Corrections to Some of Those Values
reported k
modified k
study
(298 K)
(298 K)
reason for the modification
Leu3
3.62
4.2
3.62
Margitan4
3.5
adjusted for the resolution dependence of the
absorption cross section at 277.5 nm
Schwab et al.6
Zahniser and Kaufman26
Ongstad and Birks7
Nicovich et al.21
this work
3.5
4.3
3.76
3.7
3.90
3.5
4.13
adjusted for current recommendation for k(Cl + O3)22
adjusted for current recommendation for k(O + NO2)22
3.63
3.7
3.90
average
3.7 ( 0.4 (2σ)
TABLE 3: Summary of k1 Measurements (The values are those quoted by the authors; corrections to k1(298 K) shown in Table
2 are not included here.)
[ClO]
measurement
reference
k(298K)a
f(298K)
Aa
E/R (K)
∆E/R (K)
T (K)
techniqueb,c
Basco and Dogra25
1.2
298
Indirectd
Bemand et al.24
5.3 ( 0.8
5.7 ( 2.3
5.2 ( 1.6
4.35
298
DF -RF(O)
298
DF-MS(ClO)
DF-RF(O)
Clyne and Nip23
Zahniser and Kaufman2
Leu3
1.07 ( 0.30
2.12 ( 0.11
5.0 ( 1.0
224 ( 76
-75 ( 40
96 ( 20
0
220-426
220-298
236-422
241-298
298
mass spec
DF-RF(Cl)e
DF-RF(O)
3.62
4.2
3.5 ( 0.6
3.5
indirect
Margitan4
4.2 ( 0.8
UV absorption
mass spec
LMR
DF-LFP-RF(O)
Ongstad and Birks5
Schwab et al.6
DF-CL(O)f
3.5 ( 0.5
252-347
DF-LMR(ClO)
DF-RF(O)
Ongstad and Birks7
Nicovich et al.21
this worke
3.76
3.75
3.9 ( 0.6
3.8
5
3.7
2.61 ( 0.60
-97 ( 64
-263 ( 60
-75 ( 40
-70
220-387
231-367
227-362
mass spec
DF-CL(O)f
1.55 ( 0.33
indirect
LFP-RF(O)
DF/LFP-RF(O)
recommendation
recommendation
3.0 ( 0.8
UV absorption
NASA12
1.2
1.26
1.15
3.0
5
70
250
25
-
-
CODATA27
0
our recommendation
2.5
-110
a Units of 10-11 cm3 molecule-1 s-1 b Abbreviations for various techniques used: DF, discharge flow; RF, resonance fluorescence; MS, mass
.
spectrometry; LMR, laser magnetic resonance; CL, NO2 chemiluminescence; LFP, laser flash photolysis. c The species monitored is in parentheses.
d The rate coefficient was deduced from a mechanism that accounted for the changes in [ClO] in the flash photolysis of OClO and Cl2O. e k1
measured relative to k(Cl + O3). f Values include estimated systematic error.
definitely less than the 7.26 × 10-18 cm2 used in his analysis.
Thus, we estimate that his measured rate constants are high by
∼20%, leading to a revised 298 K value of 3.5 × 10-11 cm3
molecule-1 s-1. Ongstad and Birks,5,7 measured k1 by directly
determining the concentration of ClO and also by converting
ClO to NO2 via addition of NO and measuring the rate
coefficient for the reaction of O with NO2. We can change their
measured value relative to the O + NO2 reaction by using the
currently accepted value.22 Schwab et al. 6 measured k1 in both
excess ClO and O and their value at 298 K is in excellent
agreement with ours. Zahniser and Kaufman,2 measured k1
relative to k2. They argue that it is best to use the value of k2
measured in their system. However, if we were to use the
currently accepted value of k2, their measured value of k1 is
reduced by only ∼5%. The value of k1(298 K) reported by
Nicovich et al. 8 is in excellent agreement with our value. Table
2 lists the small changes to the previously reported values of k1
because of the reasons noted above. All the previous values of
k1(298) (except the very old data of Bemand et al.23,24 and Basco
and Dogra25) agree with each other very well. The average of
these values has a standard deviation of 10% at the 95%
confidence level. Given that all these studies used very different
methods and potentially had different systematic errors, we can
confidently place an uncertainty factor of 1.15, in NASA/JPL
notation, at the 1σ level. This is clearly one of the better-
determined radical-radical reaction rate coefficients at 298 K.
A summary of the previous studies of the temperature
dependence of k1 is given in Table 3 and also shown in Figure
6. In this figure, we have not included the data of Bemand et
al. and Leu; these are the two studies that reported a positive
activation energy for this rate coefficient. Almost all the previous
determinations of the temperature dependence of k1 were carried
Figure 6. A comparison of our results (filled circles) with those
discussed in the text. The data of Zahniser and Kaufman2 (open circles)
and of Margitan4 (open diamond) have been corrected, as discussed in
the text. The data of Schwab et al.,6 (bow-tie) is that reported by the
authors as the slopes of their k1′ vs [ClO] plots. The data of Nicovich
et al.21 (open squares) and Ongstad and Birks7 (filled triangles) are
also shown. The error bounds obtained from this combined data set,
and discussed in the text, are shown for the 2σ level. The dashed line
is the current recommendation of DeMore et al.12
out over limited ranges and by measuring at just a few
temperatures. For example, Zahniser and Kaufman, Schwab et
al., and Margitan measure k1 at two temperatures other than
298 K. Leu measured k1 at five temperatures and obtained a
distinct positive temperature dependence. It is not clear as to
why he saw such a dependence. The results of the more
extensive studies of Ongstad and Birks, and of Nicovich et al.

Kinetics of the O + ClO Reaction
J. Phys. Chem. A, Vol. 105, No. 22, 2001 5409
A ) 2.4 × 10-11 cm3 molecule-1 s-1;
can be combined with our results to deduce the temperature
dependence of k1:
E/R ) 110 K; ∆E/R ) 25 K
These error bounds, at the 95% confidence level, and the mean
values are also shown in Figure 6.
k1(T) ) (2.39 ( 0.42) × 10-11
×
exp [(130 ( 50)/T] cm3 molecule-1 s-1
Acknowledgment. This work was funded in part by NASA's
Upper Atmospheric Research Program. L.G. thanks NASA for
the Global Change Research Doctoral Fellowship.
where the error bars are at the 95% confidence level and σA )
Aσln A. Also, the preexponential factor has been slightly lowered
References and Notes
to reproduce k1(298 K) ) 3.7 × 10-11 cm3 molecule-1 s-1
.
(1) Scientific Assessment of Ozone Depletion: 1998; World Meteoro-
logical Organization, 1999.
(2) Zahniser, M. S.; Kaufman, F. J. Chem. Phys. 1977, 66, 3673-
3681.
These are the three studies that were carried out over extended
temperature ranges, a necessity for quantification of small
activation energies. These three studies also clearly observed
that k1 increased with decreasing temperature. We can also
include the few points of Zahniser and Kaufman, Schwab et
al., and Margitan (after correction for the cross section at 277.5
nm). Note that Margitan measured the concentration of ClO at
298 K and, hence, we do not have to account for the temperature
dependence of the cross section of ClO at 277.5 nm. Figure 6
shows the data from these studies as an Arrhenius plot. Clearly,
the collective data shows a negative temperature dependence
and a linear least-squares analysis of ln(k1) vs 1/T from all these
studies yields
(3) Leu, M. T. J. Phys. Chem. 1984, 88, 1394-1398.
(4) Margitan, J. J. J. Phys. Chem. 1984, 88, 3638-3643.
(5) Ongstad, A. P.; Birks, J. W. J. Chem. Phys. 1984, 81, 3922-3930.
(6) Schwab, J. J.; Toohey, D. W.; Brune, W. H.; Anderson, J. G. J.
Geophys. Res. 1984, 89, 9581-9587.
(7) Ongstad, A. P.; Birks, J. W. J. Chem. Phys. 1986, 85, 3359-3368.
(8) Nicovich, J. M.; Wine, P. H.; Ravishankara, A. R. J. Chem. Phys.
1988, 89, 5670-5679.
(9) Burkholder, J. B.; Hammer, P. D.; Howard, C. J.; Goldman, A. J.
Geophys. Res. 1989, 94, 2225-2234.
(10) Kegley-Owen, C. S.; Gilles, M. K.; Burkholder, J. B.; Ravishankara,
A. R. J. Phys. Chem. 1999, 103, 5040-5048.
(11) Turnipseed, A. A.; Gilles, M. K.; Burkholder, J. B.; Ravishankara,
A. R. J. Phys. Chem. 1997, 101, 6667-6678.
(12) 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.
k1(T) ) (2.53 ( 0.45) × 10-11
×
exp [(110 ( 50)/T] cm3 molecule-1 s-1
(13) Goldfarb, L. The photochemistry and kinetics of chlorine com-
pounds important to stratospheric mid-latitude ozone destruction. Ph.D.,
University of Colorado, 1997.
(14) Goldfarb, L.; Harwood: M. H.; Burkholder, J. B.; Ravishankara,
A. R. J. Phys. Chem. 1998, 102, 8556-8563; Goldfarb, L.; Schmoltner,
A.-M.; Gilles, M. K.; Burkholder, J. B.; Ravishankara, A. R. J. Phys. Chem.
1997, 101, 6658-6666.
where the errors are the same as noted above.
Given that the studies of Nicovich et al.8 were at pressures
of tens of Torr, ours in the range of 5-20 Torr, and some of
the previous flow tube studies in the range of 1-3 Torr, we
can safely assume that k1 is essentially independent of pressure,
i.e., to within 20%. It appears that the higher pressure study of
Nicovich et al. (∼50 Torr) yields slightly more negative
activation energies than our results (6-20 Torr) and those from
the discharge flow studies (1-2 Torr). These differences are
small and there are sufficient systematic errors that we cannot
make a strong correlation between pressure and measured
activation energy. So, we do not attach much significance to
this small difference. The activation energy derived from the
combined data noted above should be appropriate for atmo-
spheric calculations since the pressures and temperatures at
which this reaction is important in the atmosphere has been
covered in the laboratory. It would be interesting to measure
this rate coefficient at higher pressures and, possibly, higher
temperatures to see if the O-ClO intermediate can be quenched
at higher pressure or if the reaction can proceed directly (i.e.,
without sampling the intermediate state) at higher temperatures.
We suggest the following expression obtained by combining
our results with those from selected previous studies for
stratospheric calculations:
(15) Burkholder, J. B.; Wilson, R. R.; Gierczak, T.; Talukdar, R.;
McKeen, S. A.; Orlando, J. J.; Vaghjiani, G. L.; Ravishankara, A. R. J.
Geophys. Res. 1991, 96, 5025-5043; Burkholder, J. B.; Talukdar, R. K.;
Ravishankara, A. R. Geophys. Res. Lett. 1994, 21, 585-588.
(16) Trolier, M.; Mauldin, R. L., III.; Ravishankara, A. R. J. Phys. Chem.
1990, 94, 4896-4907.
(17) Troe, J. J. Chem. Phys. 1977, 66, 4745-4757.
(18) Cady, G. H. Inorgan. Synth. 1957, 5, 156.
(19) Wine, P. H.; Nicovich, J. M.; Ravishankara, A. R. J. Phys. Chem.
1985, 89, 3914-3918.
(20) Hubinger, S.; Nee, J. B. J. Photochem. Photobiol A: Chem. 1995,
86, 1-7.
(21) Nicovich, J. M.; Wine, P. H.; Ravishankara, A. R. J. Chem. Phys.
1988, 89, 5670-5679.
(22) Sander, S. P.; Friedl, R. R.; DeMore, W. B.; Golden, D. M.; Kurylo,
M. J.; Hampson, R. F.; Huie, R. E.; Moortgat, G. K.; Ravishankara, A. R.;
Kolb, C. E.; Molina, M. J. Chemical Kinetics and Photochemical Data for
Use in Stratospheric Modeling, EValuation No. 13; Jet Propulsion Labora-
tory, Pasadena, CA, 2000.
(23) Clyne, M. A. A.; Nip, W. S. J. Chem. Soc., Faraday Trans. 1 1976,
72, 2211-2217.
(24) Bemand, P. P.; Clyne, M. A. A.; Watson, R. T. J. Chem. Soc.,
Faraday Trans. 1 1973, 69, 1356-1374.
(25) Basco, N.; Dogra, S. K. Proc. R. Soc. A. 1971, 323, 29; 323, 401.
(26) Zahniser, M. S.; Kaufman, F.; Anderson, J. G. Chem. Phys. Lett.
1976, 37, 226-231.
(27) Atkinson, R.; Baulch, D. L.; Cox, R. A.; Hampson, R. F., Jr.; Kerr,
J. A.; Rossi, M. J.; Troe, J. J. Phys. Chem. Ref. Data 1997, 26, 521.
k1(298 K) ) 3.7 × 10-11 cm3 molecule-1 s-1; f(298) ) 1.15