Rate constant measurements for the reaction of HO2 with O3 from 200 to 300 K using a turbulent flow reactor

Article abstract of DOI:10.1021/jp002383t

The rate constant for the reaction of HO₂ with O₃ was measured using a turbulent flow reactor with tunable diode laser absorption detection of HO₂. Two separate methods were used to determine k₁ from the pseudo-first-order decay of HO₂ in excess O₃. The isotopic labeling method used H¹⁸O₂ in excess ¹⁶O₃ to avoid reformation of the reactant. The OH scavenger method used trifluorochloroethylene to remove the OH product and prevent re-formation of HO₂. The turbulent flow reactor allowed measurements at temperatures between 297 and 197 K at total pressures from 80 to 175 Torr, spanning a wide range of stratospheric conditions. The temperature-dependent rate constant is given by the three-term expression k₁(T) = {(103 ± 57) exp[-(1323 ± 160)77] + 0.88} × 10^-15 cm³ molecule^-1 s^-1, reflecting its non-Arrhenius behavior at low temperatures. These results demonstrate that the catalytic destruction of lower stratospheric O₃ by HO_x radicals (OH and HO₂) may proceed faster than current models predict. The rate constant at 295 K is (2.0 ± 0.2) × 10^-15 cm³ molecule^-1 s^-1.

Full text of DOI:10.1021/jp002383t

J. Phys. Chem. A 2001, 105, 1583-1591
1583
Rate Constant Measurements for the Reaction of HO2 with O3 from 200 to 300 K Using a
Turbulent Flow Reactor†
Scott C. Herndon,* Peter W. Villalta,‡ David D. Nelson, John T. Jayne, and Mark S. Zahniser*
Aerodyne Research, Inc., Center for Atmospheric and EnVironmental Chemistry,
Billerica, Massachusetts 01821
ReceiVed: July 3, 2000; In Final Form: NoVember 2, 2000
The rate constant for the reaction of HO2 with O3 was measured using a turbulent flow reactor with tunable
diode laser absorption detection of HO2. Two separate methods were used to determine k1 from the pseudo-
first-order decay of HO2 in excess O3. The isotopic labeling method used H18O2 in excess 16O3 to avoid
reformation of the reactant. The OH scavenger method used trifluorochloroethylene to remove the OH product
and prevent re-formation of HO2. The turbulent flow reactor allowed measurements at temperatures between
297 and 197 K at total pressures from 80 to 175 Torr, spanning a wide range of stratospheric conditions. The
temperature-dependent rate constant is given by the three-term expression k1(T) ) {(103 ( 57) exp[-(1323
( 160)/T] + 0.88} × 10-15 cm3 molecule-1 s-1, reflecting its non-Arrhenius behavior at low temperatures.
These results demonstrate that the catalytic destruction of lower stratospheric O3 by HOx radicals (OH and
HO2) may proceed faster than current models predict. The rate constant at 295 K is (2.0 ( 0.2) × 10-15 cm3
molecule-1 s-1.
Introduction
flow reactor used in this study allows direct measurement of k1
at stratospheric temperatures and pressures. Both isotopic
labeling of reactant HO2 and an OH scavenger were used to
reduce the effects of secondary chemistry. In conventional low-
pressure laminar flow tubes, the loss of HO2 to the walls is
sufficient to complicate measurements of k1, particularly at low
temperatures. In the high-pressure turbulent flow technique,
there is limited mass exchange between the turbulent core and
the viscous sublayer, which forms near the wall. Under these
circumstances, the measured wall loss rate constants are less
dependent on wall temperature. This property of the turbulent
flow reactor allows the study of HO2 kinetics at temperatures
less than 240 K.
The cycling of HOx (HOx ) OH + HO2) through reactions
1 and 2 represents the dominant O3 loss process occurring in
k1
HO2 + O3
OH + O3
98 OH + 2O2
(R1)
(R2)
k2
98 HO2 + O2
the lower stratosphere over much of the globe.1 Reaction 1 is
the rate-limiting step in this HOx-only ozone destruction cycle.
The atmospheric impact of reaction 1 on O3 loss, however, is
coupled to the ozone destruction associated with NOx (NOx )
NO + NO2) and reactive halogen.2,3
A difficulty with direct measurements of k1 under pseudo-
first-order conditions with respect to HO2 is the rapid regenera-
tion of HO2 due to this very cycle. Previous studies4-6 have
used an OH scavenger or have used isotopic labeling7 of the
HO2 in order to determine k1.
The previous work indicates that strict Arrhenius behavior
is not observed in the temperature dependence of k1. At lower
temperatures the rate constant may be higher than would be
predicted by extrapolation of the data taken above 230 K.
Reaction 1 proceeds through two channels, corresponding to
transfer of the hydrogen atom (major) or the transfer of the
terminal oxygen atom in HO2 (minor),7,8 but the present
understanding of the associated mechanisms does not predict
the temperature dependence of k1.
Although this rate constant, k1, is most frequently utilized in
atmospheric modeling, there have been no previous direct
measurements of k1 below 233 K, even though lower strato-
spheric temperatures can be as low as 185 K. The turbulent
Experimental Section
Turbulent Flow Approach. The use of a turbulent flow
reactor to study chemical kinetics was developed by Seeley and
co-workers.9,10 The technique has been used to study the kinetics
of several reactions.11,12 It is a relatively new method, though,
and a short description of the technique will be presented here
to facilitate the discussion of these results.
Conventional laminar flow tube reactors have a radially
parabolic velocity profile where the gas velocity is greatest at
the center of the tube and decreases by the square of the radial
position to zero at the tube walls. The velocity fluctuations at
any given radial point are low relative to the overall average
velocity. This condition leads to a wide distribution of residence
times for species that traverse the length of the tube, which is
undesirable when attempting kinetics measurements. To cir-
cumvent this difficulty, the working pressure is held below ∼10
Torr in order to increase diffusional dispersion to such an extent
that the radial concentration gradients in the flow tube are
minimized. The use of molecular diffusion to overcome the
gradient due to the flow profile characterizes the conventional
low-pressure laminar flow approach to flow tube kinetics.
The Reynolds number for tubular flow is defined in eq 1.
† Part of the special issue “Harold Johnston Festschrift”.
‡ Present address: Mass Spectrometry Facility, University of Minnesota
Cancer Center, Box 806 Mayo, 420 Delaware Street SE, Minneapolis, MN
55455.
10.1021/jp002383t CCC: $20.00 © 2001 American Chemical Society
Published on Web 12/14/2000

1584 J. Phys. Chem. A, Vol. 105, No. 9, 2001
Herndon et al.
long) which has been coated with a thin layer of halocarbon
wax, emptying into a six-way cross. A sampling port with a 3
mm orifice was located inside the cross and directed a portion
of the flow, via differential pumping, into a 55 cm long, 8.5
cm i.d. multipass cell. Nitrogen (Middlesex Gases 99.998%),
from a liquid N2 gas pack, served as the main flow tube carrier
gas for the majority of the work presented here. Helium
(Middlesex Gases 99.999%) was used at the carrier gas for the
flow directed through the inner injector. Flows were measured
using Tylan mass flow meters, which were periodically cali-
brated by measuring the rate of change of pressure into a known
volume. Special care was taken to ensure these calibrations were
done isothermally. Pressures in the reactor and multipass cell
were measured using calibrated capacitance manometers. The
temperature of the gas in the flow tube was measured using
high gauge calibrated chromel-alumel thermocouples both at
the end of the injector and just prior to the sampling port shown
in Figure 1. The main carrier gas was precooled before passing
into the main flow tube, and its temperature was maintained by
a thermostated fluid which was circulated through the reactor
jacket. The amount of precooling was controlled so that the
temperature measured at the injector tip matched that measured
prior to the sampling port. The flow through the injector was
typically less than 6000 standard cubic centimeters per minute
(sccm), while the total flow through the reactor was ∼55 000
sccm. The pumping speed was adjusted such that the pressure
in the flow tube ranged from 80 to 174 Torr and the average
flow velocity of the gas varied from 850 to 2500 cm s-1. The
sampling pressure of 10 Torr was selected to optimize detection
sensitivity at the point where pressure broadening and Doppler
broadening are equal. The number density of O3 dropped by a
factor of 20 upon passage into the detector, and as a result, the
extent of reaction occurring in the multipass cell relative to that
in the flow tube was negligible. In flow tube experiments
constraints such as the minimum distance between the injector
tip and the detection region have an effect on the minimum
detectable reaction times. This is sometimes referred to as the
“end effect”. The experimental conditions employed in the
majority of this work result in an end effect, which is the
equivalent of ∼6 cm of flow tube distance.
2rνjF
η
QM
rη
Re )
) R
(1)
Equation 1 shows two representations of Re, where r is the
tube radius, νj is the average axial velocity of the gas, F is the
density, η is the viscosity of the gas, R is a contant of
proportionality, Q is the mass flow rate, and M is the molecular
mass of the gas. In the case of tubular laminar flow, the
Reynolds number is smaller than 900, usually <100 for flow
tube kinetics. As the mass flow increases, the shear flow near
the wall develops instabilities and velocity fluctuations occur
which are greater than those present in laminar flow. The flow
profiles in this regime are not well-defined, and this is referred
to as the transition region between viscous and nonviscous flow.
When Re > 2000, the flow becomes turbulent with a velocity
profile which is more uniform than the parabolic profile
characteristic of a laminar flow. The radial velocity profile of
fully developed turbulent flow within the core is flat. Near the
tube walls a viscous sublayer forms, and between the sublayer
and the core a turbulent boundary layer develops. The radial
motion in the core is the combination of molecular diffusion
and small-scale turbulent mixing. Flow visualization studies10
aimed at measuring the effective diffusivity of species in the
core have found that in a fully developed turbulent flow the
combined molecular and “eddy” diffusion is great enough to
allow for radial mixing, but not so great that appreciable axial
mixing occurs. Essentially, it is this result that allows chemical
kinetics measurements to be conducted using a turbulent flow
tube reactor.
When the one-dimensional form of the continuity equation
is integrated using the boundary conditions appropriate for a
reaction in a flow tube, the concentration of the reactive species,
cj, is given by9,13
-k1*z
cj(z) ) c0 exp
(2)
(
)
Veff
where z represents the distance down the flow tube (c(0) ) c0),
eff is the gas velocity in the turbulent core, k1* is the measured
V
pseudo-first-order rate constant, and Deff is the effective
dispersion coefficient. In eq 2, an effective velocity of the
turbulent core is used in place of the average velocity, which
differs from the result using a fully laminar flow. The total
pseudo-first-order rate coefficient is given by eq 3.9,13 For the
A movable double injector was used to create and inject HO2
to the main flow. A Beenaker style 2.5 GHz microwave
discharge cavity17 produced H atoms from trace amounts of H2
(Wesco Gases 99.9995%) in the He carrier gas, which passed
through the inner injector. This mixture flowed into the outer
injector, which contained a sufficient flow of O2 (Middlesex
Gases 99.999%) to maintain more than 1 × 1017 molecules cm3
O2 in the diluted flow. The inner injector was recessed 25 cm
from the end of the outer injector (see Figure 1), allowing the
H atoms sufficient time to mix with the O2/He and react to
produce HO2. Both the inner and outer injectors were coated
with a halocarbon wax to reduce H and HO2 losses on the
injector walls.
D
eff k1*
k1′ ) k1* 1 +
(3)
2
(
)
Veff
measurements presented here, k1 ) k1[O3] + kwall. Equation 3
is very similar to that which arises in the instance of laminar
flow. Other studies14-16 have developed methods for measuring
rate constants using a laminar flow at higher pressures either
by determining the complete radial concentration profile or by
applying corrections to an apparent temporal profile. When the
detector samples the core of a turbulent flow reactor, the simple
equations (2) and (3) may be applied; however, the effective
velocity must be measured and the magnitude of Deff must be
determined.
Two Teflon “turbulizers” were mounted on the exterior of
the outer injector as shown in Figure 1. These have the effect
of enhancing the turbulent mixing between the injector and main
gas flows. The gas velocities across the tube diameter were
determined by measuring the difference between the static
pressure and the impact pressure18 using a Pitot tube. The impact
pressure was determined using a 1 mm i.d. Pitot tube situated
parallel to the direction of the flow. The static pressure was
measured using a 1 mm i.d. Pitot tube with two holes normal
to the direction of the tube. A differential manometer, 1 Torr
full scale, was used to accurately measure ∆p. The measure-
Apparatus Description. A turbulent flow reactor coupled
to a multipass cell designed for a long path length in a small
volume has been used in the determination of k1. A schematic
of the instrument, with an expanded view of the area near the
end of the injector, is presented in Figure 1. The flow reactor
consists of a double jacketed Pyrex tube (2.5 cm i.d., 85 cm

Reaction of HO2 with O3 from 200 to 300 K
J. Phys. Chem. A, Vol. 105, No. 9, 2001 1585
Figure 1. Schematic of the turbulent flow reactor coupled to a tunable diode laser. The inset shows detail of movable double injector with “turbulizers”
mounted on the tip.
ments associated with the development of the turbulent flow
will be discussed further in the results section.
The output from the laser system was collected with a reflecting
microscope objective and directed into the multipass cell. For
the work presented here, the multipass cell was configured for
366 passes, which provided 201 m of total path length. The
exit beam was directed to an HgCdTe photoconductive detector.
The data acquisition is based on a rapid current sweep of the
diode laser with synchronous detection of the transmitted
radiation. For the work presented here, the 120 point spectra
spanning the features of interest were acquired at a rate of 2.5
kHz. The 300 kHz analog-to-digital conversion rate is fast
enough to minimize the impact of low-frequency amplitude
noise.
HO2 lines near 1407 cm-1 were used in this work with the
current ramp adjusted so that the laser frequency range spanned
∼0.15 cm-1. This was wide enough to include four H16O2 lines
and two H18O2 lines in a single laser sweep. The infrared
spectroscopy of H16O2 has been previously investigated.22-26
The line strengths determined by Zahniser et al.25 and the
pressure broadening coefficients determined by Nelson and
Zahniser26,27 are tabulated in the most recent HITRAN data
compilation.28
To measure the relative line strengths and positions for H18O2,
equal amounts of 18O2 and normal O2 were mixed and added
to trace amounts of H atoms injected into the multipass cell.
The assignments of peak positions and line strengths for H18O2
are given in Table 1. For the pseudo-first-order kinetic measure-
ments presented here the absolute accuracy of the line strengths
is of secondary importance.
Figure 2 shows two traces of the transmitted light as a
function of wavenumber: one where isotopically normal HO2
is present and another for H18O2. The spectra shown in this
figure were acquired by modulating the H atom microwave
discharge source (1 s on, 1 s off) and taking the ratio of the
spectra. This ratio method removes optical fringes due to
scattered light in the multipass cell. The absolute H16O2 and
Ozone was generated by passing O2 through a corona arc
discharge using a commercial ozone synthesizer. The ozone/
O2 mixture was passed through a 3 L glass bulb filled with
silica gel. The O3 was trapped and stored on the silica gel in an
ethanol bath cooled to ∼190 K. The ozone was delivered to
the flow tube by passing a measured N2 flow through the silica
gel. This O3/N2 mixture flowed through an absorption cell with
a path length of 0.165 cm where the concentration of O3 was
determined by ultraviolet absorption of the atomic Hg emission
line at 253.7 nm using the O3 cross section19 σ ) 1.15 × 10-17
cm2. This concentration was scaled by the appropriate ratios of
pressures and flows to arrive at a concentration of O3 in the
flow tube of typically (0.1-2.5) × 1016 molecules cm-3
.
Two experimental approaches were employed in this work,
each a separate tactic for reducing the secondary influences of
HO2 regeneration. In one, 18O2 was used in place of 16O2 making
H18O2 in place of HO2. 16O2 and 18O2 (Isotec/Matheson 98.8%)
could be switched through a low-volume solenoid valve between
experiments. The low volume of the valve preserved the purity
of the 18O2 isotope and allowed the two isotopes to be regulated
through the same needle valve. In the other approach, an OH
scavenger, trifluorochloroethylene (CFC-1113), was added to
the reaction mixture. The CFC-1113 (Matheson 99.0%) con-
centration was inferred by measuring each of the flow rates using
calibrated mass flow meters and the total pressure in the flow
tube. The CFC-1113 was mixed with the main N2 flow,
upstream from its introduction to the flow tube.
Tunable Diode Laser. The tunable diode laser (TDL) system
coupled to an astigmatic Herriot cell used in these measurements
has been described more completely elsewhere,20,21 but some
general comments on how the apparatus works are warranted.
The infrared light suitable for HO2 detection was generated by
using a Pb-salt laser diode housed in a liquid nitrogen Dewar.

1586 J. Phys. Chem. A, Vol. 105, No. 9, 2001
Herndon et al.
TABLE 1: Line Positions and Strengths for HO2 and H18O2
line wavenumber
(cm-1
line strength (10-21 cm2
transition
N′K′a,K′c r_N′K′a,K′c
species
HO2
)
molecule-1 cm-1
)
references
1407.063
1407.066
1407.069
1407.078
1407.103
1407.108
1410.968
1410.983
1411.031
1411.043
1411.101
1411.113
3.4
4.7
3.4
5.5
5.0
5.0
5.6
5.0
5.3
5.1
2.6
2.8
F1 [62,5 r 52,4
F2 [71,7 r 61,6
F1 [62,4 r 52,3
F1 [71,7 r 61,6
]
]
]
]
Zahniser et al.25
and Nagai et al.36
are tabulated in the
HITRAN28 database
this work
H18O2
Figure 2. H16O2 and H18O2 transmission spectra. The solid lines are
the data while the dashed lines are the least-squares fits. The black
data are H16O2 and the gray data are H18O2. The concentration of each
species is ∼5 × 1010 molecules cm-3 in the multipass cell. The residuals
Figure 3. Velocity profile measurements. Results of the Pitot tube
measurements of the flow velocity as a function of radial distance from
the center of the flow tube at three injector positions are displayed.
The average flow velocity is displayed as a dashed line. The solid line
represents an area-weighted average of the measured velocity profile
of the turbulent core. The gray parabolic profile corresponds to a fully
developed laminar flow and is included for comparison.
of the fit have an rms absorbance of 1 × 10-4
.
H18O2 concentrations were calculated by fitting the absorption
feature to a Voigt line shape using the line parameters given in
Table 1 and the broadening coefficients determined in earlier
work.27 The detection sensitivity with signal-to-noise ratio ) 2
persistent concentration gradient across the turbulent core. Under
experimental conditions typical of this work, the correction to
k′ due to axial diffusion, using Deff ) 250 cm2 s-1, is less than
1% and was not applied to these measurements. The magnitude
of this correction is less than the uncertainty in k′ that results
from the uncertainty in the effective velocity determination (5%).
The second consideration when using the turbulent flow
reactor lies in measuring the effective velocity of the turbulent
core. An empirical relationship was developed relating the Re
and average flow velocity to the appropriate effective velocity.
This allows the effective velocity (to be used in eq 2, above) to
be determined for a given average flow velocity. Some typical
results of the velocity profile measurements are shown in Figure
3. Two Teflon “turbulizers” shortened the distance for turbulent
flow development to less than 10 cm. The distance between
the turbulizers was greater than the tooth-gap size of the
turbulizers by a factor of 30 to prevent the second turbulizer
from sampling a persistent stream or jet caused by the first.
The initial large-scale turbulent motions caused by the second
turbulizer cause vortices in the downstream flow which are on
the order of the radius of the tube. The large-scale turbulence
is dampened to “fully developed” turbulent flow within 10 cm,
and the associated vortices are much smaller.29 Though there
are some small differences between these velocity profiles at
different injector positions, no overall trend with injector position
was observed. The flow velocity measurements at each measured
radial position for a given flow profile were averaged with area
was 20 ppb, which corresponded to an absorbtivity of 2 × 10-4
,
at 10 Torr in the multipass cell. At each injector position, data
was collected for 20 s, which increased the detection sensitivity
by a factor of 101/2 to a mixing ratio of 7 ppb. This corresponds
to a number density of 2 × 1010 molecules cm3 in the flow
tube (100 Torr, 298 K).
Results and Discussion
When employing the turbulent flow reactor technique, two
important considerations are assessing the magnitude of the
effective dispersion coefficient, Deff, and measuring the effective
velocity of the turbulent core. The experiments conducted by
Seeley et al.9 have shown that Deff under turbulent flow
conditions becomes dominated by the small-scale turbulent
mixing, and is between 50 and 500 cm2 s-1 with no dependence
on Reynolds number, when the flow is fully turbulent. The
velocity fluctuations in the axial and radial directions are larger
in fully developed turbulent flow than in laminar flow; however,
they are sufficiently small compared to the average velocity so
that, in the turbulent core, plug flow is nearly achieved. Also,
the radial velocity fluctuations cause eddies whose sizes relative
to the tube diameter are appreciable and allow for rapid radial
mixing. As a result, whatever wall effects are occurring through
the turbulent boundary layer do not manifest themselves as a

Reaction of HO2 with O3 from 200 to 300 K
J. Phys. Chem. A, Vol. 105, No. 9, 2001 1587
Figure 4. Velocity of turbulent core as a function of Reynolds number.
Results of the Pitot tube velocity profile measurements are displayed.
The ratio of the area weighted effective velocity to the average velocity
is plotted as a function of Re. Data are shown for the following
conditions: open triangles, 100 Torr; open circles, 50 Torr; open
squares, 200 Torr. The solid circles were taken at the same injector
position, but the temperature was varied, from 220 to 275 K. The spread
of the data about the empirical fit line is ∼5% and represents a source
of systematic uncertainty in the measured value of k1.
Figure 5. Temporal profiles of H16O2 and H18O2 in the presence of
O3. HO2 mixing ratios are shown as a function of injector distance for
the case where the source is H + 16O2 (depicted by the solid fit lines)
and H + 18O2 (shown by the dashed lines). The × symbols are the
H16O2 concentrations while the open circles are those of H18O2. These
experiments were conducted at 160 Torr, 295 K, and with [O3] ) 1.2
× 1016 molecules cm3.
k1
weighting to determine the effective flow velocity (νeff). The
ratio of νeff to the average flow velocity νave is shown in Figure
4. In some of the measurements the temperature of the gas was
varied, but no trend in effective velocity with temperature was
observed. The relationship between the quantity νeff/νave to Re
was found to be 1 + 25.7/Re0.661 and is depicted by the solid
line. These data show a slightly different power dependence
for the Reynolds number than that predicted by theory.30 The
empirical function is used to estimate the effective core velocity
used in eq 2. Note that for Re > 2200 (the lowest used in this
work), the difference between the average velocity and that of
the turbulent core is approximately 15%.
Kinetics Measurements H18O2 Isotope Measurements. The
first approach adopted in this work for measuring k1 without
interference from HO2 regeneration was to use isotopically
labeled HO2. The inner injector delivered H atoms to the outer
injector (see Figure 1), where either O2 or 18O2 was present,
generating HO2 or H18O2 via the following two reactions.
HO2 + O3
OH + O3
9
8 OH + 2O2
(R1)
(R2)
k2
98 HO2 + O2
k4a
9
H18O2 + O3
8 OH + 18O2 + O2
(R4a)
k4b
9
k4c
9
k4d
9
k4e
9
8 18OH + 18O16O + O2 (R4b)
8 H16O2 + 18O216O
(R4c)
(R4d)
(R4e)
8 H18O16O + 18O16O2
8 H16O18O + 18O16O2
None of the listed product channels for reaction R4 has a
plausible chance of regenerating H18O2 through an elementary
reaction with normal O3. The isotopic scrambling reactions
(R4c-R4e) have also been shown to be small7 and contribute
<2% to the total, k4. Isotopic exchange between the O3 and
18O2 in the flow reactor would complicate this measurement,
by opening a possible regeneration of H18O2 via reaction of
18OH with 18O-16O2 present in the reactor. The exchange
between molecular oxygen and ozone has been shown to be
very slow.31,32 Assuming the most favorable conditions for this
source of H18O2 regeneration, the amount of 18O16O2 would be
0.01% of the total O3. The natural abundance of the 18O isotope
(0.2%) is much greater than the amount formed by scrambling
between 18O2 and O3 in the flow tube. This source of
regeneration could make the apparent k1 less than the true k1
by less than 0.02% due to the fact that, at room temperature,8
the ratio of k4b to k4 is <12%.
The manner of data collection while using the isotope allows
for a convenient measurement of the wall loss at each O3
concentration. The ratio of H18O2 (when (R3b) is employed) to
HO2 (when (R3a) is used) yields temporal decays, due exclu-
sively to k1[O3]. The ratio of H18O2 to H16O2 taken under the
circumstances described is depicted in Figure 6 vs injector
distance. The figure legend gives the specific number densities
of O3. A coarse guide to the ozone concentration is given by
the shading of the points where the darker points indicate greater
O3 than the lighter points. This shows temporal decays of H18O2
H + O2
9
M8 HO2
M8 H18O2
(R3a)
(R3b)
H + 18O2
9
H18O2 is not regenerated by reaction R2 due to the very low
abundance of 18O in normal O3. Figure 5 depicts the observed
H16O2 and H18O2 concentrations for the two sources in the pres-
ence of 1.2 × 1016 molecules cm-3 O3, as a function of injector
distance. When the source of HO2 is (R3a), the concentration
of H18O2 is zero (lower trace). H16O2 exhibits a slow (10 s-1
)
decay (top trace) due to the cycling between HO2 and OH while
being lost to other processes (walls). This decay rate constant
is independent of the amount of O3 added. When the source of
radicals is (R3b), the H18O2 signal decays while the H16O2 signal
grows in time. In this instance, H18O2 is not being regenerated
and its loss is due to the sum of reaction with O3 and loss on
the walls. When the source of radicals is (R3a), the apparent
decay of H16O2 is due to loss of radical to the walls. The ratio
of H18O2 to H16O2, when using sources (R3b) and (R3a),
respectively, gives a temporal decay due exclusively to k1[O3].
The reactions given below show some of the relevant
processes to consider when measuring k1 using isotopically
substituted HO2.

1588 J. Phys. Chem. A, Vol. 105, No. 9, 2001
Herndon et al.
than any previously published results. One set of experiments
(solid diamonds) performed at T ) 208 K shows considerable
imprecision relative to those conducted at T ) 211 K (solid
bow ties). The initial concentration of H18O2 in these two sets
of experiments was 35 and 200 ppb, respectively. The data show
that the ratio technique leads to a zero intercept as anticipated.
The individual fit lines are shown in the legend for each
temperature. A summary of the measurements using the H18O2
isotope is given in Table 2.
In the flow tube where H18O2 is formed (R3b), a small amount
of H218O2 may be formed, which could regenerate H18O2 upon
reaction with OH. The results of a simple numerical box model
show that under the worst circumstances in the injector the
H218O2 concentration delivered to the flow tube (1 × 1012
molecules cm3) would result in an apparent rate constant which
is 0.2% greater than the true value. The box model also predicts
the intercept of the k′ vs [O3] to be less than the measured wall
loss. The observed intercepts in the plots of k′ vs [O3] are
consistent with the measured wall loss when the H18O2 decay
is considered alone. The H218O2 + OH mechanism is not a
significant source of H18O2 regeneration.
When using the isotopically labeled HO2 (R3b as the source),
the ratio technique (following H18O2 decay) is the most
straightforward way to analyze the isotope data. Product
formation, however, can also yield information about k1, i.e.,
the appearance of H16O2 when (R3b) is used as the radical
source. The temporal behavior of H16O2(t) can be solved directly
from the rate laws for H16O2, H18O2, and OH radical with a
few assumptions which greatly simplify the result. Assuming
the wall loss rate constant for HO2 is the same as that for OH,
(R4a) is the only product channel in reaction R4, and k1 ) k4,
the result is given by
Figure 6. Temporal profiles of the ratio of H18O2 (R3b source) to
H16O2 (R3a source). In the upper panel marked (a), the ratio of H18O2
when 18O2 is present in the source to H16O2 when 16O2 is present in the
source is plotted vs injector distance. The total pressure in the reactor
was 130 Torr, and the temperature was 273 K. The darker data points
are indicative of greater [O3] relative to the lighter data points. The
increase in the pseudo-first-order decay constant with increasing O3
can be seen. The initial concentration of H18O2 and H16O2 was 53 ( 4
ppb for each of the decays shown in the figure. In the lower panel
marked (b), the corresponding k′ vs [O3] plot is shown.
[H16O2]t
)
[H18O2]0
1
wall}t
(k2e-k t + k1e-{(k +k )[O ]+k }t) - e-{k [O ]+k
(4)
wall
1
2
3
wall
1
3
k1 + k2
The use of this equation to describe the HO2 temporal profile
provides another measurement of k1, to the extent that k2 is
known. Using this equation to fit k1 to the H16O2 appearance is
problematic, though because data collection was restricted to
reaction times greater than ∼5 ms with the geometry employed
in the majority of this work. Nevertheless, when k1 was used
as a fixed parameter in (4) to see how well this expression
predicts magnitude of formed [H16O2], insight was gained into
the HO2/OH/O3 cycling mechanism. Using the H18O2 decay
channel to determine k1 and kwall and taking k2 from ref 19, eq
4 overpredicts the amount of HO2 generated at the 295 and 273
K data points by 18%. At the lower temperatures, 211 and 208
K, however, the agreement is much better 5%. More of the
reaction is proceeding through the H atom abstraction (R4a)
channel at lower temperatures than at room temperature.
Alternatively, the improved agreement could be due to the wall
loss coefficients for HO2 and OH becoming closer in magnitude
at low temperatures than at room temperature. In either case,
the temporal behavior of H16O2 confirms the basic mechanism
postulated for HOx cycling.
Figure 7. k′ vs [O3]. Measured pseudo-first-order decay coefficients
vs ozone concentration are plotted for the experiments conducted
between 197 and 211 K. The solid points were measured using the
isotope technique, where the decay is determined from H18O2/H16O2,
and as a result these decays have a zero intercept. The open data points
are the total decays measured using the OH scavenger where the wall
loss of HO2 shows as the positive intercept (6-12 s-1).
relative to the null or cycling H16O2 increase with O3 concentra-
tion. The pseudo-first-order rate constants from the temporal
profiles similar to those in Figure 6 are plotted versus [O3] to
determine k1. Such a plot is given in Figure 7, which shows
results for both the isotope experiments conducted at 208 and
211 K. Figure 7 also shows the results for the OH scavenger
experiments discussed below. The data on this figure represent
the majority of the experiments conducted at temperatures lower
Discharge Flow Laminar Conditions. Some initial work
using isotopic substitution was performed on this apparatus using
a traditional low-pressure laminar flow approach. In this
instance, the multipass mirrors were physically mounted in the
six-way cross at the end of the flow tube. The wall loss rate
coefficient for H18O2 at lower temperatures was deemed too

Reaction of HO2 with O3 from 200 to 300 K
J. Phys. Chem. A, Vol. 105, No. 9, 2001 1589
TABLE 2: Temperature-Dependent Measurements of k1
rate constant
temperature
(K)
pressure
(Torr)
mass flow
(slm)
flow velocity
Reynolds
number
no. of
expts
(1 × 10-15 cm3
(cm s-1
)
molecule-1 s-1
)
technique
297
296
296
295
295
273
238
233
225
219
211
208
204
202
197
4.8
154
156
80-171
110
130
174
157
5.3
4.7
112
138
4.2
108
98-141
0.6
60
68
44-54
38
44
54
64
1
1
49
57
0.9
44
370
1042
1186
657-1492
940
837
675
870
391
435
842
792
430
755
627-902
40
3505
4
5
4
25
5
6
7
7
2
4
6
6
4
5
2.25 ( 0.2
2.17 ( 0.2
1.9 ( 0.2
2.0 ( 0.2
2.1 ( 0.6
1.5 ( 0.2
1.4 ( 0.3
1.17 ( 0.12
1 ( 1.0
1.2 ( 0.2
1.1 ( 0.2
1.1 ( 0.6
0.9 ( 0.9
1.1 ( 0.2
0.92 ( 0.2
laminar 18O2
scavenger
3933
2524-3108
2186
2715
3602
4317
70
18O2 isotope
scavenger
scavenger
18O2 isotope
scavenger
scavenger
laminar 18O2
laminar 18O2
18O2 isotope
18O2 isotope
laminar 18O2
scavenger
68
3524
4190
55
3284
3712-4467
49-59
15
scavenger
great to measure k1 using the laminar flow method with the
desired accuracy and led to the application of the turbulent flow
being sampled into the multipass cell as shown in Figure 1.
The data from these measurements are designated “laminar 18O2”
in Table 2 and are included for comparison purposes.
Kinetics Measurements: OH Scavenger. Another technique
for preventing the regeneration of HO2 from influencing the
measured k1 has been used in this work. In these experiments,
an OH scavenger has been added to the reaction mixture.
In reaction R5, [M] refers to the number density of the bath
complex and subsequent reaction with O2 to form a halogenated
peroxy radical.33 At the pressure and number density of O2
employed in this work, this is expected to be the dominant
channel and ejection of a Cl atom is unlikely.
The measured value of k1, using the isotope and scavenger
techniques, agrees at both room temperature and low temper-
atures (197-211 K). This suggests that the secondary chemistry
sources unique to each technique are negligible contributors to
the measured values of k1.
Self-Reaction. At the number densities of HO2 employed in
the flow tube, typically less than 4 × 1011 molecules cm-3, the
self-reaction could be significant. The total rate constant for
the HO2 self-reaction consists of bimolecular and termolecular
components, shown below.
[M], k5
OH + ClFCCF2
8 OH‚C2F3Cl
(R5)
gas. The results of the measurement of k at the four lowest
temperatures vs [O3] are shown in Figure 7. Over the pressures
and flow velocities employed at T ) 197 K, the figure shows
that k1 is invariant. The dashed line represents the fit to all of
the compiled data at 197 K. The results of this work at all
temperatures are summarized in Table 2, and in the technique
column, “scavenger” refers to the addition of C2F3Cl to the
reaction mixture.
The rate constant for reaction R5 in 7 Torr of He at room
temperature33 is 6 × 10-12 cm3 molecule-1 s-1. The rate constant
exhibits some bimolecular and termolecular character. The
previously measured value is assumed to be a lower limit for
k5, as these experiments employ ∼100 Torr of N2 as the bath
gas. In this work, experiments where the scavenger concentration
was varied did not indicate any dependence of the rate
coefficient measurements with the scavenger concentration >5
× 1014 molecules cm3. The concentration of C2F3Cl was at least
3 × 1015 molecules cm-3 in the scavenger experiments reported
here. At the highest concentrations of O3 and the highest
temperatures in this study, the measured rate coefficient, k,
would be influenced by regeneration by at most 2%. A
previously reported4 upper limit for the reaction of HO2 with
C2F3Cl is 2 × 10-16 cm3 molecule-1 s-1 which at the
concentrations used in these experiments would represent a
pseudo-first-order loss of HO2 of <1 s-1. This effect would
not vary with O3 and is not anticipated to have an effect on the
measurement of k1.
k6a
HO2 + HO2
98 H2O2 + O2
(R6a)
(R6b)
[M], k6b
8 H2O2 + O2
The rate constant for the self-reaction of HO2, reaction R6 at
200 Torr and 298 K, is ∼2 × 10-12 cm3 molecule-1 s-1 and
increases with decreasing temperature. A simple numerical box
model was constructed to ascertain the effects of reaction R6
on the measurement of k1. The model was used to construct
temporal HO2 data considering reactions (R1, R2, and R5), then
again with (reaction R6) added to the model. These data were
fit, neglecting the initial times due to the flow tube end effect
(approximation time < 10 ms). The result of these model trials
for the greatest amounts of initial HO2 (1 × 1012 molecules
cm-3) was that the apparent or measured k1 would be up to 5%
lower than the true k1 at 195 K (1% at 298 K). The model also
predicts that in zero O3, the apparent first-order decay rate
constant would be greater than that chosen for the wall. The
results from the isotope experiments give an effective kwall which
does not change with [O3]. As a result, reaction R6 does not
have a significant impact on the results within the precision of
these measurements. The measured values of the rate constant
have not been corrected by the small percentage predicted by
the model.
Several variations in pressure and flow velocity were
performed at room temperature using the scavenger technique.
The measured rate constant showed no systematic trend with
these variations. This lends additional confidence that the
turbulent core velocity correction function (see Figure 4) is not
introducing a systematic bias that is greater than the precision
of the apparatus.
There are two possibilities for the OH‚C2F3Cl complex. The
first possibility is the ejection of a Cl atom.33 Were this to occur,
the Cl atom would react with O3, forming ClO. The rate constant
for the reaction of ClO with HO2 is great enough to be of
concern despite the low concentration of ClO. The second
possibility for the excited adduct is the stabilization of the

1590 J. Phys. Chem. A, Vol. 105, No. 9, 2001
Herndon et al.
reference
TABLE 3: Rate Constant Parameters for This and Previous Work (measurements of k1)
temperature
range (K)
pressure
(Torr)
A (1 × 10-15 cm3
C (1 × 10-15 cm3
molecule-1 s-1
)
Ea/R (K-1
)
molecule-1 s-1
)
technique
DF-LMR
DF-LMR
245-365
243-413
14 ( 4
38 ( 24
320 ( 580
1.9 ( 0.3
18 ( 6
103 ( 57
88 ( 35
580 ( 100
820 ( 190
1730 ( 740
Zahniser et al.4
2.0-3.2
Sinha et al.7
1.2 ( 0.5
298
253-400
197-300
197-300
1.6-2.0
0.5-3.0
50-230
DF-photofragment
DF-photofragment
TFR-TDL
Manzanares et al.5
Wang et al.6
this work
680 ( 148
1323 ( 160
1303 ( 116
0.88 ( 0.1
0.9 ( 0.1
all data, T < 300 K
Conclusions
This work extends the low-temperature measurements of k1
considerably, down to 197 K. This work has used two methods,
isotopic labeling and OH scavenging, for measuring k1 directly,
without appreciable influence from the regeneration of HO2.
Within the precision of these measurements, these two schemes
agree. In the temperature range where previous measurements
were available, 233-400 K, the determination of k1 presented
in this work agrees within ∼10%.
The use of the turbulent flow reactor has allowed a direct
examination of k1 at lower temperatures and higher pressures
than traditional flow tube methods. Secondary chemistry in the
flow tube with respect to the regeneration of H16O2 and/or or
H18O2 via reaction R2 and other possibilities has been shown
to be well understood and not a significant source of error. The
turbulent flow reactor technique has been shown to be capable
of measuring rate constants for other systems.11,12 The agreement
of this work with other measurements of k1 gives further
confidence that the turbulent flow reactor technique can be
employed to measure gas-phase rate constants.
Figure 8. Arrhenius plot for k(HO2+O3). The measured rate constants
are displayed as a function of inverse temperature. The open triangles
are the data of Wang et al., the open circles are the data of Sinha et al.,
the open squares are the data of Zahniser and Howard, and the room-
temperature measurement of Manzanares et al. is depicted as a solid
square. The solid symbols represent this work. The triangles are the
laminar flow measurements, the squares are the scavenger measure-
ments, and the bow ties are the isotope measurements. The solid line
represents the best fit to all the data below 300 K, using the form
described in the text. The dashed line is the recommended rate constant
and the shaded region represents the uncertainty in the rate constant
taken from Sander et al.35
To determine an expression for use in stratospheric modeling,
a simple fit of all data was done (this work and the previous
direct measurements of k1 for T < 300 K). The empirical form
of the expression chosen is the Arrhenius expression plus a
constant and results in k1(T) ) {(88 ( 35) exp[-(1303 ( 116)/
T] + 0.9} × 10-15 cm3 molecule-1 s-1
.
These measurements show that k1 exhibits non-Arrhenius
behavior. The rate constant for the O atom transfer (terminal O
on HO2) channel has been previously shown8 to be a small
fraction of the total rate constant at low temperatures and is
not responsible for the observed curvature. It remains possible
that at low temperatures new uninvestigated channels open up
or that H atom tunneling becomes significant.
Comparison to Previous Measurements. The previous
measurements of k1 along with the rate constants determined
in this work are shown in Figure 8 and summarized in Table 3.
In Figure 8, the straight dashed line represents the current
recommended rate constant for use in stratospheric modeling
and the shaded area is the uncertainty in this rate constant.35
The solid black line represents a simple fit of the combined
data, T < 300 K, to an empirical form which is the traditional
Arrhenius form, plus a constant. A fit of the data in the present
work yields the expression k1(T) ) {(103 ( 57) exp[-(1323
( 160)/T] + 0.88} × 10-15 cm3 molecule-1 s-1, which
adequately describes the curvature in the Arrhenius expression
over the limited temperature range of this study.
The first observation that may be drawn from looking at
Figure 8 and Table 3 is that the absolute agreement between
this work and each of the previous studies is very good, within
10%. The second and most important is that k1 exhibits non-
Arrhenius behavior. At 200 K, the rate constant is 40% greater
than that predicted by an Arrhenius extrapolation of the higher
temperature measurements.
Acknowledgment. We thank the NASA Upper Atmospheric
Research Program for the support of this work under Contract
No. NAS5-32917. Invaluable discussions with Charles Kolb and
Richard Miake-Lye are gratefully acknowledged.
References and Notes
(1) Wennberg, P. O. Science 1994, 266, 398.
(2) Cohen, R. C.; Wennberg, P. O.; Stimpfle, R. M.; Koplow, J.;
Anderson, J. G.; Fahey, D. W.; Woodbridge, E. L.; Keim, E. R.; Gao, R.
S.; Proffitt, M. H.; Loewenstein, M.; Chan, K. R. Geophys. Res. Lett. 1994,
21, 2539-2542.
(3) Lanzendorf, E. J.; Hanisco, T. F.; Stimpfle, R. M.; Anderson, J.
G.; Wennberg, P. O.; Cohen, R. C.; Gao, R. S.; Margitan, J. J.; Bui, T. P.
J. Phys. Chem. A, this issue.
(4) Zahniser, M. S.; Howard, C. J. J. Chem. Phys. 1980, 73, 1620.
(5) Manzanares, E. R.; Suto, M.; Lee, L. C. J. Chem. Phys. 1986, 85,
5027-5034.
These results are also consistent with recently reported34 low-
temperature results for the sum of k1 + k2. This apparent
agreement is not a strong confirmation of the low-temperature
results, because the sum is dominated by the reaction of OH
with O3 (k2).
(6) Wang, X.; Suto, M.; Lee, L. C. J. Chem. Phys. 1988, 88, 896-
899.
(7) Sinha, A.; Lovejoy, E. R.; Howard, C. J. J. Chem. Phys. 1987, 87,
2122-2128.
(8) Nelson, D. D.; Zahniser, M. S. J. Phys. Chem. 1994, 98, 2101-
2104.

Reaction of HO2 with O3 from 200 to 300 K
J. Phys. Chem. A, Vol. 105, No. 9, 2001 1591
(9) Seeley, J. V. Experimental Studies of Gas-Phase Radical Reactions
Using the Turbulent Flow Tube Technique, Ph.D. Thesis, MIT, 1994.
(10) Seeley, J. V.; Jayne, J. T.; Molina, M. J. Int. J. Chem. Kinet. 1993,
25, 571-594.
(24) Buchanan, J. W.; Thrush, B. A.; Tyndall, G. S. Chem. Phys. Lett.
1983, 103, 167-168.
(25) Zahniser, M. S.; McCurdy, K. E.; Stanton, A. C. J. Phys. Chem.
1989, 93, 1065-1070.
(11) Seeley, J. V.; Jayne, J. T.; Molina, M. J. J. Phys. Chem. 1996,
100, 4019-4025.
(26) Nelson, D. D.; Zahniser, M. S. J. Mol. Spectrosc. 1991, 150, 527-
534.
(27) Nelson, D. D.; Zahniser, M. S. J. Mol. Spectrosc. 1994, 166, 273-
279.
(28) Rothman, L. S.; Rinsland, C. P.; Goldman, A.; Massie, S. T.;
Edwards, D. P.; Flaud, J.-M.; Perrin, A.; Camy-Peyret, C.; Dana, V.;
Mandin, J.-Y.; Schroeder, J.; McCann, A.; Gamache, R. R.; Wattson, R.
B.; Yoshino, K.; Chance, K. V.; Jucks, K. W.; Brown, L. R.; Nemtchinov,
V.; Varanasi, P. J. Quant. Spectrosc. Radiat. Transfer 1998, 60, 665-710.
(29) Sullivan, P. J. J. Fluid Mech. 1971, 49, 551-576.
(30) White, F. M. Fluid Mechanics; McGraw-Hill: New York, 1979.
(31) Anderson, S. M.; Morton, K.; Mauersberger, K.; Yung, Y. L.;
DeMore, W. B. Chem. Phys. Lett. 1992, 189, 581.
(12) Seeley, J. V.; Meads, R. F.; Elrod, M. J.; Molina, M. J. J. Phys.
Chem. 1996, 100, 4026-4031.
(13) Brenner, H. Langmuir 1990, 6, 1715-1724.
(14) Poirier, R. V.; Carr, R. W. J. J. Phys. Chem. 1971, 75, 1593-
1601.
(15) Abbatt, J. P. D.; Demerjian, K. L.; Anderson, J. G. J. Phys. Chem.
1990, 94, 4566.
(16) Keyser, L. F. J. Phys. Chem. 1984, 88, 4750-4758.
(17) Beenakker, C. I. M. Spectrochim. Acta 1976, 31B, 483-486.
(18) Emrich, R. D. Methods of Experimental Physics; Academic Press:
New York, 1981; Vol. 18A.
(19) 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 Photochemistry Data for Use in Stratospheric
Modeling;” Jet Propulsion Laboratory, 1997.
(20) McManus, J. B.; Kebabian, P. L.; Zahniser, M. S. Appl. Opt. 1995,
34, 3336-3348.
(21) Zahniser, M. S.; Nelson, D. D.; McManus, J. B.; Kebabian, P. L.
Philos. Trans. R. Soc. London A 1995, 351, 371-382.
(22) Saito, S.; Matsumura, C. J. Mol. Spectrosc. 1980, 80, 34-40.
(23) Paukert, T. T.; Johnson, H. S. J. Chem. Phys. 1972, 56, 2824-
2838.
(32) Anderson, S. M.; Morton, J.; Mauersberger, K. Geophys. Res. Lett.
1987, 14, 1285.
(33) Howard, C. J. J. Chem. Phys. 1976, 65, 4771.
(34) Nizkorodov, S. A.; Harper, W. W.; Blackmon, B. W.; Nesbitt, D.
J. J. Phys. Chem. 2000, 104, 3964-3973.
(35) 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 Photochemistry Data
for Use in Stratospheric Modeling, Supplement to Evaluation 12: Update
of Key Reactions,” Jet Propulsion Laboratory, 2000.
(36) Nagai, K.; Y., E.; Hirota, E. J. Mol. Spectrosc. 1981, 89, 520.