The CH3C(O)O2 Radical. Its UV Spectrum, Self-Reaction Kinetics, and Reaction with CH3O2

Article abstract of DOI:10.1021/jp9533234

The reactions of CH3C(O)O2 with itself and with CH3O2 are investigated using flash photolysis combined with time-resolved UV spectroscopy and transient infrared absorption.The UV spectrum of CH3C(O)O2 exhibits two bands; the stronger short wavelength component has a maximum cross section of 6.5E-18 cm² at 206 nm and the weaker one a cross section of 2.9E-18 cm² at 250 nm.These bands are used to monitor the disappearance of CH3C(O)O2 and the secondary formation of CH3O2, yielding a self-reaction rate constant of (3.0 +1.7/-1.1)E-12 e^{(504 +/- 114)/T} cm³ s^-1 and a cross reaction rate constant of (8.5 +2.6/-2.0)E-13 e^{(726 +/- 75)/T} cm³ s^-1.Transient IR monitoring of the formation of CH2O reveals CH3C(O)O2 + CH3O2 -> CH2O + CH3COOH + O2 to be the dominant channel for the cross reaction over the entire 209-358 K temperature range.This contrasts with previous studies that indicate a steep temperature dependence for the branching ratio, with a channel leading to CH3C(O)O + CH3O + O2 dominant at high temperatures.

Full text of DOI:10.1021/jp9533234

J. Phys. Chem. 1996, 100, 4507-4513
4507
The CH3C(O)O2 Radical. Its UV Spectrum, Self-Reaction Kinetics, and Reaction with
CH3O2
M. Matti Maricq* and Joseph J. Szente
Research Laboratory, Ford Motor Company, P.O. Box 2053, Drop 3083, Dearborn, Michigan 48121
X
ReceiVed: NoVember 10, 1995; In Final Form: December 12, 1995
The reactions of CH
with time-resolved UV spectroscopy and transient infrared absorption. The UV spectrum of CH
exhibits two bands; the stronger short wavelength component has a maximum cross section of 6.5 × 10
cm at 206 nm and the weaker one a cross section of 2.9 × 10 cm at 250 nm. These bands are used to
3
C(O)O
2
with itself and with CH
O
3 2
are investigated using flash photolysis combined
3
C(O)O
2
18
-
2
-18
2
monitor the disappearance of CH
3
C(O)O
2
and the secondary formation of CH
3
O
2
, yielding a self-reaction
rate constant of (3.0 ) × 10 e(504(114)/T cm s and a cross reaction rate constant of (8.5-2.0) ×
+1.7
-12
3
-1
+2.6
-
1.1
-
13 (726(75)/T
3
-1
10
e
cm s . Transient IR monitoring of the formation of CH
2
O reveals CH
3
C(O)O
2
3 2
+ CH O f
CH
temperature range. This contrasts with previous studies that indicate a steep temperature dependence for the
branching ratio, with a channel leading to CH C(O)O + CH O + O dominant at high temperatures.
2
O + CH
3
COOH + O to be the dominant channel for the cross reaction over the entire 209-358 K
2
3
3
2
I. Introduction
and at three temperatures by Moortgat et al., with the values at
98 K disagreeing by up to a factor of 7. The kinetics of the
2
Organic peroxy radicals represent an important class of
cross reaction with CH3O2 has only been investigated by
intermediates formed in the course of the atmospheric oxidation
5
Moortgat et al.
of hydrocarbon compounds.1
,2
One of the more notable
The kinetic measurements reported in the present paper make
use of time-resolved UV spectroscopy to disentangle the
secondary chemistry that complicates the study of the acetyl-
peroxy radical self-reaction. The secondary complications arise
because the CH3C(O)O product of the self-reaction,
constituents of this class of radicals is the acetylperoxy radical.
It is formed as a consequence of the photooxidation of a variety
of carbonyl compounds, either from their direct emission into
the atmosphere or from their appearance as secondary products
of hydrocarbon oxidation. The acetylperoxy radical is a
significant contributor to smog formation owing to the reaction
that it undergoes with NO2 to form peroxyacetylnitrate (PAN).
Furthermore, this reaction is important to the understanding of
the ozone photochemical steady state since, due to its relative
chemical and thermal stability, the formation of PAN effectively
sequesters NO2.
The self-reaction of acetylperoxy radicals, while of minor
consequence in the atmosphere, is often an important consid-
eration in laboratory studies. The reason is that, under the high
radical concentrations typical of laboratory measurements, the
self-reaction competes with the reaction under investigation,
whether it be PAN formation or the removal of acetylperoxy
radicals by reaction with NO, HO2, or other peroxy radicals.
Clearly, a good understanding of the self-reaction kinetics is
required to extract useful information about the reactions of
acetylperoxy radicals with these important atmospheric trace
species. Furthermore, owing to significant secondary chemistry,
the investigation of the self-reaction also yields information
about the atmospherically more relevant cross reaction between
acetylperoxy and methylperoxy radicals.
CH C(O)O + CH C(O)O f 2CH C(O)O + O (1)
3
2
3
2
3
2
rapidly dissociates, leading in the presence of molecular oxygen
to CH3O2 formation:
CH C(O)O f CH + CO
2
(2)
(3)
3
3
CH + O + M f CH O + M
3
2
3
2
Because the UV spectrum of methylperoxy radical significantly
overlaps that of acetylperoxy radical, a simple analysis of the
change in the UV absorption of the reaction mixture with time
fails to yield the desired rate constant.
A further difficulty arises from the possibility that two
channels exist for the cross reaction between acetylperoxy and
methylperoxy radicals:
CH C(O)O + CH O f CH C(O)O + CH O + O (4a)
3
2
3
2
3
3
2
CH C(O)O + CH O f CH COOH + CH O + O (4b)
3
2
3
2
3
2
2
There are relatively few previous studies of the acetylperoxy
UV spectrum, of its self-reaction, or of its reaction with
methylperoxy radicals. An examination of the literature reveals
that the previously published values for these quantities exhibit
sizable discrepancies. UV cross sections of the acetylperoxy
In fitting their UV absorption vs time profiles, Moortgat et al.5
found it necessary to postulate that reaction 4b dominates at
low temperatures (<295 K) but that, due to a steep temperature
dependence of the branching ratio, reaction 4a becomes the more
important one at high temperatures (368 K). In a subsequent
3
4
radical by Addison et al., Basco and Parmar, and Moortgat et
6
5
product study, Horie and Moortgat refined the branching ratio
al. differ both in their wavelength dependence as well as in
but still found it to be steeply temperature dependent. The
present study provides acetylperoxy UV cross sections and rate
constants for reactions 1 and 4 in good agreement with Moortgat
et al. , but the fits of the kinetic data do not reproduce the steep
temperature variation in the branching ratio of reactions 4a and
their absolute intensities. The self-reaction rate constant has
been measured at ambient temperature by the first two groups
5
*
Author for correspondence.
Abstract published in AdVance ACS Abstracts, February 15, 1996.
X
0022-3654/96/20100-4507$12.00/0 © 1996 American Chemical Society

4
508 J. Phys. Chem., Vol. 100, No. 11, 1996
Maricq and Szente
4
b. Instead, transient infrared absorption experiments reveal
that CH2O is formed immediately and suggest that channel 4b
dominates over the entire 232-348 K temperature range. A
more complete discussion of this result, as well as of the self-
reaction kinetics, is provided following a description of the
experimental techniques and the presentation of the results.
II. Experimental Section
1
4
Acetylperoxy radicals, at concentrations of 5 × 10 to 1.4
15
-3
×
10 cm , are generated by 351-nm excimer laser photolysis
of Cl2 in the presence of acetaldehyde and oxygen via the
reactions
Cl + CH CHO f CH CO + HCl
(5)
(6)
3
3
Figure 1. Time-resolved UV spectra following the photolysis of Cl
in the presence of CH CHO and O . The change in the shape of the
UV spectrum with time indicates the formation of a secondary UV
absorbing species, namely CH
2
CH CO + O + M f CH C(O)O + M
3
2
3
2
3
2
The radicals are produced in an approximately cylindrical
volume 51 cm long and roughly 1.5 cm in diameter (based on
the excimer laser beam shape), the central 0.8 cm of which is
sampled by the probe beam. For the acetylperoxy UV absorp-
tion cross-section measurements, the acetaldehyde concentration
was varied from 0.5 to 1.7 Torr and the oxygen concentration
between 4 and 47 Torr in order to ascertain conditions under
which the photolyzed chlorine atoms are stoichiometrically
converted to CH3C(O)O2. The kinetics experiments were
carried out with acetyladehyde concentrations of 0.5-1.6 Torr
and an oxygen partial pressure of ∼22 Torr, values that ensure
that reactions 5 and 6 reach completion before the radical-
radical reactions can proceed to any significant extent.
3 2
O .
HgCdTe detector having a time response of 0.3 µs. CH O is
2
monitored via vibration-rotation lines of the ν carbonyl stretch
2
-
1
-18
2
mode at 1729 cm (σIR ) 3.6 × 10 cm at 30 Torr and 345
K) and 1766 cm- (σIR ) 2.7 × 10 cm at 110 Torr and 345
1
-18
2
K). Acetaldehyde also absorbs at these frequencies (σ = 5
IR
-19
2
× 10
cm ); however, at the 30-110-Torr total pressure of
these experiments, the individual rotational transitions are not
resolved; rather, the spectrum is nearly continuous over regions
of the width of the formaldehyde lines. Therefore, a comparison
of transient absorptions on vs off the CH O resonances enables
2
the contributions from formaldehyde vs acetaldehyde to be
The reaction is followed by monitoring the absorbance due
to CH3C(O)O2 and CH3O2 using time-resolved UV spectros-
copy, a detailed description of which is provided in refs 7 and
distinguished.
Acetaldehyde was obtained from Aldrich (99%) or Fisher
(99.5%) and freeze-pump-thawed prior to use. Nitrogen
(99.999%) and oxygen (ultrazero grade) were obtained from
Michigan Airgas. Chlorine was obtained from Matheson as a
9.7% mixture in helium. Gas flows are controlled by Tylan
flow controllers, except for chlorine which is set by a needle
valve. Partial pressures are determined by timing the pressure
rise into a fixed volume. Temperature control is achieved using
a Neslab ULT-80dd recirculating chiller. The gases are
precooled/preheated prior to entering the reaction vessel.
8
. Spectra are recorded at various delay times following the
photolysis pulse using probe light from a cw deuterium lamp
and a gated diode array spectrometer. They are deconvoluted
by comparison against reference spectra of CH3C(O)O2 and
CH3O2 along with the residual absorbance found after the
reactions initiated by photolysis have gone to completion; thus,
Abs(λ,t) ) l(σ
(λ)[CH C(O)O ] +
3 2 t
CH C(O)O
3
2
σ
(λ)[CH O ] + c(t)Abs(λ,∞)) (7)
3 2 t
CH O
3
2
III. Results
where l represents the path length, σi(λ) is the cross section of
component i, and [i]t denotes the time-dependent concentration
of component i. The last term accounts for the accumulation
of UV-absorbing reaction products. Abs(λ,∞) represents the
absorbance of these products, and c(t) is proportional to their
concentration.
A. UV Spectrum. Figure 1 illustrates the variation in the
UV spectrum of a CH3CHO/Cl2/O2/N2 gas mixture with the
delay time after chlorine photolysis. The spectrum at 10 µs
reveals two broad and apparently unstructured absorption
features centered at approximately 210 and 250 nm. As time
progresses, the short-wavelength band decreases rapidly in
intensity, whereas the long-wavelength band initially remains
at nearly constant intensity and only later decreases in amplitude.
Clearly the spectra do not arise from absorption by a single
compound. Instead, the temporal behavior of the spectra can
be explained by reactions 1-4. The early spectrum is almost
entirely due to absorption by CH3C(O)O2. Each molecule lost
by self-reaction is replaced, essentially instantaneously, by
CH3O2 owing to the rapid dissociation of CH3C(O)O and the
facile addition of O2 to the CH3 fragment. Methylperoxy
radicals absorb in the UV in a single, broad, Gaussian-shaped
band centered at 238 nm. The lack of a CH3O2 feature at 210
nm explains why the absorbance in this spectral region decreases
rapidly whereas the absorbance at 250 nm does not.
Initial radical concentrations are found by substituting ethane
for acetaldehyde and measuring, via the method just described,
the formation of C2H5O2 radicals. This is a convenient and
accurate method for determining the yield of Cl atoms from
photolysis since the UV cross section of C2H5O2 has been well
8
-10
established,
the reactions forming ethylperoxy radicals are
rapid, and the removal of the radicals via self-reaction is very
1
,2
slow.
This procedure is used both to assign absolute cross
sections to the acetylperoxy UV spectrum and to provide initial
radical concentrations for fitting model predictions to the kinetic
data.
Additional experiments were carried out to measure, via a
previously described transient IR absorption method,1 the rate
and extent of formaldehyde production from the Cl-initiated
oxidation of acetaldehyde. IR light from a Pb-salt diode laser
is directed through the reaction mixture and detected by a
1
Because of the fast decay of CH3C(O)O2 and the interference
from the CH3O2 that is produced, care must be taken in
measuring the acetylperoxy spectrum and assigning to it absolute

Kinetics of the CH3C(O)O2 Radical
J. Phys. Chem., Vol. 100, No. 11, 1996 4509
TABLE 1: CH
3
C(O)O
2
Self-Reaction Mechanism
reactiona
rate constant
-
12
3
-1 b
1
2
3
4
4
9
9
.
.
.
a.
b.
a.
b.
CH
CH
CH
CH
CH
CH
CH
3
3
3
3
3
3
3
C(O)O
C(O)O f CH
2
+ CH
3
C(O)O
+ CO
2
f 2CH
3
C(O)O + O
2
k ) 3.0 × 10 e504/T cm s
6
-1
3
2
k > 1 × 10 s
k ) 4.5 × 10 31(T/300)-3 cm s
-
6
-1 15
+ O
2
+ M f CH
2
3
O
2
+ M
3
-1 b
C(O)O
C(O)O
+ CH
+ CH
3
O
O
2
f CH
f CH
3
C(O)O + CH
COOH + CH
3
O + O
2
2
k ) 0 cm s
-
3 -1 b
2
3
2
3
O + O
2
k ) 8.5 × 10 13e726/T cm s
-14 416/T
3
-1 1,2
O
O
2
+ CH
+ CH
3
O
O
2
f 2CH
f CH OH + CH
3
O + O
2
k ) (1 - b) × 9.1 × 10
e
cm s
-
14 416/T
3
-1 1,2
2
3
2
3
2
O + O
2
k ) b × 9.1 × 10
e
cm s
-
1165/T -1
b ) (1 + 25e
)
-12
3 -1 12
1
1
1
0.
1.
2.
CH
CH
CH
3
3
3
O
2
+ CH
3
O f products
O f products
k ) 1.5 × 10 cm s
-11 3 -1 12
O + CH
O + O
3
k ) 3 × 10 cm s
-14 -900/T
3 -1 12
2
f CH O + HO
2
2
k ) 3.9 × 10
e
cm s
a
Reaction numbers correspond to those used in the text. b Measured in the present study.
absorption cross sections. The measurements presented here
were accomplished in two stages: (1) ascertaining conditions
under which the chlorine atoms formed by photolysis are
quantitatively converted to acetylperoxy radicals and (2) cor-
recting the raw spectrum for the short time delay (5 µs) between
photolysis and spectral measurement.
CH3C(O)O2 is formed via reactions 5 and 6, i.e., attack of
acetaldehyde by chlorine atoms followed by oxygen addition
to the acyl radical. The first step can be made rapid by the
-
11
addition of a sufficient quantity of CH3CHO (k5 ) 7.6 × 10
cm s at 295 K implies a lifetime of 0.4 µs at 1 Torr).12 Too
much will enhance the chain reaction propagated by the reaction
between CH3CO and Cl2; however, this fortunately has little
effect on the quantity of CH3C(O)O2 produced. Similarly, the
oxygen concentration can be increased to enhance the rate of
3 -1
Figure 2. UV absorption spectrum of CH
3
C(O)O
2
. The spectrum
-
12
3
-1
the O2 addition reaction (k6 ) 5 × 10
cm s at 295 K
measured in the present work is given by the solid line. The dashed
line represents a fit of the spectrum to two Gaussian-shaped absorption
bands (see text). The symbols represents results from previous studies.
implies a lifetime of 0.6 µs at 10 Torr).12 If the [O2]/[CH3-
CHO] ratio is too high, however, peroxy radicals are formed
before all of the chlorine atoms have been consumed. As a
consequence, the very rapid reaction typically found between
peroxy radicals and chlorine atoms13 can significantly reduce
the conversion of Cl to CH3C(O)O2. The quantities of both
acetaldehyde and oxygen were varied in order to locate a plateau
region over which the CH3C(O)O2 yield shows little dependence
on changes in their concentrations. This region occurs for [CH3-
CHO] > ∼0.5 Torr and [O2] > ∼10 Torr.
B. CH3C(O)O2 Reaction Kinetics. Time-dependent con-
centrations of CH3C(O)O2 and CH3O2 are calculated from time-
resolved spectra, such as those shown in Figure 1, by fitting
them to a superposition of the UV cross sections of the peroxy
radicals plus a fraction of the residual absorbance recorded at
20 ms, as per eq 7. The residual absorbance accounts for stable
species, such as CH2O and CH3COOH, that are generated from
the photooxidation of acetaldehyde and absorb weakly in the
UV. An example of the resulting concentration vs time profiles
is provided by Figure 3. These are then fit to the reaction model
of Table 1 in order to extract values for the unknown rate
constants k1 and k4.
Spectra recorded with the above conditions undergo a
variation in time such as depicted in Figure 1. The raw
spectrum, measured at 5 µs, is corrected to zero time by the
following procedure. Using the presently measured values of
k1 and k4, predictions from the reaction model of Table 1 indicate
that, for an initial CH3C(O)O2 concentration of 1015 cm , 14%
of the molecules are converted to CH3O2 in the intervening 5
µs. The contribution from the methylperoxy radicals is
subtracted using the known wavelength-dependent cross sections
-3
The principal reactions required to explain the concentration
dependences depicted in Figure 3 are reactions 1 and 4. The
self-reaction of CH C(O)O determines the initial decay rate
3
2
of this species. On the time scale of the radical-radical
8
of this species, and the remainder is increased by 14% to
reactions, the dissociation of the CH C(O)O formed by reaction
3
compensate for the loss of CH3C(O)O2. The resulting acetyl-
peroxy spectrum is shown by the solid line in Figure 2 and
compared to previously measured spectra.
1 and the subsequent addition of O to the methyl radical
2
fragment occur very rapidly. Thus, under our experimental
conditions, each CH C(O)O lost by self-reaction appears to
3
2
The broad featureless spectra of peroxy radicals are often well
3 2
be instantly replaced by a CH O radical, as is evident in Figure
fit by a Gaussian shape of the form14
3 from the fact that the initial rate of CH O rise matches the
3
2
rate of CH3C(O)O2 decay.
2
σ(λ) ) σ exp(-R[ln(λ /λ)] )
(8)
The observation that the methylperoxy level does not rise to
the initial acetylperoxy concentration indicates that it is lost by
subsequent reactions. The self-reaction of CH3O2 alone is too
slow to account for this, and it does not explain the non-second-
order decay of CH3C(O)O2. Instead, one postulates the exist-
ence of a cross reaction between the two peroxy radicals.
m
m
Assuming that the CH3C(O)O2 spectrum consists of two such
bands, a good fit of the spectrum, as illustrated by the dashed
line in Figure 2, is obtained with the parameters σ1m ) 6.5 ×
-
18
2
-18
1
0
cm , λ1m ) 206 nm, R1 ) 119 and σ2m ) 2.9 × 10
2
5
cm , λ2m ) 250 nm, R2 ) 81. The parameterization according
to eq 8 affords a compact representation of the UV cross sections
of CH3C(O)O2.
According to the previous work of Moortgat et al. and Horie
6
and Moortgat two channels are possible: reactions 4a and 4b.
The principal difference between the two, for the purpose of

4
510 J. Phys. Chem., Vol. 100, No. 11, 1996
Maricq and Szente
Figure 4. Variation of CH
3
C(O)O
2
self-reaction rate constant with
temperature. Error bars represent deviations of 2σ. Filled circles show
the present measurements assuming b ) 1. Crosses indicate the rate
constants determined with b derived from the branching ratio of Horie
3 2 3 2
Figure 3. Time dependence of the CH C(O)O and CH O concentra-
tions obtained by deconvoluting time-resolved UV spectra. Error bars
represent 2σ of the fitting error. Both sets of concentrations are fit
1 4
simultaneously for the two unknowns k and k . Solid lines illustrate
6
and Moortgat, eq 13. Open triangles indicate the values obtained by
5
Moortgat et al. The solid line gives the best fit of the present data to
the Arrhenius expression.
best fits of the reaction model to the data using a branching fraction of
b ) 1. Dashed lines show the best fits for the value of b ) 0.06
6
-11
3
obtained from Horie and Moortgat, for which k
1
) 1.3 × 10 cm
-1
-11
3
-1
s
and k
4
) 2.2 × 10 cm s
.
fitting the RO2 concentration vs time profiles, is that channel
b causes a net loss of CH3O2, whereas channel 4a does not;
4
the methylperoxy lost on the left-hand side of (4a) is replaced,
after reactions 2 and 3, by the CH3C(O)O produced on the right
side. As demonstrated below, the branching ratio has a minor
influence on the value of k1 but a major effect on k4.
Whereas channel 4b yields stable products, channel 4a and
channel 9a of the methylperoxy self-reaction,
CH O + CH O f 2CH O + O
2
(9a)
(9b)
3
2
3
2
3
CH O + CH O f CH OH + CH O + O
2
3
2
3
2
3
2
produce methoxy radicals. These are formed too late to have
anything but a negligible effect on the acetylperoxy concentra-
tion; however, they participate in secondary reactions via
Figure 5. Variation of the CH
3
C(O)O
2
3 2
+ CH O reaction rate constant
with temperature. Error bars represent deviations of 2σ. Filled circles
show the present measurements assuming b ) 1. Crosses indicate the
rate constants determined with b derived from the branching ratio of
CH O + CH O f products
(10)
(11)
(12)
3
2
3
6
Horie and Moortgat. Open triangles indicate the values obtained by
CH O + CH O f products
3
3
5
Moortgat et al. The solid line gives the best fit of the present data to
the Arrhenius expression. Dashed line gives best fit of data (×’s) to
a sum of two Arrhenius expressions related by â from eq 13 assuming
that the branching ratio for reaction 4 is temperature dependent.
CH O + O f CH O + HO
2
3
2
2
Very little is known about reaction 10; a channel leading to
CH2O and CH3OOH may constitute about 20% of the total
100% of the reaction proceeds via channel 4a; b ) 1 implies
that it goes entirely by (4b). Allowing all three parameters to
vary always gives a best fit with b ) 1 over the entire
temperature range of the study, 209-358 K. Unfortunately, a
drawback of the three-parameter fit is that it leads to large error
bars for both b and the cross reaction rate constant. An
alternative procedure is, therefore, adopted. Two parameter fits
are performed by setting b ) 1 and, for comparison, setting b
1
2
products. Somewhat more information is available for the
methoxy self-reaction.12 At ambient temperatures and above,
the predominant products may be CH2O and CH3OH; however,
the recombination reaction could become important at low
temperature and high pressure. Fortunately, reactions 10 and
1
1 have only a minor influence on the determinations of k1 and
k4. The reaction of CH3O with O2 is well-known but is included
here only for completeness, since at the O2 concentrations of
the present study this reaction proceeds too slowly to affect the
kinetics measurements.
Fits of the reaction model to the time-dependent RO2
concentrations involves three unknowns: k1, k4, and b, the
branching fraction of reactions 4a and 4b. b ) 0 implies that
6
to the value obtained from Horie and Moortat. Section IIIC
presents evidence supporting the b ) 1 value favored by the
three-parameter fits of the RO2 concentration vs time data.
The temperature-dependent rate constants derived by the
above fitting procedure are illustrated in Figures 4 and 5. The
preferred results, with b ) 1, are also given in Table 2 along

Kinetics of the CH3C(O)O2 Radical
J. Phys. Chem., Vol. 100, No. 11, 1996 4511
TABLE 2: Measured Rate Constants
conditions
results
, 1014 cm-3
k
, 10-11 cm s
3
-1
k
, 10-11 cm s
3
-1
temp, K
CH
3
CHO, Torr
Cl
2
, Torr
O
2
, Torr
P
tot, Torr
[Cl]
o
1
4
209
216
224
226
233
253
273
295
295
327
358
1.2
1.2
1.3
1.2
1.5
0.76
1.5
1.5
0.52
1.5
1.6
0.26
0.53
0.21
0.28
0.51
0.47
0.43
0.56
0.59
0.35
0.38
22
23
23
22
23
23
22
22
13
23
24
106
110
108
105
112
110
109
112
112
112
120
7.3
13.6
5.2
7.7
13.5
6.2
10.5
12.4
7.2
7.3
6.4
3.2 ( 0.5
2.7 ( 0.4
3.5 ( 0.4
2.8 ( 0.5
2.2 ( 0.3
2.6 ( 0.5
2.0 ( 0.3
1.6 ( 0.4
1.5 ( 0.2
1.4 ( 0.3
1.2 ( 0.2
3.0 ( 0.8
2.3 ( 0.4
2.0 ( 0.4
2.0 ( 0.4
1.8 ( 0.3
1.6 ( 0.5
1.3 ( 0.2
1.0 ( 0.2
1.1 ( 0.15
0.76 ( 0.25
0.58 ( 0.15
with the experimental conditions. The CH3C(O)O2 self-reaction
Figure 4) has a negative temperature dependence and is little
(
affected by the choice of branching fraction. The largest effect,
at 358 K, is an increase of 26% upon changing b from 0 to 1.
The cross reaction (Figure 5) also exhibits a negative temper-
ature dependence for b ) 1; however, when b is set to the
previously reported temperature-dependent values, k4 increases
with increasing temperature between about 260 and 295 K
before returning to the negative temperature dependence.
Errors in the rate constants (aside from the effect of b) derive
primarily from two sources: noise and uncertainty in the peroxy
radical UV cross sections. A comparatively negligible contribu-
tion arises from uncertainties in the secondary chemistry,
reactions 9-12. The noise in the concentration vs time data
introduces on average a 15% fitting error for k1 and a 20% error
for k4. The methylperoxy and ethylperoxy UV spectra have
been well studied, and their intensities are known to within about
5
%. The uncertainty in the acetylperoxy UV cross sections is
somewhat higher, about 10%, owing to the corrections neces-
sitated by its rapid self-reaction. However, the fact that both
the CH3C(O)O2 and CH3O2 data can be fit simultaneously
suggests that these cross sections are internally consistent.
Adding statistically a roughly 10% contribution from uncertain-
ties in the reference UV spectra to the noise error yields the
overall errors of 12-22% for k1 and 14-33% for k4 listed in
Table 2.
Figure 6. Time dependence of CH O formation upon photolysis of a
2
CH
the predictions of the reaction model using b ) 1; the dashed line shows
CH O formation for b ) 0.03, assuming that the reactions of CH
with itself and with CH produce CH O at 100% yield. The dotted
line assumes no CH O is produced by reaction 10; the dot-dash line
assumes no CH O from either reaction 10 or 11. The inset compares
3 2 2 2
CHO/Cl /O /N gas mixture at 345 K. The solid line illustrates
2
3
O
O
3 2
2
C. CH3C(O)O2 + CH3O2 Product Channel. It is apparent
that an accurate determination of the acetylperoxy plus meth-
ylperoxy reaction rate constant requires information about the
branching of this reaction. The product study of Horie and
2
2
the transient IR absorption with the diode laser locked to a CH O
2
absorption line vs an off-resonant transient absorption (see text).
6
5
Moortgat and the kinetic measurements of Moortgat et al.
indicate the branching ratio to have a steep temperature
dependence, with
CH2O was probed by vibration-rotation lines of the carbonyl
stretch mode. These lines are overlapped by the corresponding
band of the acetaldehyde precursor. However, at the 30-110
Torr total pressures of these experiments, the latter absorption
features are considerably broader than the CH2O lines. The
inset to Figure 6 illustrates the raw changes in IR absorption
following Cl2 photolysis. Off the CH2O resonance there is an
abrupt increase in the transmitted light intensity which occurs
because acetaldehyde is removed by reaction with chlorine
atoms. With the IR probe beam tuned to the CH2O resonance,
there is a nearly identical abrupt increase in transmitted light;
however, this is followed by an increase in absorption with a
half-life of approximately 200 µs. The abrupt increase in light
intensity is, again, due to loss of acetaldehyde, whereas the
subsequent loss of transmitted light arises from the formation
of formaldehyde. In order to ensure that the absorption is indeed
due to CH2O and not because of an accidental spectral
coincidence with another molecule, the measurements were
6
-3870/T
â ) k /k ) 2.2 × 10 e
(13)
4
a
4b
In contrast, the present time-resolved peroxy radical measure-
-
1
ments are best fit with b ) (1 + â) ) 1. Although the fits
with b ) 1 are superior to those based on eq 13, this is not
sufficient to rule out the possibility that setting b ) 1
compensates for some other shortcoming of the reaction model
or for a systematic error in the experimental measurements. As
an independent means to investigate the branching ratio, time-
dependent measurements of CH2O formation were undertaken,
since formaldehyde is produced by channel 4b but not channel
4
a. This procedure is not as ideal as might be hoped because
reactions 9b and 10-12 also form CH2O. To make matters
worse, the increase in k4, found upon setting b to the value of
6
Horie and Moortgat, increases the CH2O contribution from the
-1
latter reactions, making it more difficult to distinguish between
reactions 4a and 4b.
repeated at two well-separated frequencies, 1729 and 1766 cm ,
with consistent results.

4
512 J. Phys. Chem., Vol. 100, No. 11, 1996
Maricq and Szente
Using measured IR line strengths, CH2O absorbance traces
ments. However, calibration differences alone cannot account
for the fact that the spectral shapes are different; i.e., the relative
intensities at 207 nm vs 260 nm are different. The latter
occurrence likely arises from uncertainites in correcting the
acetylperoxy spectrum for the conversion of CH3C(O)O2 to
CH3O2, a procedure required by both studies.
are converted to concentration vs time profiles, such as the one
illustrated in the main portion of Figure 6. The key point to
notice is that, after photolysis, CH2O formation commences with
almost no delay; the small delay that can be observed occurs
because of interference from the excimer laser pulse. This sharp
rise was noted uniformly from 232 K, at which temperature
The acetylperoxy self-reaction rate constant has been mea-
6
3
our UV experiments and Horie and Moortgat’s results agree
sured previously at room temperature by Addison et al. (k1 )
-
12
3 -1
that channel 4b predominates to 345 K, where there is
disagreement. At the high temperature, the [CH2O] time
dependence can be compared to differing model predictions
based on (i) b ) 1 and k1 and k4 taken from Table 2 vs (ii) b
from ref 6 and the alternative rate constants exhibited by x’s in
Figures 4 and 5. These are illustrated, respectively, by the solid
and dashed curves in Figure 6. In agreement with the data, the
solid line exhibits only a very small delay in rising after the
photolysis pulse. In contrast, the prediction based on a
temperature-dependent branching ratio, with b ) 0.03 at 345
K, indicates a noticeable delay in CH2O production, that, in
spite of interference from the excimer laser, is not observed in
the data. The discrepancy between the data and predictions
based on b ) 0.03 increases if it is assumed (as is likely) that
reaction 10 produces no CH2O (dotted line) or if neither reaction
2.5 × 10 cm s ) using molecular modulation and by Basco
4
-12
3
-1
and Parmar (k1 ) 8 × 10
cm s ) using flash photolysis.
5
More recently, Moortgat et al. reported rate constants at 253,
298, and 368 K. The earlier rate constants are substantially
5
smaller than both the present value and that of Moortgat et al.
at 295 K. The rate constant of Addison et al. is based on
3
measurements at 210 and 240 nm; however, the CH3C(O)O2
UV cross sections they employed are ∼50% smaller than the
present ones, and they do not account for residual absorption
by stable reaction products. Although they attempt to correct
for methylperoxy formation, these authors use a rate constant
k4 that is about 6 times smaller than the value reported herein.
While these factors might explain an underestimate of k1, that
the rate constant is over 5 times smaller than the present value
4
is difficult to rationalize. Basco and Parmar base their rate
1
0 nor 11 produces CH2O (dot-dash line). Although not
constant on the time dependence of the absorbance in the 198-
208-nm region assuming no contribution from methylperoxy
radicals, whereas, in fact, it contributes roughly 25%. This
and the lack of accounting for the cross reaction between
CH3C(O)O2 and CH3O2 are likely reasons for the low value of
conclusive because CH2O is formed by reactions other than (4b),
the CH2O formation measurements agree with the UV data that
channel 4b represents the principal CH3C(O)O2 + CH3O2
reaction pathway over the 210-350 K temperature range.
4
k1 reported by Basco and Parmar.
IV. Discussion
As is apparent from Figure 4, the self-reaction rate constants
measured by Moortgat et al. agree extremely well with the
5
Figure 2 compares the UV spectrum of CH3C(O)O2 presented
in this work to previous measurements. All the spectra reveal
two bands: an intense one at about 207 nm and a somewhat
weaker one at 240 nm. The intensities and exact positions of
the peaks, however, vary considerably, except that the spectrum
present results based on the preferred branching fraction for
reaction 4b of b ) 1. Even if we set b to the temperature-
6
dependent values from Horie and Moortgat, the agreement for
k1 remains very good. Of course, this is somewhat fortuitous;
5
Moortgat et al. derived rate constants based on their original
5
of Moortgat et al. agrees well with the present work. The
“
unscaled” CH3C(O)O2 UV cross sections, and they did not
3
spectrum of Addison et al. was recorded in the presence of
account for the residual absorbance due to stable products in
their kinetic simulations. Still, these shortcomings should not
influence the rate constant determinations by more than about
NO2 and, therefore, required subtraction of the UV absorption
contributed by PAN. Since PAN absorbs more strongly over
the short-wavelength portion of the CH3C(O)O2 spectrum than
the long-wavelength portion, this may explain why the spectrum
2
0%, implying that the agreement would remain very good.
The principal issue concerning the cross reaction of
3
of Addison et al. agrees relatively well with ours above 225
CH3C(O)O2 with CH3O2 is the branching between the radical
channel yielding CH3C(O)O and CH3O vs the molecular channel
that produces formaldehyde and acetic acid. From flash
nm but much more poorly below 225 nm. The spectrum of
4
Basco and Parmar differs from ours in both overall intensity
and in the relative intensity of the 207- and 240-nm peaks. The
discrepancy in overall intensity may result from their use of
Cl2 loss as a means of determining radical production, a method
inherently less sensitive than the present one based on C2H5O2
5
photolysis/UV absorption data, Moortgat et al. determined self-
reaction and cross reaction rate constants by fitting model
simulations to absorbance vs times traces at a few fixed
wavelengths in the 200-260-nm range. Fits of their traces gave
a branching ratio of
4
production. Although Basco and Parmar claim to measure their
spectrum “immediately” after photolysis, as shown by the
present work, even a small delay causes a substantial change
in the shape of the spectrum because of the rapid conversion of
acetylperoxy radicals to methylperoxy radicals. It is likely that
this is responsible for the enhanced absorption recorded by
5
-3900/T
â ) 4.4 × 10 e
Unfortunately, this determination must be viewed with some
skepticism because, as Figure 3 shows, the peroxy radical
concentration vs time data cannot with accuracy provide unique
4
4
Basco and Parmar in the 230-260-nm region relative to the
2
07-nm peak.
values for â and k . On the basis of a product study of biacetyl
5
6
The spectrum of Moortgat et al. was measured relative to
photolysis, Horie and Moortgat later revised this to the
the absorption of HO2 at 210 nm and has, therefore, been
rescaled to account for the change in the accepted HO2 cross
section at this wavelength since their work.1 This rescaling
gives very good overall agreement with the present spectrum.
Close comparison reveals small discrepancies; at 207 nm, the
present spectrum is less intense by about 10%, whereas it is
more intense above 250 nm. The discrepancy at 207 nm is
well within the error bars of the respective intensity measure-
expression given by eq 13. The major effect of the revision is
to decrease the temperature at which channels 4a and 4b
contribute equally from 300 to 265 K.
The results of the present study indicate that channel 4b is
the major channel over the entire 209-358 K temperature range.
We base this on two types of experimental measurements and
on two observations. First, consider the experiments. The fits
of the UV data for [CH3C(O)O2]t and [CH3O2]t invariably

Kinetics of the CH3C(O)O2 Radical
J. Phys. Chem., Vol. 100, No. 11, 1996 4513
produce best agreement with b ) 1. However, when fitting
simultaneously for k1, k4, and b, we cannot rule out b ) 0 at
the 2σ confidence level. More importantly, infrared measure-
ments reveal on onset of formaldehyde production with negli-
gible delay when b ) 1. This is confirmed by experiment over
the 232-345 K temperature range. In contrast, predictions
based on the branching ratio of eq 13 reveal at 345 K a
significant delay preceding formaldehyde formation that is not
observed experimentally (see Figure 6).
point that more work is needed to fully understand the reaction
between acetylperoxy and methylperoxy radicals.
V. Conclusion
The techniques of time-resolved UV spectroscopy and
transient infrared absorption have been used to investigate the
CH3C(O)O2 self-reaction and its reaction with CH3O2. The UV
spectrum of CH3C(O)O2 has been corrected for secondary
production of methylperoxy radicals, unlike some earlier
3
,4
Next, consider two observations concerning the issue of
branching ratio. First, experiments have been conducted to
measure the self-reaction rate constant of the CF3C(O)O2 radical
using the same time-resolved UV spectroscopy technique as
measurements. After recalibrating the previous spectrum of
5
Moortgat et al., it is found to be in good agreement with the
present work. Similarly, very good agreement is found for both
the absolute magnitudes and the temperature dependence of the
CH3C(O)O2 self-reaction rate constants between the present
1
6
described here.
Fits of the data in that case revealed a
-
12
3
-1
5
CF3C(O)O2 + CF3O2 rate constant of <2 × 10
cm s .
results and the work of Moortgat et al. In contrast, two earlier
Presumably, the reason for the small rate constant is that, as
with acetylperoxy, the radical channel is slow and that, due to
fluorine substitution, the molecular channel becomes unacces-
sible. Second, we can examine the consequence of the
branching ratio given by eq 13 and the values of k4 determined
under this assumption (marked by x’s in Figure 5). A best fit
of these rate constants to a sum of two Arrhenius expressions
studies that lacked as careful an analysis of the secondary
3
,4
chemistry reported much smaller rate constants.
The magnitudes and temperature dependence of the cross
reaction between CH3C(O)O2 and CH3O2 found in the present
5
work also agree well with those of Moortgat et al. However,
the present results are based on channel 4b being the dominant
pathway over the entire 209-358 K temperature range, whereas
the latter results have channel 4a dominant above 300 K. The
present paper concludes that reaction 4 proceeds nearly exclu-
sively via channel 4b and supports this by IR measurements of
CH2O production and the fact that the values of k4 measured in
the present work are inconsistent with two product channels
both exhibiting Arrhenius temperature dependences, but related
-
9
-1870/T
3
-1
yields k4a ) 6.8 × 10 e
cm s and k4b ) 3.1 ×
-
15 2000/T
3
-1
1
0
e
cm s and the dashed curve in Figure 5. As is
apparent, the curve gives a poor fit to the data, suggesting that
the branching ratio is not consistent with the measured overall
rate constants and the assumption of Arrhenius behavior for
each channel. Moreover, the preexponential factors are un-
physical for both channels. In contrast, fixing b ) 1 over the
temperature range of the present experiments leads to a
consistent interpretation of both the UV and IR data.
6
by the branching ratio of Horie and Moortgat.
Note Added in Proof. After submission of this paper we
became aware of the measurements of CH3C(O)O2 kinetics by
Roehl, Bauer, and Moortgat [J. Phys. Chem., in press]. Their
UV cross sections are virtually identical to the measurements
presented here. Their values of k1 and k4, at 295 K, are in very
good agreement with the ones in this work.
A comparison between the CH3C(O)O2 + CH3O2 rate
constants measured in the present work and those determined
5
by Moortgat et al. is presented in Figure 5. Except at 368 K,
the agreement is extremely good when the branching fraction
for channel 4b is set to b ) 1. Again, this is in part fortuitous
since Moortgat et al.5 employed a temperature-dependent
branching ratio in their fits. For comparison, poor agreement
is obtained above 270 K when we force the temperature
References and Notes
(1) Lightfoot, P. D.; Cox, R. A.; Crowley, J. N.; Destriau, M.; Hayman,
G. D.; Jenkin, M. E.; Moortgat, G. K.; Zabel, F. Atmos. EnViron. 1992,
26A, 1805.
6
dependence of Horie and Moortgat on the fits of our UV data.
(
67.
2) Wallington, T. J.; Dagaut, P.; Kurylo, M. J. Chem. ReV. 1992, 92,
The question remains as to the origin of the discrepancy in
the branching ratio of reactions 4a and 4b. The present work
has shown that, even if time-resolved UV spectra are recorded,
fits of the data to k1, k4, and b produce rather large error bars
6
(3) Addison, M. C.; Burrows, J. P.; Cox, R. A.; Patrick, R. Chem. Phys.
Lett. 1980, 73, 283.
(
(
4) Basco, N.; Parmar, S. S. Int. J. Chem. Kinet. 1985, 17, 891.
5) Moortgat, G.; Veyret, B.; Lesclaux, R. J. Phys. Chem. 1989, 93,
5
for the branching ratio. Moortgat et al. found that, in order to
2
362.
(6) Horie, O.; Moortgat, G. K. J. Chem. Soc., Faraday Trans. 1992,
8, 3305.
fit their UV absorbance vs time traces, they needed to introduce
into their model a third adjustable parameter, a temperature-
dependent branching ratio for reactions 4a and 4b. Because
their data were recorded at just a few fixed wavelengths, it is
likely that another effect, such as residual absorption by stable
products or a systematic discrepancy in the UV cross sections
employed in their model, could have been introduced instead
with equal success. The reason for the discrepancy with the
8
(
(
7) Maricq, M. M.; Szente, J. J. J. Phys. Chem. 1992, 96, 10862.
8) Maricq, M. M.; Wallington, T. J. J. Phys. Chem. 1992, 96, 986.
(9) Bauer, D.; Crowley, J.; Moortgat, G. K. J. Photochem. Photobiol.
A: Chem. 1992, 65, 329.
10) Fenter, F. F.; Catoire, V.; Lesclaux, R.; Lightfoot, P. D. J. Phys.
(
Chem. 1993, 97, 3530.
(11) Maricq, M. M.; Szente, J. J.; Kaiser, E. W. J. Phys. Chem. 1993,
9
7, 7970.
12) Mallard, W. G.; Westley, F.; Herron, J. T.; Hampson, R. F.; Frizzell,
(
6
product study of Horie and Moortgat is less clear. They base
D. H. NIST Chemical Kinetics Database: Version 6.0; NIST: Gaithersburg,
their results on measurements of the temperature dependence
of a variety of products from the photooxidation of biacetyl;
however, the mechanism is rather complex. For the analogous
reaction between CH3C(O)O2 and HO2, they are able, rather
directly, to base the branching ratio on the ratio of certain
products. That procedure does not work for reactions 4a and
1994.
(13) Maricq, M. M.; Szente, J. J.; Kaiser, E. W.; Shi, Jichun. J. Phys.
Chem. 1994, 98, 2083.
(
14) For examples, see: Francisco, J. S.; Maricq, M. M. AdV. Photochem.
1995, 20, 79.
(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
Modelling. JPL Publication 94-26, 1994.
4
b. Instead, they had to employ an indirect means to deter-
mine the branching ratio. Interestingly enough, they find a
much smaller temperature dependence for the reaction of
CH3C(O)O2 with HO2 as opposed to CH3O2. It appears at this
(16) Maricq, M. M.; Szente, J. J.; Khitrov, G. A.; Francisco, J. S. J.
Phys. Chem., 1996, 100, 4514.
JP9533234