Mechanism of HOx formation in the gas-phase ozone-alkene reaction. 1. Direct, pressure-dependent measurements of prompt OH yields

Article abstract of DOI:10.1021/jp002121r

The gas-phase reaction of ozone with alkenes is known to be a dark source of HO_x radicals (such as OH, H, and R) in the troposphere, though the reaction mechanism is currently under debate. It is understood that a key intermediate in the reaction is the carbonyl oxide, which is formed with an excess of vibrational energy. The branching ratios of the ozone-alkene reaction products (and thus HO_x yields) depend critically on the fate of this intermediate: it may undergo unimolecular reaction (forming either OH or dioxirane) or be collisionally stabilized by the bath gas. To investigate this competition between reaction and quenching, we present direct, pressure-dependent measurements of hydroxyl radical (OH) yields for a number of gas-phase ozone-alkene reactions. Experiments are carried out in a high-pressure flow system (HPFS) equipped to detect OH using laser-induced fluorescence (LIF). Hydroxyl radicals are measured in steady state, formed from the ozone-alkene reaction and lost to reaction with the alkene. Short reaction times (usually approximately 10 ms) ensure negligible interference from secondary and heterogeneous reactions. For all substituted alkenes covered in this study, low-pressure yields are large but decrease rapidly with pressure, resulting in yields at 1 atm which are significantly lower than current recommendations and indicating the important role of collisional stabilization in determining OH yield. The influence of alkene size and degree of substitution on pressure-dependent yield is consistent with the influence of collisional stabilization as well as the accepted reaction mechanism.

Full text of DOI:10.1021/jp002121r

1554
J. Phys. Chem. A 2001, 105, 1554-1560
Mechanism of HOx Formation in the Gas-Phase Ozone-Alkene Reaction. 1. Direct,
Pressure-Dependent Measurements of Prompt OH Yields†
Jesse H. Kroll,* James S. Clarke,‡ Neil M. Donahue,§ and James G. Anderson
Department of Chemistry and Chemical Biology, HarVard UniVersity, Cambridge, Massachusetts 02138
Kenneth L. Demerjian
Department of Earth and Atmospheric Sciences, and Atmospheric Sciences Research Center, SUNY-Albany,
Albany, New York 12203
ReceiVed: June 12, 2000; In Final Form: October 18, 2000
The gas-phase reaction of ozone with alkenes is known to be a dark source of HOx radicals (such as OH, H,
and R) in the troposphere, though the reaction mechanism is currently under debate. It is understood that a
key intermediate in the reaction is the carbonyl oxide, which is formed with an excess of vibrational energy.
The branching ratios of the ozone-alkene reaction products (and thus HOx yields) depend critically on the
fate of this intermediate: it may undergo unimolecular reaction (forming either OH or dioxirane) or be
collisionally stabilized by the bath gas. To investigate this competition between reaction and quenching, we
present direct, pressure-dependent measurements of hydroxyl radical (OH) yields for a number of gas-phase
ozone-alkene reactions. Experiments are carried out in a high-pressure flow system (HPFS) equipped to
detect OH using laser-induced fluorescence (LIF). Hydroxyl radicals are measured in steady state, formed
from the ozone-alkene reaction and lost to reaction with the alkene. Short reaction times (usually ∼10 ms)
ensure negligible interference from secondary and heterogeneous reactions. For all substituted alkenes covered
in this study, low-pressure yields are large but decrease rapidly with pressure, resulting in yields at 1 atm
which are significantly lower than current recommendations and indicating the important role of collisional
stabilization in determining OH yield. The influence of alkene size and degree of substitution on pressure-
dependent yield is consistent with the influence of collisional stabilization as well as the accepted reaction
mechanism.
Introduction
The gas-phase reaction of ozone and alkenes has received
considerable attention in recent years due to its unique potential
as a dark source of tropospheric HOx radicals. Over the past
decade, numerous indirect scavenger studies1-8 have provided
strong evidence for significant hydroxyl (OH) radical formation
from this class of reaction. Recently our laboratory presented
the first direct evidence of prompt OH formation, using laser-
induced fluorescence to detect OH in the reaction mixture.9 The
amount of hydroxyl radical formed from such reactions is
thought to be important in the atmosphere: it has been shown
that measured yields are substantial enough to be a significant,
even dominant, contributor of total HOx in rural and urban
environments.10,11
Figure 1. Criegee mechanism of carbonyl oxide formation: ozone-
alkene cycloaddition (R1) and subsequent dissociation (R2) to form
the carbonyl oxide.
The mechanism by which OH is believed to be formed is
shown in Figures 1 and 2. Ozone adds to the alkene (reaction
R1) to form a vibrationally excited primary ozonide, which
quickly dissociates (R2) into a carbonyl oxide (the Criegee
intermediate) and a carbonyl species. If the substituents on either
side of the alkene double bond are different, this dissociation
has two possible sets of products.
Figure 2. Unimolecular reaction channels available to the substituted
carbonyl oxide: isomerization (R3a) to an excited hydroperoxide, which
will quickly dissociate (R3b) to OH or ring closure to form dioxirane
(R4).
The vibrationally excited carbonyl oxide may react via further
unimolecular reactions or be collisionally stabilized by the bath
gas. There are two primary reaction pathways, dissociation to
OH (R3) and ring closure to dioxirane (R4),12,13 as shown in
Figure 2 for a substituted carbonyl oxide. OH formation takes
place via an excited hydroperoxide intermediate, which is
formed via a five-membered transition state. The fate of the
† Part of the special issue “Harold Johnston Festschrift”.
* Corresponding author. E-mail kroll@fas.harvard.edu.
‡ Present Address: Laboratorium fu¨r Organische Chemie, ETH Zu¨rich,
CH-8092 Zu¨rich, Switzerland.
§ Present Address: Departments of Chemistry and Chemical Engineering,
Carnegie Mellon University, Pittsburgh, PA 15213-3890.
10.1021/jp002121r CCC: $20.00 © 2001 American Chemical Society
Published on Web 12/05/2000

Gas-Phase Ozone-Alkene Reaction
J. Phys. Chem. A, Vol. 105, No. 9, 2001 1555
Experimental Section
dioxirane, which is also formed with excess vibrational energy,
is not clear. If the carbonyl oxide is unsubstituted (that is, the
methyl group is replaced by an H), OH will form not via a
hydroperoxide but rather in a single concerted step. This step
occurs via a four-membered transition state and so has a higher
barrier to reaction;14 consequently, the degree of substitution is
expected to play an important role in OH yield.
As reported previously,9 we observe radicals in the steady
state. OH and H are formed in the ozone-alkene reaction (with
yields of R and â, respectively) and are lost to reaction with
the alkene:
O3 + alkene f ROH + âH + products
(R5)
Thus far, most OH yield measurements have relied on the
use of a radical scavenger added to the reaction mixture. Most
such experiments are performed in environmental chambers, and
due to the slow rate of reaction between ozone and alkene, they
are carried out over long periods of time, typically several
minutes to hours. Thus, unwanted secondary chemistry may
interfere. In these studies, OH yield is determined in one of
two ways: either from the yield of a product of the OH-
scavenger reaction (such as production of cyclohexanone from
OH + cyclohexane and cyclohexanol2,3 or production of CO2
from OH + CO1) or from the decrease in the scavenger
concentration.5,7,8 A third tracer technique, in which OH yield
is determined by the change in rate of alkene loss when a
scavenger is added to the system, has been recently introduced.6
In all scavenger studies save one15 (which is believed to have
significantly underestimated experimental error6,8), the loss of
scavenger or formation of products is consistent with production
of OH. Yield results from various studies are generally in good
agreement; for a comprehensive list of OH yield measurements
see Paulson et al.8
OH + alkene f products
H + alkene f products
(R6)
(R7)
When reaction R5 is the sole source and reaction R6 the sole
sink of OH, the steady-state concentration of OH is
kO [O3]
3
[OH]ss ) R
(1)
kOH
where OH concentration is linear with ozone concentration but
independent of alkene. Therefore, a yield may be obtained by
a single ozone and radical measurement. However, it is more
precise to take a derivative approach:
k
OH ∂[OH]
R )
(2)
kO3
∂[O3]
We take the latter approach in this study, obtaining yields by
measuring steady-state radical concentrations at a number of
ozone concentrations. This approach is also useful in that it
provides a check on our understanding of the OH chemistry:
nonlinearities or nonzero intercepts in [OH] versus [O3] suggest
additional chemistry not included in reactions R5 through R7.
The H-atom yield â may be found using steady-state equations
similar to eqs 1 and 2.
For the above equations to hold, a number of experimental
conditions must be met. The first is that the radicals must be
observed in the steady state. Since the alkene serves as the
radical sink and thus controls the approach to steady state, this
condition can be met by using a high concentration of alkene.
In addition, it is necessary that the above reactions (R5-R7)
completely describe the chemistry of OH and H in the system.
Even in a simple N2/O2/O3 environment, there exist numerous
additional sources and sinks of OH, including
Recently we reported the first direct observation of OH
radicals from the ozone-alkene reaction.9 Radicals were
detected using laser-induced fluorescence (LIF) and were
measured in steady state, so that no tracer was necessary. In
that experiment, OH was detected only 50 ms after reaction
initiation, thus minimizing interference by secondary reactions.
Formation of OH by secondary reactions could not be entirely
ruled out, but prompt formation was suggested by the fact that
for one reaction, [OH] remained linear with [O3] down to low
ozone concentrations. OH production has since been confirmed
using a different spectroscopic technique.16
While our previous study unequivocally confirmed OH as a
product of the ozone-alkene reaction, the pressure range was
low (4-6 Torr). In this study we extend the OH yield
measurements up to hundreds of Torr, allowing for extrapolation
to atmospheric pressures with much less uncertainty. Yields
presented here are also considerably more precise than those
presented before, and reaction times are significantly shorter,
so the influence of secondary reactions is negligible. In addition,
in our previous study we reported detecting H atoms by
resonance fluorescence (RF); here we have calibrated our RF
measurement and therefore present H-atom yields as well.
H + O3 f OH + O2
H + O2 + M f HO2 + M
OH + O3 f HO2 + O2
HO2 + O3 f OH + 2O2
(R8)
(R9)
(R10)
(11)
All of these may be safely neglected so long as all OH and H
radicals react with alkene only. Thus, for this study large alkene
concentrations and small ozone concentrations are required.
However, this does not necessarily eliminate all interferences
from secondary chemistry. Sources of OH (or H) that are
independent of [O3], such as bimolecular reactions of stabilized
intermediates, may still exist. Thus there is the additional
requirement that radicals formed via reaction R5 only be formed
from unimolecular decomposition of the vibrationally excited
intermediates (via reactions 1-4) and not by intermediate
bimolecular channels; this may be accomplished by making the
experimental time scale as small as possible.
Our objective is to understand the role of collisional stabiliza-
tion in OH production, including the influence of substitution
on the carbonyl oxide. Consequently, we have selected a series
of symmetric alkenes whose reaction pathways are as simple
as possible. In all cases only a single carbonyl oxide is produced,
though in some cases stereoisomers may play a role. The alkenes
in this study are ethene, which produces an unsubstituted
carbonyl oxide; trans-2-butene, trans-3-hexene, and trans-4-
octene, which produce monosubstituted carbonyl oxides; and
2,3-dimethyl-2-butene (also called tetramethylethylene, or TME)
and 3,4-dimethyl-3-hexene, which produce disubstituted car-
bonyl oxides. As we shall see, this series is sufficient to elucidate
and separate the competing roles of size and substitution on
collisional stabilization and OH production.
Flow systems, in which experimental time scales are con-
trolled by flow velocity, are well-suited for such measurements.
The high-pressure flow system (HPFS) used in this study was

1556 J. Phys. Chem. A, Vol. 105, No. 9, 2001
Kroll et al.
pressure in the cell to obtain [O3] in the tube. In a separate
HPFS, we have compared concentrations obtained using this
technique with FTIR measurements;22 concentrations agree to
within a few percent over a wide range of pressures.
Since this is a bulk sampling technique, measurements cannot
be made from within the reaction zone without the risk of
interfering with the radical concentrations, spectroscopic mea-
surements, or flows. Thus the UV cell is placed near the end of
the tube; this requires that by the time alkene is injected into
the flow, the ozone profile has stopped evolving, so that the
[O3] in the reaction zone equals that further downstream. In
other words, the ozone must be of uniform radial concentration;
using the UV cell we have verified that in the reaction zone
this is the case, over a wide range of experimental conditions.
Alkene is injected into the center of the flow via a quartz
injector, located 2.4 m downstream of the beginning of the tube.
As mentioned above, high alkene concentrations are generally
desired; however, care is taken that flow through the alkene
injector is not so high as to interfere with the flat ozone profile
(see ref 19). Alkenes used in this study were ethene (99.5+%,
Matheson); trans-2-butene, 2,3-dimethyl-2-butene, trans-3-
hexene (all 99+%, Aldrich); 3,4-dimethyl-3-hexene (99+% cis/
trans, Chemsampco); and trans-4-octene (90+%, Aldrich). We
do not measure alkene concentration directly, but it may be
estimated on the basis of flows through calibrated flow
controllers; a typical alkene concentration in the flow tube is
1015-1016 molecules/cm3.
Hydroxyl radicals are detected using laser-induced fluores-
cence (LIF), 5-10 cm (corresponding to 5-10 ms) downstream
of the alkene injector. While at this point the alkene is a plume,
the alkene is well-mixed within the very small volume of the
LIF measurement. LIF detection of OH is discussed in ref 23,
and its adaptation for the HPFS is described in ref 18. A
frequency-doubled, tunable dye laser pumps the A2Σ+ (V ) 1)
r X2Π (V ) 0) transition in OH (at 282.2 nm). Some of the
excited OH fluoresces back to the ground state, but most is
collisionally stabilized to the V ) 0 state and then fluoresces or
is collisionally stabilized to the ground state. This red-shifted
fluorescence at 309 nm is measured using a PMT (Hamamatsu)
placed normal to the laser beam. A photodiode placed at the
end of the beam monitors average UV laser power. The PMT
signal, corrected for undercounting23 and scaled by average
power, yields a signal which is proportional to OH number
density. The LIF axis used in this experiment is more sensitive
than previous HPFS designs, due to better optical filters (Omega
Optical) as well as a wider solid angle.
Figure 3. The high-pressure flow system (HPFS) used in this study.
For H-atom yield measurements, the RF axis is moved upstream
(replacing the LIF axis) to ensure the measurement of prompt H
formation.
initially designed for measuring rates of radical-molecule
reactions,17 and flow conditions have been extensively ana-
lyzed.18,19 Here major features of the system necessary for yield
measurements are outlined. The HPFS is shown in Figure 3; it
consists of a stainless steel flow tube, 3.5 m long and 12.36 cm
in diameter. The diameter is sufficiently large such that, within
the reaction zone, species do not have time to diffuse from the
center of the tube to the wall, so that the system is effectively
“wall-less”,19 and thus, potential heterogeneous reactions are
categorically eliminated from the observed chemistry. In addi-
tion, the t ) 0 of the experiment is well-defined, with extremely
good time resolution, so that we may make measurements a
known time (on the order of milliseconds) after reaction
initiation.
Nitrogen is used as the bath gas in this experiment and is
regulated by a 400 slm flow controller (MKS). Flow velocity
is monitored with a pitot-static tube attached to a 0.2 Torr
differential capacitance manometer (MKS) located at the end
of the flow tube and is typically 10 m/s. The pressure of the
system may be varied from under 1 Torr to an atmosphere;
however, pressures over 50 Torr are controlled not by the flow
controller but by the valve to the vacuum pump, so that at higher
pressures, flow velocities are typically smaller (down to 1 m/s).
To measure the H-atom yield, we replace the LIF detection
axis with a resonance fluoresence (RF) detection axis. Measure-
ment of H atoms in the HPFS is described in detail in ref 17.
A sealed lamp consisting of UH3 in Ne is heated and subjected
to an rf plasma, sending Lyman-R (121.6 nm) light into the
tube, again normal to the flow. H atoms in the flow absorb the
Lyman-R light, and a PMT perpendicular to the lamp monitors
the subsequent fluorescence; the fluorescence signal is then
scaled by the lamp power, measured by a photodiode opposite
the lamp.
Ozone is injected at the beginning of the tube to ensure a
uniform radial concentration. Ozone is generated by passing
ultrahigh-purity oxygen through a glass tube and subjecting it
to a high-voltage electric field. Ozone concentration is modu-
lated by varying the voltage through the oxygen; at maximum
voltage (∼12 000 V ac) approximately 7% of the O2 is converted
to O3. The O2/O3 mixture is then sent into the flow tube at 100
sccm.
Both the LIF and RF have been calibrated by titrating NO2
in an excess of H atoms:
Ozone concentration is measured using a UV White cell.20
A mechanical pump draws gas from the center of the flow tube
1
through /4 in. Teflon tubing and into the cell. Light from the
NO2 + H f OH + NO
(R12)
254 nm Hg line is multipassed through the cell, for a total path
length of 600 cm. The UV cross-section of ozone is well-known
at this wavelength (1.157 × 10-17 cm2 21), so a highly precise
ozone concentration within the cell may be obtained; this
number is then scaled by the ratio of pressure in the HPFS to
Trace amounts (109-1011/cm3) of NO2 in the HPFS are
maintained in an advective steady state through the addition of
a standard NO2/He mixture, determined to be 36.4 ppm NO2
by FTIR spectroscopy. The NO2 is well-mixed by the time it

Gas-Phase Ozone-Alkene Reaction
J. Phys. Chem. A, Vol. 105, No. 9, 2001 1557
Figure 4. Typical run, for ozone + TME at 10 Torr. Ozone is modulated at five different concentrations, with the resultant OH concentration
varying accordingly. OH yield is the slope of the [OH] vs [O3] line, scaled by kOH/kO3
.
reaches the reaction zone, and its concentration is calculated
on the basis of the measured flows. Our H-atom source,
described elsewhere,17 is essentially a H2/He/Ar mixture sent
through a microwave plasma and injected as a plume into the
center of the flow through a quartz injector. We verify that NO2
is titrated by changing [H] within the plume and finding no
change in LIF signal. Calibrations are performed by modulating
NO2 flow (and thus concentration); the calculated NO2 con-
centration is equal to loss of H and formation of OH, so we
may correlate our change in RF and LIF signals with concentra-
tion.
at least a factor of 3 and often a factor of 10 or more. A yield
measurement for a given alkene and pressure takes between 10
and 45 min.
Because of the reduced sensitivity of LIF and RF at high
pressures (>50 Torr), higher concentrations of ozone, and
therefore alkene, are often used. Also in these cases, as well as
at low pressures (<3 Torr), velocity can be slower as well.
Therefore it is possible that a significant fraction of the ozone
will have reacted between the reaction zone and the UV cell.
To compensate for this, we measure ozone concentrations
without alkene and use these values for calculating the yield;
this correction is generally of the order of a few percent and is
never higher than 15%.
Calibrations have been performed over a range of pressures
(5-50 Torr). At high pressures both the RF and LIF calibrations
vary with pressure inversely, as expected, since in both cases
collisional quenching competes with fluorescence from the
excited state. The full pressure dependence of the LIF measure-
ment is understood, and the rates of the individual processes
(collisional stabilization vs fluorescence of the excited OH) are
known (see ref 23). Our LIF calibration data fit this known
pressure dependence well. We have some reason to question
the accuracy of our calibration (see Results below), but precision
is very high (∼10%), and the measured pressure dependences
of the yields are reliable. Soon we plan to more accurately
calibrate our LIF system by measuring the N2 Raman shift at
302 nm.23
Results
A typical run, for TME at 10 Torr, is shown in Figure 4.
Ozone is held at five different concentrations, and the resulting
OH signal is proportional to ozone concentration. The detection
of OH is unequivocal; detuning the laser reveals the 3:1 doublet
characteristic of the OH transition. Laser-generation of OH, by
photolysis of O3 to O(1D) followed by H-atom abstraction, is a
potential problem, but our experimental design greatly mini-
mizes this interference. We draw on the proven design of the
NASA ER2 HOx instrument,23 using very high laser pulse
repetition rates to keep the peak laser power (and thus secondary
production) low. Two separate tests have established that none
of the OH observed is laser-generated. First, measured OH yield
is independent of laser power, suggesting that no two-photon
processes (namely ozone photodissociation and subsequent OH
detection) are occurring. Second, negligible LIF signal is
observed when a large amount of isobutane is added to the ozone
in the system. Isobutane has a tertiary hydrogen and thus is
expected to be significantly more susceptible to H-atom
abstraction than any of the alkenes used in this study. We thus
conclude that laser-generated OH does not interfere with our
experiment.
Before a yield measurement is made, a test is done in which
ozone and alkene are both injected into the system and alkene
flow is modulated. By eq 1, steady-state radical concentration
should be independent of alkene concentration, so if the LIF
(or RF) signal does not change with a change in alkene flow,
the radical is assumed to be in steady state and the yield
measurement is performed.
Yield measurements are performed by sending a constant flow
of alkene into the reaction zone and modulating ozone on and
off at various concentrations. During the “off” cycle the O2/O3
flow is diverted downstream of the UV cell, while background
LIF (or RF) and UV signals are taken. In a typical run, this is
repeated for four to six ozone concentrations, spanning [O3] by
Plotting [OH] vs [O3] shows an extremely tight correlation,
as predicted by the steady-state expression (eq 2). The yield, as

1558 J. Phys. Chem. A, Vol. 105, No. 9, 2001
Kroll et al.
Figure 5. OH yields versus pressure (in Torr) for the six alkenes covered in this study. Data are fit to an empirical function which duplicates
master equation results well; yields may be greater than 1 due to uncertainties in kOH, kO3, and LIF calibration (see text).
TABLE 1: Rate Constants for Reactions of Alkenes with O3,
determined by the slope of the line multiplied by kOH/kO3, is
1.86. The measured yield above unity is probably not real but
OH, and Ha
alkene
kO3/(10-17
0.159((0.048) p-depb
19((6.7)
113((40)
13((1.5)
)
kOH/(10-12
)
kH/(10-12
p-depc
)
instead caused by errors in kOH, kO3, or the LIF calibration, all
of which scale the raw data. Such errors do not affect the
precision of the measurements, nor do they significantly affect
the reported pressure dependence of the yields: as mentioned
above, the pressure dependence of LIF sensitivity is well-
constrained. Uncertainties in the pressure dependences of the
rate constants (kOH, kO3) may be a factor, but there is little
evidence that those rates are significantly pressure-dependent.
An important exception is the reaction of OH with ethene, but
the pressure dependence of that reaction is well-characterized.24
ethene
trans-2-butene
TME
trans-4-octene
64.0((12.8) 0.825((0.006)
110((22)
69((17)
na
1.34((0.010)
na
1.05((0.006)
3,4-dimethyl-3-hexene 37
a Values for kOH and kO3 are from ref 1 and kH are from ref 17. Rates
are in molecules/cm3/s; “na” indicates no value is available. b Pressure-
dependent rate constant, from ref 24. c Pressure-dependent rate constant,
from ref 17.
We have found that [OH] remains linear with [O3] as low as
1011 molecules O3/cm3. This demonstrates that the OH is indeed
a prompt product of the reaction and is not formed by secondary
processes: the lifetime of a species which reacts at the collision
rate (k ) 3× 10-10 cm3 molecule-1 s-1) with ozone or a reaction
product (with concentration of e1011 molecules/cm3) is 33 ms,
longer than the time scale of the experiment. Furthermore,
bimolecular rate constants for carbonyl oxides are believed to
be much slower than the collision rate;24 secondary reactions
with ozone or reaction products are thus not significant in this
experiment. Our alkene concentration (1015-1016/cm3) is much
higher than [O3], so secondary reactions with the alkene are at
least possible on the time scale of our experiment. However,
this is unlikely to lead to any interference, as there are few
alkene reactions that could produce OH; those that might (such
as reactions with HO2, or carbonyl oxide + alkene cycloaddi-
tions) have high barriers and thus long time scales well beyond
that of our experiment. We also cannot completely rule out
bimolecular reactions with O2, which is present in higher
concentrations than O3; however, by varying [O2] we see no
change in the measured yield.
Table 1 presents the values of the rate constants used for
determining yield; kOH and kO3 are from the recommendations
in ref 24 and kH are from the values determined in this
laboratory, presented in ref 17. For the six alkenes in this study,
two kOHs (trans-3-hexene and 2,3-dimethyl-3-hexene) and two
kHs (trans-3-hexene and trans-4-octene) have not been mea-
sured; in these cases we approximate the kOH/kO3 or kH/kO3 ratio
as that for trans-2-butene or TME, depending on the level of
alkene substitution.
Pressure-dependent OH yields for the six alkenes in this study
are shown in Figure 5. The fact that for most alkenes maximum
yields exceed unity suggests that the LIF calibration is imperfect;
however, errors in rate constants may also contribute. Data are
fit to the empirical function yield(P) ) a0 exp[-a,Pa2] (in which
a0, a1, and a2 are adjustable parameters), which we have found

Gas-Phase Ozone-Alkene Reaction
J. Phys. Chem. A, Vol. 105, No. 9, 2001 1559
TABLE 2: OH Yields as Measured by Scavenger Studies at 1 Atm
study
ethene
trans-2-butene
trans-3-hexene
TME
Atkinson et al.1,2
Chew and Atkinson3
Gutbrod et al.4
0.12(+0.06/-0.04)
0.64(+0.32/-0.21)
1.00(+0.50/-0.33)
0.80((0.12)
0.36((0.02)
0.89((0.22)
1.00
0.08((0.01)
0.14((0.07)
0.18((0.06)
0.24((0.02)
0.54((0.14)
0.64
Rickard et al.5
Paulson et al.7,8
0.47((0.07)
TABLE 3: Low-Pressure Hydrogen Atom Yields
alkenes with the same number of vibrational modes but different
degree of substitution (for example, TME and trans-3-hexene),
the observed pressure dependences are not drastically different,
even though the disubstituted carbonyl oxide (from TME) is
significantly more stable than the monosubstituted one. This
suggests that only the excess energy of the ozonideswhich is
not expected to vary much from reaction to reactionsdetermines
the vibrational energy content of the carbonyl oxide.
alkene
yield
ethene
0.076 ( 0.060
trans-2-butene
trans-3-hexene
trans-4-octene
TME
0.045 ( 0.0028
0.021 ( 0.0041
0.0072 ( 0.0026
0.0022 ( 0.0010
0.0024 ( 0.0021
3,4-dimethyl-3-hexene
In addition, we present the first direct measurements of
H-atom yields from gas-phase ozonolysis. The H yields are
extremely small (<1%) for the fully substituted alkenes (TME
and 3,4-dimethyl-3-hexene), which is consistent with the
hydroperoxide channel being the dominant reaction mechanism.
For ethene, H yield is larger (at low pressures at least), on the
order of a few percent. This suggests an alternate reaction
mechanism; one possible source is the “hot acid” channel,26 in
which the dioxirane isomerizes to highly excited formic acid,
via a dioxymethane intermediate. This acid can then dissociate
to one of many sets of products, including H atoms; the observed
H may also arise from the dissociation of HCO. For the
disubstituted alkenes, H yields are small but not insignificant,
ranging from 4.5% for trans-2-butene to 0.7% for trans-4-
octene. This too may be from dissociation of the “hot acid,” to
form H + RCO2. However, the observed H may be formed
instead via the hydroperoxide channel (reaction R6b): the
resulting HCOCR2 fragment may further dissociate, to a ketene
and an H atom.
Table 2 presents a summary of OH indirect yield measure-
ments, made at 1 atm. With the exception of ethene, extrapola-
tion of the data in this study to 1 atm suggests OH yields
significantly lower than those determined in most scavenger
studies. Recently, the Paulson laboratory has determined ethene
OH yields to be pressure-dependent, with significantly higher
yields at pressures below 100 Torr.27 Because of the limited
pressure range of our measurements, as well as the large error
bars in our calculations, we cannot rule out the possibility of a
pressure dependence. However, the same study reports no
pressure dependence for trans-2-butene, trans-3-hexene, or
TME,27 in sharp contrast with the results of this work.
It is of course extremely important to understand the source
of this discrepancy. An obvious candidate is the different method
of OH detection: direct, spectroscopic detection versus indirect,
scavenger-based measurements. But this is unlikely to introduce
any major errors, as the measurements in scavenger studies are
most certainly of OH and not some other oxidant. Thus the
difference likely lies in an additional source of OH, not present
in our system but present in the scavenger studies.
fits master equation results well (we discuss our use of the
master equation in a separate paper25). For the fit, data points
are weighted by their respective uncertainties.
Measurements of OH yields for ethene above 10 Torr were
not made due to very low precision: when pressure is increased,
not only does LIF sensitivity decrease but the OH-ethene rate
constant (kOH) increases, decreasing the steady-state concentra-
tion of OH.
Due to the poor signal-to-noise ratio, H-atom yields are
determined at low pressures: 10 Torr for the substituted alkenes
and 5 Torr for ethene. Results are presented in Table 3. For all
alkenes, yields are small but nonzero; in general, yield decreases
with size and degree of substitution.
Discussion
For the reaction of ozone with all the alkenes studied, OH is
found to be produced promptly. In all cases save ethene, yields
are found to be highly pressure-dependent, indicating competi-
tion between collisional stabilization and dissociation of a
vibrationally excited intermediate, presumably the carbonyl
oxide. This conclusion is supported by the fact that the data
are generally fit well by our empirical function; one notable
exception is the high-pressure TME data.
Given the competition between stabilization and dissociation,
we would expect that the larger alkenes (those with more
vibrational modes) would exhibit the greatest pressure depen-
dence, as they are the most susceptible to collisional stabiliza-
tion. Within the series of alkenes studied, this effect seems to
be weak but existent: trans-2-butene, for example, certainly
exhibits a weaker pressure dependence than trans-4-octene.
Yields from larger alkenes (including many species which are
important in the troposphere) are expected to be even more
strongly pressure-dependent.
As is to be expected, OH yield is closely related to degree of
alkene substitution. TME is found to have the highest yields,
consistent with the reaction mechanism shown in Figure 2, in
which OH is formed via the hydroperoxide channel. On the
other hand, ethene is found to have the lowest yields, consistent
with isomerization to dioxirane dominating over dissociation
to OH. The three disubstituted alkenes (trans-2-butene, trans-
3-hexene, and trans-4-octene) may react by either mechanism,
and correspondingly their OH yields are intermediate.
However, while the degree of substitution may control
absolute yields, it does not seem to have a significant influence
on the pressure dependences of such yields. The degree of
substitution affects not only which reaction pathways are
available but also the stability of the carbonyl oxide, since alkyl
groups serve to stabilize the electron-poor carbon. For two
This additional source may arise from generation of OH via
secondary chemistry in the scavenger studies that could not
occur within the time scale of our experiments. OH has been
suggested as a possible product of the reaction of carbonyl oxide
with carbonyls28 or water.29,30 In addition, RO2 chemistry is
highly uncertain and may also be an OH source, via either
unimolecular or bimolecular channels. Still, there is little
evidence that any of these potential sources could produce
enough OH to account for the large differences in yields
measured in this study and in scavenger studies.

1560 J. Phys. Chem. A, Vol. 105, No. 9, 2001
Kroll et al.
In another paper25 in this series, we present an alternate
explanation for this discrepancy, using the time-dependent
master equation. We show that the carbonyl oxide, assuming it
does not undergo fast bimolecular reactions, may dissociate to
OH thermally on the time scales of the scavenger studies, but
not of our experiments.
(3) Chew, A. A.; Atkinson, R. J. Geophys. Res. 1996, 101, 28649.
(4) Gutbrod, R.; Meyer, S.; Rahman, M. M.; Schindler, R. N. Int. J.
Chem. Kinet. 1997, 29, 717.
(5) Rickard, A. R.; Johnson, D.; McGill, C. D.; Marston, G. J. Phys.
Chem. A 1999, 103, 7656.
(6) Neeb, P.; Moortgat, G. K. J. Phys. Chem. A 1999, 103, 9003.
(7) Paulson, S. E.; Fenske, J. D.; Sen, A. D.; Callahan, T. W. J. Phys.
Chem. A 1999, 103, 2050.
(8) Paulson, S. E.; Chung, M. Y.; Hassan, A. S. J. Phys. Chem. A
1999, 103, 8125.
(9) Donahue, N. M.; Kroll, J. H.; Anderson, J. G.; Demerjian, K. L.
Geophys. Res. Lett. 1998, 25, 59.
Conclusions
We have presented spectroscopic measurements of OH yields
from the gas-phase ozone-alkene reaction for a series of simple,
symmetric alkenes over a wide range of pressures. These
constitute the first direct gas-phase OH measurements from this
class of reaction at pressures above a few Torr. In all cases we
observe prompt OH formation, and in all cases (save ethene)
the yield is pressure-dependent. This is consistent with a
competition between unimolecular reaction and collisional
stabilization of the vibrationally excited carbonyl oxide inter-
mediate. As expected, the larger alkenes exhibit a greater
pressure dependence, but this effect is not particularly strong.
The relationship between yield and degree of substitution is more
pronounced, suggesting our understanding of the reaction
mechanism (reactions R1-R4) is qualitatively accurate, even
if the branching ratios are not well constrained. In another
paper25 we use statistical-dynamical calculations to estimate
these branching ratios, further constraining the reaction mech-
anisms. Yields of H atoms were also measured directly for the
first time, and may be understood in terms of the “hot acid”
channel, available only for ethene and disubstituted alkenes.
Finally, for the substituted alkenes, our measured yields,
extrapolated to 1 atm, are considerably lower than those
determined by scavenger studies; this point, too, will be
addressed in the companion paper.
(10) Paulson, S.; Orlando, J. Geophys. Res. Lett. 1996, 23, 3727.
(11) Ariyam, P. A.; Sander, R.; Crutzen, P. J. J. Geophys. Res. 2000,
105, 17721.
(12) Martinez, R. I.; Herron, J. T.; Huie, R. E. J. Am. Chem. Soc. 1981,
103, 3807.
(13) Niki, H.; Maker, P. D.; Savage, C. M.; Breitenbach, L. P.; Hurley,
M. D. J. Phys. Chem. 1987, 91, 941.
(14) Gutbrod, R.; Schindler, R. N.; Kraka, E.; Cremer, D. Chem. Phys.
Lett. 1996, 252, 221.
(15) Scha¨fer, C.; Horie, O.; Crowley, J. M.; Moortgat, G. K. Geophys.
Res. Lett. 1997, 24, 1611.
(16) Pfeiffer, T.; Forberich, O.; Comes, F. J. Chem. Phys. Lett. 1998,
298, 351.
(17) Clarke, J. S.; Donahue, N. M.; Kroll, J. H.; Rypkema, H. A.;
Anderson, J. G. J. Phys. Chem. 2000, 104, 5254.
(18) Abbatt, J. P. D.; Demerjian, K. L.; Anderson, J. G. J. Phys. Chem.
1990, 94, 4566.
(19) Donahue, N. M.; Clarke, J. S.; Demerjian, K. L.; Anderson, J. G.
J. Phys. Chem. 1996, 100, 5821.
(20) Weinstock, E. M.; Schiller, C. M.; Anderson, J. G. Geophys. Res.
Lett. 1986, 91, 5237.
(21) DeMore, W. B.; Sander, S. P.; Golden, D. M.; Hampson, R. F.;
Kurylo, M. J.; Howard, C. J.; Ravishankara, A. R.; Kolb, C. E.; Molina,
M. J. Chemical kinetics and photochemical data for use in stratopsheric
modeling. Technical Report 97-4, Jet Propulsion Laboratory, 1997.
(22) Donahue, N. M.; Demerjian, K. L.; Anderson, J. G. J. Phys. Chem.
1996, 100, 17855.
(23) Wennberg, P.; Cohen, R. C.; Hazen, N. L.; Lapson, L. B.; Allen,
N. T.; Hanisco, T. F.; Oliver, J. F.; Lanham, N. W.; Demusz, J. N.;
Anderson, J. G. ReV. Sci. Instr. 1994, 65, 1858.
(24) Calvert, J. G.; Atkinson, R.; Kerr, J. A.; Madronich, S.; Moortgat,
G. K.; Wallington, T. J.; Yarwood, G. The Mechanisms of Atmospheric
Oxidation of the Alkenes. Oxford University Press: New York, 2000.
(25) Kroll, J. H.; Sahay, S. R.; Anderson, J. G.; Demerjian, K. L.;
Donahue, N. M. J. Phys. Chem. A, submitted.
(26) Herron, J. T.; Huie, R. E. J. Am. Chem. Soc. 1977, 99, 5430.
(27) Fenske, J. D.; Hasson, A. S.; Paulson, S. E.; Kuwata, K. T.; Ho,
A.; Houk, K. N. J. Phys. Chem. A 2000, 104, 7821.
(28) Neeb, P.; Horie, O.; Moortgat, G. K. J. Phys. Chem. A 1998, 102,
6778.
Acknowledgment. This work was supported by US EPA
STAR grant R825258010 (University at Albany-SUNY and
Harvard University), as well as NSF grant 9414983 (Harvard
University). J.H.K. gratefully acknowledges support from the
NSF Graduate Student Fellowship program.
References and Notes
(29) Paulson, S. E.; Flagan, R. C.; Seinfeld, J. H. Int. J. Chem. Kinet.
1992, 23, 103.
(1) Atkinson, R.; Aschmann, S. M.; Arey, J.; Shorees, B. J. Geophys.
Res. 1992, 97, 6065.
(2) Atkinson, R.; Aschmann, S. M. EnViron. Sci. Technol. 1993, 27,
1357.
(30) Neeb, P.; Sauer, F.; Horie, O.; Moortgat, G. K. Atmos. EnViron.
1997, 31, 1417.