Absolute rate constant of the OH + ClO reaction at temperatures between 218 and 298 K

Article abstract of DOI:10.1021/jp012426l

The total absolute rate constant (k_1a + k_1b) for the reactions OH + ClO → Cl + HO₂ (k_1a) → HCl + O₂ (k_1b) has been determined in the temperature range from 218 to 298 K at a total pressure

Full text of DOI:10.1021/jp012426l

10544
J. Phys. Chem. A 2001, 105, 10544-10552
Absolute Rate Constant of the OH + ClO Reaction at Temperatures between 218 and
298 K
Jin Jin Wang and Leon F. Keyser*
Atmospheric Chemistry Element, Earth and Space Sciences DiVision, Jet Propulsion Laboratory,
California Institute of Technology, Pasadena, California 91109
ReceiVed: June 26, 2001; In Final Form: August 28, 2001
The total absolute rate constant (k1a + k1b) for the reactions OH + ClO f Cl + HO2 (k1a) f HCl + O2 (k1b)
has been determined in the temperature range from 218 to 298 K at a total pressure of 1 Torr of helium.
Pseudo-first-order conditions were used with ClO in large excess over initial OH. ClO was produced by
reacting Cl atoms with an excess of O3; both ClO and O3 were quantitatively determined by UV
spectrophotometry between 210 and 310 nm. Two sources of OH were used: the reaction of F atoms with
excess water vapor and the reaction of H atoms with an excess of NO2. OH was monitored by resonance
fluorescence near 308 nm. The absolute rate constant was determined from plots of ln [OH] vs time and
absolute ClO concentrations. At 298 K, the result is (2.22 ( 0.33) × 10-11 cm3 molecule-1 s-1. The temperature
dependence expressed in Arrhenius form is (7.2 ( 2.2) × 10-12 exp[(333 ( 70)/T] cm3 molecule-1 s-1. The
uncertainties are at the 95% confidence limits and include statistical errors from the data analysis and estimates
of systematic experimental errors. Numerical simulations show that under the experimental conditions used
secondary reactions did not interfere with the measurements. Numerical simulations and experimental checks
also show negligible loss of ClO between the time when it is determined by UV absorption and the start of
the OH + ClO reaction.
Introduction
have shown that stoichiometric conversion to ClO cannot be
assumed.7,9,10 Also, some of the early studies3-5,7 used excess
Cl atoms, which requires a correction for the regeneration of
OH by reaction 5b.
The reaction of OH with ClO (eq 1) is important in
partitioning reactive and nonreactive forms of chlorine in the
OH + ClO f Cl + HO2
OH + ClO f HCl + O2
(1a)
(1b)
Cl + HO2 f HCl + O2
Cl + HO2 f OH + ClO
(5a)
(5b)
stratosphere. Reaction pathway 1a interconverts reactive species
in the HOx and ClOx families and is a chain propagation step.
Pathway 1b converts two radical species to relatively inert
reservoir species and serves as a chain termination step. The
effect of reaction 1 on stratospheric chlorine chemistry is very
sensitive to the branching ratio of these two pathways and to
the absolute rate constant for the total reaction.1 We recently
measured the HCl yield, k1b/(k1a + k1b), and found a value of
(9.0 ( 4.8)%.2 In the present study, we determine the absolute
total rate constant for this reaction.
These issues could explain some of the disagreement among
the results. To avoid the correction for reaction 5b, several more
recent studies6-9 reacted Cl atoms with excess O3 to form ClO.
Three of the previous studies6,8,9 reported a slight negative
temperature dependence, and two studies4,5 found the reaction
to be temperature-independent.
In the present study, a discharge-flow system is used to
determine the absolute rate constant of reaction 1 in the
temperature range from 218 to 298 K. Pseudo-first-order
conditions are used with [ClO] . [OH]. The rate constant is
determined directly from the slopes of ln [OH] vs time plots
and absolute [ClO]. Numerical models are used to assess the
importance of secondary chemistry. The results are compared
with previous measurements of this rate constant.
Previous measurements of k1 at 298 K range from 9.1 × 10-12
to 24.4 × 10-12 cm3 molecule-1 s-1 3-9
.
ClO was produced by
reacting Cl atoms with O3, OClO, or Cl2O:
Cl + O3 f ClO + O2
(2)
Experimental Section
Cl + OClO f 2 ClO
(3)
(4)
The apparatus used in the present study has been described
in detail recently,2,11 and only a brief summary is presented here.
Reactor. The temperature-controlled Pyrex main reactor has
an internal diameter of 5.04 cm and is 62 cm in length. At the
downstream end, it is connected to a stainless steel resonance
fluorescence cell. Pressure is measured by using a 10 Torr
capacitance manometer connected to a port between the reactor
Cl + Cl2O f ClO + Cl2
In the earliest studies,3,4 absolute concentrations of ClO were
determined by equating [ClO] to initial O3 or Cl2O; later studies
* To whom correspondence should be addressed. Fax: (818) 354-5148.
E-mail: Leon.F.Keyser@jpl.nasa.gov.
10.1021/jp012426l CCC: $20.00 © 2001 American Chemical Society
Published on Web 10/25/2001

Absolute Rate Constant of the OH + ClO Reaction
J. Phys. Chem. A, Vol. 105, No. 46, 2001 10545
and the fluorescence cell. At the upstream end are connections
to a fixed ClO source and a movable OH source described in
separate sections below. The inner surface of the reactor and
outer surface of the movable OH source are coated with
halocarbon wax to minimize wall loss of reactive species.
Temperatures in the reaction zone are maintained within (2 K
by using refrigerated bath circulators to pass heat-exchange
fluids (water or methanol) through a cooling jacket. Tempera-
tures are monitored by two thermocouples located inside each
end of the cooling jacket. Upstream of the main reaction zone,
the reactants pass through a precooling region the length of
which depends on the position of the movable source; average
lengths are about 50 cm, which gives residence times near 50
ms. This region allows the reactants to cool from room
temperature to the main reactor temperature before the reaction
starts.
reactor, their concentrations measured in the absorption cell are
adjusted by using eq 8
[M]m ) (Pm/Pa)(Fa/Fm)(Ta/Tm)[M]a
(8)
where M is ClO or O3, P is the total pressure, F is the total
flow rate, and T is the absolute temperature; the subscripts m
and a refer to the main reactor and the absorption cell,
respectively. To minimize calibration errors, the same flow
meter and pressure gauge are used for the measurements in the
absorption cell and the main reactor. Production efficiencies
based on the fraction of initial Cl2 converted to ClO average
between 60% and 75%.
OH Sources. At temperatures between 260 and 298 K, a
movable injector is used to produce OH by adding F atoms to
an excess of water vapor (eq 9). The F atoms are produced in
ClO Source. ClO is formed by reacting Cl atoms with an
excess of O3 (eq 2) in a fixed 5.0 cm diameter reactor located
upstream of the main reaction zone; all surfaces except the Cl2
discharge region are coated with halocarbon wax. Cl atoms are
produced in a 2.45 GHz discharge (50 W) of dilute mixtures of
Cl2 in He. Ozone is formed by passing O2 through a high voltage
discharge ozonator and trapping the O3 on silica gel at 195 K;
the excess O2 is pumped off and the O3 is added to the source
by passing a stream of He through the cold silica gel trap.
Conditions are set so that the Cl + O3 reaction is complete
before the ClO exits the reactor.
ClO and O3 Determination. After formation in the source
reactor, the ClO is passed through a quartz absorption cell 3.0
cm in diameter and 50.1 cm in length; the cell walls, windows,
and all connecting tubes downstream are also coated with
halocarbon wax. ClO and O3 are quantitatively determined by
using absorption spectrophotometry in the UV between 210 and
310 nm. The collimated output of a 30 W deuterium lamp passes
through the absorption cell and is focused on the entrance slit
of a 0.3 m imaging spectrograph. A 600 groove/mm grating is
used with an entrance slit width of 20 µm to give a resolution
of 0.26 nm (fwhm) as determined by using the Hg line at 253.65
nm. The absorption spectrum is recorded by using a photodiode
array situated at the exit focal plane of the spectrograph. To
avoid any saturation effects, the exposure time is set to 0.25 s,
which is at the middle of the linear response region of the diodes
under our experimental conditions. A low-pressure Hg lamp is
used to calibrate the wavelength scale of the spectrograph-
diode array detector by recording several atomic lines between
253.65 and 334.15 nm.
F + H2O f OH + HF
(9)
a 2.45 GHz discharge of dilute F2 in helium. An alumina tube
is used for the F2 discharge, which is operated at a total power
of about 20 W. All surfaces except the discharge tube are coated
with halocarbon wax. Water vapor is added by passing a stream
of helium through a water bubbler maintained at 18 °C. Flow
rates of H2O are determined from the He flow rate, the total
pressure, and the vapor pressure of H2O. Typically, H2O
concentrations in the source are about 2.8 × 1014 molecules
cm-3. The reaction length is set to give a reaction time near
2.5 ms. Using k9 ) 1.4 × 10-11 cm3 molecule-1 s-1 18
,
we
calculate that reaction 9 is complete for all the experimental
conditions used. On the basis of the initial concentrations of F2
added, reaction efficiencies to form OH average 55%.
At 218 and 239 K, OH is generated by adding H atoms to an
excess of NO2 (eq 10). The H atoms are formed in a microwave
H + NO2 f OH + NO
(10)
discharge of dilute mixtures of H2 in He; a quartz tube is used
at powers near 20 W. Again, all surfaces except the discharge
region are wax coated. In the source, concentrations of NO2
and O2 are each about 3 × 1013 molecules cm-3. The reaction
time is set at 3 ms. Using k10 ) 4.0 × 10-10 exp(-340/T) cm3
molecule-1 s-1 18
we estimate the reaction is complete before
,
OH is added to ClO in the main reactor. On the basis of initial
H2 concentrations, reaction efficiencies to form OH by reaction
10 average about 70%.
OH Detection. Hydroxyl radicals are detected by resonance
fluorescence at a fixed point downstream of the reaction zone.
A resonance lamp operated at 50 W of microwave power is
used to excite the OH fluorescence. A stream of He saturated
with water vapor is passed through the lamp at a total pressure
near 4.5 Torr. Hydroxyl radical fluorescence near 308 nm is
detected at right angles to the lamp by using an interference
filter, photomultiplier tube, amplifier-discriminator, and dual
counter-timer interfaced to a computer for data acquisition and
analysis. The filter is placed between the photomultiplier and a
suprasil quartz window that makes the vacuum seal to the flow
tube, so the filter is never in contact with any of the reagents
used. A Corning filter (0-53) is placed in front of the OH lamp
to cut off radiation at wavelengths shorter than about 290 nm.
OH Calibration. Although we do not need absolute con-
centrations of OH to determine the rate constant, we want to
know them as well as possible to do accurate modeling of the
reaction. The system is calibrated by generating specific amounts
of OH by adding known concentrations of NO2 to an excess of
The observed spectrum is a sum of O3 and ClO spectra; to
separate them, a spectral subtraction method is used. This
technique has been described in detail2,9 and will not be
discussed here. Both ClO and O3 are determined at 253.65 nm
by using eq 6, where M is ClO or O3, A is the absorbance
defined by eq 7, σM is the absorption cross section, and L is the
absorption path length.
[M] ) A/(σML)
(6)
(7)
A ) ln[(Io - Ib)/(It - Ib)]
In eq 7, Io, Ib, and It are the incident, background, and transmitted
light intensities, respectively. The cross sections used are 11.58
12
× 10-18 cm2 molecule-1 for O3 and 4.23 × 10-18 cm2
molecule-1 for ClO.13-17
Generally the total flow rates, pressures, and temperatures
in the absorption cell differ from those in the main reactor.
Therefore, to determine ClO and O3 concentrations in the main

10546 J. Phys. Chem. A, Vol. 105, No. 46, 2001
Wang and Keyser
expand into a 5 L flask to which He is then added. NO2 for use
in the OH source is produced by adding an excess of O2 to NO
and storing the mixture in a thermally insulated Pyrex flask of
known volume (4821 cm3). The main carrier He is further
purified by passage through a molecular sieve (Linde 3A) trap
at 77 K just prior to use.
Results
Experimental Conditions. Concentrations of ClO are in the
range of 2.8 × 1012 to 2.2 × 1013 cm-3 with initial OH
concentrations from 2 × 1011 to 1 × 1012 cm-3. Initial
stoichiometric ratios range from 7 to 52 and average greater
than 25. Concentrations of O3 are between 6 × 1012 and 7 ×
1013 cm-3. Temperatures are in the range of 217.5 to 297.8 K
at a total pressure of 1 Torr of helium. ClO is added to the
system by turning on the Cl2 flow to the fixed microwave
discharge upstream of the ClO source reactor. ClO is removed
by turning off the Cl2 flow with all other flows held constant.
Data Analysis. Under these conditions, loss of OH is pseudo-
first-order. When ClO is added to the system, the loss of OH
may be written
Figure 1. OH resonance fluorescence intensity vs [OH]. The response
is linear over the range of [OH] used in this study; the line is a linear
least-squares fit to the data.
H atoms (eq 10). H atoms are generated in a fixed microwave
discharge (60 W) of dilute mixtures of H2 in He. NO2 in He
mixtures of specific composition are prepared by using a
calibrated capacitance manometer; the mixtures are stored in a
5 L Pyrex flask for later use. A calibrated flow meter is used to
add known flows of the NO2 in He mixture through the movable
injector. Initial OH concentrations are corrected for wall loss
and for self-reaction (eq 11) by using k11 ) 4.2 × 10-12
d[OH]+/dt ) k1[ClO][OH] + k+L [OH]
(12)
Because the OH resonance fluorescence intensity, IOH, varies
linearly with [OH] (see Figure 1), we can write
k+I ≡ -dln(IO+H)/dt ) k1[ClO] + k+L
(13)
where k+I is the total pseudo-first-order rate constant for the
loss of OH with added ClO. The parameter k+L is the loss of
OH other than by reaction with ClO; it includes wall loss,
secondary reactions with species in the OH and ClO sources,
and reactions with species formed by the OH + ClO reaction
itself. In the absence of added ClO, we may write
exp(-240/T) cm3 molecule-1 s-1 18
.
OH + OH f O + H2O
(11)
Typical OH fluorescence sensitivities are about 2.3 × 10-8
counts s-1/(molecules cm-3) with background signals of 1000
counts s-1. For the 50 s counting times used, the minimum
detectable [OH] is about 3 × 108 molecules cm-3 at a signal-
to-noise ratio of unity. A plot of OH fluorescence intensity vs
[OH] is shown in Figure 1. The signal is linear with respect to
[OH] up to at least 2 × 1012 molecules cm-3. This range covers
the [OH] used in the present study.
Meter Calibrations. Mass flow meters and controllers are
calibrated directly by using the volume change at constant
pressure (bubble meter) method or by using the pressure rise at
constant volume method. Pressure gauges are calibrated by using
an oil manometer. All thermocouples used in the experiments
are calibrated at 273 and near 195 K by using ice plus water
and CO2 plus methanol baths, respectively. Barometric correc-
tions are used to obtain the CO2 equilibrium temperature. The
temperature in the reaction zone is measured by using a
thermocouple probe in place of the movable inlet. At low
temperatures, the probe reading is 1-2 K lower than the two
thermocouples in the cooling jacket, while at 298 K, the probe
temperature is within 0.2 K. All temperatures reported are based
on the probe readings.
Reagents. Gases used are research grade He and H2, each at
99.9999%, research grade Cl2 (99.999%), ultrahigh purity O2
(99.8%), 1% and 5% mixtures of F2 in He, and CP grade NO
(99% min). As described above (see ClO source), O3 is formed
by passing O2 through a high-voltage discharge. NO2 for OH
calibrations is prepared from NO by adding an excess of O2
and letting the mixture stand overnight. The excess O2 is
removed by pumping the mixture through a trap at 195 K, where
the NO2 condenses as a white solid; then mixtures in He are
prepared by warming the solid and allowing the NO2 gas to
k-I ≡ -dln(IO-H)/dt ) kL-
(14)
where k-L includes wall loss, secondary reactions with species
in the OH source, and reaction with O3 from the ClO reactor.
One experimental run consists of a pair of OH-loss measure-
ments to determine k+I and kI- under similar conditions. The
difference between the two decay rates (k+I - kI-) and a
measurement of the absolute concentration of ClO gives a value
for k1
k1 ) (k+I - k-I )/[ClO]
(15)
This assumes that OH losses due to secondary reactions are
small or do not change appreciably when ClO is added: (k+L -
k-) , k1[ClO]. Evidence that secondary chemistry is unim-
L
portant is provided by using two different OH sources and by
numerical simulations. These points are discussed below.
Values of k+I and k-I are determined from the slopes of ln-
(IOH) vs l plots by linear least-squares analysis. Typical plots
are shown in Figure 2. The reaction length, l, under plug flow
conditions determines the reaction time, t ) l/V, where V is the
average flow velocity. Reaction lengths are varied from about
4 to 27 cm, and reaction times range from 2 to 18 ms. Observed
first-order rate constants are corrected for axial and radial
diffusion by using a method described previously.19,20 For kI+,
these corrections range from 3% to 33% and average 15%; for
k-I , they range from 0.3% to 2% and average 1%. A value of
750 cm2 s-1 is used for the diffusion coefficient of OH in He
at 300 K and 1 Torr; a temperature dependence of T1.5 is used

Absolute Rate Constant of the OH + ClO Reaction
J. Phys. Chem. A, Vol. 105, No. 46, 2001 10547
TABLE 1: Rate Constant Data for OH + ClOa
(k+I
[ClO]/ k-I ),
[OH]0
s-1
-
10-12 [ClO], 10-13 [O3], 10-11 [OH]0,
1011 k1, cm3
molecule-1
s-1
molecule
molecule
molecule
cm-3
cm-3
cm-3
T ) 297.8 K, P ) 1.00 Torr, V ) 1590 cm s-1, OH from F + H2O
2.82
4.05
5.85
7.17
8.21
8.46
9.27
9.32
2.06
3.32
5.38
2.41
3.11
1.84
2.44
3.37
3.28
1.16
2.65
3.71
3.76
1.23
4.17
2.13
3.13
5.56
3.64
5.86
3.35
3.26
6.31
6.14
3.48
5.27
5.43
5.79
3.74
5.19
13.2
12.9
10.5
19.7
14.0
25.2
28.4
14.8
21.8
41.7
29.6
29.6
28.7
45.4
35.4
63.6
95.1
2.26
2.35
2.36
2.31
2.34
2.23
2.25
1.99
2.30
2.11
2.33
1.89
2.27
1.89
2.35
138.2
165.5
191.8
188.8
208.8
185.7
308.9
306.1
363.9
304.7
376.1
321.9
432.2
13.4
14.5
15.6
16.1
16.6
17.0
18.4
av ) 2.22 ( 0.33
T ) 278.0 K, P ) 0.986 Torr, V ) 1489 cm s-1, OH from F + H2O
5.65
7.18
8.91
2.58
3.58
1.49
1.57
2.11
0.938
3.41
4.04
1.11
3.47
3.69
5.11
4.94
4.72
5.37
5.43
5.98
5.24
4.52
5.16
15.3
14.0
18.0
23.3
24.8
25.0
24.4
31.7
42.7
37.4
131.0
166.2
194.3
251.3
322.2
322.4
398.5
414.5
509.8
506.7
2.32
2.31
2.18
2.28
2.42
2.37
2.73
2.50
2.64
2.62
11.0
13.3
13.6
14.6
16.6
19.3
19.3
av ) 2.44 ( 0.36
T ) 259.8 K, P ) 0.987 Torr, V ) 1413 cm s-1, OH from F + H2O
3.10
3.42
3.69
6.43
9.27
3.72
3.17
2.83
2.41
2.64
1.16
1.12
2.67
2.28
1.88
0.603
4.52
2.79
2.55
4.71
5.49
2.40
6.52
9.34
4.64
4.94
6.63
6.9
12.3
14.5
13.6
16.9
52.5
19.9
14.7
31.0
34.6
29.4
87.1
85.6
85.5
2.81
2.50
2.32
2.77
2.61
2.66
2.64
2.63
2.03
2.60
2.30
178.0
241.8
335.5
342.8
360.3
292.3
444.5
449.3
Figure 2. OH decay curves: (O) without added ClO (k-I ); (b) with
added ClO (k+I ). The lines through the data are linear least-squares
fits; panel a conditions were 297.8 K, [ClO] ) 8.5 × 1012 cm-3, [O3]
) 1.8 × 1013 cm-3; panel b conditions were 217.5 K, [ClO] ) 1.0 ×
12.6
13.0
13.7
14.4
17.1
19.5
1013 cm-3, [O3] ) 5.6 × 1013 cm-3
.
to estimate the diffusion coefficient at other temperatures.20 No
corrections for viscous pressure drop are made because they
are less than 1%.
av ) 2.53 ( 0.46
T ) 238.6 K, P ) 1.00 Torr, V ) 1269 cm s-1, OH from H + NO2
3.45
7.57
11.1
13.5
21.5
22.5
6.84
2.70
3.72
2.43
4.24
4.44
4.80
3.94
4.27
4.78
4.37
4.48
7.2
19.2
26.0
28.2
49.2
50.2
89.8
2.60
3.14
2.55
2.27
3.20
3.01
237.3
282.6
306.4
687.4
677.0
A summary of experimental conditions and observed rate
constants is given in Table 1 and a plot of (k+I - kI-) vs [ClO]
at the two extreme temperatures used in this study is given in
Figure 3. The last column of Table 1 gives the values for k1
obtained from eq 15. Values for k1 were also obtained from the
slopes of the (k+I - kI-) vs [ClO] plots, and the results are
shown in Table 2. Comparison of columns three and four of
Table 2 shows that no significant differences are observed
between the values obtained from the averages of individual
(kI+ - kI-)/[ClO] points and from the slopes of (kI+ - k-I ) vs
[ClO] plots by linear least-squares fitting. The results reported
for the present study are obtained from the averages. An
Arrhenius plot of the data is shown in Figure 4 and results in
the following expresion:
av ) 2.80 ( 0.75
T ) 217.5 K, P ) 1.02 Torr, V ) 1132 cm s-1, OH from H + NO2
3.43
5.34
5.57
5.72
6.63
6.34
6.73
5.82
5.10
6.29
5.65
5.94
5.51
5.84
5.03
3.20
4.98
3.65
5.73
3.48
3.84
3.99
4.56
3.26
4.34
10.7
10.7
15.3
10.0
19.0
26.8
39.1
35.1
52.4
45.8
119.5
200.5
196.5
215.7
215.0
291.7
633.5
548.0
485.3
664.9
3.48
3.76
3.53
3.77
3.24
2.83
4.06
3.42
2.84
3.34
10.3
15.6
16.0
17.1
19.9
av ) 3.43 ( 0.78
k1 ) (7.2 ( 2.2) × 10-12 exp[(+333 ( 70)/T] (16)
a k+I and k-I are corrected for axial and radial diffusion; errors given
are two standard deviations.
for 217.5 e T e 297.8 K. The overall measurement error is
(30% obtained from the root-mean-square average of the
statistical error ((28%) at the 95% confidence level plus the
estimated systematic error ((6%). The statistical error is
obtained from an unweighted linear least-squares analysis; the
systematic error is discussed below.
Discussion
Numerical Modeling and Secondary Chemistry. At the
concentrations of OH and ClO and at the [ClO]/[OH]0 ratios
used in the present study (see Table 1), secondary reactions are
not expected to interfere. To check this, the reaction systems

10548 J. Phys. Chem. A, Vol. 105, No. 46, 2001
Wang and Keyser
integrator, CHEMRXN, was used to model each of the ClO
and OH sources separately; concentrations, temperatures, pres-
sures, and reaction times were close to those used in actual
experimental runs. The results of the source models were then
combined to simulate the OH + ClO reaction conditions.
Because there was about a 75 ms transit time between the
absorption cell and the start of the OH + ClO reaction,
simulations of the ClO source were used to assess the magnitude
of ClO loss due mainly to self-reaction (eq 17)
ClO + ClO f Cl2 + O2
ClO + ClO f 2Cl + O2
(17a)
(17b)
ClO + ClO f OClO + Cl
(17c)
(17d)
ClO + ClO + He f (ClO)2 + He
In the model, wall loss of ClO was set to zero. Experimental
checks for loss of ClO due to wall reaction are discussed below.
The modeled ClO concentration at the center of the UV
absorption cell, corrected for flow, pressure, and temperature
ratios (eq 8), would be the ClO concentration at the start of the
OH + ClO reaction if no loss of ClO occurred; it is termed
[ClO]uv. The loss of ClO is estimated by running the model for
75 ms; the resulting ClO concentration, [ClO]model, is then
compared to [ClO]uv to assess the magnitude of the loss. From
218 to 298 K, the differences between [ClO]uv and [ClO]model
Figure 3. Plot of pseudo-first-order rate constant vs [ClO]. All of the
data have been corrected for axial and radial diffusion (see text); lines
are linear least-squares fits of the data; temperatures are given next to
each plot.
TABLE 2: Summary of Observed Rate Constants for OH +
ClOa
1011 k1, cm3 molecule-1 s-1
T, K
runs
averageb,c
slopeb,d
297.8
278.0
259.8
238.6
217.5
15
10
11
6
2.22 ( 0.33
2.44 ( 0.35
2.53 ( 0.46
2.80 ( 0.75
3.43 ( 0.78
2.18 ( 0.24
2.52 ( 0.24
2.47 ( 0.32
2.93 ( 0.64
3.36 ( 0.60
are small: the values range from less than 1% for [ClO]uv
)
.
3.0 × 1012 cm-3 to less than 6% for [ClO]uv ) 2.4 × 1013 cm-3
Loss of ClO lowers the observed rate constant. However,
because the calculated loss is much less than our measurement
error of (30% and because these corrections would act in the
opposite direction from those due to secondary chemistry in
the OH + ClO reaction system (see below), no corrections were
made. Under the conditions used in this study, only small
concentrations of the ClO dimer are expected to be produced.
The model results show that, even at 218 K, the dimer
concentrations are less than 2% of the ClO concentrations at
the start of the OH + ClO reaction.
10
a Corrected for axial and radial diffusion. b Errors are two standard
deviations. c Average of individual values of k1. d From plots of k1[ClO]
vs [ClO].
In addition to their effect on ClO concentrations, secondary
reactions could contribute to the measured time profile of OH
and thereby affect the observed rate constants. Secondary loss
of OH would increase the observed rate constants; the most
likely interfering reactions of this type are reactions with HO2
(eq 18), with O3 (eq 19), and with Cl2 (eq 20).
OH + HO2 f H2O + O2
OH + O3 f HO2 + O2
OH + Cl2 f HOCl + Cl
(18)
(19)
(20)
Secondary reactions that produce OH could decrease the
observed rate constant; the most important reactions in this
category are eqs 5b, 21, and 22. HO2 and Cl are the primary
NO + HO2 f OH + NO2
HO2 + O3 f OH + 2O2
(21)
(22)
Figure 4. Arrhenius plot of present results compared to earlier studies
that also used excess O3 to produce ClO. The line is the unweighted
linear least-squares fit of the present results only; the error bars are
two standard deviations.
products of reaction 1, O3 and Cl2 are present in the ClO source,
and NO is a product of the H + NO2 reaction (eq 10) used to
produce OH in some of the measurements. Because the amount
of HO2 in the system depends on [OH]0, interference from
reactions 5b, 18, 21, and 22 is expected to increase as [OH]0
were simulated by numerical modeling. The procedure and
reactions used have been described previously2 and only an
outline is given here. A Gear-based differential equation

Absolute Rate Constant of the OH + ClO Reaction
J. Phys. Chem. A, Vol. 105, No. 46, 2001 10549
increases. We observe no significant dependence on [OH]0; this
may be taken as evidence that these four secondary reactions
do not interfere strongly with the measurements. Interference
from reaction 19 can be estimated by using k19 ) 1.5 × 10-12
exp(-880/T) cm3 molecule-1 s-1 21 and the maximum [O3] at
each temperature. These calculations show that loss of OH by
reaction 19 was at most 3% of that by reaction 1 and reaction
19 should not seriously interfere with the rate constant measure-
ments. By using k20 ) 1.4 × 10-12 exp(-900/T) cm3 molecule-1
s-1 18
, we estimate that loss of OH by reaction 20 is negligible
at all temperatures studied.
To more fully check for the effects of secondary chemistry,
the OH + ClO reaction system was modeled by combining the
model results for the ClO source and the appropriate OH source
for the temperature studied. Each experimental run at 218 and
298 K was simulated separately by using ClO, OH, and O3
concentrations and reaction times that were close to the actual
experimental values. Input rate constants, k1(in), were the
experimental values obtained for each run. Wall-loss rate
constants for OH, Cl, and ClO were set at 10, 5, and 0 s-1
,
respectively. The model output consists of [OH] vs reaction time
profiles. These were treated in the same way as experimental
data to obtain the model value for k1: plots of ln [OH] vs
reaction time were fit by linear regression over time ranges
similar to those used in the experiments; the slope gives a value
for -dln [OH]/dt; then from eq 15 and the value for [ClO] in
the model, we can obtain k1(out). At both temperatures, values
of k1(out) are mostly within about (10% of k1(in) and the
average k1(out) is 4-6% higher than k1(in). Because these
differences are well within our experimental uncertainty of
(30% and because they tend on average to cancel the effect of
ClO loss (see above), no corrections were made. The model
results confirm that the major loss of OH in our system is by
reaction with ClO with only minor losses from reactions with
HO2 and O3 (eqs 18 and 19); loss by reaction with Cl2 (eq 20)
is negligible. In addition, the model shows that some minor loss
of OH can occur by reaction with the ClO dimer in low-
temperature experiments with high initial [ClO]. The simulations
show that production rates of OH from reactions 5b, 21, and
22 are small compared to OH loss by reaction with ClO; they
also show that in some runs minor amounts of OH can be
generated from the reaction of Cl with HOCl (eq 23). HOCl is
produced mainly by the reaction of ClO with HO2 (eq 24) with
a small contribution from reaction 20.
Figure 5. (a) Absorption cross sections of ClO in the continuum region
from 225 to 265 nm; (b) expanded view of the region from 250 to 255
nm. References are Johnston et al.13 (dashed line), Mandelman &
Nicholls14 (dashed-dot-dot line), Sander & Friedl15 (dashed-dot line),
Simon et al.16 (dotted line), Bloss17 (solid line). Vertical line is at 253.65
nm.
ClO is determined at 253.65 nm in the continuum of the
separated spectrum where the cross sections are independent
of temperature.22 The residual spectrum is calculated by
subtracting the separated ClO and O3 spectra from the original
combined spectrum. The residual absorbance at all wavelengths
is less than 0.1% of the absorbance of the separated spectra;
this shows that O3 makes no significant contribution to the
separated ClO absorbance that is used to obtain [ClO]. We can
estimate the precision of the method by repeating the same
subtraction several times; the resulting [ClO] varies about (4%.
There have been several measurements of the ClO absorption
cross section in the continuum region.13-17 Figure 5 is summary
plot of these studies. The results using different measurement
techniques are all in very good agreement and yield an average
value of (4.23 ( 0.13) × 10-18 cm2 molecule-1 at 253.65 nm;
the errors given are twice the standard deviation of the average.
From this, we estimate an uncertainty of about 3-4% in the
Cl + HOCl f OH + Cl2
ClO + HO2 f HOCl + O2
(23)
(24)
Absolute ClO Concentrations. The accuracy of the present
measurement depends critically on how well the absolute ClO
concentrations are known. ClO concentration is determined by
UV absorption as described above and calculated from experi-
mental quantities given by eq 6: A, the absorbance; σClO, the
absorption cross section; and L, the absorption path length.
Uncertainties in each of these are discussed below.
To obtain an experimental value for the absorbance of ClO,
an experimental spectrum of ClO plus O3 is separated by
subtracting a scaled ClO reference spectrum.2,9 The reference
spectrum of ClO containing no O3 (excess Cl atoms) is recorded
separately. The scale factor that minimizes the ClO structure
in the banded region (265-310 nm) and that also minimizes a
calculated residual spectrum is taken as the best value. A
separated ClO spectrum is calculated by multiplying the ClO
reference spectrum by this scale factor; then the absorbance of
value of σClO
.
The accuracy in determining the absorption path length, L,
is about (1%. Combining this with the uncertainties in the
absorbance and cross section gives an estimated uncertainty of
(6% in the ClO concentration measurements in the UV
absorption cell.
Loss of ClO can occur between the absorption cell and the
start of the OH + ClO reaction; ClO can also be lost over the

10550 J. Phys. Chem. A, Vol. 105, No. 46, 2001
Wang and Keyser
measurement time of the OH decay. Between the cell and the
point where the reaction starts, this loss is due mainly to self-
reaction (eq 17) and to reaction at the wall. As discussed above,
numerical models of the ClO source show that the losses due
to self-reaction are less than 6%. Models also show that at the
stoichiometric ratios used in the present experiments, loss of
ClO during the measurement time is small (generally about
2-5%). Experimental tests described below show that ClO wall
loss is small in our system. Combining the uncertainties in the
absorbance ((4%), cross section ((4%), path length ((1%),
and reaction losses (about (10%), we estimate the uncertainty
in the absolute ClO concentration measurements to be about
(12%.
ClO Wall Loss. In low surface to volume reactors, wall loss
of ClO generally ranges from less than 0.1 to 2 s-1 even at
very low temperatures.7,18,23 In our system, the surface to volume
ratio changes by about 45% when the OH source is moved along
the main reactor axis; this can serve as a rough experimental
check for large ClO wall losses. If significant ClO wall losses
occurred, the [ClO] would strongly depend on the position of
the OH source (the reaction length) and nonlinear ln(IOH) vs l
plots should result. All of the OH decay plots were linear, and
this is consistent with no major loss of ClO on the main reactor
surfaces. To further check for ClO wall loss between the
absorption cell and the main reactor, we converted ClO to Cl
atoms by adding an excess of NO (eq 25) in the main reactor
75 ms after ClO was determined by UV absorption.
respectively. We have discussed above the error estimates for
σ
ClO ((4%) and L ((1%). Taking the root-mean-square average
of the systematic errors, we estimate the overall systematic error
at (6%. Combining the systematic error with the measurement
(statistical) error of (28% in k1, we estimate an overall
uncertainty of (30% in k1 at the 95% confidence level.
Comparison with Other Studies. A summary of rate
constant measurements for the OH + ClO reaction is presented
in Table 3. The first three entries3-5 in the table used excess Cl
atoms to produce ClO from reactions 2, 3, or 4 and assumed
that [ClO] was equal to initial [O3], [Cl2O], or [OClO].
Subsequent studies have shown that this assumption is not
always true; for the measurement to be useful, experimental
checks are needed to verify the assumption.7,9,10 Use of excess
Cl atoms also creates a secondary source of OH from reaction
5b and necessitates a correction procedure. These complications
probably account for the relatively low results reported by these
early studies, and they are not considered further in this
discussion. The next five entries6-9,24 plus the present study used
excess O3 to produce ClO; the studies by Poulet et al.7 and
Bedjanian et al.24 used both excess Cl atoms and excess O3 (or
OClO). Except for the study by Lipson et al.,8 the agreement
among these studies is reasonably good: k1 at 298 K ranges
from 1.8 × 10-11 to 2.4 × 10-11 cm3 molecule-1 s-1 with an
average value of 2.1 × 10-11 cm3 molecule-1 s-1; the Arrhenius
parameters are also in good agreement over the temperature
ranges studied.
Because all of the studies used ClO in excess, differences in
the results may be due to the methodology and accuracy of the
various techniques used to determine absolute ClO concentra-
tions. Hills and Howard6 measured k1 with a slight excess of
O3 over Cl atoms but calibrated their ClO detection sensitivity
by using reaction 2 with excess Cl atoms and assumed [ClO]
) [O3]0. However, they checked this assumption by titrating
the ClO with NO (eq 25), and the titration agreed with the
calibration results within 10%.
ClO + NO f Cl + NO2
(25)
This is close to the timing in the actual rate measurements.
Assuming one-to-one stoichiometry for the ClO to Cl conversion
from reaction 25, [ClO] inferred from [Cl] agrees with [ClO]
calculated from the UV absorption measurements within about
15%. The overall accuracy of these measurements is estimated
at (20%, mainly because of uncertainties in the Cl lamp
calibrations. The results show that there is no large wall loss of
ClO between the absorption cell and the start of the OH + ClO
reaction. No correction for wall loss is made because these
experiments cannot determine it precisely and the estimate is
well within the experimental uncertainty.
Poulet et al.7 used both excess Cl atoms and excess O3 to
measure k1. In both cases, they determined absolute [ClO] by
the NO titration method and obtained similar results (see Table
3).
Lipson et al.8 detected ClO by using chemi-ionization mass
spectrometry following a charge-transfer reaction (eq 30). ClO
Error Analysis. An estimate of systematic and overall
experimental errors in determining k1 can be made by consider-
ing that under plug flow conditions eqs 13 and 14 become
ClO + SF6- f ClO- + SF6
(30)
k+I ≡ -V{dln(IO+H)/dl} ) k1[ClO] + k+L
k-I ≡ -V{dln(IO-H)/dl} ) kL-
(26)
(27)
was calibrated as NO2 by adding an excess of NO (eq 25).
Because excess O3 was used, they added C2H6 to scavenge (eq
31) the Cl atoms produced by reaction 25 to prevent reformation
Cl + C2H6 f HCl + C2H5
(31)
In a cylindrical reactor, the average flow velocity, V, is given
by
of ClO from Cl + O3 (eq 2). They modeled this system to
correct for the small amount of NO2 lost by reaction with C2H5;
these corrections were less than 15%. Their result at 298 K is
about 40% lower than the average of the five other measure-
ments using excess O3. Although they do not explicitly discuss
NO2 formation from the reaction of NO with excess O3 (eq
32), we have tried to estimate the amount of added NO2 formed
V ) (760/P)(T/273.2)(F/(60πa2))
(28)
where P is the flow pressure in Torr, T is the absolute
temperature, F is the mass flow in standard cm3 min-1, and a
is the reactor radius in cm. From eqs 6, 15, 26, and 27, we
have
NO + O3 f NO2 + O2
(32)
k1 ) -V{dln(I+OH)/dl - dln(IO-H)/dl}(σClOL/A) (29)
by reaction 32 in their system and find that it should be less
than 2% of the NO2 formed from ClO + NO (eq 25) and thus
cannot explain the difference in our results. They also did not
correct their results for the effects of turbulent flow, which they
estimate lowered their k1 measurement by about 8%. Making
Systematic errors in k1 include errors in V, σClO, and L; errors
in {dln(I+OH)/dl - dln(IO-H)/dl} and A are measurement or
statistical errors. The errors in V comprise errors in P, T, F, and
a, which we estimate at (2%, (1%, (4%, and (2%,

Absolute Rate Constant of the OH + ClO Reaction
J. Phys. Chem. A, Vol. 105, No. 46, 2001 10551
TABLE 3: Summary of OH + ClO Rate Constant Measurements
1012 k1,a
cm3 molecule-1 s-1
9.1 ( 2.6
T, K
P, Torr
methodb
commentsc
ref
298
1.0-3.5 He DLF-RF
ClO from O3 (Cl2O) +
excess Cl;
Leu, 19793
OH from H + NO2
11.7 ( 3.3
independent of T
248-335 1 He
DLF-RF
DLF-RF
ClO from O3 + excess Cl;
correct for Cl + HO2;
OH from H + NO2
Ravishankara et al., 19834
Burrows et al., 19845
11.9 ( 1.8
independent of T
243-298 1-5 He
ClO from O3 + excess Cl;
correct for Cl + HO2;
OH from H + NO2
ClO from Cl + excess O3;
OH from H + NO2
ClO from Cl + O3 (ClO2);
ClO determined as NO2 after
NO + ClO;
use both excess O3 (ClO2)
and excess Cl;
(8.0 ( 1.4) exp[(235 ( 46)/T]; 219-373 1.0-1.1 He DLF-LMR
17.5 ( 3.1 at 298 K
19.9 ( 2.5 (excess Cl),
18.9 ( 2.1 (excess O3),
19.4 ( 3.8 (all data)
Hills and Howard, 19846
Poulet et al., 19867
298
0.5-0.9 He DLF-LIF (OH),
EIMS (ClO)
correct for Cl + HO2;
OH from H + NO2
(5.5 ( 1.6) exp[(292 ( 72)/T]; 205-298 100 N2
14.6 ( 2.3 at 298 K
DTF-CIMS
ClO from Cl + excess O3;
ClO determined as NO2 after
ClO + NO;
Lipson et al., 19978
OH from H + NO2
(8.9 ( 2.7) exp[(295 ( 95)/T]; 234-356 4.2-15.4 He DLF-UVA (ClO), ClO from Cl + excess O3;
Kegley-Owen et al., 19999
Bedjanian et al., 200124
24.4 ( 0.7 at 298 K
FP-LIF (OH)
ClO by UV absorption;
OH from photolyis of O3 + H2O
or HNO3
ClO from Cl + O3;
ClO determined as NO2
after ClO + NO;
use both excess O3 and excess Cl;
OH from H + NO2
ClO from Cl + excess O3;
ClO by UV absorption;
OH from F + H2O and H + NO2
(6.7 ( 1.8) exp[(360 ( 90)/T]; 230-360 1.0 He
22 ( 4 at 298 K
DLF-EIMS
(7.2 ( 2.2) exp[(333 ( 70)/T]; 218-298 1.0 He
22.2 ( 3.3 at 298 K
DLF-RF-UVA
present study
a Uncertainties are two standard deviations. b CIMS ) chemical ionization mass spectrometry; DLF ) discharge laminar flow; DTF ) discharge
turbulent flow; EIMS ) electron impact mass spectrometry; FP ) flash photolysis; LIF ) laser induced fluorescence; LMR ) laser magnetic
resonance; RF ) resonance fluorescence; UVA ) ultraviolet absorption spectrometry. c Pseudo-first-order in OH ([ClO] . [OH]).
this correction would improve agreement, but their result would
still be lower than the other excess O3 measurements by about
30%. The reason for this discrepancy remains unclear at this
time.
to detect ClO as NO2 following the NO + ClO reaction (eq
25). The mass spectrometer signal was calibrated by using an
excess of Cl atoms to generate a known amount of ClO from
Cl + O3 (eq 2). This calibration was checked under excess O3
conditions by measuring the fraction of Cl2 dissociated to form
Cl atoms. The two methods agreed within a few percent. The
rate constant was measured under pseudo-first-order conditions
([ClO] . [OH]0) by using both excess O3 and excess Cl atoms
to produce ClO. Numerical simulations were used to correct
the observed decay rates for secondary chemistry. Under excess
O3, the main interference was OH + HO2 (eq 18); under excess
Cl atoms, the main interference was Cl + HO2 (eq 5b). After
making the corrections, the results obtained under the two
reaction conditions agree very well.
Our results agree very well with the recent measurements of
Kegley-Owen et al. (KO).9 Although both studies used UV
absorption to monitor ClO, the measurements differ in some of
the detailed parameters needed to determine absolute concentra-
tions of ClO. KO measured ClO in the actual reactor tube where
the OH + ClO reaction occurred. In their experiments, the
absorption path length could not be measured directly because
of uncertainties as to how well the flow dynamics confined the
ClO radicals in the UV absorption path and the possible presence
of stagnation regions. The absorption path length was calibrated
by comparing the optical density of O3 measured in a cell of
known path length in series with the OH + ClO reactor with
the O3 optical density in the reactor. In the present study, ClO
was measured in an absorption cell of known path length
upstream of the main OH + ClO reactor and absolute ClO
concentrations were obtained by adjusting for mass flow,
pressure, and temperature differences between the absorption
cell and the main reactor (eq 8). The two studies differ also in
the methods used to generate OH radicals. KO used flash
photolysis of O3 plus H2O mixtures or of HNO3; the present
study used F + H2O (eq 9) and H + NO2 (eq 10) as described
above.
The excellent agreement among the latest three studies, which
used a variety of OH sources and entirely different detailed
experimental methods to determine absolute [ClO], is good
evidence that secondary chemistry did not interfere and that
[ClO] was accurately measured.
Acknowledgment. The research described in this article was
performed at the Jet Propulsion Laboratory, California Institute
of Technology, under a contract with the National Aeronautics
and Space Administration. We thank William Bloss for com-
municating his results on ClO absorption cross sections and Yuri
Bedjanian for sending his results on the OH + ClO reaction.
Our results also agree very well with a recent measurement
by Bedjanian et al.24 In this study, mass spectrometry was used

10552 J. Phys. Chem. A, Vol. 105, No. 46, 2001
Wang and Keyser
(15) Sander, S. P.; Friedl, R. R. J. Phys. Chem. 1989, 93, 4764.
(16) Simon, F. G.; Schneider, W.; Moortgat, G. K.; Burrows, J. P. J.
Photochem. Photobiol. A: Chem. 1990, 55, 1.
References and Notes
(1) Chance, K.; Traub, W. A.; Johnson, D. G.; Jucks, K. W.; Ciarpallini,
P.; Stachnik, R. A.; Salawitch, R. J.; Michelsen, H. A. J. Geophys. Res.
1996, 101, 9031.
(2) Wang, J. J.; Keyser, L. F. J. Phys. Chem. A 2001, 105, 6479.
(3) Leu, M. T.; Lin, C. L. Geophys. Res. Lett. 1979, 6, 425.
(4) Ravishankara, A. R.; Eisele, F. L.; Wine, P. H. J. Chem. Phys.
1983, 78, 1140.
(5) Burrows, J. P.; Wallington, T. J.; Wayne, R. P. J. Chem. Soc.,
Faraday Trans. 2 1984, 80, 957.
(6) Hills, A. J.; Howard, C. J. J. Chem. Phys. 1984, 81, 4458.
(7) Poulet, G.; Laverdet, G.; Le Bras, G. J. Phys. Chem. 1986, 90,
159.
(8) Lipson, J. B.; Elrod, M. J.; Beiderhase, T. W.; Molina, L. T.;
Molina, M. J. J. Chem. Soc., Faraday Trans. 1997, 93, 2665.
(9) Kegley-Owen, C. S.; Gilles, M. K.; Burkholder, J. B.; Ravishankara,
A. R. J. Phys. Chem. A 1999, 103, 5040.
(10) Burkholder, J. B.; Hammer, P. D.; Howard, C. J.; Goldman, A. J.
Geophys. Res. 1989, 94, 2225.
(11) Wang, J. J.; Keyser, L. F. J. Phys. Chem. A 1999, 103, 7460.
(12) Molina, L. T.; Molina, M. J. J. Geophys. Res. 1986, 91, 14501.
(13) Johnston, H. S.; Morris, E. D., Jr.; Van den Bogaerde, J. J. Am.
Chem. Soc. 1969, 91, 7712.
(14) Mandelman, M.; Nicholls, R. W. J. Quant. Spectrosc. Radiat.
Transfer 1977, 17, 483.
(17) Bloss, W. J., Ph.D. Thesis, University of Cambridge, 1999.
(18) DeMore, W. B.; Sander, S. P.; Howard, C. J.; Ravishankara, A.
R.; Golden, D. M.; Kolb, C. E.; Hampson, R. F.; Kurylo, M. J.; Molina,
M. J. Chemical Kinetics and Photochemical Data for Use in Stratospheric
Modeling, EValuation No. 12; JPL Publication 97-4; Jet Propulsion
Laboratory, California Institute of Technology: Pasadena, CA, 1997.
(19) Brown, R. L. J. Res. Natl. Bur. Stand. (U. S.) 1978, 83, 1.
(20) Keyser, L. F. J. Phys. Chem. 1984, 88, 4750.
(21) Sander, S. P.; Friedl, R. R.; DeMore, W. B.; Ravishankara, A. R.;
Golden, D. M.; Kolb, C. E.; Kurylo, M. J.; Hampson, R. F.; Huie, R. E.;
Molina, M. J.; Moortgat, G. K. Chemical Kinetics and Photochemical Data
for Use in Stratospheric Modeling, EValuation No. 13; JPL Publication 00-
3; Jet Propulsion Laboratory, California Institute of Technology: Pasadena,
CA, 2000.
(22) Trolier, M.; Mauldin, R. L., III; Ravishankara, A. R. J. Phys. Chem.
1990, 94, 4896.
(23) Lee, Y.-P.; Stimpfle, R. M.; Perry, R. A.; Mucha, J. A.; Evenson,
K. M.; Jennings, D. A.; Howard, C. J. Int. J. Chem. Kinet. 1982, 14, 711.
(24) Bedjanian, Y.; Riffault, V.; Le Bras, G. Int. J. Chem. Kinet., in
press. We became aware of these measurements after we submitted the
present study for publication.