Kinetics and Mechanism of the Acetylperoxy + HO2 Reaction

Article abstract of DOI:10.1021/jp983150t

The reaction of HO₂ with CH₃C(O)O₂ is examined using flash photolysis and FTIR smog chamber techniques. Time-resolved UV spectroscopy is used to follow the transient peroxy species. It yields reasonable concentration versus time profiles for CH₃C(O)O₂ and HO₂, but indicates anomalously high levels of secondary CH₃O₂ radicals. Transient IR diode laser absorption confirms the HO₂ decay rates; however, the anticipated reaction model substantially underestimates the observed decay. The model is augmented by assuming that, in analogy with formaldehyde, there exists a reaction between HO₂ and acetaldehyde (the precursor for CH₃C(O)O₂). Consistent with this, the fitted rate for the hypothesized reaction increases with increasing initial acetaldehyde level. Relative rate measurements reveal that chlorine atoms remove more CH₃CHO relative to CH₂OH in air as compared to nitrogen diluent. This supports the hypothesis since, in the presence of oxygen, HO₂ is formed and presents an additional acetaldehyde removal pathway. Employing the augmented model, analyses of HO₂ decay traces yield a CH₃C(O)O₂ + HO₂ rate constant of k₁ = (3.9_-2.3^+5.0) × 10^-13e^{(1350±250)/T} cm³ s^-1. Reasons are discussed for why the present rate constants are 2 - 3 times larger than previously reported. FTIR - smog chamber studies reveal the reaction to proceed via two channels to (a) peracetic acid and O₂ and to (b) acetic acid and O₂, with a branching fraction at 295 K that is less than half of the literature value. Time-resolved UV absorption measurements support this smaller fraction; averaged together the two methods give k_1b/k₁ = 0.12 ± 0.04. As part of this work, relative rate techniques are used to measure k(Cl+CH₃C- (O)OH) = (2.5 ± 0.3) × 10^-14 cm³ s^-1and k(Cl+CH₃C(O)OOH) = (4.5 ± 1.0) × 10^-15 cm³ s^-1 at 295 K.

Full text of DOI:10.1021/jp983150t

J. Phys. Chem. A 1999, 103, 365-378
365
Kinetics and Mechanism of the Acetylperoxy + HO2 Reaction
Mary A. Crawford,† Timothy J. Wallington, Joseph J. Szente, and M. Matti Maricq*
Research Laboratory, Ford Motor Company, P.O. Box 2053, Drop 3083, Dearborn, Michigan 48121
Joseph S. Francisco
Department of Chemistry, Purdue UniVersity, West Lafayette, Indiana 47907-1397
ReceiVed: July 23, 1998; In Final Form: NoVember 5, 1998
The reaction of HO2 with CH3C(O)O2 is examined using flash photolysis and FTIR smog chamber techniques.
Time-resolved UV spectroscopy is used to follow the transient peroxy species. It yields reasonable concentration
versus time profiles for CH3C(O)O2 and HO2, but indicates anomalously high levels of secondary CH3O2
radicals. Transient IR diode laser absorption confirms the HO2 decay rates; however, the anticipated reaction
model substantially underestimates the observed decay. The model is augmented by assuming that, in analogy
with formaldehyde, there exists a reaction between HO2 and acetaldehyde (the precursor for CH3C(O)O2).
Consistent with this, the fitted rate for the hypothesized reaction increases with increasing initial acetaldehyde
level. Relative rate measurements reveal that chlorine atoms remove more CH3CHO relative to CH3OH in air
as compared to nitrogen diluent. This supports the hypothesis since, in the presence of oxygen, HO2 is formed
and presents an additional acetaldehyde removal pathway. Employing the augmented model, analyses of
HO2 decay traces yield a CH3C(O)O2 + HO2 rate constant of k1 ) (3.9+-25..30) × 10-13e(1350(250)/T cm3 s-1.
Reasons are discussed for why the present rate constants are 2-3 times larger than previously reported.
FTIR-smog chamber studies reveal the reaction to proceed via two channels to (a) peracetic acid and O2 and
to (b) acetic acid and O3 , with a branching fraction at 295 K that is less than half of the literature value.
Time-resolved UV absorption measurements support this smaller fraction; averaged together the two methods
give k1b/k1 ) 0.12 ( 0.04. As part of this work, relative rate techniques are used to measure k(Cl+CH3C-
(O)OH) ) (2.5 ( 0.3) × 10-14 cm3 s-1 and k(Cl+CH3C(O)OOH) ) (4.5 ( 1.0) × 10-15 cm3 s-1 at 295 K.
I. Introduction
surprising since at night [NO2]/[NO] is generally high and [NO]
is relatively low; thus, one would surmise that both the net loss
Acetylperoxy radicals appear in the atmosphere as intermedi-
ates in the photooxidation of a variety of anthropogenic and
biogenic compounds. It has been well established that this
radical plays an important role in determining ambient peroxy-
acetylnitrate (PAN) concentrations.1-3 PAN, a ubiquitous
compound found in photochemical smog, is formed by the
reaction of CH3C(O)O2 with NO2 . The stability of this adduct
against thermal decomposition exhibits a strong temperature
dependence. At low temperatures, PAN effectively sequesters
both NO2 and a peroxy radical and, thereby, manifests a
significant influence on ozone formation. Furthermore, owing
to its relative stability, PAN provides a vehicle for long-range
transport of reactive nitrogen from urban areas where NOx
concentrations tend to be high to rural areas having relatively
low NOx levels.
As a result of its importance as a member of the odd nitrogen
group, there have been a number of measurements of urban
and rural PAN concentrations. Recently, Shepson et al.4 and
Hartsell et al.5 have explored the relationships between PAN,
ozone, and NOy in sites in Ontario, Canada and as part of the
Southern Oxidants Study in the Southeast United States,
respectively. Both studies observe diurnal variations in the [O3]/
[PAN] ratio that indicate nighttime PAN concentrations to
decrease faster than those of ozone. They find this somewhat
of PAN via thermal decomposition and the loss of ozone via
titration with NO would be slow. Both groups suggest that the
nighttime increase in the [O3]/[PAN] ratio arises from a higher
deposition velocity for PAN as compared to ozone.
Atmospheric simulations by Stockwell et al.6 suggest a
different explanation. Using a box model, they found that PAN
concentrations are sensitive to a number of peroxy radical
reactions, notably the acetylperoxy radical/HO2 reaction, along
with the CH3C(O)O2 reactions with itself, NO, and CH3O2 .
Stockwell et al.6 further indicate that PAN decomposition
through the CH3C(O)O2 intermediate can be important even if
the [NO2] to [NO] ratio is high and conclude that this is a viable
alternative to deposition for explaining the observed diurnal
variation in the [O3]/[PAN] ratio. However, despite their
potential significance for understanding tropospheric PAN
chemistry, the reactions of the acetylperoxy radical have
received relatively little attention, particularly those reactions
with other peroxy radicals.
Recent measurements of PAN and NO2 over the southern
United States suggest that the concentration of CH3C(O)O2
radicals in the free troposphere is typically (1-10) × 104 cm-3 7
.
The most important of the atmospheric peroxy radical/peroxy
radical reactions of CH3C(O)O2 is its reaction with HO2 .6 Niki
et al.8 were the first to examine this reaction and their product
study showed it to proceed via two channels:
* Author for correspondence.
† On leave from Department of Chemistry, Purdue University. Present
address: Knox College, Galesburg, IL 61401.
10.1021/jp983150t CCC: $18.00 © 1999 American Chemical Society
Published on Web 01/05/1999

366 J. Phys. Chem. A, Vol. 103, No. 3, 1999
CH3C(O)O2 + HO2 f CH3COO2H + O2
CH3C(O)O2 + HO2 f CH3COOH + O3
Crawford et al.
(1a)
(1b)
with branching fractions of 0.75 and 0.25 for channels 1a and
1b, respectively. In a subsequent kinetic study, Moortgat et al.9
reported the somewhat higher yield of 0.33 for channel 1b. Horie
et al.10 also investigated the CH3C(O)O2 + HO2 reaction by
photolyzing biacetyl in the presence of O2 and analyzing the
reaction products utilizing matrix-isolation FTIR spectroscopy.
They confirmed the production of ozone and determined the
temperature dependence of the branching ratio over the 263-
333 K range. The only direct kinetic investigation of the CH3C-
(O)O2 + HO2 reaction is that of Moortgat et al.9 Although based
on limited data, the results suggest that the reaction is fast and
has a steep negative temperature dependence with Ea = 1040
K-1
.
The rate constants and branching ratios measured in the
present work differ significantly from those reported previously.
Our time-resolved UV spectra could not be deconvoluted to
yield sensible levels of the secondary methylperoxy radical.
Possibly, this was due the difficulty in separating the contribu-
tions of three peroxy radicals, HO2, CH3C(O)O2, and CH3O2,
ozone, and long-lived reaction products to the UV spectra of
the reaction mixture. Therefore, independent measurements of
the [HO2] decay were undertaken using transient IR absorption.
The IR data confirmed the UV determinations of the HO2
concentrations, but the resultant [HO2] decay was as much as 7
times faster than anticipated and could only be fit upon
augmenting the standard kinetic model with a reaction between
HO2 and acetaldehyde. Data from the smog chamber experi-
ments also suggest that such a reaction takes place. The present
experiments, techniques, and analysis, are discussed in detail
in the following two sections. A comparison of our results to
those of Moortgat et al.9 is given in Section IV.
Figure 1. UV reference spectra of the principal species contributing
to the absorbance spectra of the CH3C(O)O2 + HO2 reaction mixture.
Reference spectra are obtained as follows: CH3C(O)O2, ref 23; HO2,
ref 16; CH3O2, ref 11; O3, unpublished; “products”, present work.
Two methods are employed to follow the reaction, time-
resolved UV spectroscopy and transient IR absorption. Both
have been described previously.11-13 In the former technique,
broadband UV light from a D2 lamp counterpropagates though
the reaction cell and is dispersed by a 0.32 m monochromator
with a 147 groove/mm grating onto a gated diode array detector.
This configuration allows the UV absorption of the peroxy
radicals to be monitored over the 200-300 nm wavelength
range. Transient spectra are recorded by gating the detector at
various times in the 0-1 ms range following the photolysis
pulse. A time resolution of 10 µs is typical.
Time-dependent peroxy radical concentrations are determined
by deconvolution of the absorbance versus wavelength traces
according to
II. Experimental Section
abs(t) ) ∑[xi]tσ(xi)l
(I)
A. Kinetic Measurements. The kinetics of the CH3C(O)O2
+ HO2 reaction is investigated over the 270-360 K temperature
range. The peroxy radical reactants are generated by the
photolysis of molecular chlorine in Cl2/CH3OH/CH3CHO/O2/
N2 gas mixtures using 10 ns, 300-600 mJ, 351 nm light pulses
from an excimer laser. The gas mixture flows through a
temperature-controlled cylindrical cell having a diameter of 3.2
cm and a path length of 51 cm. Gas flow rates are controlled
using Tylan flow controllers with the exception of Cl2 whose
flow rate is regulated through a needle valve. Nitrogen is added
to achieve the desired total pressure.
where l represents the path length and the index i ) 1-5
accounts for the hydroperoxy, acetylperoxy, methylperoxy,
ozone, and “products” contributions to the overall absorbance.
The relevant reference spectra are illustrated in Figure 1. While
recommended cross sections for HO2, O3, and CH3O2 are
available from literature reviews,14 the recommended CH3C-
(O)O3 spectrum is not up to date, and that for “products” is
unavailable. Thus, for consistency, all the reference spectra are
collected with the same spectrometer as used for the kinetic
measurements. A contribution to the absorbance from meth-
ylperoxy is included, since these radicals are secondary products
of the acetylperoxy self-reaction. Ozone is included because it
is formed by reaction 1b and absorbs strongly at 254 nm.
“Products” accounts for UV absorption by reaction products
such as H2O2 , CH3OOH, and CH3C(O)OOH, the first two of
which begin to absorb weakly below about 230 nm15 and the
last of which is assumed to absorb in the same region by
analogy. It is important to include the “products” term in eq I
to distinguish its contribution to the absorbance from that of
HO2, which, having a peak cross section at 203 nm, also
contributes in this spectral region;16 thus, omission of the
“products” term invariably leads to apparent HO2 concentrations
that decay to a nonzero level. The “products” reference spectrum
is obtained by subtracting the ozone contribution from the
absorbance of the reaction mixture at 20 ms, by which time the
radical concentrations have decayed to 0.
The reactant acetylperoxy and hydroperoxy radicals are
formed after chlorine atoms released by photolysis abstract
hydrogen atoms from the acetaldehyde and methanol precursors;
thus,
Cl + CH3CHO f CH3CO + HCl
(2)
(3)
CH3CO + O2 + M f CH3C(O)O2 + M
Cl + CH3OH f CH2OH + HCl
CH2OH + O2 f CH2O + HO2
(4)
(5)
The precursor and oxygen levels are made sufficiently high to
ensure that these reactions are complete on a microsecond time
scale, fast compared to the ensuing peroxy radical reactions.

Acetylperoxy + HO2 Reaction
J. Phys. Chem. A, Vol. 103, No. 3, 1999 367
Because of potential difficulties in deriving reliable concen-
trations by deconvoluting the rather complex UV absorbance
spectra obtained for the reaction mixture (compare Figure 1 and
Figure 3 below), the second method of transient IR absorption
was used for independent measurement of the HO2 concentration
following its generation by flash photolysis. These experiments
were, in a number of cases, run back-to-back under the same
conditions as a corresponding UV measurement. With this
technique, narrow band IR radiation from a Pb salt diode laser
replaces the UV light provided by the D2 lamp and is directed
onto a HgCdTe detector with a 0.3 µs response time. A
vibration-rotation line in the ν3 O-O stretching band that
conveniently falls between a pair of ammonia lines at 1117.45
and 1117.64 cm-1 is used to monitor the HO2 concentration.
The cross section of the HO2 probe line is measured in a
companion experiment in which the acetaldehyde is omitted
from the reaction mixture; thus, all of the radicals are converted
into HO2 . Because the experiments are carried out in the
pressure-broadening regime, the cross section varies with
temperature, pressure, and the nature of the diluent gas. As a
point of reference, it is 1.0 × 10-18 cm2 at 295 K and 52 Torr
of N2 . Although the measurement precision is about 5%, the
cross sections are calibrated against the concentration of
ethylperoxy radicals; thus, the overall accuracy is about 10%.
Concentration versus time data are extracted from the transient
absorption measurements via
OOH are studied using a relative rate method. Second, the
reaction between HO2 radicals and CH3CHO is investigated
using the irradiation of CH3CHO/13CH3OH/Cl2 mixtures in 700
Torr of either O2 or N2 diluent. Finally, the yields of CH3C-
(O)OH and CH3C(O)OOH are quantified following the irradia-
tion of CH3CHO/CH3OH/Cl2/air mixtures to determine the
branching ratio for reaction 1.
The apparatus for these experiments consists of an FTIR
spectrometer interfaced to a 140-L Pyrex reactor.17 As with the
kinetics experiments, radicals are generated by the UV irradia-
tion (λ > 300 nm) of molecular chlorine in the presence of
CH3CHO and CH3OH, in this case using 22 fluorescent
blacklamps (GTE F40BLB). The experiments are performed at
295 K in the presence of 700 Torr of ultrahigh purity air or N2
diluent. With only nitrogen present, hydrogen abstraction by
the chlorine atoms produces CH3CO and CH2OH radicals. When
O2 is also present, these are rapidly converted to CH3C(O)O2
and HO2 radicals. Initial concentrations of the gas mixtures used
in this study were 30-230 mTorr of CH3CHO, 200-300 mTorr
of Cl2, and 0-140 mTorr of CH3OH in 700 Torr of air diluent
(ultrahigh purity). The [CH3OH]/[CH3CHO] concentration ratio
was varied over the range 0-6.7. The rate constant ratio k2/k4
) 1.5; thus, the initial rate of HO2 radical production is 0-4.5
times that of CH3C(O)O2 radicals.18
The loss of CH3CHO and the formation of products are
monitored using FTIR spectroscopy. Products are quantified by
fitting reference spectra of the pure compounds to the observed
product spectra using integrated absorption features over the
800-2000 cm-1 range. The IR path length is 28 m, the spectral
resolution is 0.25 cm-1, and spectra are derived from 32 coadded
interferograms. Reference spectra are acquired by expanding
known volumes of the reference compounds into the chamber.
The acquisition of reference spectra for CH3C(O)OH and
CH3C(O)OOH is complicated by dimerization of acetic acid
and the fact that the peracetic acid is supplied as a 32 wt %
solution in acetic acid. The vapor above the liquid peracetic
acid sample is =70% peracetic acid and =30% acetic acid dimer
and monomer. There is no evidence for dimerization of the
peracetic acid. The vapor above liquid acetic acid contains both
monomer and dimer. The monomer exhibits IR absorptions
centered at 991, 1184, 1275, 1385, and 1790 cm-1, while the
dimer has features at 944, 1294, 1426, 1731 cm-1. Experiments
were performed to study the equilibrium between monomer and
dimer using a small (300 cm3) Pyrex reaction cell filled with
0.27-3.95 Torr of CH3C(O)OH vapor (both monomer and
dimer). As expected, the concentration of dimer varies in
proportion to the square of the monomer concentration. A value
of Keq ) [dimer]/[monomer]2 ) 2.5 ( 0.3 Torr-1 is derived at
295 K and 700 Torr of N2 diluent. This result is consistent with
the available literature data.19 Quantification of the CH3C(O)-
OH and CH3C(O)OOH absorption intensities is achieved using
dV(t)
dt
d
2
t HO
2
) V
e-[HO ] σ l - kdetV(t)
(II)
0 dt
a modified version of Beers Law that accounts for the ac
coupling of the detector. Here, V(t) represents the detector signal,
l is the path length, and σHO is the IR cross section of the HO2
2
vibration-rotation line. Because the detector is an AC-coupled
device, there is a decay constant of kdet ) 300 s-1 for the detector
output to return to zero after its initial response to a step function
change in light intensity. This term makes a small but noticeable
contribution to the HO2 decay over the 1 ms time scale of the
present experiments, which primarily affects the determination
of the HO2 + CH3CHO reaction rate.
As has been done previously,11-13 the concentration of
chlorine atoms generated by the photolysis pulse is calibrated
by substitution of ethane for the acetaldehyde and methanol
precursors under otherwise identical experimental conditions and
recording the number of ethylperoxy radicals that are generated.
The same procedure is used for measuring the total initial radical
concentration in the CH3C(O)O2 + HO2 kinetics experiments
and for determining the HO2 IR cross section. In this study, the
total radical concentration varies from 5 × 1014 to 11 × 1014
cm-3 and the [HO2]0/[CH3C(O)O2]0 ratio ranges from 0.7 to 3.
The temperature of the cell and reaction mixture is controlled
by a Neslab ULT-80dd recirculating chiller that precools/
preheats the gases, with the exception of acetaldehyde, before
their entrance into the reaction vessel. Acetaldehyde is intro-
duced just prior to the cell entrance to prevent its condensation
on the walls of the gas manifold at the lowest temperatures
employed. Acetaldehyde (99%) and methanol (99.9%) are
acquired from Fisher. Nitrogen (99.999%), oxygen (ultrazero
grade, THC < 0.5 ppm), and ethane (99.5%) are purchased from
Michigan Airgas. Chlorine (4.8% in nitrogen) is obtained from
Matheson. All the reagents are used without further purification.
B. FTIR-Smog Chamber System. Three sets of experi-
ments using the FTIR system are reported. First, the kinetics
of the reaction of Cl atoms with CH3C(O)OH and CH3C(O)-
σ(CH3C(O)OH monomer) ) 5.9 × 10-19 cm2 at 1177 cm-1
,
σ(CH3C(O)OH dimer) ) 1.80 × 10-18 cm2 at 1295 cm-1, and
σ(CH3C(O)OOH) ) 1.81 × 10-19 cm2 at 1251.5 cm-1
.
Systematic uncertainties associated with these calibrations are
estimated to be <10% for CH3C(O)OH and <15% for CH3C-
(O)OOH. The reference spectra used in this work are given in
Figure 2.
III. Results
A. Kinetic Measurements of k1. This section describes the
real-time kinetics measurements of the HO2 + CH3C(O)O2
reaction rate constant. The results of the product study are
presented in Section III.B. Figure 3 shows a typical UV

368 J. Phys. Chem. A, Vol. 103, No. 3, 1999
Crawford et al.
absorbance. By 500 µs, the acetylperoxy and hydroperoxy
radicals have largely disappeared and the spectrum is dominated
by methylperoxy, ozone, and “product” absorptions. As ex-
plained in Section II, omitting the “products” contribution to
the absorbance leads incorrectly to apparent nonnegligible
concentrations of HO2 at long times.
Although Figure 1 shows that the spectra of the five
contributing species overlap considerably, recording the com-
plete wavelength dependence of the absorbing reaction mixture
affords a reasonable prospect of extracting concentration versus
time profiles for the individual constituents. Deconvolution of
the spectral surface in Figure 3 according to eq I yields the
concentration traces illustrated in Figure 4. Predictions from
the model of Table 1 and based on the rate constants described
below are in good agreement with the rapid decay observed in
CH3C(O)O2 and HO2 levels and with the production of O3;
however, they underestimate by about a factor of 2 the extent
of CH3O2 formation. It is possible that this problem originates
from difficulties in separating the contributions of the five
constituents to the overall absorbance. However, the fact that a
large discrepancy appears for the methylperoxy concentration,
whereas the others are correctly predicted, suggests that
something may be missing from the postulated reaction mech-
anism.
The IR diode laser probe provides an independent assessment
of the amount of HO2 in the reaction mixture and permits us to
ascertain the accuracy of concentrations extracted from the time-
resolved UV absorbances. Consistency between the IR and UV
determinations of HO2 has been demonstrated by our previous
study of the HO2 + c-C5H9O2 reaction.21 Figure 5 illustrates
the results for the present reaction with CH3C(O)O2 . The inset
reveals good agreement between the UV (symbols) and IR (line)
derived HO2 levels. Although the acetylperoxy radical is not
independently probed in the IR, the predictions based on the
value of k1 found from fitting the HO2 IR data are in good
agreement with the UV-based CH3C(O)O2 measurements and
indirectly corroborate the UV data.
The main part of Figure 5 compares two model predictions
to the IR-derived HO2 concentrations. The dashed line represents
the best fit using a model consisting of reactions 1 and 6-9
and the secondary reactions of the methylperoxy radical with
HO2 , CH3C(O)O2 , and itself. This model cannot account for
the shape of the hydroperoxy radical decay for any of the
experiments in this study. If, as shown in the figure, the initial
decay rate is in reasonable agreement with the data, then the
overall loss is underestimated. Alternatively, the overall loss
of HO2 could be brought into agreement by raising k1 , but then
the model would considerably overestimate the initial decay rate.
One feasible explanation for this behavior is that an unan-
ticipated secondary species reacts with HO2 causing larger
removal at longer times than predicted by the model. This
explanation, however, falls short in two respects. First, at the
1014 cm-3 radical concentrations of these experiments, sufficient
HO2 removal by a secondary species would require an unreal-
istically large rate constant. Second, this hypothesis fails to
account for the anomalously high methylperoxy levels observed
from the UV absorbance measurements.
Figure 2. IR reference spectra used for the FTIR product study.
absorbance-wavelength-time surface obtained following pho-
tolysis of a CH3CHO/CH3OH/Cl2/O2/N2 gas mixture. Qualita-
tively, the spectrum undergoes the expected changes in time.
Initially, it exhibits a dominant peak at about 210 nm with a
shoulder at 250 nm, which, according to the reference spectra
in Figure 1, is consistent with a mixture of CH3C(O)O2 and
HO2 radicals (essentially formed instantaneously by the pho-
tolyzed Cl atoms). As time progresses, the 210 nm peak intensity
decreases rapidly, since both the acetylperoxy and hydroperoxy
radicals exhibit strong absorption features in this wavelength
region and both are lost due to radical/radical reactions. The
principal loss pathways include the CH3C(O)O2 + HO2 reaction
under investigation, as well as the self-reactions
HO2 + HO2 f H2O2 + O2
(6)
CH3C(O)O2 + CH3C(O)O2 f 2 CH3C(O)O + O2 (7)
In contrast, the absorbance at 250 nm decreases more slowly,
giving the spectrum a flatter appearance at longer times. Unlike
at 210 nm, the loss of intensity in this spectral region does not
depend on the HO2 concentration. Intensity lost due to a decrease
in CH3C(O)O2 concentration is partly compensated by absorp-
tion due to CH3O2 radicals formed from the products of reaction
7 via
CH3C(O)O f CH3 + CO2
(8)
(9)
Potentially, another explanation for the higher than expected
CH3O2 levels involves the possible recycling of CH3C(O)O2
radicals by the reaction of secondary methoxy radicals with
acetaldehyde (k ) 7.4 × 10-14 cm3 s-1),22
CH3 + O2 + M f CH3O2 + M
and from the ozone produced by reaction 1b. Ozone and the
secondary methylperoxy radicals complicate analysis of the UV
absorption data. A generic “products” category is added to
account for the peroxide products of reactions 1a and 6 and
from secondary reactions described in Table 1,20 bringing to
five the number of species contributing to the overall UV
CH3O + CH3CHO f CH3CO + CH3OH
(10)
In the present mechanism (Table 1), methoxy radicals arise from
the cross reaction of acetylperoxy radicals with CH3O2 and from

Acetylperoxy + HO2 Reaction
J. Phys. Chem. A, Vol. 103, No. 3, 1999 369
Figure 3. Time-resolved UV absorbance of a CH3CHO/CH3OH/Cl2/O2/N2 reaction mixture following 351 nm photolysis.
TABLE 1: CH3C(O)O2 + HO2 Reaction Mechanism
reaction
rate constanta
Principal
(1a) CH3C(O)O2 + HO2 f CH3C(O)OOH + O2
(1b) CH3C(O)O2 + HO2 f CH3COOH + O3
(6) HO2 + HO2 f H2O2 + O2
k1 ) 3.9 × 10-13e1350/T cm3 s-1 b
k1b/k1 ) 0.12b
k6 ) 2.8 × 10-13e(594/T) cm3 s-1 16
k7 ) 3.0 × 10-12e504/T cm3 s-1 13
k13 = 5 × 10-14 cm3 s-1 b
(7) CH3C(O)O2 + CH3C(O)O2 f 2 CH3C(O)O + O2
(13) HO2 + CH3CHO f CH3C(OH)HO2
Secondary
(8) CH3C(O)O f CH3 + CO2
(9) CH3 + O2 + M f CH3O2 + M
k8 > 1 × 106 s-1
k9 ) 4.5 × 10-31(T/300)-3 cm6 s-1 15
k11a ) 0 cm3 s-1 23,c
(11a) CH3C(O)O2 + CH3O2 f CH3C(O)O + CH3O + O2
(11b) CH3C(O)O2 + CH3O2 f CH3COOH + CH2O + O2
HO2 + CH3O2 f CH3OOH + O2
k11b ) 8.5 × 10-13e726/T cm3 s-1 23
k ) 4.1 × 10-13e(790/T) cm3 s-1
assumed ) kCH O +HO
HO2 + CH3C(OH)HO2 f products
CH3C(O)O2 + CH3C(OH)HO2 f products
3
2
2
assumed ) kCH C(O)O +CH O
3
2
3 2
Tertiary
HO2 + CH2O f CH2(OH)O2
k ) 9.71 × 10-15e(625/T) cm3 s-1
(12a) CH3O2 + CH3O2 f 2 CH3O + O2
(12b) CH3O2 + CH3O2 f CH3OH + CH2O + O2
k12a ) (1 - b) × 9.1 × 10-14e416/T cm3 s-1
k12b ) b × 9.1 × 10-14e416/T cm3 s-1
-1
b ) (1 + 25 × e-1165/T
k ) 3 × 10-11 cm3 s-1
)
CH3O2 + CH3O f CH2O + CH3OOH
CH3O + CH3O f CH2O + CH3OH
CH3O + O2 f CH2O + HO2
k ) 1.5 × 10-12 cm3 s-1
k ) 3.9 × 10-14e-900/T cm3 s-1
a Rate constants taken from ref 15 unless otherwise noted. b Measured in the present study. c Branching fraction varied from 0 to 1 to establish
sensitivity on k1 determination.
the self-reaction of the latter species,
experiments. There is currently disagreement in the literature
over the branching of reaction 11.10,23,24 However, model
simulations show that the maximum recycling is less than about
5% of the initial CH3C(O)O2 concentration, even if reaction 11
proceeds entirely via channel 11a; thus, the recycling mechanism
cannot explain the difficulty in fitting the HO2 decay in Figure
5.
CH3C(O)O2 + CH3O2 f CH3C(O)O + CH3O + O2 (11a)
f CH3COOH + CH2O + O2 (11b)
CH3O2 + CH3O2 f 2CH3O + O2
(12a)
Another possibility is that HO2 reacts with acetaldehyde. This
reaction was postulated by Moortgat et al.25 in analogy with
the known reaction between HO2 and formaldehyde.26 It would
form an adduct that rearranges to the 1-hydroxy ethylperoxy
radical; thus,
f CH3OH + CH2 + O2 (12b)
Because it is slow and involves two secondary species, the
methylperoxy self-reaction produces only a negligible quantity
of methoxy radicals over the 1000 µs time scale of the present

370 J. Phys. Chem. A, Vol. 103, No. 3, 1999
Crawford et al.
Figure 4. Concentration versus time profiles for HO2 , CH3C(O)O2 , CH3O2 , and O3 as obtained from deconvolution of the spectral surface
illustrated in Figure 3. The solid lines represent predictions from the reaction model of Table 1. The dashed lines are obtained assuming the value
for k1 reported by Moortgat et al.9 The error bars are 95% confidence limits for the deconvolution. The actual errors are somewhat larger, ∼10%,
owing to the systematic uncertainty in the reference cross sections.
this product contribute significantly to the anomalously high
absorption in the CH3O2 spectral range. An accurate assessment
of the contribution of the 1-hydroxy ethylperoxy radical to the
UV absorbance spectra is not feasible since a reference spectrum
of this radical is not available. Simply assuming it to be identical
to that of methylperoxy leads to a total “CH3O2 + CH3CH-
(OH)O2” concentration that falls about halfway between the
methylperoxy prediction and the “methylperoxy” data in Figure
4. The remaining discrepancy could easily arise from the fact
that the CH3O2 and CH3CH(OH)O2 spectra are not identical
and, therefore, that the deconvolution procedure of eq I
incorrectly attempts to separate the overlapping contributions
of six species into five components.
In principle, a “direct” measurement of reaction 13 should
be feasible. Using the present technique, this would involve
increasing the acetaldehyde concentration sufficiently to ensure
that its reaction with HO2 represents the dominant HO2 loss
Figure 5. The HO2 concentration in a CH3CHO/CH3OH/Cl2/O2/N2
mechanism and adjusting the methanol concentration to ensure
reaction mixture following 351 nm photolysis. The inset compares the
that HO2 radicals dominate over CH3C(O)O2 following Cl2
HO2 concentrations derived from deconvolution of the UV spectra
photolysis. Unfortunately, this ideal situation is difficult to
achieve in practice. The rapid reaction between chlorine atoms
and acetaldehyde necessitates the use of high levels (10-20
Torr) of CH3OH. Two difficulties are incurred: (1) the
introduction of the methanol into the reaction cell; (2) the
dependence of the HO2 self-reaction on methanol concentra-
tion.27 While not achieving ideal conditions, a number of
experiments were performed at high initial acetaldehyde con-
centrations (∼2 Torr) and with sufficient methanol to provide
[HO2]0/[CH3C(O)O2]0 ratios of 2-3.
Figure 6 demonstrates the results. Consistent with the example
of Figure 5, a good fit of the data is obtained with a value of
k13 that is commensurate with the rate constant for HO2 +
CH2O.26 In contrast, omitting reaction 13 and setting k1 to the
value of Moortgat et al.9 severely underestimates the HO2 decay
(symbols) to those measured using IR probing (noisy line); also shown
are the UV-based CH3C(O)O2 data. In the main figure, the solid line
represents the best fit prediction from the model of Table 1, whereas
the dotted lines show the limits imposed by the 2σ error bounds on k1.
The dashed line exhibits the best fit in the absence of a reaction between
HO2 and CH3CHO.
HO2 + CH3CHO f CH3CH(OH)O2
(13)
Including this reaction in the model (see Table 1) and allowing
both k1 and k13 to vary as fitting parameters lead to excellent
fits of the IR measurements of [HO2], as illustrated by the solid
line in Figure 5. Furthermore, the 1-hydroxy ethylperoxy radical
generated by reaction 13 is expected, by analogy with HOCH2O2,
to absorb UV radiation with a spectrum similar to that of
CH3O2;26 thus, a reaction between HO2 and CH3CHO could by

Acetylperoxy + HO2 Reaction
J. Phys. Chem. A, Vol. 103, No. 3, 1999 371
Figure 7. The variation in the HO2 + CH3CHO reaction rate versus
initial acetaldehyde concentration. Data are shown at two temperatures,
295 and 360 K.
since reactions 4 and 5 lead to HO2 formation in the presence
of oxygen and this is not possible in the N2 mixture. Typical
gas mixtures used to test the hypothesis consisted of 31-51
mTorr of CH3CHO, 2-218 mTorr of 13CH3OH, and 200-500
mTorr of Cl2, with [13CH3OH]/[CH3CHO] varied over the range
0.04-4.4. Isotopically labeled methanol is used as a precursor
to distinguish it from CH3OH formed as an oxidation product
of acetaldehyde.
Figure 8A depicts the CH3CHO loss compared to that of
13CH3OH in the N2 buffer. The rate constant ratio deduced from
a linear least-squares analysis of these data is k2/k4 ) 1.48 (
0.07, independent of the [CH3OH]0/[CH3CHO]0 ratio. This result
is in excellent agreement with the recent measurement of k2/k4
) 1.50 ( 0.10 by Tyndall et al.,18 who used mixtures of 35
mTorr of CH3CHO and 9 mTorr of 13CH3OH and found the
ratio to be independent of total pressure of N2 diluent over the
1-700 Torr range. It also compares well with the ratio of the
recommended absolute rate constants, namely k2/k4 ) 7.2 ×
10-11/5.3 × 10-11 ) 1.36 ( 0.29 (ref 28).
Figure 6. HO2 concentration versus time at high [CH3CHO]0 . The
transient IR data are compared to the best fit prediction based on the
present model, a prediction using the rate constant of Moortgat et al.,9
and a fit to the present model assuming that k13 ) 0. Predictions from
a model modified to include CH3C(O)O2 regeneration via reaction 10
are indistinguishable from the best fit prediction.
rate (this remains the case regardless of the value used for the
branching ratio of reaction 11). By substantially increasing k1,
the initial HO2 loss can be modeled, but not the long-time
behavior.
The variation in the rate for the HO2 + CH3CHO reaction as
a function of initial acetaldehyde concentration is demonstrated
in Figure 7 at two temperatures. While they exhibit considerable
scatter, the rates increase with increasing CH3CHO level as
would be expected for reaction 13. The determination of the
error bars is described below; however, it is relevant to note
that they include both statistical errors and systematic uncertain-
ties, e.g., the initial total radical concentration. Yet the data
scatter is outside the anticipated error bounds. A plausible
explanation is as follows: the HO2 absorbance is a weak signal
recorded by the ac-coupled infrared detector out to relatively
long times (∼1 ms). We attempt to correct the natural tendency
of the detector to decay to zero output, regardless of light level,
by assuming an equivalent RC circuit; however, this is only an
approximation. It is possible that this correction is inadequate
and, furthermore, that it is affected by a detector baseline that
is not always perfectly flat. Both affect the long-time appearance
of the HO2 decay, on which the value of k13 depends. In contrast,
the fitting of k1 is much less affected because, as shown in Figure
6, the acetylperoxy radicals are largely gone by 200 µs.
FTIR-smog chamber experiments provide further support
for a hydroperoxy radical reaction with acetaldehyde. The
evidence comes from a comparison of the Cl atom initiated loss
of CH3CHO relative to 13CH3OH in 700 Torr of N2 versus air.
If the hypothesis of a reaction between HO2 and CH3CHO holds
true, then the loss of acetaldehyde relative to methanol should
be higher in the air mixture as compared to the N2 mixture,
The situation is different when the experiments are carried
out in air. In this case, rate constant ratios measured at a low
[13CH3OH]0/[CH3CHO]0 ratio are indistinguishable from those
obtained in N2. However, as evident in Figure 8B, increasing
[CH3OH]0/[CH3CHO]0 produces a small but distinct deviation
of k2/k4 from the values obtained in N2 (indicated by the dotted
line). A linear least-squares regression to the data obtained in
air gives k2/k4 ) 1.89 ( 0.10, which is distinctly different from
the value of k2/k4 ) 1.48 ( 0.07 in N2. The ratio in air is
higher, consistent with the existence of a second CH3CHO
removal pathway, and thus, supports the hypothesis of a reaction
between HO2 radicals and CH3CHO.
The measured rate constants for the CH3C(O)O2 + HO2
reaction are collected in Table 2 along with the values obtained
for k13 , the initial radical concentrations, and the experimental
conditions. The rate constants are determined from the best fit
of the model in Table 120 to HO2 concentration versus time
traces recorded by transient IR absorption. The IR data were
used because of the complications, discussed above, in decon-

372 J. Phys. Chem. A, Vol. 103, No. 3, 1999
Crawford et al.
initial partitioning of radicals into HO2 and CH3C(O)O2 popula-
tions, leads to the bulk of the error in k1 , up to 25%. Including
a 10% uncertainty in the total initial radical concentration
contributes an additional 6% to the error. Uncertainties in the
self-reaction rate constants of HO2 and CH3C(O)O2 are smaller,
adding about 1% each to the error. Combining these contribu-
tions statistically, the total 2σ error bars for k1 are approximately
(25% in most cases.
Considerations for the error bars of k13 are similar. In this
case, the fitting error, which includes the effects of uncertainties
in k1 and [HO2]0/[CH3C(O)O2]0 , varies more (10%-125%) and
is the major source of error. At the lower end of this range, an
additional 14% error contributed by uncertainty in total radical
concentration and the 3% errors from uncertainties in the peroxy
radical self-reaction rate constants add nonnegligible contribu-
tions to the overall error. As explained with reference to Figure
7, however, there are additional errors associated with the
detector that we are unable to account for; thus, the fitted values
of k13 fall into the range of (1.2-6) × 10-14 cm3 s-1 at 295 K.
For comparison, the literature rate constant for the analogous
HO2 + HCHO reaction varies from 5.5 × 10-14 cm3 s-1 at 360
K to 9.8 × 10-14 cm3 s-1 at 270 K.28 The values obtained here
for k13 are of the same order.
In contrast to k13, the HO2 + CH3C(O)O2 rate constant
remains essentially unaffected by changes in [CH3CHO]0 and
variation of [HO2]0/[CH3C(O)O2]0 between 0.7 and 3.0 (see
Table 2). The data displayed in Figure 9 indicate consistent
values for this rate constant at each temperature (with one outlier
at 359 K; Table 2). The temperature dependence of the CH3C-
(O)O2 + HO2 rate constant follows the Arrhenius expression
Figure 8. The loss of CH3CHO versus 13CH3OH in 700 Torr of either
(A) N2 or (B) air diluent. Initial concentration ratios of [13CH3OH]/
[CH3CHO] are 0.04-4.4 in the N2 experiments and 0.04 (]), 0.17
(0), 1.7 (4), and 2.4 (3) for the experiments in air. The solid line in
panel A is a linear least-squares fit of the data, which is reproduced as
the dotted line in panel B.
k1 ) (3.9+5.0) × 10-13e(1350(250)/T cm3 s-1
(III)
-2.3
The negative temperature dependence is typical of HO2 + RO2
reactions and suggests that the reaction proceeds through a rate-
limiting adduct formation.
voluting the time-resolved UV absorbances. As indicated in
Table 2, however, both techniques were used under a subset
of experimental conditions with generally good agreement found
for the HO2 concentrations, as illustrated in the inset to Figure
5.
B. Product Studies. Reactions of Cl Atoms with CH3C(O)-
OH and CH3C(O)OOH. Prior to investigating the products
arising from the reaction of CH3C(O)O2 and HO2 radicals,
relative rate experiments were performed using the FTIR system
to investigate the kinetics of the reaction of Cl atoms with CH3C-
(O)OH and CH3C(O)OOH. The relative rate technique is
described in detail elsewhere.29,30 Photolysis of molecular
chlorine provides the source of Cl atoms. The kinetics of reaction
14 are measured relative to reaction 16, while reaction 15 was
measured relative to reactions 17 and 18.
Fitting the HO2 concentration traces consists of the following
steps. The total initial radical concentration is calibrated against
the number of ethylperoxy radicals formed upon the substitution
of ethane for acetaldehyde and methanol under otherwise
identical experimental conditions. Where available, the UV
absorbance data at early times, when HO2 and CH3C(O)O2
contributions dominate, are used to partition of the initial radicals
into CH3C(O)O2 and HO2 populations; otherwise, this is based
on the initial HO2 concentration determined by the IR absorp-
tion. The model is then fit to the entire [HO2] versus time trace,
using k1 and k13 as fitting parameters and allowing small
adjustments in the initial partitioning of radicals. Reactions of
the CH3C(OH)HO2 product of reaction 13 with HO2 and CH3C-
(O)O2 are included in the model by assuming the rate constants
to be the same as those for the corresponding reactions with
CH3O2. The effects of these reactions on the model predictions
are small, less than indicated by the dashed error bars in Figure
5.
Cl + CH3C(O)OH f products
Cl + CH3C(O)OOH f products
(14)
(15)
Cl + CH4 f HCl + CH3
(16)
Cl + CD4 f DCl + CD3
(17)
(18)
Cl + HCFC-141b f products
Several factors contribute to the error bars for k1. These
include signal-to-noise, uncertainties in the total number of
radicals produced, and uncertainties in the rate constants of the
principal reactions in the model, namely the HO2 and CH3C-
(O)O2 self-reactions and the HO2 + CH3CHO reaction. The
fitting error, which includes the uncertainties in k13 and in the
To test for a loss of reactants via photolysis or heterogeneous
reactions, mixtures of CH3C(O)OH and CH3C(O)OOH in air
were prepared without Cl2, left to stand in the dark for 10 min,
and then irradiated for 10 min. There was no loss (<2%) of
either compound, showing that photolysis and heterogeneous
loss are insignificant.

Acetylperoxy + HO2 Reaction
J. Phys. Chem. A, Vol. 103, No. 3, 1999 373
TABLE 2: CH3C(O)O2 + HO2 Rate Constants
conditions
resultsa
temp
(K)
Ptot
(Torr)
O2
(Torr)
Cl2
(Torr)
CH3CHO
(Torr)
CH3OH
(Torr)
[HO2]0
[CH3C(O)O2]0
kHO +CH CHO
kCH CO +HO
3 3 2
2
3
(1014 cm-3
)
(1014 cm-3
)
(10-14 cm-3 s-1
)
(10-1 cm-3 s-1
)
269b
269
269
297b
293
295
295
295b
295
295
295
295
325b
328
329b
359b
359
359
359
363
50.2
50.3
50.7
51.3
50.2
52.0
54.3
48.8
57.0
61.0
56.9
56.0
50.5
50.4
50.1
54.3
50.7
50.8
50.9
51.0
40.6
40.8
16.2
26.2
15.2
22.8
23.2
14.9
45.2
48.3
44.8
45.2
25.3
42.8
41.7
43.7
41.2
39.9
37.9
43.2
0.37
0.36
0.44
0.40
0.53
0.42
0.41
0.51
0.34
0.35
0.34
0.34
0.40
0.30
0.32
0.44
0.42
0.42
0.42
0.31
0.60
0.70
0.80
0.69
0.75
0.13
1.10
0.72
0.97
1.43
1.68
2.04
0.66
0.63
0.33
0.41
0.35
1.00
1.98
0.74
1.27
1.25
.90
1.56
.90
4.0
3.9
4.2
4.7
5.3
5.6
4.2
4.5
6.3
5.4
5.5
4.9
4.8
2.5
3.8
3.7
3.7
3.1
2.9
1.9
1.6
2.8
3.3
2.0
3.8
1.8
3.4
3.5
2.1
2.5
2.6
3.5
1.8
2.4
1.2
1.5
2.8
3.4
4.0
2.9
20 ( 3
5.9 ( 2.6
5.1 ( 1.3
4.8 ( 1.4
4.4 ( 1.6
4.0 ( 0.9
4.9 ( 1.5
3.2 ( 0.9
3.3 ( 0.7
4.0 ( 0.6
4.7 ( 0.8
4.6 ( 0.7
3.6 ( 0.7
2.5 ( 0.9
2.3 ( 0.6
2.1 ( 0.6
1.4 ( 0.6
1.8 ( 0.3
1.3 ( 0.2
0.6 ( 0.2
2.0 ( 0.6
8.0 ( 1.4
16 ( 3
4.3 ( 1.2
4.8 ( 1.3
5.9 ( 3.5
1.5 ( 0.6
1.3 ( 0.6
2.3 ( 0.4
1.2 ( 0.3
2.1 ( 0.3
5.0 ( 0.9
4.6 ( 0.9
7.4 ( 2.2
2.8 ( 0.7
4.9 ( 1.0
5.9 ( 1.1
3.5 ( 0.6
3.8 ( 0.7
2.1 ( 2.6
0.16
1.38
0.88
3.77
4.03
3.70
3.77
1.50
2.25
1.28
1.02
0.39
1.14
2.25
0.65
a Error bars are (2σ and include fitting and systematic uncertainties. b Indicates that a UV experiment was done under the same conditions.
Figure 9. Temperature dependence of the CH3C(O)O2 + HO2 rate
constant. The solid lines represent least-squares fits of the rate constants
to the Arrhenius expression.
Figure 10A shows the observed loss of CH3C(O)OH versus
that of CH4 following UV irradiation of a CH3C(O)OH/CH4/
Cl2 mixture in 700 Torr total pressure of N2. Figure 10B shows
CH3C(O)OOH loss versus those of CD4 and CFCl2CH3 upon
the UV irradiation of CH3C(O)OOH/CD4/Cl2 and CH3C(O)OOH/
CFCl2CH3/Cl2 mixtures in 700 Torr total pressure of N2. Linear
least-squares analyses of these data give k14/k16 ) 0.25 ( 0.03,
k15/k17 ) 0.81 ( 0.06, and k15/k18 ) 2.06 ( 0.24. Inserting the
literature values k16 ) 1.0 × 10-13 [ref 15], k17 ) 6.1 × 10-15
[ref 30], and k18 ) 2.0 × 10-15 cm3 s-1 [ref 30] yields k14
)
(2.5 ( 0.3) × 10-14, k15 ) (4.9 ( 0.6) × 10-15, and k15 ) (4.1
( 0.5) × 10-15 cm3 s-1, respectively. Unless otherwise noted,
errors quoted for relative rates are 2σ based on the scatter in
the data. When these are converted to rate constants via
comparison to the reference reaction, the uncertainty in the latter
quantity is also incorporated into the overall error. There being
no inherent reason to prefer one reference reaction over the
other, we take for k15 the average of the two relative rate
determinations and set the error limits to encompass the extremes
Figure 10. Plots of the loss of CH3C(O)OH versus CH4 and the loss
of CH3C(O)OOH versus HCFC-141b (CFCl2CH3) and CD4 when
mixtures containing these compounds are exposed to Cl atoms in 700
Torr of N2 diluent at 296 K. Solid lines are linear least-squares fits.
of the individual determinations; hence, k15 ) (4.5 ( 1.0) ×
10-15 cm3 s-1. Our measurement of k14 ) (2.5 ( 0.3) × 10-14
cm3 s-1 agrees well with the previous determination of k14
)

374 J. Phys. Chem. A, Vol. 103, No. 3, 1999
Crawford et al.
Figure 12. The formation of CH3C(O)OOH and CH3C(O)OH versus
CH3CHO loss following UV irradiation of a mixture of 52.1 mTorr of
CH3CHO, 26.1 mTorr of CH3OH, and 290 mTorr of Cl2 in 700 Torr
of air diluent.
shown in panel C. The large feature at 1104 cm-1 arises from
HCOOH, which is a well-established photooxidation product
of CH3OH. The loss of CH3CHO is 42.7 mTorr. Comparison
with the reference spectra of CH3C(O)OOH and CH3C(O)OH,
shown in parts D and E of Figure 11, confirms the species
identification. Figure 12 depicts the observed formation of
CH3C(O)OOH and CH3C(O)OH versus the loss of CH3CHO
in this experiment.
As with all product studies, the possibility of product loss
via secondary reactions deserves careful attention. Potential
unwanted secondary processes include photolysis, heterogeneous
losses, and reaction with Cl atoms. As discussed above, no
discernible photolytic or heterogeneous loss of CH3C(O)OOH
or CH3C(O)OH was detected over the time scale of the present
experiments. Furthermore, considering that CH3C(O)OH and
CH3C(O)OOH are, respectively, 3000 and 16000 times less
reactive toward Cl atoms than is CH3CHO, their loss via Cl
atom attack is insignificant.
The observed yields of CH3C(O)OOH and CH3C(O)OH are
plotted versus [CH3OH]0/[CH3CHO]0 in Figure 13. This figure
demonstrates that increasing the initial concentration ratio from
0 to 2 raises both the CH3C(O)OOH and CH3C(O)OH yields,
as expected from an increased importance of reaction 1 relative
to reactions 7 and 11. Both product yields reach plateaus for
[CH3OH]0/[CH3CHO]0 > 2, indicating that the CH3C(O)O2 loss
occurs almost entirely via reaction 1. Averaging the product
levels in this regime provides a CH3C(O)OOH yield of (75 (
6)% and a CH3C(O)OH yield of (8.5 ( 0.9)%, including 2σ
fitting errors. We estimate that potential systematic uncertainties
in quantifying the CH3C(O)OOH and CH3C(O)OH yields are
20% and 15%, respectively. Using conventional error propaga-
tion to incorporate these uncertainties leads to yields for CH3C-
(O)OOH and CH3C(O)OH of (75 ( 16)% and (8.5 ( 1.6)%,
respectively. Together, CH3C(O)OOH and CH3C(O)OH account
for (84 ( 17)% of the loss of CH3CHO.
Figure 11. IR spectra acquired before (A) and after (B) 2 min
irradiation of a mixture containing 52.1 mTorr of CH3CHO, 26.1 mTorr
of CH3OH, and 290 mTorr of Cl2 in 700 Torr of air diluent. Subtraction
of IR features attributable to CH3CHO and CH3OH gives the product
spectrum shown in panel C (note the change in y-axis scale). Reference
spectra of CH3C(O)OOH and CH3C(O)OH are shown in panels D and
E.
2.8 × 10-14 cm3 s-1 by Koch and Moortgat31. There are no
literature values against which to compare our measurement of
k15.
Product Study of the CH3C(O)O2 + HO2 Reaction. The
products of reaction 1 were studied using the FTIR-smog
chamber system and UV irradiation of CH3CHO/CH3OH/Cl2/
air mixtures. Atomic chlorine formed by the photolysis of
molecular chlorine is converted into CH3C(O)O2 and HO2
radicals via reactions 2-5. The acetylperoxy self-reaction (7),
its reaction with HO2 (1), and its reaction with CH3O2 (11) then
compete for the available CH3C(O)O2 radicals. The experimental
approach to ascertain the branching ratio of reaction 1 is to
measure CH3C(O)OOH and CH3C(O)OH yields as a function
of the initial concentration ratio, [CH3OH]0/[CH3CHO]0 . As
this ratio increases, so does the [HO2]/[CH3C(O)O2] ratio.
Reaction 1 thereby becomes a more important CH3C(O)O2 loss
pathway relative to reactions 7 and 11, and at sufficiently high
ratios, it becomes essentially the sole loss mechanism. Increasing
[CH3OH]0/[CH3CHO]0 beyond this point will not change the
observed CH3C(O)OOH and CH3C(O)OH yields; thus, this
asymptotic regime reflects the relative importance of channels
1a and 1b.
Both CH3C(O)OOH and CH3C(O)OH are observed as
products following the UV irradiation of CH3CHO/CH3OH/Cl2/
air mixtures. Figure 11 compares spectra acquired before (A)
and after (B) a 2 min irradiation of a mixture of 52.1 mTorr of
CH3CHO, 26.1 mTorr of CH3OH, and 290 mTorr of Cl2 in 700
Torr of air diluent. Subtraction of IR features attributable to
the CH3CHO and CH3OH reactants gives the product spectrum
There are two factors that account for a carbon balance of
less than 100%. First, although reaction 1 dominates the loss
of CH3C(O)O2 radicals under conditions where [CH3OH]0/[CH3-
CHO]0 > 2, a small loss of CH3C(O)O2 radicals via self-reaction
and reaction with CH3O2 is unavoidable. Second, as discussed

Acetylperoxy + HO2 Reaction
J. Phys. Chem. A, Vol. 103, No. 3, 1999 375
Figure 13. Observed molar product yields for CH3C(O)OOH and
CH3C(O)OH versus the initial concentration ratio [CH3OH]0/[CH3-
CHO]0 following the UV irradiation of CH3CHO/CH3OH/Cl2/air
mixtures at 700 Torr total pressure at 296 K. The lines are averages of
the data for [CH3OH]0/[CH3CHO]0 > 2.
Figure 14. The formation of CH3C(O)OOH versus that of CH3C(O)-
OH following the irradiation of CH3CHO/CH3OH/Cl2/air mixtures with
[CH3OH]0/[CH3CHO]0 > 2 (filled symbols) and [CH3OH] 0/[CH3CHO]o
< 2 (open symbols). The solid line is a linear least-squares fit that has
a slope of 9.2 ( 0.7.
in Section III.A, we have evidence that some CH3CHO is
removed via reaction with HO2 , presumably producing CH3C-
(OH)HO2 radicals. By analogy to C2H5O2 radicals,32 we expect
that they will react with HO2 to give the corresponding
hydroperoxide; thus, it is unlikely that CH3C(OH)HO2 is
converted into either peracetic or acetic acid under our
experimental conditions. An estimate of the magnitude of CH3-
CHO loss via reaction with HO2 can be derived from the data
shown in Figure 8. Linear least-squares analysis of the data in
Figure 8B, using [CH3OH]0/[CH3CHO]0 ) 2.4, gives an
apparent value of k2/k4 )1.8, which is 20% greater than the
“true” value of k2/k4 ) 1.48 ( 0.07 obtained from Figure 8A.
Thus, for the experiment with [CH3OH]o/[CH3CHO]o ) 2.4,
the CH3CHO loss is approximately 20% greater than can be
explained solely on the basis of Cl atom chemistry. This
additional loss of CH3CHO is of the correct magnitude to
explain why the combined CH3C(O)OOH and CH3C(O)OH
yield of (84 ( 17)% does not account for 100% of the
acetaldehyde loss.
Figure 14 shows the observed formation of CH3C(O)OOH
versus that of CH3C(O)OH following UV irradiation of CH3-
CHO/CH3OH/Cl2/air mixtures. Results from experiments with
[CH3OH]0/[CH3CHO]0 > 2 are indicated by the filled symbols,
while those with [CH3OH]0/[CH3CHO]0 < 2 are indicated by
open symbols. There is no discernible impact of the [CH3OH]0/
[CH3CHO]0 ratio on the relative yields of CH3C(O)OOH and
CH3C(O)OH, suggesting that reaction 1 is the dominant source
of both acids. Assuming that CH3C(O)OOH and CH3C(O)OH
are formed solely via reaction 1, a linear least-squares analysis
of the data in Figure 14 gives k1a/(k1a+k1b) ) 0.90 ( 0.02 and
k1b/(k1a+k1b) ) 0.10 ( 0.02.
Figure 15. Observed formation of CH3C(O)OOH ([), CH3C(O)OH
(2), and O3 (b) versus loss of CH3CHO following irradiation of CH3-
CHO/Cl2/air mixtures. For clarity, the data for CH3C(O)OOH and
CH3C(O)OH have been shifted vertically by 10 units.
O3. A correction is necessary to account for the loss of ozone
via
Cl + O3 f ClO + O2
(19)
As reported by Niki et al.8 and Horie and Moortgat,10 ozone
is formed as a coproduct with CH3C(O)OH in channel 1b.
Unfortunately, the IR spectrum of ozone is overlapped by that
of CH3OH, so it is not possible to confirm the formation of O3
in the experiments using CH3CHO/CH3OH/Cl2/air mixtures
described above. An additional set of experiments was con-
ducted using mixtures of 73-228 mTorr of CH3CHO and 0.3-
0.4 Torr of Cl2 in 700 Torr of air to quantify ozone formation.
Figure 15 shows the yields of CH3C(O)OOH, CH3C(O)OH, and
This is computed using the Acuchem program with k2 ) 7.2 ×
10-11 cm3 s-1 and k19 ) 1.2 × 10-11 cm3 s-1 [ref 15]. The
corrections are in the range 1-17% and have been applied to
the ozone data in Figure 15. The ClO radicals react with the
various peroxy radicals in the system and represent a minor
loss pathway for these species.
A linear least-squares analysis of the data in Figure 15
provides yields of 0.28 ( 0.03 for CH3C(O)OOH, 0.034 ( 0.005
for CH3C(O)OH, and 0.027 ( 0.004 for O3. Within the

376 J. Phys. Chem. A, Vol. 103, No. 3, 1999
Crawford et al.
TABLE 3: Ozone Channel Branching Fraction
temp
(K)
Ptot
(Torr)
O2
(Torr)
Cl2
(Torr)
CH3CHO
(Torr)
CH3OH
(Torr)
[HO2]0
[CH3C(O)O2]0
[O3] yield
(1014 cm-3
)
(1014 cm-3
)
(10-14 cm3 s-1
)
k1b/k1
a
265
266
269
297
295
295
325
329
359
132
30.8
23.8
40.6
26.2
29.0
14.9
25.3
41.7
43.7
0.33
0.41
0.37
0.40
0.31
0.51
0.40
0.32
0.44
0.98
0.67
0.60
0.69
1.17
0.72
0.66
0.33
0.41
1.07
2.91
1.27
1.56
1.11
0.88
1.50
1.28
1.04
3.7
4.3
3.1
5.2
4.5
3.6
4.6
4.7
4.6
4.6
3.1
2.0
2.4
4.1
4.2
1.6
0.80
1.3
0.63
0.47
0.29
0.35
0.40
0.35
0.20
0.11
0.14
0.21 ( 0.04
0.20 ( 0.04
0.15 ( 0.05
0.18 ( 0.04
0.16 ( 0.04
0.14 ( 0.03
0.15 ( 0.05
0.11 ( 0.05
0.13 ( 0.07
50.5
50.2
51.3
132
48.8
50.5
50.1
54.3
a Error bars are (2σ.
experimental uncertainties, O3 formation is indistinguishable
from that of CH3C(O)OH. It should be noted that no correction
has been made for O3 loss via its reaction with HO2 radicals,
although given the similarity between the O3 and CH3C(O)OH
yields, this reaction appears to be of minor importance.
The flash photolysis/time-resolved UV spectroscopy experi-
ments provide additional data on the branching fraction for
reaction 1. Of the six species that contribute to the absorbance
of the reaction mixture, the five shown in Figure 1 and the
1-hydroxy ethylperoxy product of reaction 13, only two remain
in the reaction cell at long times, namely ozone and “products”.
The others are lost via the reactions in Table 1. Judging from
Figure 1, it is easy to separate the contributions of these two
components from the absorbance spectra at long time and,
thereby, obtain an ozone yield. This yield cannot be used directly
to compute the branching fraction for reaction 1 since, unlike
with the FTIR-smog chamber system, CH3C(O)O2 removal via
self-reaction and reaction with CH3O2 remains competitive with
reaction 1, owing to the much higher radical concentrations
relevant to the flash photolysis experiments.
The branching can be deduced by comparing the ozone yield
to predictions based on the kinetic model of Table 1 along with
the values of k1 and k13 listed in Table 2. An example of a fit
to ozone data is provided in Figure 4. The results of such
comparisons are presented in Table 3 along with the pertinent
experimental conditions and are illustrated by Figure 16. The
average value of k1b/k ) 0.16 ( 0.04 at 295 K is somewhat
larger than the result of k1b/k ) 0.10 ( 0.02 obtained from the
smog chamber experiments. Because of the complexity of the
reaction model used to extract the ozone branching fraction from
the kinetic data, this level of agreement should be considered
reasonable. A weighted average of k1b/k ) 0.12 ( 0.04 is
consistent with both sets of experiments. It appears from Figure
16 that the branching for the ozone channel has a slight negative
temperature dependence; however, a quantitative determination
of this effect is not warranted owing to the scatter in the data.
Figure 16. Temperature dependence of the branching fraction for
reaction 1 to produce acetic acid and ozone. The solid lines represent
linear regressions.
260 K to 2.1 at 360 K, as determined from the ratio of the
respective least-squares regressions. The “true” discrepancy is
actually about 25% larger than this because the “best” HO2 and
CH3C(O)O2 UV cross sections available to Moortgat et al.9 were
about 25% larger than the currently accepted values. On the
basis of the difficulties that we encountered in separating the
contributions of the various UV absorbing species present in
this reaction system, even when full spectra are recorded, we
suspect that the reliance on fixed single wavelength absorption
measurements in the earlier work is, at least partly, the cause
of the disagreement in rate constant values.
IV. Discussion
Rate constants for the reaction of HO2 with CH3C(O)O2 have
been reported previously by Moortgat et al.9 A comparison of
their values with the present results is presented in Figure 9.
Moortgat et al. used flash photolysis/UV absorption to measure
the rate constants and generated the peroxy radicals via reactions
2-5 from methanol and acetaldehyde precursors. Two important
differences exist, however, between their study and ours. First,
they made absorbance measurements at only a few fixed
wavelengths in contrast to the full spectra recorded in the present
study. Second, transient IR absorption is employed in the present
study to confirm the HO2 decay and determine k1.
The difficulties inherent in the fixed wavelength approach
are illustrated by Figure 17. Here, the symbols represent
absorbance versus time slices taken at three wavelengths from
the full spectral surface in Figure 3. Solid lines show the
predictions from the model of Table 1, whereas dashed lines
give the predictions based on the smaller value of k1 reported
by Moortgat et al.9 and with no reaction between HO2 and
acetaldehyde. The absorbances are calculated from the predicted
concentrations of HO2 , CH3C(O)O2 , CH3O2 , and O3 , using
the currently recommended UV cross sections.14 The absorbance
data at these three wavelengths seem to be fit better when using
the rate constant reported by Moortgat et al.9 as compared to
The comparison in Figure 9 shows our value of k1 to be larger
than the previous determination by a factor ranging from 2.8 at

Acetylperoxy + HO2 Reaction
J. Phys. Chem. A, Vol. 103, No. 3, 1999 377
fitting procedure and report the values for k1 listed in Table 2
and shown in Figure 9.
The branching of reaction 1 into peracetic acid and acetic
acid plus ozone channels has been investigated previously at
295 K by three groups, all of which report values of k1b/k1 2-3
times larger than those found here. Niki et al.,8 using a FTIR-
smog chamber system, reported k1b/k1 ) 0.25 ( 0.05. By
varying the ozone contribution to the total UV absorbance at
210, 245, and 265 nm, Moortgat et al.9 obtained k1b/k1 ) 0.33
( 0.07, independent of temperature. Finally, Horie and Moort-
gat10 derive a ratio of k1b/k1 ) 0.28 ( 0.05 from experiments
involving the photolysis of biacetyl. In contrast, the FTIR-
smog chamber and UV absorbance experiments reported here
give ratios of k1b/k1 ) 0.10 ( 0.02 and 0.16 ( 0.04, respectively.
It is also worth noting (Figure 16) that the branching fraction
of Horie and Moortgat10 exhibits a significantly stronger
temperature dependence than that found in the present study.
It is unclear why our determination of the branching fraction
differs from that of Niki et al.8 The discrepancy between the
present ratio and that of Moortgat et al.9 probably exists for the
same reasons as discussed above with regard to k1. Horie and
Moortgat10 derived the branching ratio for reaction 1 from the
ratio of the peracetic acid to ozone yield recorded from the
photolysis of biacetyl. This differs from the present determi-
nation in two ways. First, they sampled their products from a
flowing reaction mixture and analyzed the product yields by
matrix-isolated IR absorption, whereas the present experiments
relied on in situ measurements. Thus, it is possible that the
relative amounts of peracetic acid to ozone were affected by
the sampling. Second, they relied on the photooxidation of
biacetyl to produce HO2, whereas in the present study, the HO2
to CH3C(O)O2 ratio is varied by changing the relative concen-
trations of the methanol and acetaldehyde precursors. As
discussed in Section III.B, this is important because of the
complex secondary chemistry involved.
Figure 17. Comparison of predicted and measured values of the total
UV absorbance at three fixed wavelengths. Solid lines represent
predictions of the absorbance from the present model based solely on
the HO2, CH3C(O)O2, CH3O2, and O3 contributions. Dashed lines
represent predictions based on the value of k1 reported by Moortgat et
al.,9 with k13 ) 0 and k1b/k1 ) 0.33.
the presently measured value. However, when the full spectra
are deconvoluted, as in Figure 4, the situation is reversed and
our value of k1 gives a superior fit to the HO2 concentration.
Furthermore, the present determination of k1 fits the IR
measurements of HO2 concentrations, as shown in Figures 5
and 6. The principal reason for the reversal in quality of fit
between the absorbance data in Figure 17 and the concentration
data in Figures 4, 5, and 6 is that, when relying simply on
absorbances at a limited number of wavelengths, there is no
indication that the “apparent” methylperoxy contribution is
anomalously high. The contribution from CH3C(OH)HO2, which
is expected to have a UV spectrum similar to that of CH3O2, is
not accounted for; instead, it is compensated for by the lower
value of k1 . Likewise, the contribution by the “products” to
the overall absorbance was ignored in calculating the dashed
lines in Figure 17, and this, too, affects what value of k1 gives
the best fit to the absorbance data.
The preceding example is meant as an illustration of potential
problems encountered when using fixed wavelengths to analyze
a complex absorbing reaction mixture and not as a reanalysis
of the data of Moortgat et al.9 It shows that our UV absorption
data are consistent with that of Moortgat et al.; if we were to
employ the analysis at fixed wavelengths used by them, we
would arrive at essentially the same value of k1 that they did.
However, our IR measurements of HO2 (Figures 5 and 6) and
those based on deconvolution of the UV data (Figure 4) show
it to decay substantially faster than can be accounted for by the
value of k1 reported by Moortgat et al.9 In the absence of the
postulated reaction between HO2 and acetaldehyde, a rate
constant roughly 5 times larger becomes necessary to describe
at least the initial HO2 decay. We believe that our kinetics
measurements made as a function of [CH3CHO]0 and the smog
chamber observations of a higher rate of CH3CHO loss relative
to CH3OH in air versus N2, while not conclusive, are sufficient
to support the hypothesis of a reaction between HO2 radicals
and acetaldehyde. Therefore, we include this reaction in our
A simple estimate can be made regarding the atmospheric
importance of a reaction between hydroperoxy radicals and
acetaldehyde. The major atmospheric removal mechanism for
acetaldehyde is expected to be via hydrogen abstraction by OH
radicals, which proceeds with a rate constant of 1.4 × 10-11
cm3 s-1 at 295 K.15 Except under unusual circumstances, the
reaction with HO2 is not expected to change this. Whereas the
tropospheric HO2 concentrations can exceed those of OH (which
range33 from 0.4 to 9 × 106 cm-3) by 2 orders of magnitude,
the rate constant for acetaldehyde removal by HO2 is ap-
proximately 500 times smaller than that for its loss via reaction
with OH. Thus, the HO2 reaction should represent a minor
removal pathway for acetaldehyde, and the CH3C(OH)HO2
radical should be a minor degradation product. Since the
reactions with OH and HO2 both exhibit negative temperature
dependences, this would remain the case as a function of
temperature and, hence, altitude.
The rate constant for the HO2 + CH3C(O)O2 reaction differs
considerably from the version employed by Stockwell et al.6 to
model atmospheric PAN concentrations. In their second genera-
tion regional acid deposition mechanism34 (RADM), these
authors take the expression 7.7 × 10-14e1300/T cm3 s-1 to describe
the kinetics of reaction 1. Interestingly, the temperature
dependence coincides very closely with our measurement.
However, the magnitudes of the rate constants reported here
are over 5 times larger than those used in their RADM. This is
not the only acetylperoxy reaction for which newly measured
rate constants differ from those employed by Stockwell et al.34
Recent data for the competing reaction with NO give rate

378 J. Phys. Chem. A, Vol. 103, No. 3, 1999
Crawford et al.
constants about twice as large as used in their model,13,35,36
whereas a recent investigation of the acetylperoxy self-reaction
leads to a significantly stronger temperature dependence and
rate constants ranging from 5 to 8 times larger than those in
the RADM.34 How these changes affect model predictions of
the nighttime PAN decomposition relative to ozone loss awaits
detailed calculations that are beyond the scope of the present
paper. However, the increased reactivity of the acetylperoxy
radical with HO2 and other peroxy radicals that has been
measured recently suggests that closer scrutiny be given to the
suggestion by Stockwell et al.6 that acetylperoxy/peroxy chem-
istry as opposed to differential deposition velocities could
explain the relative PAN versus O3 removal rates.
(15) DeMore, W. B.; Sander, S. P.; Golden, D. M.; Hampson, R. F.;
Kurylo, M. J.; Howard, C. J.; Ravishankara, A. R.; Kolb, C. E.; Molina,
M. J. Chemical Kinetics and Photochemical Data for Use in Stratospheric
Modeling; JPL: Pasadena, CA, 1997; Publication 97-4.
(16) Maricq, M. M.; Szente, J. J. J. Phys. Chem. 1994, 98, 2078.
(17) Wallington, T. J.; Japar, S. M. J. Atmos. Chem. 1989, 9, 399.
(18) Tyndall, G. S.; Orlando, J. J.; Wallington, T. J.; Hurley, M. D. Int.
J. Chem. Kinet. 1997, 29, 655.
(19) Chao, J.; Zwolinski, B. J. J. Phys. Chem. Ref. Data 1978, 7, 363.
(20) Unnumbered reactions in this table are of minor importance to the
determination of k1 . They are primarily included for completeness and for
attempts to understand the anomalous nature of the “methylperoxy”
concentrations determined from the UV absorbance measurements.
(21) Crawford, M. A.; Szente, J. J.; Maricq, M. M.; Francisco, J. S. J.
Phys. Chem. 1997, 101, 5337.
(22) Delcroix et al. CMD meeting, Zurich, 1997.
(23) Maricq, M. M.; Szente, J. J. J. Phys. Chem. 1996, 100, 4507.
(24) Roehl, C. M.; Bauer, D.; Moortgat, G. K. J. Phys. Chem. 1996,
100, 4038.
(25) Moortgat, G. K.; Cox, R. A.; Schuster, G.; Burrows, J. P.; Tyndall,
G. S. J. Chem. Soc., Faraday Trans. 2 1989, 85, 809.
(26) Veyret, B.; Lesclaux, R.; Rayez, M.-T.; Rayez, J.-C.; Cox, R. A.;
Moortgat, G. K. J. Phys. Chem. 1989, 93, 2368.
(27) Andersson, B. Y.; Cox, R. A.; Jenkin, M. E. Int. J. Chem. Kinet.
1988, 20, 283.
(28) Atkinson, R.; Baulch, D. L.; Cox, R. A.; Hampson, R. F.; Kerr, J.
A.; Troe, J. J. J. Phys. Chem. Ref. Data. 1997, 26, 521.
(29) Wallington, T. J.; Hurley, M. D.; Shi, J.; Maricq, M. M.; Sehested,
J.; Nielsen, O. J.; Ellermann, T. Int. J. Chem. Kinet. 1993, 25, 651.
(30) Wallington, T. J.; Hurley, M. D. Chem. Phys. Lett. 1992, 189, 437.
(31) Koch, S.; Moortgat, G. K. Chem. Phys. Lett. 1990, 173, 151.
(32) Wallington, T. J.; Japar, S. M. Chem. Phys. Lett. 1990, 166, 495.
(33) Wayne, R. P. Chemistry of Atmosphere; Clarendon Press: Oxford,
1991.
References and Notes
(1) Stephens, E. R. AdV. EnViron. Sci. Technol. 1969, 1, 119.
(2) Singh, H. B.; O’Hara, D.; Herlth, D.; Bradshaw, J. D.; Sandholm,
S. T.; Gregory, G. L.; Sachse, G. W.; Blake, D. R.; Crutzen, P. J.;
Kanakidou, M. A. J. Geophys. Res. 1992, 97, 16511.
(3) Roberts, J. M. Atmos. EnViron. 1990, 24A, 243.
(4) Shepson, P. B.; Hastie, D. R.; So, K. W.; Schiff, H. I.; Wong, P.
Atmos. EnViron. 1992, 26A, 1259.
(5) Hartsell, B. E.; Aneja, V. P.; Lonneman, W. A. J. Geophys. Res.
1994, 99, 21033.
(6) Stockwell, W. R.; Milford, J. B.; Dongfen, G.; Yang, Y.-J. Atmos.
EnViron. 1995, 29, 1591.
(7) Gaffney, J. S.; Marley, N. A.; Drayton, P. Private communication.
(8) Niki, H.; Maker, P. D.; Savage, C. M.; Breitenbach, L. P. J. Phys.
Chem. 1985, 89, 588.
(9) Moortgat, G. K.; Veyret, B.; Lesclaux, R. Chem. Phys. Lett. 1989,
160, 443.
(10) Horie, O.; Moortgat, G. K. J. Chem. Soc., Faraday Trans. 1992,
88, 3305.
(11) Maricq, M. M.; Wallington, T. J. J. Phys. Chem. 1992, 96, 986.
(12) Maricq, M. M.; Szente, J. J.; Kaiser, E. W. J. Phys. Chem. 1993,
97, 7970.
(13) Maricq, M. M.; Szente, J. J. J. Phys. Chem. 1996, 100, 12380.
(14) Lightfoot, P. D.; Cox, R. A.; Crowley, J. N.; Destriau, M.; Hayman,
G. D.; Jenkin, M. E.; Moorgat, G. K.; Zabel, F. Atmos. EnViron. 1992,
26A, 1805.
(34) Stockwell, W. R.; Middleton, P.; Chang, J. S.; Tang, X. J. Geophys.
Res. 1990, 95, 16343.
(35) Villalta, P. W.; Howard, C. J. J. Phys. Chem. 1996, 100, 13624.
(36) Sehested, J.; Christensen, L. K.; Møgelberg, T. E.; Nielsen, O. J.;
Wallington, T. J.; Guschin, A.; Orlando, J. J.; Tyndall, G. S. J. Phys. Chem.,
in press.