Absorption Spectrum and Kinetics of the Acetylperoxy Radical

Article abstract of DOI:10.1021/jp9526298

The ultraviolet absorption spectrum of acetylperoxy radical was remeasured in the 195-280 nm range.Acetylperoxy radicals were generated by laser flash photolysis of Cl2/CH3CHO/O2 mixtures.The absorption cross sections were calibrated against the ethylperoxy radical, generated in the flash photolysis of Cl2/C2H6/O2 mixtures.The acetylperoxy spectrum is bimodal, with a strong maximum at 207 nm, ? = 6.67E-18 cm² molecule^-1, and a weaker maximum near 240 nm, ? = 3.21E-18 cm² molecule^-1.Newly obtained cross section were used along with absorption-time profiles, obtained over a range of radical concentrations, to determine a rate constant for the acetylperoxy self reaction at 298 K, 2CH3C(O)O2 -> 2CH3C(O)O + O2, of k₁ = (1.36 +/- 0.19)E-11 cm³ moleclule^-1 s^-1, and to determine the rate coefficients for the acetyl- and methylperoxy cross reactions at 298 K, CH3C(O)O2 + CH3O2 -> CH3C(O)O + CH3O + O2 and CH3C(O)O2 + CH3O2 -> CH3C(O)OH + HCHO + O2, of k_4a = (8.8 +/- 1.5)E-12 cm³ molecule^-1 s^-1 and k_4b = (1.0 +/- 0.5)E-12 cm³ molecule^-1 s^-1, respectively.New cross sections are reported for CH3O2 and HO2 and the absorption spectrum of CH3COCl is presented.

Full text of DOI:10.1021/jp9526298

4
038
J. Phys. Chem. 1996, 100, 4038-4047
Absorption Spectrum and Kinetics of the Acetylperoxy Radical
Coleen M. Roehl,* Dieter Bauer, and Geert K. Moortgat*
Max-Planck-Institut f u¨ r Chemie, Atmospheric Chemistry DiVision, Postfach 3060, 55020 Mainz, Germany
X
ReceiVed: September 6, 1995; In Final Form: December 4, 1995
The ultraviolet absorption spectrum of acetylperoxy radical was remeasured in the 195-280 nm range.
Acetylperoxy radicals were generated by laser flash photolysis of Cl
cross sections were calibrated against the ethylperoxy radical, generated in the flash photolysis of Cl
2
/CH
3
CHO/O
2
mixtures. The absorption
/C
2
2 6
H /
-
18
O
2
mixtures. The acetylperoxy spectrum is bimodal, with a strong maximum at 207 nm, σ ) 6.67 × 10
2 -1 -18 2 -1
cm molecule , and a weaker maximum near 240 nm, σ ) 3.21 × 10 cm molecule . Newly obtained
cross sections were used along with absorption-time profiles, obtained over a range of radical concentrations,
to determine a rate constant for the acetylperoxy self reaction at 298 K, 2CH
3
C(O)O
2
3 2
f 2CH C(O)O + O ,
-
11
3
-1 -1
of k
methylperoxy cross reactions at 298 K, CH
CH f CH C(O)OH + HCHO + O
1.0(0.5) × 10 cm molecule s , respectively. New cross sections are reported for CH
and the absorption spectrum of CH COCl is presented.
1
) (1.36 ( 0.19) × 10 cm molecule s , and to determine the rate coefficients for the acetyl- and
3
C(O)O
2
+ CH
3
O
2
f CH
3
C(O)O + CH
3
O + O
2
and CH
3
C(O)O
2
-12
3
-1 -1
+
(
3
O
2
3
12
2
, of k4a ) (8.8 ( 1.5) × 10
cm molecule
s
and k4b
)
2
-
3
-1 -1
O
3 2
and HO
3
Introduction
account for a subsequent redetermination of the HO2 cross
section at 210 nm, as recommended by two recent organic
Following the discovery of the role of peroxy radicals in
balancing the photostationary state of NOx/O3 in polluted
environments, extensive studies of oxidative degradation pro-
cesses of hydrocarbons in the atmosphere have been under-
10,11
9
peroxy radical reviews.
However, since Moortgat et al.
used computer simulations, which utilized the erroneously high
HO2 cross sections, to fit absorbance traces and to concurrently
optimize both the CH3C(O)O2 cross section and the CH3C(O)-
O2 self reaction rate constant, such a simple rescaling is not
adequate.
Determinations of the kinetic parameters for the acetylperoxy
system have been based on concentration profiles calculated
from measured absorbances using the available cross sections.
Hence, discrepancies in the CH3C(O)O2 cross sections have, in
part, also created large differences in the calculated reaction
rate constants. CH3C(O)O2 is not the only absorbing species
present during such measurements; CH3O2 and HO2 radicals
grow in as CH3C(O)O2 decays. CH3C(O)O2 radicals self react
1
taken. Acetylperoxy radicals, in particular, are known to be
formed in the photooxidation of higher carbonyl compounds
(i.e., acetaldehyde, acetone, methyl vinyl ketone, methylglyoxal,
etc.) and, hence, are important in many atmospheric photooxi-
dation processes. CH3C(O)O2 is also a precursor to peroxy-
acetyl nitrate (PAN), which acts as a temporary reservoir and
2
,3
transporter for NOx within cooler regions of the troposphere.
Furthermore, thermal decomposition of PAN is believed to be
a significant source of nighttime peroxy radicals, which take
part in the important RO2 + NO3 nighttime chemistry.4 Finally,
peroxy radical self and cross reactions become important in NOx-
free or clean environments. For example, the cross reaction
between acetylperoxy radicals and HO2 is thought to be
responsible for the significant organic acid buildup in the
quickly according to reaction 1. The resulting CH3C(O)O
12,13
radicals then undergo rapid thermal decomposition
methyl radicals, which add O2 to produce CH3O2:
to yield
5
,6
troposphere.
Yet, despite its atmospheric importance, the
2
CH C(O)O f 2CH C(O)O + O
2
(1)
(2)
(3)
3
2
3
spectrum and kinetics of the acetylperoxy radical are not well
established.
The UV absorption spectrum of the acetylperoxy radical was
CH C(O)O + M f CH + CO + M
3
3
2
first determined in a modulated photolysis study by Addison et
CH + O + M f CH O + M
3
2
3
2
7
al. Cl2/CH3CHO/O2/NO2 mixtures were photolyzed to generate
the radicals and the formation of PAN was observed. The
CH3C(O)O2 spectrum was obtained by correcting the composite
spectrum for contributions due to PAN and by subtracting the
components due to CH3O2. In two newer studies, the acetyl-
peroxy radicals were produced by flash photolysis of Cl2/
Besides self reacting, the two peroxy radicals also react with
each other via
CH C(O)O + CH O f CH C(O)O + CH O + O (4a)
3
2
3
2
3
3
2
8
CH3CHO/O2 mixtures. Basco and Parmar did an absolute
CH C(O)O + CH O f CH C(O)OH + HCHO + O
2
3
2
3
2
3
calibration by measuring the amount of chlorine dissociated in
the flash. Moortgat et al.9 measured the absorption cross
sections relative to the absorption cross section of HO2 at 210
nm. Although the general shapes of the three spectra are the
same, the magnitudes of the two maxima are not in agreement.
The best available acetylperoxy spectrum is the Moortgat et
(4b)
The methoxy radicals, CH3O, generated in reaction 4a further
react with the excess O2 via
CH O + O f HO + HCHO
(5)
3
2
2
9
al. data rescaled (downward 21 and 14%, respectively) to
to form HO2 radicals. And finally, HO2 radicals also undergo
self reactions and reactions with CH3C(O)O2 and CH3O2:
X
Abstract published in AdVance ACS Abstracts, February 15, 1996.
0022-3654/96/20100-4038$12.00/0 © 1996 American Chemical Society

Absorption Spectrum and Kinetics of CH3C(O)O2
J. Phys. Chem., Vol. 100, No. 10, 1996 4039
CH C(O)O + HO f CH (O)OOH + O
2
(6a)
(6b)
(7)
The light intensity from a deuterium (D2, Hamamatsu) probe
lamp passing through the photolysis cell was monitored, before
3
2
2
3
(I0) and at known times after the flash (It), at specific
CH C(O)O + HO f CH (O)OH + O
3
3
2
2
3
wavelengths. Attenuation of this light beam upon production
of absorbing radicals was used to calculate the absorbance at
that wavelength (λ) as a function of time with the equation
CH O + HO f CH OOH + O
2
3
2
2
3
7
Addison et al. analyzed their modulated absorption curves
as a function of time to obtain the rate constants k1 ) (2.5 (
Absλ,t ) -ln[I /I ]
(A)
t
0
-12
3
-1 -1
-12
3
1
.0) × 10
cm molecule
s
and k4 ) 3.0 × 10
cm
molecule s-1 (k4 ) k4a + k4b). Basco and Parmar calculated
-
1
8
The absorbances were converted into cross sections, σλ(i),
using the Beer-Lambert relationship
-
12
3
-1 -1
a larger k1 value of (8.0 ( 1.5) × 10
cm molecule
s
from second-order fits of decay traces obtained near 200 nm.
-
11
λ,t
σλ(i) ) Abs /(l[i]t)
(B)
Rate constants of k1 ) (1.7 ( 0.4) × 10
and k4 ) (1.4 (
-11
3
-1 -1
0
.3) × 10 cm molecule
s
and a branching ratio of R )
where l is the pathlength and [i]t is the concentration of a single
absorbing species, i, at time, t. Once measured, the cross
sections of acetylperoxy, along with the Beer-Lambert equation
for a multiabsorber system,
k4a/k4 ) 0.48 at 298 K were deduced from the computer
simulations of Moortgat et al.9 A strong temperature depen-
dence in the CH3C(O)O2 + CH3O2 branching ratio was noted
9
14
by Moortgat et al. and led to a further study of the products
and branching ratio between 223 and 333 K. The temperature
dependence of the cross reactions was determined to be â )
Absλ,t /l )
σλ(i)[i]
(C)
∑
t
6
n
k4a/k4b ) 2.2 × 10 exp(-3870/T), yielding R ) â/(1 + â) )
0
.83 ( 0.17 at 298 K in contradiction to their earlier measure-
were used to convert absorption-time profiles to absorber
concentration-time profiles (n is the number of absorbers). A
simulation program was thereafter employed to fit the measured
absorbances and to simulate concentration-time profiles, thus
deriving information on the kinetics of the radicals involved.
The apparatus used in these experiments is shown in Figure
ment. Since our work began, another flash photolysis experi-
1
5
ment was initiated to study k4. The CH3O2 radical precursor,
CH4, was added to the photolysis mixtures of Cl2/CH3CHO/O2
and the radical concentrations were monitored by UV absorp-
-12
3
tion. Preliminary reports give k4 ) (7.6 ( 3.8) × 10
cm
molecule-
1
s
-1
at 298 K using R ) 0.48.
1
. A rectangular quartz cell of 40 cm length was used, and
In an attempt to resolve the existing discrepancies in the
absorption spectrum of acetylperoxy radical and in the kinetic
parameters, especially the CH3C(O)O2 + CH3O2 branching ratio,
a new laser flash photolysis experiment was constructed and
employed. A new ultraviolet absorption spectrum of acetyl-
peroxy radical is reported in the 195 -280 nm range. The rate
coefficients for the acetylperoxy self reaction and cross reaction
with methylperoxy were determined at 298 K from absorption-
time profiles using the newly obtained cross sections.
photolysis was performed from the side by expansion of the
output from the pulsed excimer laser with a set of cylindrical
mirrors. The light from the D2 lamp was passed lengthwise
through the cell and was folded by internal White optics,
resulting in eight passes with a total path length of 326 cm,
before it entered a 0.5 m monochromator (Acton Research
-1
Corporation, grating with 300 lines mm blazed at 300 nm),
which was equipped with both a photomultiplier tube (PMT,
Hamamatsu) and a gated photodiode array detector (EG&G
Princeton Applied Research). The entrance slit of the mono-
chromator and the slit between the monochromator and the PMT
were fixed at 0.5 mm, resulting in a PMT resolution of about
3.3 nm. The photodiode array resolution, calculated as the full
width at half-maximum FWHM of the 282.4 nm line of the Zn
lamp, was 1.0 nm. All experiments were conducted at or near
Experimental Section
For all experiments described here, CH3C(O)O2 radicals were
generated by laser flash photolysis of Cl2, using the 351 nm
XeF line of a Lambda Physik LPX-300 excimer laser, in the
presence of CH3CHO and O2 via the reactions
2
98 K.
Cl + hν f 2Cl
(8)
(9)
2
The gated photodiode array detector was used to measure
relative cross sections over the entire spectral region between
195 and 280 nm within a small time window. The gate timing
of the photodiode array measurements, or the time during which
light impinged on the diode array, was established by preset
pulses sent by the delay generator with typical duration times
of ∼10 µs. Measurements consisted of the coaddition of
multiple scans (∼5000) taken both immediately before (Io) and
over the time period between 1 and 11 µs after the laser pulses
Cl + CH CHO f CH C(O) + HCl
3
3
CH C(O) + O + M f CH C(O)O + M
(10)
3
2
3
2
Methylperoxy radicals, CH3O2, evolved as secondary products
of the rapid self reaction of acetylperoxy radicals according to
reactions 1-3. In some experiments, CH4 was added to the
photolysis mixtures and CH3CHO was reduced in order to
generate initial CH3O2 radicals and thereby enhance the reaction
flux through the cross reaction channels, reactions 4a and 4b.
This initial CH3O2 is formed from the reactions
(It). The wavelength scale was calibrated with an accuracy of
∼
0.2 nm with the output from a Zn lamp.
Single wavelength absorption measurements, used to deter-
mine both absorption cross sections and rate constants, were
carried out as a function of time using the PMT. The output of
the PMT, a small dc signal, was amplified, digitized, and
transferred to a personal computer for analysis. As with the
photodiode array measurements, I0 was obtained by averaging
the PMT signal immediately before the laser pulse. The It values
were typically recorded between 0 and 4 ms after the laser flash
with a 5 µs resolution and between 4 ms and 2 s after the laser
flash with a 50 µs resolution. The final absorption-time profiles
Cl + CH f CH + HCl
(11)
(3)
4
3
CH + O + M f CH O + M
3
2
3
2
The respective precursor concentrations were adjusted according
to their rate constants with Cl to ensure that complete conversion
to CH3C(O)O2 and CH3O2 occurred. (See Results.)

4
040 J. Phys. Chem., Vol. 100, No. 10, 1996
Roehl et al.
Figure 1. Schematic diagram of the instrument used in the CH
3
C(O)O
2
absorption cross sections and kinetic experiments.
were produced per laser pulse. CH3CHO concentrations were
usually consisted of averaged signals of three or four sets of
00 laser pulses. When relative wavelength measurements were
1
7
-3
5
kept high, about (1-3) × 10 molecule cm . O2 addition to
made, the wavelength was alternated after each set of 500
flashes.
CH C(O) was ensured by maintaining high O concentrations
3
2
1
8
-3
of about (8-9) × 10 molecule cm , and the maximum
CH3C(O)O2 radical concentration was achieved within <1 µs.
Preventing significant buildups of CH3C(O) is important since
it is known to have large cross sections with a maximum of
The Cl2, O2, N2 bath gas and in certain experiments CH4,
C2H6, or gaseous CH3CHO were introduced through calibrated
mass flowmeters. For higher acetaldehyde concentrations, an
additional regulated flow of N2 was bubbled through liquid
CH3CHO which was contained in a temperature-controlled
vessel. The CH3CHO temperature and N2 flow were varied to
attain a suitable CH3CHO concentration. With total pressures
of ∼400 Torr and flow rates between 4000 and 5000 sccm, the
residence time in the cell was about 5-6 s. The repetition rate
of the laser was greater than or equal to the residence time,
thus ensuring that the photolysis products formed in one laser
flash were flushed out and that a fresh gas mixture was flowed
into the cell before the next laser flash. The stated purities of
the compounds used were Cl2, 99.95% in a 1.0% mix in
-17
2
-1
8
about 3.8 × 10
cm molecule at 216 nm. High O2
concentrations also minimized the CH C(O) + Cl reaction
3
2
forming acetyl chloride (CH C(O)Cl). An absorbance spectrum
3
of this product, displaying a maximum near 240 nm, was
8
recorded by Basco and Parmar; however, cross sections were
not reported. Since CH C(O)Cl is a stable molecule which
3
would continue to absorb even after the radical concentrations
had been greatly reduced, its existence could lead to errors in
the kinetic analysis. The significance of CH3C(O)Cl absorption
in this system depends on the magnitude of the CH3C(O)Cl cross
sections relative to those of the peroxy radicals. For this reason,
UV spectra of 0.9-1.6 Torr samples of pure CH3C(O)Cl were
measured in a static system using the photodiode array detector.
Results given in Figure 2 show that CH3C(O)Cl absorbs only
weakly in the wavelength range of interest, with a maximum
9
9
9.999% N2; O2, 99.99%; N2, 99.999%; CH3CHO, 99%; CH4,
9.995%; and C2H6, 99.95%.
Results
In this work, several parameters were obtained: the UV
absorption cross sections of CH3C(O)O2, the rate constant for
the self reaction of CH3C(O)O2, and the reaction rate constant
of CH3C(O)O2 + CH3O2 along with its branching ratio. During
the course of this study, the spectra of CH3C(O)Cl, CH3O2, and
HO2 were also measured.
-
19
2
-1
of 1.30 × 10 cm molecule at 241 nm. CH3C(O)Cl cross
sections, given as squares in Figure 2, were also calculated from
the absorbance spectrum of Basco and Parmar and were
8
normalized to our value at 240 nm. Under our experimental
conditions, no evidence of either CH3C(O) at short time scales
or CH3C(O)Cl at long time scales was observed.
Absorption Cross Sections. Absorption cross sections of
CH3C(O)O2 were measured using both the PMT and the gated
photodiode array systems. For each of these measurements,
the initial radical concentration and therefore the initial decay
rate were minimized by photolyzing only small amounts of Cl2.
To avoid doing complex spectral subtractions of secondary
absorbing species, CH3O2 and HO2, it was necessary to measure
the CH3C(O)O2 spectrum immediately upon formation. O3,
produced in reaction 6b, also absorbs in the wavelength region
of interest, but its concentration and hence its contribution to
the total absorbance was negligible throughout the time period
analyzed. The ability to record the entire spectrum in a very
16
-3
Cl2 concentrations of (2-3) × 10 molecule cm were
employed and typically [Cl]i ) (2-3) × 10 molecule cm-3
13

Absorption Spectrum and Kinetics of CH3C(O)O2
J. Phys. Chem., Vol. 100, No. 10, 1996 4041
is derived for calculating the absorption cross section of
CH3C(O)O2 at λ ) 240 nm, where Abs240nm,0(CH3C(O)O2) and
Abs240nm,0(C2H5O2) are the extrapolated absorbances at t ) 0
in the CH3C(O)O2 and C2H5O2 systems, respectively. The
1
0
C2H5O2 cross section recommended by Lightfoot et al.
(
-18
2
-1
σ240nm(C2H5O2) ) (4.36 ( 0.22) × 10
preferred over the 4.8% lower value of Wallington et al.
because its derivation included a more recent measurement
cm molecule ) is
1
1
1
6
performed in this laboratory. Since the publication of the
reviews, Maricq and Wallington completed another study on
1
7
C2H5O2 cross sections in which they found σ240nm(C2H5O2) )
-
18
2
-1
4
.6 × 10
cm molecule , a value 5% higher than the
Lightfoot review. A value of σ240nm(CH3C(O)O2) ) (3.21(0.32)
-18
2
-1
×
10
cm molecule was determined.
Figure 2. UV absorption cross sections of CH
3
C(O)Cl measured near
The single wavelength PMT absorption values and the
2
98 K. This work is shown as the solid line; squares represent cross
photodiode array spectrum were normalized to the absorption
cross section value determined at 240 nm and are displayed in
Figure 3. As is observed in Figure 3, the ratio R ) σ210nm/
σ240nm measured with the photodiode array detector is smaller
than that from the PMT. Further investigations suggested that
this ratio R determined with the photodiode array detector was
dependent on the initial radical concentration. Photodiode array
measurements were conducted in which the initial radical
sections calculated from the absorbance spectrum of ref 8 and
normalized to our value at 240 nm.
short time interval (1-11 µs) was achieved by the combination
of using a relatively large setting for the entrance slit of the
monochromator and the gated photodiode array detector. On
average, only about 0.5% of the CH3C(O)O2 decayed during
this short measurement time interval, and therefore significant
concentrations of secondary absorbers were not produced during
the measurement. The shape of the spectral feature was
calculated with eq A.
The UV absorption features were also mapped out in 10 nm
intervals through repeated determinations of Absλ,0 /Abs240nm,0
with the PMT system. The absorbance values at time t ) 0
were obtained by extrapolating second order fits of the measured
absorption-time profiles, extending out to times of ∼1 ms, back
to zero time. Because of secondary chemistry, fits of these
absorption-time profiles (as such) cannot be used to derive
information about the decay rate of CH3C(O)O2. However, they
can be used to determine the initial CH3C(O)O2 absorbance.
At t ) 0, no absorbing products have yet been formed and the
extrapolated absorbances are only due to CH3C(O)O2.
Calculating the absolute UV absorption cross sections of
CH3C(O)O2 from either the photodiode array or the PMT
absorbance values requires knowledge of the concentration of
CH3C(O)O2 present at the time of the measurement. The
CH3C(O)O2 concentration depends on the Cl atom concentration
formed in the photolysis flash, which in turn is determined by
the Cl2 concentration and the laser intensity. Due to the
difficulty in accurately measuring the absolute laser intensity,
the Cl atom concentrations (and therefore the CH3C(O)O2
concentrations) were instead calibrated against the ethylperoxy
radical, which was generated in separate flash photolysis
experiments of Cl2/C2H6/O2 mixtures. CH3C(O)O2 and C2H5O2
absorption measurements were made back-to-back at 240 nm
with the PMT, and initial absorbances were calculated by the
extrapolation technique described above. Care was taken to
maintain the same Cl2 concentrations and relative laser power
in the two measurements, so that the Cl atom concentrations
and, hence, the initial peroxy radical concentrations were equal,
1
3
concentrations were stepwise reduced from 3 × 10 to 7 ×
1
2
-3
1
0
molecule cm , by decreasing the Cl2 concentration and/
or limiting the laser power. The ratio R calculated from the
resulting spectra increased with decreasing radical concentration,
approaching the ratio measured with the PMT. Extrapolating
a plot of these ratios versus radical concentration back to zero
initial radical concentration resulted in a ratio of 2.02, exactly
equal to the PMT measured ratio. Test measurements of the
C2H5O2 spectrum, which has smaller cross sections in the low
wavelength region, did not reveal a similar problem. These
experiments suggested that large changes in the light intensity,
occurring at short wavelengths, where the intensity from the
D2 lamp is drastically reduced, caused a nonlinearity problem
in the photodiode array detector, resulting in erroneously low
absorbances. Unfortunately, avoiding the nonlinearity hindrance
by reducing the radical concentration results in a much noisier
spectrum. To maintain the better signal-to-noise level, the
photodiode array spectrum shown in Figure 3 was corrected by
dividing it into smaller wavelength ranges and then recalculating
the cross sections, scaling them to the PMT measurements in
the same wavelength range. The final spectrum is given as the
solid line in Figure 4 and the cross sections are reported in Table
1
.
Error Analysis for the CH3C(O)O2 Cross Sections. In
calculating the statistical error in the CH3C(O)O2 cross sections,
the errors associated with the parameters in eq D were first
considered. An error limit of 5% was used for σ240nm(C2H5O2),
based on the errors quoted in ref 16. The signal-to-noise ratios
obtained from the averaged absorption profiles are sufficiently
high such that errors in the extrapolations of these curves are
expected to be 2-3% at 240 nm, increasing to 5% at 200 and
2
80 nm. Including the small errors (e1%) associated with the
fluctuations in the laser power and Cl2 concentration, statistical
errors of 6% and 8% are calculated for the cross sections at
240 nm and at 200 and 280 nm, respectively. This error does
not include systematic errors which may have evolved as a result
of problems in the apparatus, the complete Cl atom to peroxy
radical conversion, or secondary chemistry. Taking these into
account, the total errors on the CH3C(O)O2 cross sections are
estimated to be 10% at 240 nm and 15% at 200 and 280 nm.
[
CH3C(O)O2]0 ) [C2H5O2]0. Small corrections (usually <3%)
were made when changes in Cl2 concentration and/or laser
power did occur. By solving eq B for [C2H5O2]0 and
[CH3C(O)O2]0, setting these equations equal, and rearranging,
the equation
σ240nm(CH C(O)O ) )
3
2
Abs240nm,0(CH C(O)O ) σ
(C H O )
2 5 2
3
2
240nm
(D)
Determination of the Rate Constants for the CH3C(O)O2
Self Reaction and for the CH3C(O)O2 Cross Reaction with
Abs2
40nm,0
(C H O )
2 5 2

4
042 J. Phys. Chem., Vol. 100, No. 10, 1996
Roehl et al.
Figure 3. Comparison of the absolute (uncorrected) CH
3
C(O)O
2
absorption cross sections determined with the PMT (points) and the photodiode
array (line).
3 2
Figure 4. Comparison of the absolute CH C(O)O absorption cross sections measured here (solid line) and those of ref 7 (squares), ref 8 (circles),
and ref 9 (triangles). The ref 9 spectrum, reduced 17.5%, is given as the black diamonds. The absorption cross sections reported here were corrected
for a nonlinearity in the photodiode array detector, as described in the text.
CH3O2. To determine the CH3C(O)O2 self reaction rate
constant, it was first necessary to deduce the CH3C(O)O2
concentrations over a particular reaction time period. Although
the initial radical concentrations were easily obtained by
extrapolating measured absorbance profiles back to t ) 0 and
by substituting these values into eq B, the derivation of
CH3C(O)O2 concentrations at times > 0 was slightly more
complex, as a result of the presence of other absorbing species,
namely CH3O2 and HO2.
The concentrations of n absorbers can be calculated as a
function of time from absorbance measurements made at n
different wavelengths, if the equations generated from the
appropriate substitutions into eq C are linearly independent. The
values of the CH3C(O)O2, CH3O2, and HO2 cross sections at
3 2
TABLE 1: Absorption Cross Sections of CH C(O)O in
18 2
-
-1
Units of 10
wavelength, nm
95
cm molecule
σ
wavelength, nm
σ
1
200
205
2
2
2
2
2
230
235
3.75
5.34
6.46
6.67
6.47
5.53
4.36
3.49
3.10
3.13
240
245
250
255
260
265
270
275
280
3.21
3.20
3.04
2.89
2.53
2.25
1.75
1.32
1.13
07
10
15
20
25
the wavelengths 210, 225, and 240 nm are unique, thus
satisfying this criterion. An equation written generally as

Absorption Spectrum and Kinetics of CH3C(O)O2
J. Phys. Chem., Vol. 100, No. 10, 1996 4043
Abs /l ) σ(CH C(O)O )[CH C(O)O ] +
t
3
2
3
2 t
σ(CH O )[CH O ] + σ(HO )[HO ] (E)
3
2
3
2 t
2
2 t
was generated at each of the three wavelengths, and the
concentrations of each absorber were calculated by simulta-
neously solving the three equations.
The absorption profiles, used in eq E, were obtained with
the PMT in photolysis experiments performed on gaseous
mixtures which can be divided into three categories according
to the hydrocarbon concentration(s) used. In the first set of
experiments, designated as “low CH3CHO” data sets, CH3CHO
was the only hydrocarbon initially present and its concentration
1
4
-3
was (3-30) × 10 molecule cm . CH3CHO was again the
only hydrocarbon initially present in the second set of measure-
ments, called “high CH3CHO”, but in these measurements the
concentrations were about 2-3 orders of magnitude larger or
1
7
-3
(
2.1-3.5) × 10 molecule cm . CH3C(O)O2 and CH3O2 were
simultaneously produced in the third set of measurements by
adding CH4 to the usual Cl2/CH3CHO/O2 photolysis mixtures.
Since H abstraction from CH4, via reaction 11, is ∼660 times
slower than from CH3CHO at 298 K, CH3CHO concentrations
14
-3
were kept low, (3-200) × 10 molecule cm , to further
enhance the CH3O2 production. CH4 concentrations were
18
-3
relatively high, (2-7) × 10 molecule cm . The Cl2 and O2
concentrations were typically (4-5) × 10 and (0.6-1.0) ×
molecule cm , respectively, throughout all measurements,
yielding radical concentrations of (2-5) × 10 molecule cm .
16
19
-3
1
0
13
-3
The newly measured absorption cross sections for
CH3C(O)O2 were utilized in eq E, along with CH3O2 and HO2
cross sections also measured with our flash photolysis/PMT
system. The CH3O2 and HO2 measurements were made relative
to C2H5O2, and the cross sections were calculated in the same
manner as previously described for the CH3C(O)O2 cross
sections. Because the reactions
Figure 5. Examples of typical (a) measured absorption-time profiles
and (b) concentration-time profiles calculated from the absorption-
time profiles in (a). “Low acetaldehyde” conditions were used; [O
2
-3
]
i
1
8
-3
16
) 5.7 × 10 molecule cm and [Cl ] ) 5.0 × 10 molecule cm
2
i
.
The solid lines are simulated by FACSIMILE from parameters obtained
in the absorption data fitting.
TABLE 2: Absorption Cross Sections of CH
CH , and HO Measured Here and Used in the
FACSIMILE Fits and Simulations in Units of 10
3 2
C(O)O ,
Cl + CH O f CH O + ClO
(12a)
(12b)
3
2
3
3
O
2
2
-
18
cm2
-
1
Cl + CH O f CH O + HCl
molecule
wavelength, nm
10
3
2
2
2
σ(CH
3
C(O)O
2
)
σ(CH
3
2
O )
2
σ(HO )
1
8
were recently shown to be very rapid at room temperature,
2
6.47
3.49
3.21
1.97
3.50
4.12
4.19
2.94
1.24
-
10
3
-1 -1
k12 ) (1.5 ( 0.2) × 10
cm molecule s , and therefore
225
40
2
potentially problematic to CH3O2 cross section determinations,
special care was taken to ensure that the CH4 concentration was
sufficiently high. Measurements conducted with various initial
chemical mechanism and associated rate constants given in
Table 3, the cross sections given in Table 2, and the wavelength
specific absorbance equations (i.e., eq E). Selected parameters,
usually one or more rate constants, were allowed to vary while
the FACSIMILE program simultaneously fit the three absor-
bance profiles from 0 to 4 ms. As seen in Figure 5b, the
CH3C(O)O2 concentration is nearly zero at 4 ms, indicating that
most of the CH3C(O)O2 chemistry is over within that time
period.
1
7
18
CH4 concentrations (ranging from 8 × 10 to 2 × 10 molecule
-3
cm ) yielded virtually identical cross sections, indicating that
the Cl + CH3O2 reaction was negligible under our experimental
conditions. The remeasured cross sections of CH3O2 and HO2
are listed in Table 2. Examples of typical measured absorption-
time profiles and the corresponding concentration-time profiles,
derived from the absorption-time profiles, the cross sections
in Table 2, and eq E are shown in Figure 5.
The rate constants k1, k4a, and k4b were allowed to vary
simultaneously during initial fits, while all other parameters were
held constant. Excellent fits were achieved with the “low
A concentration decay profile of CH3C(O)O2 as a function
of time reflects not only the self reaction loss of CH3C(O)O2,
but also losses due to cross reactions. It is therefore not
adequate to simply perform a second order fit on the
CH3C(O)O2 decay curve to determine the self reaction rate
constant. Instead, the FACSIMILE program of Harwell, which
uses numerical techniques to solve differential chemical reaction
equations, was employed. Each FACSIMILE input file con-
sisted of the three absorbance profiles measured at 210, 225,
and 240 nm, the initial gas concentrations (i.e., Cl2, CH3CHO,
and O2), initial CH3C(O)O2 radical concentrations (calculated
from simple fits of the CH3C(O)O2 concentration profiles), the
-
11
3
CH CHO” data sets, and k ) (1.36 ( 0.02) × 10
3
1
cm
molecule- s , k ) (8.4 ( 0.2) × 10 cm molecule s ,
1
-1
-12
3
-1 -1
4
a
-13
3
-1 -1
and k4b ) (8.2 ( 0.2) × 10
cm molecule
s
were
19
determined by averaging the fit results. The errors given here
represent the standard deviation of the various measurements.
Absorbance fits at wavelengths 210, 225, and 240 nm and the
corresponding CH3C(O)O2, CH3O2, and HO2 concentration
simulated using the fitted parameters are compared to the
measured absorbances, and calculated concentrations from a
typical “low CH3CHO” data set in Figure 5.

4
044 J. Phys. Chem., Vol. 100, No. 10, 1996
Roehl et al.
TABLE 3: Reaction Mechanisms and Rate Constants Used in FACSIMILE Fits and Simulations (First Order Rate Constants
in Units of s- ; Second Order Rate Constants in Units of cm molecule
1
3
-1 -1
s
)
2
CH C(O)O f 2CH C(O)O + O
3
2
3
2
varied
CH
CH
CH
CH
CH
CH
CH
CH
CH
CH
HO
HO
Cl + CH
CH
Cl + CH
3
3
3
3
3
3
3
3
3
3
C(O)O + M f CH + CO + M
+ M
3
2
2.2E+5
9.2e-13
varied
estd
+ O
2
+ M f CH
3
O
f CH
f CH
2
ref 20, 390 Torr, 298 K
C(O)O
C(O)O
2
+ CH
+ CH
3
O
O
2
3
C(O)O + CH
C(O)OH + HCHO + O
+ M
3
O + O
2
2
3
2
3
2
varied
C(O) + O
O + O
2
+ M f CH
3
C(O)O
2
5.0E-12
1.9E-15
1.2E-13
2.5E-13
1.0E-11
0.4E-11
2.3E-12
5.8E-12
6.6E-11
1.0E-11
1.0E-13
estd from ref 1
ref 20
ref 10
2
f HCHO + HO
2
O
O
C(O)O
C(O)O
2
+ CH
+ CH
3
O
O
2
f CH
3
O + CH
3
O + O
OH + O
C(O)OOH + O
2
2
3
2
f HCHO + CH
f CH
f CH
3
2
ref 10
2
2
2
+ HO
+ HO
2
3
2
refs 10, 14
refs 10, 14
ref 20, 390 Torr 298 K
ref 10
ref 21
ref 21
2
3
C(O)OH + O
3
2
+ HO
+ CH
CHO f CH
+ M f H
2
O
2
+ O
2
+ M
2
2
3
O
2
f CH
3
OOH + O
3
3
C(O) + HCl
C(O)Cl + Cl
+ HCl
3
C(O) + Cl
2
f CH
3
4
f CH
3
ref 20
TABLE 4: Sensitivity Calculations for k
1
, k4a, and k4b in Units of 10-11, 10-12, and 10-13 cm molecule s-1, respectivelya
3
-1
rate constant or cross sections varied
error range
k
1
(+,-)
k
4a(+,-)
k4b(+,-)
2
2
2
CH
CH
CH
CH
CH
3
3
3
C(O)O
2
f 2CH
3
3
C(O)O + O
O + O
OH + O
2
(0.2E-11
(0.2E-13
(0.3E-13
(0.4E-11
(0.3E-11
(0.6E-12
(0.8E-12
(10%
-,-
9.57, 7.23
8.54, 8.54
8.54, 8.54
9.09, 8.00
8.98, 8.12
8.54, 8.54
8.77, 8.54
9.01, 5.43
8.98, 8.07
7.71, 5.86
10.1, 8.90
4.71, 13.5
8.47, 8.26
8.68, 8.26
8.77, 8.13
8.47, 8.16
8.26, 8.26
8.47, 8.26
27.6, -
O
O
2
f CH O + CH
f HCHO + CH
3
2
1.38, 1.38
1.38, 1.38
1.41, 1.37
1.41, 1.37
1.38, 1.38
1.38, 1.38
1.31, 1.39
1.42, 1.35
1.55, 1.18
1.49, 1.34
2
3
2
3
C(O)O
C(O)O
2
+ HO
2
f CH
f CH
+ O
f CH OOH + O
cross sections
3
C(O)OOH + O
2
3
2
+ HO
2
3
C(O)OH + O
3
2
HO
HO
absolute CH
absolute HO
absolute CH
all absolute cross sections
2
+ M f H
2
O
2
2
+ M
2
+ CH
3
O
2
3
2
3
O
2
2
cross sections
C(O)O cross sections
(10%
8.68, 8.14
-, 41.4
3
2
(10%
(5%
10.9, 8.52
a
Typical data set refitted with one reaction rate constant or one set of σ varied within given error limits.
Analysis of the “high CH3CHO” data sets yielded slightly
with caution.) Since k13 is so slow, it was undetermined by the
data and is therefore unimportant in this system. Other reactions
of CH3CHO or the radical CH3CH(OH)O2 may explain the
remaining discrepancy in the fits, but further investigations were
not made here. Instead, the rate constants obtained from the
“high CH3CHO” data fits were not incorporated into the reported
cross reaction rate constants.
-
11
different values, k1 ) (1.51 ( 0.03) × 10 , k4a ) (5.3 ( 2.4)
-
12
-12
3
-1 -1
×
10 , and k4b ) (1.0 ( 0.4) × 10
cm molecule s ,
with larger errors. Deviations from the measured absorbance
curves were observed at times greater than about 1.5 ms, and
acceptable fits were not attained with the simulations. Refitting
the “high CH3CHO” data sets up to only 1 ms resulted in better
-11
3
fits and an average rate constants of k1 ) 1.37 × 10
cm
For FACSIMILE fits performed on data sets with CH4 added,
molecule- s , k4a ) 8.3 × 10
1
-1
-12
cm molecule s , and k4b
3
-1 -1
-11
3
-1 -1
k1 was held at a constant 1.36 × 10 cm molecule
s
and
13
3
-1 -1
)
9.1 × 10- cm molecule s , in good agreement with
only k4a and k4b were allowed to vary. k1 should be better
determined by the “low CH3CHO” data, since the ratio of the
CH3C(O)O2 radicals lost by self reaction to those lost by CH3O2
cross reactions is initially larger in the “low CH3CHO” data
sets than in systems with added CH4. As before, the initial
CH3C(O)O2 and CH3O2 concentrations were determined by
fitting the respective concentration profiles (derived from eq
E) back to time ) 0. Best fit values of k4a ) (9.1 ( 2.1) ×
the “low CH3CHO” data sets. Secondary chemistry between
CH3CHO and CH3C(O)O2 or some product of the CH3C(O)O2
decomposition was suspected in these “high CH3CHO” mea-
surements, and attempts were made to refit these data, incor-
porating the reactions
CH CHO + CH C(O)O f CH C(O)OOH + CH CO (13)
3
3
2
3
3
-12
3
-1 -1
-12
3
1
0
cm molecule
s
and k4b ) (1.2 ( 0.6) × 10
cm
-
1
-1
CH CHO + CH O f CH OH + CH CO
(14)
molecule
s
were determined from these data.
3
3
3
3
Sensitivity Analysis. The errors given for k1, k4a, and k4b
thus far reflect only the variation in these parameters as
calculated by the FACSIMILE program. They do not take into
account possible uncertainties in the input cross sections used
to calculate the radical concentrations or in the “known” rate
constants. To better understand which of these input parameters
are the most influential in determining k1, k4a, and k4b,
FACSIMILE runs were performed in which a particular rate
constant or set of cross sections (see Table 4) was systematically
varied within estimated error limits and a typical “low
CH3CHO” data set was refitted. The error limits of the rate
constants were taken directly from the references given in Table
3 when possible. In cases where the branching ratio and rate
constants were determined in separate studies, the error limits
were calculated from the reported errors of each value. The
effects of errors in the CH3C(O)O2 + CH3O2 rate constants on
CH CHO + HO T CH CH(OH)O
2
(15)
3
2
3
-
17
3
-1
The K15(eq) was held fixed at 1 × 10
cm molecule (a
value determined in ref 22) and the rate constants were allowed
-
16
3
to vary during the fits; initial values, k13 ) 1.2 × 10
cm
molecule- s , k14 ) 1.5 × 10
1
-1
-13
cm molecule s , k15 )
3
-1 -1
-
15
3
-1 -1
-1
1
.0 × 10
cm molecule s , and k-15 ) 100 s , were
estimated from literature.
22-24
Even though the agreement
between the measured and FACSIMILE generated absorbance
profiles at 210 nm was not perfect, the fits were greatly
improved by the addition of reactions 14 and 15. The optimized
-
13
3
-1 -1
fitted values were k14 ) 2.0 × 10
cm molecule s , k15
-15
3
-1 -1
-1
)
(
3.8 × 10
cm molecule s , and k-15 ) 380 s .
Note: Because the fits were still not acceptable, these optimized
parameters could be incorrect and, therefore, should be viewed

Absorption Spectrum and Kinetics of CH3C(O)O2
J. Phys. Chem., Vol. 100, No. 10, 1996 4045
k1 were not evaluated since these rate constants depend on k1.
Uncertainties in the absolute scales of the CH3O2 and HO2
spectra of 10% were estimated from the maximum deviations
between the cross sections measured here (Table 2) and the
Discussion
Absorption Cross Sections.
A
comparison of the
CH3C(O)O2 spectrum obtained here and the previously mea-
sured spectra is shown in Figure 4. Also shown in Figure 4 is
10,11,17,25
literature values.
(See Discussion.) Possible errors due
9
the Moortgat et al. spectrum scaled down 17.5%. (The scaling
to incomplete Cl to radical conversion or unknown radical losses
yielded an estimated uncertainty in the absolute scale of the
CH3C(O)O2 spectrum of 10%. Finally, the effect of a C2H5O2
calibration error was examined by varying the absolute cross
sections of all absorbers (5%. The original fitted parameters
for the particular “low CH3CHO” data set used were k1 ) 1.38
factor is an average of the recommended corrections given by
the two recent peroxy radical reviews.
greater than 225 nm, our data are in reasonably good agreement
with the averaged review data and the data of Addison et al.
1
0,11
) At wavelengths
7
Large deviations in the shape and magnitude (as much as 19%
at 207 nm) are noted when our data and those of Addison et
-
11
-12
-13
3
×
10 , k4a ) 8.54 × 10 , and k4b ) 8.26 × 10
cm
7
al. are compared at wavelengths less than 225 nm, however,
-1
-1
molecule s , and the results are shown in Table 4.
suggesting that the longer wavelength agreement between these
two data sets is perhaps fortuitous. The absolute cross sections
and relative peak heights of Basco and Parmar are notably
different from ours. (It should be pointed out that the position
of the shorter wavelength peak and its cross section are not
exactly clear in the Basco and Parmar paper; in the text, the
absorption maximum is cited to occur at 207 nm, while in their
Figure 6, the spectrum is shifted toward longer wavelengths.
Shown here are the values taken from their figure.) Even though
The sensitivity calculations show, quite clearly, that k1 is
unaffected by the stated changes in the rate constants, indicating
that errors in these rate constants will have little or no influence
on the determination of k1. k1 is sensitive to changes in the
cross sections, and these fluctuations define the real error in
8
8
-
11
the k1 determination. A value of k1 ) (1.36 ( 0.19) × 10
3
-1 -1
cm molecule is hence concluded from this analysis. The
s
largest variations in k4a and k4b coincide with changes in the
fitted k1, suggesting that k4a and k4b are sensitive to changes in
both the varied rate constant or cross sections and k1. Since
the variations in k4a and k4b given in Table 4 are not independent
of errors in k1, real errors in these rate constants cannot be
assigned from this sensitivity analysis. Averaging the “low
CH3CHO” and “CH4 added” results yields k4a ) (8.8 ( 1.5) ×
9
our data most closely resembles the rescaled Moortgat et al.
spectrum, the ratios R ) σ210nm/σ240nm still differ. Identical
sources of CH3C(O)O2, namely the photolysis of Cl2 in the
presence of CH3CHO and O2, were employed in every study,
so discrepancies between the data sets cannot be attributed to
any systematic errors in the radical production mechanism.
-
12
3
-1 -1
-12
3
7
1
0
cm molecule
s
and k4b ) (1.0 ( 0.5) × 10
cm
Addison et al. did add NO2 to the photolysis mixtures in
-
1
-1
molecule s , where the quoted errors are the calculated
standard deviations of measurements from both sets.
order to scavenge the CH3C(O)O2 and 95% yields of PAN were
reported. The absorption cross sections of PAN used in the
spectral subtraction were determined in the same laboratory,
but were not cited in the paper. The PAN data were said to be
substantially confirmed by previously reported values of
The possibility that our CH3C(O)O2 spectrum contains
contributions from CH3O2 was also considered by adjusting the
relative cross sections of CH3C(O)O2. If the CH3C(O)O2 had
partially self reacted to form CH3O2 at the time of our
measurements, then the shape of our CH3C(O)O2 spectrum
would be altered; the shorter wavelength feature of the
CH3C(O)O2 spectrum would be too low, the longer wavelength
feature would be too high, and thus our R ) σ210nm/σ240nm ratio
would be too low. The magnitude of error in this ratio would
depend on the amount CH3C(O)O2 lost. For the sake of
comparison, two runs were performed with R ) 2.5, the σ210nm/
2
6
27
Stephens , which has since proven to be 20-30% too high.
Such an error would have a considerable effect in the
CH3C(O)O2 spectrum, particularly in the short wavelength
region, where the PAN cross sections are the largest.
A calibration error might explain the difference in the absolute
8
scale of the Basco and Parmar spectrum, but this will not
account for the disparity in ratio of the peak heights. The
apparent mislabling of the wavelength axis in their Figure 6
does create some uncertainty in the calculation of the peak height
ratio; however, this uncertainty is too small to justify such a
large difference. The presence of CH3O2 in their measurement
could have resulted in a cross section at 240 nm that is too
large. Calculations indicate that their initial CH3C(O)O2 radical
concentration was an order of magnitude larger than ours,
meaning that the decay rate of CH3C(O)O2 to produce CH3O2
was faster in their experiments. It is therefore possible that their
CH3C(O)O2 spectrum contains some contribution of CH3O2.
9
σ240nm ratio of the Moortgat et al. data, and either σ240nm )
-
18
-18
3
.21 × 10
(this work) or σ240nm ) 3.01 × 10
(Moortgat
9
et al. data rescaled down 17.5%). FACSIMILE was unable
to fit the data using either set of parameters, suggesting that
the 2.5 ratio is incorrect or that another parameter should
simultaneously be changed in order to compensate. Due to the
tremendous number of possible combinations, sensitivity tests
with multiple parameter changes were not performed.
A generalized sensitivity analysis of all data sets obtained
with added CH4 was not possible because the fluxes through
the various channels, and hence the sensitivities to these
channels, varied with the initial CH3C(O)O2/CH3O2 concentra-
tions. For measurements in which the CH3O2 concentration is
low, the results should resemble those shown in Table 4,
calculated in the absence of CH4. As the CH3O2 concentration
increases, the relative flux through reaction 4 will increase, while
that through reaction 1 will decrease. At the same time, the
absolute flux through the cross reaction channel will decrease
because the CH3C(O)O2 concentration is being reduced. In
addition, the influence of the CH3O2 self reactions will increase
with the CH3O2 concentration and the fitted parameters, k4a and
k4b, will become more sensitive to the values (or errors) of the
CH3O2 self reactions.
9
Although rescaling the spectrum of Moortgat et al. does
correct for the calibration error in initial CH3C(O)O2 concentra-
tions, it does nothing to rectify the erroneously high HO2 cross
sections which were input parameters in the computer simula-
tions of absorbance profiles. Furthermore, the CH3O2 cross
9
sections used by Moortgat et al. also appear too high. The
10,11
recommended CH3O2 cross sections
are 5-8% lower than
9
those employed by Moortgat et al. The effect of using incorrect
cross sections in the simulations can best be understood by
examining eq E (also used in Moortgat et al. simulations). The
9
contribution of any one absorber to the total absorption (Absλ,t)
is dependent on the product of the absorber’s cross section and
concentration; hence, the largest effects of incorrect cross
sections will occur at wavelengths at which the cross sections
of the particular absorber are large and over the time periods

4
046 J. Phys. Chem., Vol. 100, No. 10, 1996
Roehl et al.
when its concentration is large. The σ(HO2)’s increase with
decreasing wavelength, suggesting that errors in σ(HO2) might
become important at shorter wavelengths. The overall influence
of incorrect σ(HO2) on the simulations is expected to be small,
however, since the HO2 concentrations are low throughout the
experiments and the shape of the absorbance profiles are
primarily determined by CH3C(O)O2 and CH3O2. At wave-
lengths shorter than about 225 nm, the absorbance profiles
closely resemble rapid second order decay curves of
CH3C(O)O2 since σ(CH3C(O)O2) . σ(CH3O2). Errors in the
σ(CH3O2) at these wavelengths are, therefore, not expected to
introduce large errors in the simulations. At longer wavelengths
though, the CH3O2 cross sections are larger than the
CH3C(O)O2 cross sections; the shapes of the absorbance profiles
are strongly dependent on σ(CH3O2) and actually appear to
initially increase as the CH3O2 concentration increases. Simula-
tions of the measured absorbance profiles conducted at longer
wavelength, using σ(CH3O2) values which are too high, are
expected to result in σ(CH3C(O)O2) values which are too low.
Due to the complexity of the fitting scheme, no attempt was
made to resimulate the data.
experimental flaws and excluded three measurements from 1991,
which had not yet been published.
The CH3C(O)O2 self reaction rate constant k1 determined in
the analysis of the “low CH3CHO” data sets is in perfect
agreement with that obtained from the “high CH3CHO” data
sets when the latter data sets are only fit to 1 ms. Secondary
chemistry between CH3CHO and products of the CH3C(O)O2
decomposition seems likely in the "high CH3CHO" measure-
ments and explains the deviations in the fits at times greater
than 1 ms.
The k4a and k4b obtained from the “low CH3CHO” data sets
are lower than those from measurements with added CH4 by a
factor of 1.1 and 1.5, respectively. Although the errors on the
“
low CH3CHO” values are smaller, fits of these experiments
are not expected to be as sensitive to the cross reaction rate
constants as fits of the experiments with added CH4. Since
CH3O2 is regenerated in reactions 2 and 3, there is no net change
in CH3O2 concentration due to reaction 4a and k4a is solely
determined by changes in the CH3C(O)O2 concentrations, after
the loss due to the dominating self reaction is accounted for.
Reaction flux calculations conducted with typical “low
9
Another possible problem in the Moortgat et al. work might
CH CHO” conditions have shown that reaction 4a only becomes
3
have arisen from insufficient time resolution of their absorbance
traces. Although it is not stated in their paper, it appears from
the plots that the time resolution of their measurements was on
the order of 50 µs. If the time resolution of the measurement
is inadequate, the shape of the rapidly decreasing absorbance
profiles at shorter wavelength will be distorted and the
extrapolation to zero time will be incorrect. (Note: our data
acquisition was a factor of 10 faster and hence should have a
smaller error associated with it.) Of course the curvature in the
absorbance trace depends not only on the wavelength of the
measurement, but also on the initial CH3C(O)O2 concentration.
competitive with reaction 1 at times greater than 1.5 ms and at
that point the reaction flux through reaction 4a is small. Also,
even though reaction 4b is derived from changes in both CH O
3
2
and CH C(O)O concentrations, the rate constant is relatively
3
2
small and reaction 4b remains a minor channel throughout the
entire reaction time period. Initial production of CH O by the
3
2
addition of CH4 increases the flux through the cross reaction,
and hence, the sensitivity to the cross reaction rate constant is
expected to be greater.
The large error associated with the CH4 added data was
unanticipated, but can perhaps be explained by changes in the
relative CH3CHO and CH4 concentrations, and thus the ratio
of initial radical concentrations, during the course of the
measurements. Small changes in the absorbance profiles could
have been masked in the noise of the individual data sets of
500 laser pulses used in the final averaging. In the “low
CH3CHO” sets, the initial CH3C(O)O2 concentrations only
depend on the Cl concentration (and laser power) and slight
changes in the CH3CHO concentration, which is in excess over
the Cl, would have no effect.
9
Initial radical concentrations of Moortgat et al. were around
1
4
-3
1
.1 × 10 molecule cm , a factor of 2.2-5.5 larger than the
concentrations used here.
9
Finally, examination of Moortgat et al. absorption traces
reveals that negative dips directly follow the flashes and that
absorbances do not reach an immediate maximum. The length
of their flash pulse (not stated in their paper) is assumed to be
much longer than our laser flash and perhaps the detector does
not fully recover from the influence of the flash lamp.
Determinations of the initial CH3C(O)O2 concentrations might
be affected by the long delay in the signal recovery.
Due to the limitations of each of the data sets (both with and
without CH ) discussed above, neither set was preferred over
4
Rate Constants: k1, k4a, and k4b. As was demonstrated by
the sensitivity analysis, correct cross sections of all absorbers
are critical to the determination of accurate rate constants. In
the hopes of eliminating systematic differences resulting from
monochromator resolution, path length, etc., the CH3O2 and HO2
cross sections were remeasured along with the CH3C(O)O2
rather than being taken from the literature. The CH3O2 cross
sections measured in this work are 10% lower than the
recommended values of Lightfoot et al;10 however, they clearly
lie within the range of data used in arriving at this recom-
the other and the final rate constants k4a and k4b were derived
from averaging both the “low CH3CHO” and “CH4 added”
-
12
3
-1 -1
results. Our k4 ) (9.8 ( 2.0) × 10 cm molecule
s
does
fall within the error limits of the most recent measurement by
1
5
-12
3
-1
Lesclaux et al., k4 ) (7.6 ( 3.8) × 10
cm molecule
-1
s . It should be noted, however, that the branching ratio
15
employed by Lesclaux et al. was assumed to be k4a/k4 ) 0.48,
as determined earlier by Moortgat et al. A branching ratio of
9
k4a/k4 ) 0.90 ( 0.24 was obtained in this study in good
agreement with the most recently quoted value of k4a/k4 ) 0.83
-18
2
14
mendation (σ240nm ranged from 4.03 and 4.97 × 10
cm
( 0.17 by Horie and Moortgat. Different CH3C(O)O2 cross
-1
7
molecule ). Our values are only 7% lower than the Wallington
sections were used in the older works of Addison et al., Basco
1
1
8
9
et al. recommendations. Since the reviews were published,
and Parmar, and Moortgat et al., and thus agreement between
the rate constants determined in this work and theirs is
coincidental at best. Different cross sections would mean
different concentration profiles and, therefore, different rate
constants, indicating that comparisons between our rate constant
and those of the above-mentioned works, as such, are not
meaningful.
17,25
two new CH3O2 spectra have been reported
with σ(240nm)
cm molecule , which support our
lower values. The HO2 data measured here are in excellent
-
18
2
-1
)
4.40 and 4.22 × 10
1
0
agreement with the recommended values of Lightfoot et al.,
varying less than 1% at 210 nm, but are about 8% lower than
11
the recommended cross sections of Wallington et al. It should
be pointed out, however, that the Wallington et al.11 recom-
mendation included several older data sets which have noted
Using the temperature dependence of the relative cross
1
4
reaction rate constants of Horie and Moortgat â ) k4a/k4b )

Absorption Spectrum and Kinetics of CH3C(O)O2
J. Phys. Chem., Vol. 100, No. 10, 1996 4047
6
-12
2
.2 × 10 exp(-3870/T) and our k4 ) (9.8 ( 2.0) × 10 cm3
have become aware of a similar study submitted recently to J.
Phys. Chem. by Maricq and Szente. The UV cross sections of
CH3C(O)O2 and the CH3C(O)O2 + CH3O2 rate constants (k4)
presented here and in the Maricq and Szente work are in
excellent agreement, and the CH3C(O)O2 self-reaction rate
constants (k1) agree within the error limits. An unexplained
discrepancy exists, however, between the branching ratios of
k4 determined in the two studies.
-
1
-1
molecule s , the temperature dependent rate constants can
be calculated:
-
12
6
k ) 9.8 × 10 /(1 + (1/2.2 × 10 exp(-3870/T)) (F)
4a
-12
6
k ) 9.8 × 10 /(2.2 × 10 exp(-3870/T) + 1) (G)
4b
References and Notes
(
(
(
1) Atkinson, R. Atmos. EnViron. 1990, 24A, 1.
These equations are valid only over the temperatures (263-
2) Roberts, J. M. Atmos. EnViron. 1990, 24A, 243.
-12
3
3
33 K) used in ref 14 and with the value k4 ) 9.8 × 10 cm
3) Singh, H. B.; O’Hara, D.; Herlth, D.; Bradshaw, J. D.; Sandholm,
-1 -1
molecule s , assuming k4 is independent of temperature as
S. T.; Gregory, G. L.; Sachse, G. W.; Blake, D. R.; Crutzen, P. J.;
Kanakidou, M. A. J. Geophys. Res. 1992, D15, 16511.
suggested in ref 9.
(4) Platt, U.; LeBras, G.;Poulet, G.; Burrows, J. P.; Moortgat, G. Nature
1
990, 348, 147.
(5) Madronich, S.; Calvert, J. G. J. Geophys. Res. 1990, 95, 5697.
6) Madronich, S.; Chatfield, R. B.; Calvert, J. G.; Moortgat, G. K.;
Conclusions
(
Veyret, B.; Lesclaux, R. Geophys. Res. Lett. 1990, 17, 2361.
7) Addison, M. C.; Burrows, J. P.; Cox, R. A.; Patrick, R. Chem. Phys.
Lett. 1980, 73, 283.
A laser flash photolysis experiment was used to measure the
ultraviolet absorption spectrum of acetylperoxy radical in the
(
(
(
8) Basco, N.; Parmar, S. S. Int. J. Chem. Kinet. 1985, 17, 891.
9) Moortgat, G. K.; Veyret, B.; Lesclaux, R. J. Phys. Chem. 1989,
1
95-280 nm range. The spectrum is bimodal, with a strong
-
18
2
-1
maximum at 207 nm, σ ) 6.67 × 10
cm molecule , and
9
3, 2362.
-
18
2
a weaker maximum near 240 nm, σ ) 3.21 × 10
cm
(10) Lightfoot, P. D.; Cox, R. A.; Crowley, J. N.; Destriau, M.; Hayman,
-1
G. D.; Jenkin, M. E.; Moortgat, G. K.; Zabel, F. Atmos. EnViron. 1993,
6A, 1.
11) Wallington, T. J.; Dagaut, P.; Kurylo, M. J. Chem. ReV. 1992, 92,
667.
(12) Kenley, R. A.; Traylor, T. G. J. Am. Chem. Soc. 1975, 97, 4700.
13) Weaver, J.; Meagher, J.; Shortridge, R.; Heicklen, J. J. Photochem.
molecule . CH3O2 and HO2 cross sections were remeasured
as a part of this study at several wavelengths within the range
mentioned above. The values σ240nm(CH3O2) ) (4.12 ( 0.41)
2
(
-
18
2
-1
×
10
cm molecule and σ210nm(HO2) ) (4.19 ( 0.42) ×
(
-
18
2 -1
1
0
cm molecule were obtained. A rate constant of (1.36
1
975, 4, 341.
11
3
-1 -1
(
0.19) × 10- cm molecule
s
was determined for the
(14) Horie, O.; Moortgat, G. K. J. Chem. Soc., Faraday Trans. 1992,
8, 3305, and personal communication.
8
acetylperoxy self reaction at 298 K, using the newly obtained
(15) Lesclaux, R.; Boyd, A.; Noziere, B.; Villenave, E. LABVOC
cross sections, absorption-time profiles, and the FACSIMILE
Report, Project EV5V-CT91-0038: Second Annual Report, Aug 4, 1994.
(16) Bauer, D.; Crowley, J. N.; Moortgat, G. K. J. Photochem. Photobiol.
A: Chem. 1992, 65, 392.
-12
3
program. Rate coefficients of k4a ) (8.8 ( 1.5) × 10
molecule cm molecule
and k4b ) (1.0 ( 0.5) × 10
were similarly determined for the acetyl- and methylperoxy
cm
-1
-1
-12
3
-1
s
(
(
17) Maricq, M. M.; Wallington, T. J. J. Phys. Chem. 1992, 96, 986.
18) Maricq, M. M.; Szente, J. J.; Kaiser, E. W. J. Phys. Chem. 1994,
-1
s
cross reactions. The resulting k4 is in good agreement with the
98, 2083.
(19) Curtis, A. R.; Sweetenham, W. P. AERE Rep. R-12805 1987 (U.K.
Atomic Energy Research Establishment).
20) DeMore, W. B.; Sander, S. P.; Golden, D. M.; Hampson, R. F.;
recent measurement of Lesclaux et al.,15 and the branching ratio
1
4
(0.90) is close to that reported by Horie and Moortgat.
(
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; Evaluation No. 11, JPL Publication 94-26; JPL: Pasadena, CA,
Acknowledgment. The authors wish to thank Dr. R.
Lesclaux for helpful comments on the manuscript. This work
was supported by the Commission of the European Community
as a part of Project EV5V-CT-0038 (LABVOC).
1
994.
(21) Mallard, W. G.; Westley, F.; Herron, J. T.; Hampson, R. F. NIST
Standard Reference Database 17; NIST Chemical Kinetics DatabasesVersion
6
.0; NIST: Gaithersburg, MD, 1994.
Note Added in Proof. While examining the literature for
(22) Kelly, N.; Heicklen, J. J. Photochem. 1978, 8, 83.
(
23) Weaver, J.; Meagher, J.; Shortridge, R.; Heicklen, J. J. Photochem.
975, 4, 341.
24) Moortgat, G. K.; Cox, R. A.; Schuster, G.; Burrows, J. P.; Tyndall,
G. S. J. Chem. Soc., Faraday Trans. 2 1989, 85, 809.
25) Wallington, T. J.; Maricq, M. M.; Ellermann, T.; Nielsen, O. J. J.
Phys. Chem. 1992, 96, 982.
26) Stephens, E. R. AdV. EnViron. Sci. Technol. 1969, 1, 119.
(27) Senum, G. I.; Lee, Y.-N.; Gaffney, J. S. J. Phys. Chem. 1984, 88,
1269.
the manuscript, several printing errors were noted in Section
1
1
0
III. B.3sAcetylperoxy radicals of the recent Lightfoot et al.
(
review. First, σ(205 nm) for ref 1 in Table III.12 should be
deleted; no such measurement exists. Three cross sections of
ref 2 are in error; the corrected values should be σ(205 nm) )
(
(
6
.99, σ(207 nm) ) 8.37, and σ(230 nm) ) 4.97. Finally,
-
18
σ(207 nm) ) 8.37 × 10
of ref 2 is missing from the plot in
Figure III.11. Since the acceptance of this work, the authors
JP9526298