Direct kinetics study of the temperature dependence of the CH2O branching channel for the CH3O2 + HO2 reaction

Article abstract of DOI:10.1002/kin.1030

A direct kinetics study of the temperature dependence of the CH₂O branching channel for the CH₃O₂ + HO₂ reaction has been performed using the turbulent flow technique with high-pressure chemical ionization mass spectrometry for the detection of reactants and products. The temperature dependence of the CH₂O-producing channel rate constant was investigated between 298 and 218 K at a pressure of 100 Torr, and the data were fitted to the following Arrhenius expression. 1.6_-0.7^+1.0 × 10^-15 × exp[(1730 ± 130)/T] cm³ molecule^-1 s^-1. Using the Arrhenius expression for the overall rate of the CH₃O₂ + HO₂ reaction and this result, the 298 K branching ratio for the CH₂O producing channel is measured to be 0.11, and the branching ratio is calculated to increase to a value of 0.31 at 218 K, the lowest temperature accessed in this study. The results are compared to the analogous CH₃O₂ + CH₃O₂ reaction and the potential atmospheric ramifications of significant CH₂O production from the CH₃O₂ + HO₂ reaction are discussed.

Full text of DOI:10.1002/kin.1030

Direct Kinetics Study of the
Temperature Dependence
of the CH2O Branching
Channel for the CH3O2 ؉
HO2 Reaction
MATTHEW J. ELROD, DANA L. RANSCHAERT, NICHOLAS J. SCHNEIDER
Department of Chemistry, Hope College, Holland, Michigan, 49423
Received 8 September 2000; accepted 26 January 2001
ABSTRACT: A direct kinetics study of the temperature dependence of the CH2O branching
channel for the CH3O2 ϩ HO2 reaction has been performed using the turbulent flow technique
with high-pressure chemical ionization mass spectrometry for the detection of reactants and
products. The temperature dependence of the CH2O-producing channel rate constant was
investigated between 298 and 218 K at a pressure of 100 Torr, and the data were fitted to the
1.0
15
1
1
following Arrhenius expression: 1.6ϩϪ ϫ 10Ϫ ϫ exp[(1730 Ϯ 130)/T] cm3 moleculeϪ sϪ
.
0.7
Using the Arrhenius expression for the overall rate of the CH3O2 ϩ HO2 reaction and this
result, the 298 K branching ratio for the CH2O producing channel is measured to be 0.11, and
the branching ratio is calculated to increase to a value of 0.31 at 218 K, the lowest temperature
accessed in this study. The results are compared to the analogous CH3O2 ϩ CH3O2 reac-
tion and the potential atmospheric ramifications of significant CH2O production from the
CH3O2 ϩ HO2 reaction are discussed. ᭧ 2001 John Wiley & Sons, Inc. Int J Chem Kinet 33: 363–376,
2001
INTRODUCTION
on the levels of the nitrogen oxides (NOx). Under high
NOx conditions (generally, lower tropospheric condi-
tions), CH3O2 reactions lead to the production of
ozone, the most deleterious constituent of photochem-
ical smog.
The methylperoxy radical (CH3O2) is an important in-
termediate species formed in the oxidation of methane
in the atmosphere [1].
CH3O2 ϩ NO !: CH3O ϩ NO2 (4a)
CH4 ϩ OH !: CH3 ϩ H2O
CH4 ϩ O(1D) !: CH3 ϩ OH
CH3 ϩ O2 ϩ M !: CH3O2 ϩ M
(1)
(2)
(3)
NO2 ϩ h␯ (␭ Ͻ 380 nm) !: NO ϩ O
O ϩ O2 ϩ M !: O3 ϩ M
(5)
(6)
CH3O2 can also be temporarily removed from the
ozone production cycles by the formation of reservoir
species.
Ozone levels in the atmosphere are directly affected
by CH3O2 reactions, which themselves are dependent
Correspondence to: M. J. Elrod (elrod@hope.edu)
᭧ 2001 John Wiley & Sons, Inc.
CH3O2 ϩ NO ϩ M !: CH3ONO2 ϩ M (4b)

364
ELROD ET AL.
CH3O2 ϩ HO2 !: CH3OOH ϩ O2
(7a)
rate constant k7b (using the isotopically substituted
CD3O2 reactant and detecting the HDO product from
reaction 7b) [8]. These studies are in fairly wide dis-
agreement, with estimates for the secondary CH2O-
producing channel branching ratio [k7b/(k7a ϩ k7b)]
ranging from as low as zero [16] to as high as 0.40
[8]. Indeed, because of this lack of agreement, the JPL
recommendation for the CH3O2 ϩ HO2 reaction spe-
cifically calls for new studies using direct CH2O
detection techniques [14]. Because the products
CH3OOH (from reaction 7a) and CH2O (from reaction
7b) are characterized by quite different photochemical
reactivity (with a potentially varying impact on HOx
cycling and O3 production efficiency), it is very im-
portant to quantitatively establish the atmosphericrel-
evance of reaction 7b. In addition, it has been sug-
gested by Ayers et al. [17] and Weller et al. [18] that
the underprediction by atmospheric models (which
generally assumes a zero branching ratio value for re-
action (7b)) of CH2O levels in the remote troposphere
may be a consequence of a significant CH2O-forming
product channel for the CH3O2 ϩ HO2 reaction.
In this article we describe our investigation of the
kinetics of the CH3O2 ϩ HO2 reaction conducted at
pressures near 100 Torr and at a range of temperatures
extending to those found in the upper troposphere us-
ing a turbulent flow (TF) tube coupled to a high-pres-
sure chemical ionization mass spectrometer (CIMS).
It has been previously shown that TF technique can
be used to accurately determine the rate constants
of reactions at pressures ranging from 50 to 760 Torr
and at temperatures as low as 180 K [19]. As in our
previous kinetics studies of the CH3O2 ϩ NO and
C2H5O2 ϩ NO reactions using the coupled TF-CIMS
approach [4,20], we are able to directly access atmo-
sphericpressure and temperature conditions and sen-
sitively monitor many of the relevant reactants and
products for the CH3O2 ϩ HO2 reaction. In contrast to
previous work on this reaction, we are able to unam-
biguously monitor both reactants (CH3O2 and HO2) as
well as a species (CH3OOH or CH2O) from each of
the potential product channels 7a and 7b. This capa-
bility allows for the first direct product study of reac-
tion 7b.
!: CH2O ϩ H2O ϩ O2 (7b)
A potential product of reaction 4, CH3ONO2 (from
reaction 4b), has been directly measured in the atmo-
sphere and suggested as a tracer of the photochemical
age of air masses [2], but the chemical processes lead-
ing to its presence in the atmosphere remain unclear
[3], despite recent attempts to establish the atmo-
spheric significance of reaction 4b [4]. Recently,
Wennberg et al. [5] reported the measurement of
higher HOx levels than predicted in the upper tropo-
sphere of the northern hemisphere, suggesting that this
region of the atmosphere is more susceptible to NOx-
catalyzed ozone production than previously thought.
This is an important result since it significantly im-
pacts models that predict the effect of aviation on
ozone levels in the atmosphere. Therefore, it is critical
to obtain the relative rates of reaction 4 and reaction
7 and to understand the product distributions of these
reactions in order to address the impact of CH3O2
chemistry on this issue.
In the first phase of addressing these issues for
CH3O2 chemistry, we investigated reaction 4 by meas-
uring the temperature dependence of the overall rate
constant for the CH3O2 ϩ NO reaction and establish-
ing an upper limit for the CH3ONO2-producing
branching channel [4]. This article details the second
phase of this work: an investigation of the kinetics of
the CH3O2 ϩ HO2 reaction. The overall rate constant
k7 for this reaction has received considerable previous
study by flash photolysis (FP) techniques [6–13]. As
the JPL recommendation for the overall rate constant
for this reaction states [14], the agreement among the
various studies is not very good. The lack of a con-
sensus result for this rate constant is most likely due
to the near universal use of overlapping HO2 and
CH3O2 UV absorption detection methods (for which
different UV cross sections have been used in the de-
termination of the absolute rate constants) and the
complicated secondary chemistry resulting from the
peroxy radical self-reactions (CH3O2 ϩ CH3O2 and
HO2 ϩ HO2).
As for the CH3O2 ϩ NO reaction, a secondary
branching channel for the CH3O2 ϩ HO2 reaction has
also been proposed:
EXPERIMENTAL
CH3O2 ϩ HO2 !: CH2O ϩ H2O ϩ O2 (7b)
Turbulent Fast-Flow Tube Kinetics
There have been several investigations of the
CH3OOH-producing rate constant k7a [9,15,16] as well
as one indirect investigation of the CH2O-producing
A schematic of the experimental apparatus is pre-
sented in Figure 1 and is similar to that used in a pre-
vious study of HO2 ϩ BrO [21] with the BrO source

TEMPERATURE DEPENDENCE OF THE CH2O BRANCHING CHANNEL FOR THE CH3O2 ϩ HO2 REACTION
365
1
with a 0.5–5.0 STP ml minϪ flow of a 1% F2/He
mixture (Excimer grade), which then passed through
a microwave discharge produced by a Beenakker cav-
ity operating at 50 W. To generate CH3, the fluorine
atoms were then injected into a sidearm and mixed
with an excess of CH4 (CP grade, ϳ1015 molecule
3
cmϪ ) in order to ensure that no fluorine atoms were
introduced into the main flow. CH3O2 was then pro-
duced by the addition of an excess of O2 (99.995%;
Figure 1 Experimental apparatus.
3
ϳ1 ϫ 1016 molecule cmϪ ) just downstream of the
production of CH3. Absolute CH3O2 concentrations
were determined by the titration reaction:
replaced by the CH3O2 source described in Scholtens
et al. [4] The flow tube was constructed with 2.2 cm
i.d. Pyrex tubing and was 60 cm in total length. A large
flow of nitrogen carrier gas (approximately 30 STP
CH3O2 ϩ NO !: CH3O ϩ NO2
(10)
1
(k10 ϭ 7.8 ϫ 10Ϫ12 cm3 moleculeϪ1 sϪ ) [14] and sub-
sequent calibration of the NO2 mass spectrometer sig-
nal. Computer modeling of these titration conditions
indicates that slightly more NO2 is produced than
CH3O2 initially present because of the following sec-
ondary reactions:
1
liter minϪ ) was injected at the rear of the flow tube.
The gases necessary to generate CH3O2 were intro-
duced through a 10-cm-long, 12.5-mm diameter side-
arm located at the rear of the flow tube. HO2 was gen-
erated in a triple-nested movable injector. The outer
encasement (made from corrugated Teflon tubing) was
used so that the injector could be moved to various
injector positions without breaking any vacuum seals,
as well as to prevent ambient gases from condensing
on cold portions of the injector. A fan-shaped Teflon
device was placed at the end of the injector in order
to enhance turbulent mixing. The polonium-210,
alpha-emitting ionization source was placed between
the temperature-regulated flow tube and the inlet to
the quadrupole mass spectrometer. Most of the flow-
tube gases were removed at the CIMS inlet by a 31-L
sϪ1 roughing pump. All gas flows were monitored with
calibrated mass flow meters. The flow-tube pressure
was measured upstream of the ionization source using
a 0–1,000 Torr capacitance manometer. The temper-
ature was determined at both the entrance and exit
points of the temperature-regulated region of the flow
tube using Cu–constantan thermocouples.
CH3O ϩ O2 !: CH2O ϩ HO2
HO2 ϩ NO !: OH ϩ NO2
(11)
(12)
15
1
1
(k11 ϭ 1.9 ϫ 10Ϫ cm3 moleculeϪ sϪ and k12
ϭ
8.2 ϫ 10Ϫ cm3 moleculeϪ sϪ ) [14]. Therefore, the
measured NO2 concentrations must be adjusted to de-
termine the correct [CH3O2] 0 value. However, for typ-
ical O2 and NO concentrations, the conversion factor
was close to unity ([NO2]titration ϭ 1.08 [CH3O2]0) be-
cause the following reaction was dominant over reac-
tion (11):
12
1
1
CH3O ϩ NO ϩ M !: CH3ONO ϩ M (13)
11
1
1
(k13 ϭ 1.3 ϫ 10Ϫ cm3 moleculeϪ sϪ at 100 Torr)
[14]. Reaction 13 has a smaller branching channel to
produce HNO and CH2O, but since neither of these
products produces or consumes NO2 under our flow
reactor conditions, it is not expected to impact the ti-
tration results. NO2 impurities in the NO (CP grade)
used in reaction (10) were removed by the use of a dry
ice/methanol-cooled silica gel trap placed between the
NO reservoir (usually a 3% mixture in N2) and the
flow tube. For this study, CH3O2 concentrations
Reactant Preparation
CH3O2 was generated using the following reactions:
CH4 ϩ F !: CH3 ϩ HF
(8)
(9)
CH3 ϩ O2 ϩ M !: CH3O2 ϩ M
3
ranged from about 3 to 10 ϫ 1011 molecule cmϪ .
(k8 ϭ 6.7 ϫ 10Ϫ11 cm3 moleculeϪ1 sϪ1 and k9 ϭ 4.9 ϫ
HO2 was generated from the following reaction:
13
1
1
10Ϫ cm3 moleculeϪ sϪ at 100 Torr) [14]. Fluorine
H ϩ O2 ϩ M !: HO2 ϩ M
(14)
atoms were produced by combining a 2.0 STP liter
1
minϪ flow of helium (99.999%), which had passed
13
1
1
through a silica gel trap immersed in liquid nitrogen,
(k14 ϭ 2.0 ϫ 10Ϫ cm3 moleculeϪ sϪ at 100 Torr)

366
ELROD ET AL.
[14]. Because HO2 was introduced through a movable
injector where the corresponding concentrations are
ϳ30 times higher than in the main flow tube, the dis-
proportionation reaction:
mass-spectrometric calibration. This procedure was
somewhat troublesome in that the mass-spectrometer
signal depended on the time elapsed since the sample
was prepared. This effect was probably due to a drop
in the gaseous CH2O concentration as a result of the
reformation of paraformaldehyde on the glass bulb
walls or on the surfaces of the mass flow meter. Al-
though a regular remixing of the CH2O standard sam-
ple solved this problem, we choose to employ a more
convenient, alternative CH2O calibration method. This
second method takes advantage of the chemistry of the
CH3O2/NO calibration:
HO2 ϩ HO2 (ϩM) !: H2O2 ϩ O2 (ϩM) (15)
12
1
1
(k15 ϭ 1.8 ϫ 10Ϫ cm3 moleculeϪ sϪ at 100 Torr)
[14] is a concern in the production of large quantities
of HO2. By taking advantage of the long lifetime of
H-atoms, this difficulty was surmounted by using a
nested injector that kept the hydrogen atoms (entrained
in the inner 3-mm alumina tube) and oxygen mole-
cules (entrained in the outer 6-mm Pyrex tube) sepa-
rate throughout all but the last 1 cm of the injector.
The hydrogen atoms were allowed to mix with a very
large excess of O2 (ϳ3 ϫ 1017 molecule cmϪ3 inside
the injector) for only about 1 ms, allowing reaction 14
to virtually go to completion, but preventing signifi-
cant self-reaction of HO2. Hydrogen atoms were gen-
erated by combining a 5.0 STP L minϪ1 flow of helium
(99.999%, which had passed through a liquid-nitro-
gen-cooled silica gel trap) with a 0.5 to 4.0 STP mL
CH3O2 ϩ NO !: CH3O ϩ NO2
CH3O ϩ O2 !: CH2O ϩ HO2
(10)
(11)
By using high O2 concentrations, most of the CH3O2
reactant can be converted to CH2O, with the result that
a secondary calibration for CH2O can be made by ref-
erence to the NO2 calibration that is part of the CH3O2
procedure. In order to effect the most advantageous
conditions, we used O2 as the carrier gas for the
CH3O2/NO titration/calibration experiments (for ex-
periments at 100 Torr total pressure, [O2] Ͼ 3.0 ϫ
1
minϪ flow of a 10% H2(99.9%)/He mixture, which
was then sent through a microwave discharge pro-
duced by a Beenaker cavity operating at 70 W. Ab-
solute HO2 concentrations were determined in a fash-
ion similar to that for CH3O2:
3
1018 molecule cmϪ ), so that nearly all of the CH3O2
initially present is converted to CH2O (for typical con-
ditions, [CH2O] ϭ 0.8[CH3O2]) . Since the CH2O for
this calibration process is generated in situ, the prob-
lems with paraformaldehyde formation are avoided.
Therefore, the CH3O2/NO titration/calibration ap-
proach was routinely used for CH2O calibration, while
the direct standard sample CH2O method was used
only occasionally to ensure the consistency of the two
approaches.
HO2 ϩ NO !: OH ϩ NO2
(12)
followed by calibration of the NO2 mass spectrometer
signal.
CH2O Branching-Channel Measurements
In these studies, the production of CH2O from reaction
(7b) was monitored directly over a reaction time of
ϳ25 ms. The absolute reaction times were calculated
from the flow velocity and the distance between the
injector and the mass spectrometer sampling aperture.
Computer modeling was used to extract the rate con-
stant k7b from the observed production of CH2O and
the initial concentrations of all relevant chemical spe-
cies. In order to determine the detection sensitivity of
the mass spectrometer for CH2O, two different cali-
bration methods were employed. The first method in-
volved the preparation of a standard sample of CH2O
and the introduction of a metered flow of CH2O to the
system. The standard CH2O sample was prepared by
heating paraformaldehyde and transferring the vapor
to a carefully dried glass bulb. The CH2O was then
diluted with N2 and added to the system for direct
Chemical Ionization Mass Spectrometric
Detection
A positive ion chemical ionization scheme (with
H3Oϩ(H2O)3 as the reagent ion) was used to detect
CH3O2, CH3OOH, and CH2O and a negative ion
chemical ionization scheme (with SF6Ϫ as the reagent
ion) was used to detect HO2 and NO2 with the quad-
rupole mass spectrometer. H3Oϩ was produced in the
ion source by passing a large O2 flow (5 STP L minϪ )
1
through the polonium-210 alpha-emitting ionization
source (with H2O impurities being sufficiently abun-
dant to produce sufficient quantities of reagent ions).
SF6Ϫ was produced in the ion source by passing a large
1
N2 flow (5 STP L minϪ ) and 1.0 STP mL minϪ1 of a
10 % SF6/N2 mixture through the ionization source.
The commercial ionization source consisted of a hol-

TEMPERATURE DEPENDENCE OF THE CH2O BRANCHING CHANNEL FOR THE CH3O2 ϩ HO2 REACTION
367
low cylindrical (length: 69 mm; diameter: 12.7 mm)
The temperature was controlled in the reaction region
to within 1 K.
aluminum body with 10 mCurie (3.7 ϫ 108 disinte-
1
grations sϪ ) of polonium-210 coated on the interior
walls.
Ions were detected with a quadrupole mass spec-
trometer housed in a two-stage differentially pumped
vacuum chamber. Flow tube gases (neutrals and ions)
were drawn into the front chamber through a 0.1-mm
aperture, which was held at a potential of Ϯ210 V.
The ions were focused by three lenses constructed
from 3.8-cm i.d., 4.8-cm o.d. aluminum gaskets. The
RESULTS AND DISCUSSION
Overview of Measurements
The use of relatively low concentrations of reactants
(to minimize peroxy–peroxy self-reaction), relatively
large flow velocities (necessary to ensure turbulent
flow conditions) and a relatively small overall rate
constant prevented the carrying out of a pseudo-first-
order kinetics study of the overall rate constant. In-
stead, reactants and products were monitored as a
function of time, and kinetics models were used to
compare and/or fit the desired rate constants. The rel-
evant reactions for kinetic modeling of the system are
given in Table I. In this study, absolute concentration
calibrations were obtained for CH3O2 and CH2O, thus
allowing direct use of the kinetics data for these spe-
cies. Because of the critical importance of the accuracy
of the CH2O calibration approach, a comparison of the
two CH2O calibration methods is presented in Figure
2, illustrating the consistency of the two techniques.
1
front chamber was pumped by a 6-inch, 2,400 L sϪ
diffusion pump. The gases entered the rear chamber
through a skimmer cone with a 1.0-mm orifice (held
at Ϯ130 V), which was placed approximately 5 cm
from the front aperture. The rear chamber was pumped
1
by a 250-L sϪ turbomolecular pump. Once the ions
passed through the skimmer cone, they were mass fil-
tered and detected with a quadrupole mass spectrom-
eter.
Chemical Ionization Schemes
The following chemical ionization schemes were used
to detect the species CH3O2, CH3OOH, CH2O, HO2,
and NO2:
CH3O2 ϩ H3Oϩ(H2O)3 !:
CH3OOHϩ(H2O)3 ϩ H2O (16)
CH3OOH ϩ H3Oϩ(H2O)3 !:
ϩ
CH3OOH2 (H2O)3 ϩ H2O (17)
CH2O ϩ H3Oϩ(H2O)3 !:
CH2OHϩ(H2O)3 ϩ H2O (18)
HO2 ϩ SF6Ϫ !: SF4O2Ϫ ϩ other products (19)
NO2 ϩ SF6Ϫ !: NO2Ϫ ϩ SF6
(20)
The rates of reactions (18) and (20) have been mea-
9
1
sured previously (k18 ϭ 1.4 ϫ 10Ϫ cm3 moleculeϪ
sϪ1 and k20 ϭ 1.4 ϫ 10Ϫ10 cm3 moleculeϪ1 sϪ ) [22,23],
reaction (16) has been used in our previous study of
the CH3O2 ϩ NO reaction [4], reaction (19) has been
used in a previous study of the HO2 ϩ BrO reaction
[21], and reaction (17) had been predicted to be ther-
modynamically feasible on the basis of electronic
structure calculations [24].
1
For the low-temperature studies, liquid-nitrogen-
cooled silicone oil was used as the coolant for the jack-
eted flow tube, and the nitrogen carrier gas was pre-
cooled by passing it through a copper coil immersed
in a liquid N2 reservoir followed by resistive heating.
Figure 2 Comparison of CH2O calibration methods (᭿:
standard sample method; ᭹: CH3O2 ϩ NO ϩ O2 titration
method).

368
ELROD ET AL.
Table I Kinetics Parameters for Branching Ratio Determination
A
Ϫ
Ϫ1
Reactiona
(cm3 s 1 molecule
)
Ea/R (K)
Ϫ
13
13
CH3O2 ϩ HO2 !: products
3.8 ϫ 10
Ϫ800
Ϫ190
Ϫ190
Ϫ190
Ϫ600
Ϫ1000
900
Ϫ
CH3O2 ϩ CH3O2 !: CH3OOCH3 ϩ O2
CH3O2 ϩ CH3O2 !: CH3O ϩ CH3O ϩ O2
CH3O2 ϩ CH3O2 !: CH2O ϩ CH3OH ϩ O2
HO2 ϩ HO2 !: H2O2 ϩ O2
HO2 ϩ HO2 ϩ M !: H2O2 ϩ O2 ϩ M
CH3O ϩ O2 !: CH2O ϩ HO2
CH3O ϩ CH3O2 !: CH2O ϩ CH3OOHc
CH3O ϩ CH3O !: CH2O ϩ CH3OHd
HO2 ϩ CH2O !: adduct
2.5 ϫ 10 ϫ 0.1
Ϫ
13
2.5 ϫ 10 ϫ f(T)b
Ϫ13
2.5 ϫ 10 ϫ (0.9 Ϫ f(T))
Ϫ
Ϫ
13
2.3 ϫ 10
33
1.7 ϫ 10 ϫ [M]
Ϫ14
Ϫ12
Ϫ11
Ϫ15
3.9 ϫ 10
2.6 ϫ 10
1.3 ϫ 10
6.7 ϫ 10
0
0
Ϫ600
a Rate constants calculated from k(T) ϭ AeϪE /RT. All rate constant parameters from the JPL compilation (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, JPL Publication 97-4; Jet Propulsion Laboratory: Pasadena, California, 1997), unless
a
Ϫ
1
otherwise indicated.
1131
Ϫ
ͫ ͬ
294
ͩ
ͪ
T
b Temperature-dependent branching ratio calculated from f(T) ϭ 1 ϩ e
c Rate constant parameters from Heicklen, J.; Adv Photochem 1988, 14, 177.
d Rate constant parameters from Biggs, P.; Canosa-Mas, C. E.; Fracheboud, J.-M.; Shallcross, D. E.; Wayne, R. P. J Chem Soc Faraday
Trans. 1997, 93, 2481.
An approximate calibration for CH3OOH (constructed
by using the same mass spectrometer sensitivity factor
as was determined for CH3O2) was used in order to
allow an estimate of the rate of formation of this prod-
uct. Typical concentration vs. time profiles for the spe-
cies CH3O2 and CH3OOH are presented in Figures 3
and 4, respectively. In order to facilitate discussion of
background contributions to the observed CH2O sig-
nal, the raw data (signal vs. injector distance) collected
for the CH2O profiles are presented in Figure 5, and
the time-dependent production of CH2O (with several
different assumed values for k7b overlaid) is presented
in Figure 6. HO2 profiles were not routinely collected
because HO2 concentrations were generally not sig-
3
nificantly varied (ϳ1.2 ϫ 1012 molecule cmϪ ), and
the decay profiles of HO2 were significantly influenced
by the HO2 ϩ HO2 self-reaction.
Overall Rate-Constant Evaluation
Figure 3 contains a kinetics plot ([CH3O2] vs. time)
for an experiment performed at 100 Torr and 298 K
(squares), as well as the predicted CH3O2 decay (solid
line) from a model calculation using the calibrated ini-
tial concentrations of CH3O2 and HO2 and the reac-
tions listed in Table I. As the graph indicates, there is
good agreement between our experimental data and
modeling results that employ the JPL recommendation
Figure 3 [CH3O2] vs. time for the CH3O2 ϩ HO2 reaction
Ϫ
3
at 100 Torr, 298 K, [CH3O2]0 ϭ 3.8 ϫ 1011 molecule cm
[14] for the overall rate constant for CH3O2 ϩ HO2
(which is based on the earlier FP studies). We also
find that our temperature-dependence results for
Ϫ
3
and [HO2]0 ϭ 1.0 ϫ 1012 molecule cm (᭿: actual data;
: model).

TEMPERATURE DEPENDENCE OF THE CH2O BRANCHING CHANNEL FOR THE CH3O2 ϩ HO2 REACTION
369
vious studies, in that our CIMS detection and calibra-
tion methods for CH3O2 and HO2 are completely in-
dependent of one another (in contrast to the difficult
UV detection problems presented by the overlapping
peroxy radical spectra in those same flash-photolysis
studies). Therefore, our TF-CIMS measurements of
CH3O2 decay can be considered as complementary to
the earlier flash-photolysis experiments in that, despite
the fact that our fast-flow approach suffers by com-
parison in terms of extent of reaction observed, our
chemical ionization detection technique is not hin-
dered by the same detection complexities that plague
UV detection of peroxy species.
Branching Channel Rate-Constant
Determination
Figure 4 contains a kinetics plot ([CH3OOH]approx vs.
time) for an experiment performed at 100 Torr and
298 K (squares), as well as the predicted CH3OOH
rise (solid line) from a model calculation using the
Figure 4 [CH3OOH]approx vs. time for the CH3O2 ϩ HO2
reaction at 100 Torr, 298 K, [CH3O2]0 ϭ 3.8 ϫ 1011 mole-
Ϫ
3
Ϫ3
cule cm and [HO2]0 ϭ 1.0 ϫ 1012 molecule cm (᭿:
actual data;
: model).
CH3O2 decay are well predicted by the same model.
However, as we pointed out in the previous paragraph,
experimental conditions constrain our measurements
to conditions of relatively slow rates of reactant decay
(the decay presented in Figure 3 represents only about
a 12% loss of CH3O2 over the reaction time accessed).
Therefore, from this kinetics analysis standpoint, this
first use of a fast-flow method for the CH3O2 ϩ HO2
reaction is at a disadvantage compared to earlier flash-
photolysis studies in which larger peroxy radical de-
cays were observed. We estimate that our measured
decays are consistent with values between a factor of
0.7 (lower bound) and 1.3 (upper bound) of the JPL
recommended value (which currently indicates an es-
timated upper-bound uncertainty factor of 2.0 at 298
K for this reaction [14]). However, if the uncertainties
in the absolute concentrations of HO2 and CH3O2 (as
well as other potential systematicuncertainties) are
considered, we estimate an overall upper-bound un-
certainty factor for the overall rate constant of 1.7,
which is a slight improvement over the current uncer-
tainty value. However, it is important to point out that
our detection methodology is more reliable than pre-
Figure 5 Observed production of CH2O (᭹) from the
CH3O2 ϩ HO2 reaction at 100 Torr, 298 K, [CH3O2]0 ϭ
Ϫ
6.3 ϫ 1011 molecule cm 3 and [HO2]0 ϭ 1.0 ϫ 1012 molecule
Ϫ
3
cm as compared to the background signal (᭿) obtained
with the HO2 source turned off. Linear least squares fits to
both data sets are plotted as solid lines to guide the eye, and
the 95% confidence interval for the background data is plot-
ted as dashed lines.

370
ELROD ET AL.
CH2O is directly produced via reaction (21b) and in-
directly produced via reaction (21a) in the presence of
oxygen by coupling to reaction (11).
CH3O ϩ O2 !: CH2O ϩ HO2
(11)
At 298 K, the JPL recommendation for reaction (21)
indicates the following branching ratios for channels
a, b, and c, respectively: 0.3, 0.6, and 0.1 [14]. There-
fore, 60% of the total CH3O2 ϩ CH3O2 reactive events
lead to CH2O; in the presence of sufficient O2, the
fraction increases to 90% (also note that channel 21a
produces two CH2O molecules per reactive event,
while channel 21b produces only one). For our exper-
3
imental conditions ([O2] ϳ 3.0 ϫ 1016 molecule cmϪ ),
the CH2O-producing fraction is predicted to be near
the limiting 90% value. Because main flow concentra-
3
tions of CH3O2 are low (Ͻ1.0 ϫ 1012 molecule cmϪ )
compared to the overall rate constant, it might be ex-
pected that reaction (21) would lead to small CH2O
background levels. However, CH3O2 concentrations
are about 15 times higher in the sidearm of the flow
tube where it is produced, thus increasing the rate of
the self-reactions given above before they enter the
main flow tube. In fact, the sidearm reaction time was
optimized (by varying the position in the sidearm
where the F atoms were injected) to achieve the most
favorable conditions (i.e., the largest CH3O2/CH2O ra-
tio possible). Using the relevant concentrations and
times for CH3O2 self-reaction in the sidearm and a
kinetic model incorporating the reactions given above,
we estimate that for the largest main flow-tube con-
centrations of [CH3O2]0 used (1.0 ϫ 1012 molecule
Figure 6 Sensitivity of fitted value of k7b to the data given
in Figure 5.
calibrated initial concentrations of CH3O2 and HO2
and the reactions listed in Table I. Although the agree-
ment between the experimental data and the model
curve appears excellent, the lack of a formal calibra-
tion for CH3OOH precludes the quantitative determi-
nation of the rate constant for this branching channel
(k7a). However, Figure 4 does provide confirming
evidence that the CH3OOH-producing branching
channel is in fact the major product pathway for the
CH3O2 ϩ HO2 reaction. We chose not to pursue the
quantitative determination of k7a because the partition-
ing between two competing channels is best deter-
mined by measurements of the minor channel (k7b, in
the case of CH3O2 ϩ HO2), which is described in the
following paragraph.
3
cmϪ ), a “background” main flow-tube concentration
3
of [CH2O] ϳ 1.5 ϫ 1011 molecule cmϪ should be
observed. This background source of CH2O is a con-
stant with respect to the kinetics data collected (the
CH3O2 source is spatially fixed, and thus all reactions
involving species originating from it have the same
total reaction time regardless of the HO2 injector po-
sition). Since none of the CH2O-producing reactions
listed in Table I depends on the presence of HO2, we
expect that any CH2O produced from side reactions
should not depend on the HO2 injector position. How-
ever, it is possible that the additional O2 added to the
flow system through the HO2 injector could lead to
additional production of CH2O via reactions (21a) and
(11) that would appear to depend on the position of
the HO2 injector. In Figure 5, the raw CH2O signal vs.
injector distance data is presented for two experi-
ments: The data depicted as circle symbols was col-
lected with the HO2 source on, and the data depicted
as square symbols was collected with the HO2 source
off (all gases flowing, but the microwave discharge
The major complication in the rate-constant deter-
mination for the CH2O producing channel (k7b) is the
CH2O-producing side reaction resulting from the self-
reaction of CH3O2:
CH3O2 ϩ CH3O2 !: CH3O ϩ CH3O ϩ O2
(21a)
!: CH2O ϩ CH3OH ϩ O2
(21b)
!: CH3OOCH3 ϩ O2
(21c)
1
(k21 ϭ 4.7 ϫ 10Ϫ13 cm3 moleculeϪ1 sϪ ) [14]. Note that

TEMPERATURE DEPENDENCE OF THE CH2O BRANCHING CHANNEL FOR THE CH3O2 ϩ HO2 REACTION
371
was turned off). It is apparent from these experiments
that no additional CH2O is formed by the introduction
of gases used to synthesize HO2 (helium, hydrogen,
and oxygen). It is also apparent that the CH2O pro-
duced with the HO2 source operating is statistically
significant, as the CH2O signal is observed rising
above the 95% confidence interval for the CH2O time-
dependent profile obtained with the HO2 source turned
off. Using the CH2O calibration factor and the total
CH2O signal observed for the experiments depicted in
Figure 5, a background CH2O level of about 1.7 ϫ
1011 molecule cmϪ3 is calculated. As discussed above,
this background level is consistent with kinetic model
results for the production of CH2O via CH3O2 self-
reaction chemistry. In summary, despite the fact that
a relatively large CH2O background signal is observed
(which is consistent with kinetic modeling of the
CH3O2 self-reaction chemistry occurring in the side-
arm reactor), we are able to observe CH2O production
as a function of HO2 contact time that can be positively
attributed to the CH3O2 ϩ HO2 reaction.
experimental initial concentrations of CH3O2, HO2,
and O2 to calculate the rate of CH2O production for
comparison to the experimentally determined rate
of CH2O production. The value of k7a in the fitting
process was constrained by the relationship k7a
ϭ
k7 Ϫ k7b, where k7 was calculated from the JPL-rec-
ommended Arrhenius parameters [14]. In Figure 6,
along with fitted value for k7b (solid line), k7b values
that are a factor of 2 higher (dashed line) and a factor
of 2 lower (dotted line) are used to predict CH2O pro-
files to illustrate the sensitivity of the fitted value of
k7b to the actual CH2O profile observed. It is apparent
from this plot and the statistical uncertainties reported
in Table II that the fitted k7b values are determined to
a precision of about 50%.
The experiments were carried out at several differ-
ent [CH3O2]0 and [HO2]0 concentrations, which pro-
vide a test of consistency for the kinetic model and the
method of data analysis. Table II lists the experimental
conditions and fitted k7b values (and associated branch-
ing ratios) for all experiments performed. In particular,
several different [CH3O2]0 values (at nearly fixed HO2
concentrations) were used to test the consistency of
the k7b determination method. Figure 7 shows that very
similar k7b values were obtained for these experiments
with differing initial CH3O2 concentrations. Although
this test does not rule out other systematicerrors in
the fitting of k7b (such as HO2 calibration errors or
Figure 6 shows the CH2O production observed
(squares) for the reaction conditions given in Figure
5, with a solid line fit through the data to determine
the rate of CH2O production, which is used in the fit-
ting process to determine k7b. The value of k7b was
fitted by using a kineticmodel that included the re-
actions and rate parameters given in Table I and the
Table II Branching Ratio Data for the CH3O2 ϩ HO2 : CH2O ϩ H2O ϩ O2 Reaction at 100 Torr Pressure
[CH3O2]0
(1010 molecule cm
[HO2]0
(1010 molecule cm
k7b (and 1␴ error)
Branching Ratio
k7b/(k7a ϩ k7b)
Ϫ
3
Ϫ
3
Ϫ
Ϫ
Ϫ1
T (K)
)
)
(10 14 cm3 molecule 1 s
)
298
298
298
298
298
273
273
263
263
253
253
253
253
243
243
243
237
230
230
218
218
39
38
92
63
42
80
88
80
80
77
77
77
77
82
82
82
60
48
50
52
52
125
100
100
100
115
100
100
76
56 Ϯ 22
56 Ϯ 39
45 Ϯ 39
73 Ϯ 28
67 Ϯ 34
0.10
0.10
0.08
0.13
0.12
0.08
0.10
0.13
0.13
0.20
0.16
0.16
0.15
0.15
0.15
0.12
0.21
0.35
0.29
0.33
0.30
57 Ϯ 50
71 Ϯ 71
104 Ϯ 40
104 Ϯ 32
180 Ϯ 54
144 Ϯ 63
144 Ϯ 63
135 Ϯ 36
150 Ϯ 40
150 Ϯ 30
120 Ϯ 50
231 Ϯ 55
431 Ϯ 130
357 Ϯ 86
492 Ϯ 130
447 Ϯ 270
76
139
139
139
139
140
140
140
139
79
79
79
79

372
ELROD ET AL.
for the CH2O calibration method. In addition, uncer-
tainties in the rate constants used in the fitting process
can also introduce systematic error to the determina-
tion of k7b (in particular, the uncertainty in k7, which
propagates nearly linearly into the value of k7b). There
is also some uncertainty in the determination of ab-
solute reaction time from the flow velocity and dis-
tance from the injector tip to the mass spectrometer
sampling aperture. However, this uncertainty is prob-
ably fairly low since the flow velocity depends directly
on the accuracy of the mass flow meters (calibrated to
at least 5% accuracy) and the distance is uncertain to
less than 1 cm (the distance between the ionization
region and the mass spectrometer aperture), or about
2% of the longest reaction time. Nonetheless, it is clear
that relatively large systematicerrors (as muhc as
50%, 1␴) are potentially present in our fitted values
for k7b.
Based on our five fitted k7b values at 298 K and 100
Torr, we calculate a value (and 1␴ statistical uncer-
tainty) for the branching ratio [k7b/(k7a ϩ k7b)] of
0.11 Ϯ 0.02. Table III lists the product monitored,
experimental conditions, and inferred values for the
branching ratio from the previous indirect (CH2O is
not directly detected) studies as well as the present
direct (CH2O is directly detected) one. The kinetics
study of Moortgat et al. [9] involved the simultaneous
monitoring of peroxy radicals and CH3OOH at 298 K
and 700 torr, which allowed the determination of an
Figure 7 Dependence of the fitted CH3O2 ϩ HO2
CH2O ϩ H2O ϩ O2 branching channel rate constant on ini-
tial CH3O2 concentration.
:
unknown sources of CH2O originating from species
produced in the HO2 injector), it does suggest that the
CH3O2 calibration is not subject to large systematic
error, and thus the fitted values of k7b are fairly reliable
from the point of view of the modeled CH3O2 chem-
istry. The error given in Table II represents the statis-
tical uncertainty determined in the fitting process and
does not include contributions from systematic error.
We have previously estimated that the largest source
of systematicerror in these types of experiments re-
sults from the titration/calibration methods used for
HO2 and CH3O2 [4]. We estimate that this error is on
the order of Ϯ15% (1␴) for these species as well as
1
overall rate constant (4.8 ϫ 10Ϫ12 cm3 moleculeϪ1 sϪ ;
compared to the current JPL recommended value of
5.6 ϫ 10Ϫ12 cm3 moleculeϪ1 sϪ1 used in this work) as
well as a specific CH3OOH-producing rate constant
12
1
1
(3.5 ϫ 10Ϫ cm3 moleculeϪ sϪ ). From these rate
constants, a CH2O-producing branching ratio of 0.27
may be inferred. In the work of Jenkin et al. [8],
CD3O2 was used as a reactant and the production of
HDO was followed at 298 K and 11 torr. The branch-
ing ratio determined by Jenkin et al. was determined
in a manner similar to the present work; the rate con-
stant for HDO (formed along with CD2O and O2) pro-
Table III Comparison of Branching Ratio Determinations for the CH3O2 ϩ HO2 Reaction
Product
Study
Monitored
Temperature (K)
Pressure (Torr)
k7b/(k7a ϩ k7b)
Moortgat et al. [9]
Jenkin et al. [8]
Wallington [16]
This work
CH3OOH
HDOa
CH3OOH
CH2O
298
298
298
298
700
11
15–700
100
0.27
0.40
0.08 Ϯ 0.05
0.11 Ϯ 0.02
Ϫ1160
Ϫ
1
ϩ
6.21
ͩ
ͪ
ͫ
T
ͬ
1 ϩ e
This work
CH2O
218–298
100
a From measurements of the reaction CD3O2 ϩ HO2 : CD2O ϩ HDO ϩ O2.

TEMPERATURE DEPENDENCE OF THE CH2O BRANCHING CHANNEL FOR THE CH3O2 ϩ HO2 REACTION
373
atures, we performed several measurements of k7b at
temperatures between 218 and 298 K in order to de-
termine the temperature dependence of the CH2O
branching channel. From the data listed in Table II and
plotted in Figure 8, we obtained the following Arrhe-
nius expression (with 1␴ statistical uncertainties):
1.0
3
k7b(T) ϭ 1.6ϩϪ ϫ 10Ϫ15 ϫ exp[(1730 Ϯ 130)/T] c m
0.7
1
moleculeϪ1 sϪ . It is clear from the data and the fitted
Arrhenius expression that both k7b and the branching
ratio (k7b increases even more quickly than k7 as the
temperature decreases) show a steep negative temper-
ature dependence: k7 increases by a factor of ϳ7, and
the branching ratio increases by a factor of ϳ3 as the
temperature is decreased from 298 to 218 K. The neg-
ative temperature dependence of k7 is consistent with
the formation of a reactive intermediate. Figure 9 de-
picts a potential mechanism for reaction (7b) that is
based on a six-membered ring intermediate structure
that rearranges to produce the observed products. The
more dramaticnegative temperature dependence of k7b
suggests that the CH2O-producing product channel is
even more dependent on efficient thermal stabilization
of the intermediate than is the CH3OOH-producing
channel. Given the very specific intermediate geome-
try required for the CH2O-forming mechanism shown
in Figure 9, the steep negative temperaturedependence
of k7b seems consistent with the proposed mechanism.
Since the CH3O2 ϩ CH3O2 reaction has similar
product channels to the CH3O2 ϩ HO2 reaction, it is
of interest to compare the temperature dependences of
the branching ratios for these reactions. There have
been several previous studies of the temperature de-
pendence of the branching channels for the CH3O2 ϩ
CH3O2 reaction (for a review, see ref. [25]). Although
other workers have reported the existence of a
CH3OOCH3-producing channel (reaction (21c)), the
most recent work of Tyndall et al. indicates no evi-
dence for that channel (an upper limit of 6% is esti-
mated) [26]. Accordingly, one can consider the tem-
perature dependence of just the two major product
channels for the CH3O2 ϩ CH3O2 reaction:
Figure 8 Arrhenius plot of the temperature dependence of
the rate constant at 100 Torr pressure of the CH3O2
HO2 : CH2O ϩ H2O ϩ O2 branching channel (᭿: actual
data; : fit to data).
ϩ
duction was directly determined, and the branching
ratio was determined by dividing this value by their
12
value for the overall rate constant (5.4 ϫ 10Ϫ cm3
1
moleculeϪ1 sϪ ). As Jenkin et al. point out, their CD2O-
producing branching ratio result of 0.40 for CD3O2
must be interpreted with care, as kinetic isotope effects
are likely to make the CH2O-producing results for
CH3O2 somewhat different. Wallington [16] deter-
mined the fraction of CH3OOH produced from the
CH3O2 ϩ HO2 reaction at 298 K and several pressures
by performing a carbon mass balance analysis. From
the “missing carbon,” a CH2O-producing branching
ratio of 0.08 may be inferred. Our direct result is in
excellent agreement with the inferred result of Wal-
lington [16], while it is substantially lower than the
values inferred from the work of Moortgat et al. [9]
and Jenkin et al. [8].
CH3O2 ϩ CH3O2 !:
CH3O ϩ CH3O ϩ O2 (21a)
!: CH2O ϩ CH3OH ϩ O2 (21b)
Temperature Dependence of the CH2O
Branching Channel Rate-Constant
Determination
In order to establish a set of branching ratio values for
the whole range of tropospherically relevant temper-
Figure 9 Proposed mechanism for the CH3O2 ϩ HO2 :
CH2O ϩ H2O ϩ O2 branching channel.

374
ELROD ET AL.
In these experiments, a branching ratio, defined as
␣ ϭ [k21a/(k21a ϩ k21b)], was measured as a function of
temperature. In order to allow an Arrhenius analysis,
the branching ratio data were converted to the ratio
defined as ␤ ϭ k21a/k21b. In order to generalize this ratio
to the CH3O2 ϩ HO2 analysis as well, this term can
be thought of as the ratio of the “direct” O2 elimination
channel to the “indirect” (hydrogen transfer) O2
elimination channel. For both CH3O2 ϩ CH3O2 and
CH3O2 ϩ HO2, the “indirect” channel produces CH2O.
Based on all available data, Lightfoot et al. obtained
the following Arrhenius-type expression for CH3O2 ϩ
CH3O2: ln ␤ ϭ 3.22 Ϫ1165/T [25]. In order to directly
compare the temperature dependence of the branching
ratio obtained for the CH3O2 ϩ HO2 reaction with that
for CH3O2 ϩ CH3O2, we also recast our data in terms
of ␤ (ϭ k7a/k7b for CH3O2 ϩ HO2) and obtained the
following Arrhenius-type expression: ln ␤ ϭ 6.21 Ϫ
1160/T. These expressions reveal remarkable agree-
ment in the temperature dependence of the CH2O pro-
ducing channels for the CH3O2 ϩ CH3O2 and CH3O2
ϩ HO2 reactions. The C2H5O2 ϩ C2H5O2 reaction pro-
ceeds through analogous product channels to produce
CH3CHO, albeit at smaller branching ratio at 298 K
(0.3 for the direct CH3CHO-producing channel) as
well as a less steep negative temperature dependence
(ln ␤ ϭ 2.32 Ϫ 533/T) than for CH3O2 ϩ CH3O2 [25].
Current experimental evidence obtained at 298 K
seems to indicate that the C2H5O2 ϩ HO2 reaction does
not have a significant CH3CHO-producing channel
[27]. However, our finding that the CH2O-producing
channels for CH3O2 ϩ CH3O2 and CH3O2 ϩ HO2 have
similarly strong negative temperature dependences
may indicate that the C2H5O2 ϩ HO2 reaction has the
potential to have a significant CH3CHO-producing
channel at low temperatures. Finally, from our defi-
nition of ␤ ϭ k7a/k7b, we can derive a convenient term
for the temperature dependence of the CH2O branch-
ing ratio for the CH3O2 ϩ HO2 reaction:
circumvent CH2O production via either CH3O2 ϩ HO2
pathway [reactions (4a) and (11) instead determine the
rates of CH2O production], we expect that reaction
(7b) is of minor significance in highly polluted envi-
ronments. However, we propose two potential impacts
on the prediction accuracy of atmospheric models in
environments (relatively low NOx) where significant
direct CH2O production via reaction (7b) can occur.
The atmosphericfate of CH OOH (the main channel
3
product of the CH3O2 ϩ HO2 reaction) is also the
eventual formation of CH2O:
CH3OOH ϩ h␯ !: CH3O ϩ OH
CH3O ϩ O2 !: CH2O ϩ HO2
(23)
(11)
An obvious potential effect is that if the reaction (7b)
pathway is faster than the reaction (7a/23/11) pathway
currently used in atmospheric models, one would ex-
pect that observed CH2O levels might be higher than
predicted. From the atmospheric lifetimes for the rel-
evant species (derived using background NOx levels)
presented in Lawrence et al., it does appear that the
reaction (7b) pathway (ϳhours) should be signifi-
cantly faster than the reaction (7a/23/11) pathway
(ϳdays), which is limited by the relatively long life-
time of CH3OOH against photolysis [28]. In fact, Ay-
ers et al. had earlier found that observed CH2O levels
at Cape Grim, Tasmania (a location characterized by
very low NOx levels), were underpredicted by atmo-
sphericmodels, which assumed a k7b value of zero
[17]. However, Ayers et al. found that a branching
ratio of 0.40 was required to reconcile the observed
CH2O levels with the model results, a value signifi-
cantly larger than the one reported here. More recently,
Weller et al. reported the results of CH2O measure-
ments in the marine boundary layer of the Atlanticand
came to a similar conclusion: Observed CH2O levels
were greater than the predictions from a photochemi-
cal model using a k7b value of zero [18]. Following the
earlier work of Ayers et al., Weller et al. added reac-
tion (7b) to their model (again with a branching ratio
of 0.40) and found that the predicted CH2O concen-
trations were in closer (although not full) agreement
with their atmosphericobservations. Therefore, al-
though our present study presents evidence for a sig-
nificant CH2O-forming product channel from the
CH3O2 ϩ HO2 reaction, it appears that the k7b value
determined here will not fully reconcile model predic-
tions and atmosphericobservations of CH 2O.
1160/Tϩ6.21) Ϫ1
k7b/(k7a ϩ k7b) ϭ 1/(1 ϩ ␤) ϭ [1 ϩ e(Ϫ
]
(22)
Atmospheric Significance of CH2O
Production from CH3O2 ؉ HO2 Reaction
As we have established that the CH3O2 ϩ HO2 reac-
tion possesses a significant CH2O-producing branch-
ing channel (which increases in significance as the
temperature is lowered), it is of interest to consider
how this result may impact atmospheric models, which
generally assume a branching ratio of zero for the
CH2O-producing channel. Because high NOx levels
A less obvious effect of nonnegligible CH2O pro-
duction via reaction (7b) might involve a modification
of the efficiency of HOx cycling. CH2O itself has only

TEMPERATURE DEPENDENCE OF THE CH2O BRANCHING CHANNEL FOR THE CH3O2 ϩ HO2 REACTION
375
the Camille and Henry Dreyfus Foundation, the American
Chemical Society–Petroleum Research Fund, and Research
Corporation.
a short atmosphericlifetime against photolysis
(ϳhours) [28]:
CH2O ϩ h␯ !: H2 ϩ CO
!: H ϩ HCO
(24a)
(24b)
BIBLIOGRAPHY
with the two branching channels being roughly equal
for overhead solar conditions [29]. Both products of
reaction (24b) lead to the formation of HO2:
1. Finlayson-Pitts, B. J.; Pitts, J. N. Chemistry of the Upper
and Lower Atmosphere; Academic Press: San Diego,
1999.
2. Bertman, S. B.; Roberts, J. M.; Parrish, D. D.; Buhr,
M. P.; Goldan, P. D.; Kuster, W. C.; Fehsenfeld, F. C.;
Montzka, S. A.; Westberg, H. J Geophys Res 1995, 100,
22805.
H ϩ O2 ϩ M !: HO2 ϩ M
HCO ϩ O2 !: HO2 ϩ CO
(14)
(25)
3. Flocke, F.; Atlas, E.; Madronich, S.; Schauffler, S. M.;
Aikin, K.; Margitan, J. J.; Bui, T. P. Geophys Res Lett
1998, 25, 1891.
4. Scholtens, K. W.; Messer, B. M.; Cappa, C. D.; Elrod,
M. J. J Phys Chem A 1999, 103, 4378.
5. Wennberg, P. O.; Hanisco, T. F.; Jaegle´, L.; Jacob,
D. J.; Hintsa, E. J.; Lanzendorf, E. J.; Anderson, J. G.;
Gao, R.; Keim, E. R.; Donnelly, S. G.; Negro, L. A.;
Fahey, D. W.; McKeen, S. A.; Salawitch, R. J.; Webster,
C. R.; May, R. D.; Herman, R. L.; Proffitt, M. H.; Mar-
gitan, J. J.; Atlas, E. L.; Schauffler, S. M.; Flocke, F.;
McElroy, C. T.; Bui, T. P. Science 1998, 279, 49.
6. Cox, R. A.; Tyndall, G. S. Chem Phys Lett 1979, 65,
357.
Therefore, the existence of a significant CH2O-pro-
ducing channel for the CH3O2 ϩ HO2 reaction could
lead to faster HOx regeneration simply because CH2O
is produced faster via reaction (7b) than via reactions
(7a/23/11). This effect may be a contributing factor in
the search for an answer to the observation that ob-
served HOx levels in the upper troposphere are greater
than those predicted by current atmospheric models
(which assume a value of zero for k7b) [5].
CONCLUSIONS
7. Cox, R. A.; Tyndall, G. S. J Chem Soc, Faraday Trans
2 1980, 76, 153.
The results presented here represent the first direct
measurement of the CH2O-producing channel of the
CH3O2 ϩ HO2 reaction, as well as the first product-
specific temperature-dependence study of any kind for
this reaction. The kinetics of the reactant CH3O2 and
the main branching channel product CH3OOH were
also monitored and found to be consistent with JPL
recommendation for the overall rate constant [14].
This study indicates the CH3O2 ϩ HO2 reaction should
be a significant source of CH2O in the troposphere, as
the branching ratio [k7b/(k7a ϩ k7b)] at 100 Torr was
found to be 0.11 at 298 K and increased to 0.31 as the
temperature was lowered to 218 K. This work should
help improve the kinetics database for methane oxi-
dation chemistry by placing more stringent constraints
on CH2O and HOx formation and destruction rates. In
particular, these findings may help explain larger-than-
predicted (by models which assume a k7b value of
zero) CH2O observations in the clean troposphere, as
well as larger-than-predicted HOx levels in the upper
troposphere.
8. Jenkin, M. E.; Cox, R. A.; Hayman, G. D.; Whyte,
L. J. J Chem Soc, Faraday Trans 2 1988, 84, 913.
9. Moortgat, G. K.; Cox, R. A.; Schuster, G.; Burrows,
J. P.; Tyndall, G. S. J Chem Soc, Faraday Trans 2 1989,
85, 809.
10. McAdam, K.; Veyret, B.; Lesclaux, R. Chem Phys Lett
1987, 133, 39.
11. Kurylo, M. J.; Dagaut, P.; Wallington, T. J.; Neuman,
D. M. Chem Phys Lett 1987, 139, 513.
12. Dagaut, P.; Wallington, T. J.; Kurylo, M. J. J Phys
Chem 1988, 92, 3833.
13. Lightfoot, P. D.; Veyret, B.; Lesclaux, R. J Phys Chem
1990, 94, 708.
14. DeMore, W. B.; Sander, S. P.; Howard, C. J.; Ravis-
hankara, 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 Mod-
eling JPL Publication 97-4, Jet Propulsion Laboratory:
Pasadena, California, 1997.
15. Kan, C. S.; Calvert, J. G.; Shaw, J. H. J Phys Chem
1980, 84, 3411.
16. Wallington, T. J. J Chem Soc, Faraday Trans 1991, 87,
2379.
17. Ayers, G. P.; Gillett, R. W.; Granek, H.; de Serves, C.;
Cox, R. A. Geophys Res Lett 1997, 24, 401.
18. Weller, R.; Schrems, O.; Boddenberg, A.; Gab, S.; Gau-
trois, M. J Geophys Res 2000, 105, 14401.
This research was funded by grants from the National Sci-
ence Foundation (ATM-9874752), the Towsley Foundation,

376
ELROD ET AL.
19. Seeley, J. V.; Jayne, J. T.; Molina, M. J. Int J Chem
Kinet 1993, 25, 571.
20. Ranschaert, D. L.; Schneider, N. J.; Elrod, M. J. J Phys
Chem A 2000, 104, 4449.
21. Elrod, M. J.; Meads, R. F.; Lipson, J. B.; Seeley, J. V.;
Molina, M. J. J Phys Chem 1996, 100, 5808.
22. Midey, A. J.; Arnold, S. T.; Viggiano, A. A. J Phys
Chem A 2000, 104, 2706.
23. Huey, L. G.; Hanson, D. R.; Howard, C. J. J Phys Chem
1995, 99, 5001.
25. Lightfoot, P. D.; Cox, R. D.; Crowley, J. N.; Destriau,
G. D.; Hayman, G. D.; Jenkin, M. E.; Moortgat, G. K.;
Zabel, T. Atmos Env 1992, 26A, 1805.
26. Tyndall, G. S.; Wallington, T. J.; Ball, J. C. J Phys
Chem A 1998, 102, 2547.
27. Wallington, T. J.; Japar, S. M. Chem Phys Lett 1990,
166, 495.
28. Lawrence, M. G.; Crutzen, P. J.; Rasch, P. J.; Eaton,
B. E.; Mahowald, N. M. J Geophys Res Atmos 1999,
104, 26245.
24. Messer, B. M.; Stielstra, D. E.; Cappa, C. D.; Scholtens,
K. W.; Elrod, M. J. Int J Mass Spectrom 2000, 197,
219.
29. Seinfeld, J. H.; Pandis, S. N. AtmosphericChemistry
and Physics; John Wiley and Sons, Inc.: New York,
1998.