CH3C(O)O2 + NO Reaction
J. Phys. Chem., Vol. 100, No. 30, 1996 12383
Errors in the UV measurements of k1 arise primarily from
signal noise (∼15%), i.e., from uncertainties in extracting
concentrations from the UV absorbances. Smaller errors of 7%
and 8% respectively arise from the 10% uncertainties in the
UV cross sections of CH3C(O)O2 and CH3O2. In spite of the
considerable secondary chemistry that takes place, uncertainties
in the rate constants used in the reaction model of Table 1
contribute relatively little to the error bars. This is primarily
due to the excess of NO as compared to the radical concentration
and the fact that the reaction under investigation is the first one
of the series. Uncertainties of 20% in the rate constants k2, k5,
k6, k11, and k14 contribute <2% each to the overall error.
Assuming that the individual uncertainties are statistically
independent leads to an overall error of roughly (20%.
c. IR Measurements. UV spectroscopy provides a means
to probe two of the participants in the title reaction, CH3C(O)-
O2 and the product CH3C(O)O, the latter being detected via
the methylperoxy radical into which it is rapidly converted. A
more complete investigation of this reaction is provided by IR
probes of the other two participants, namely, NO and NO2.
Examples are presented in Figure 4 of transient IR absorption
measurements that have been converted, via eq 9, into concen-
tration versus time profiles. Consistent with expectations from
the reaction mechanism, the quantity of NO consumed exceeds
the NO2 produced, which, in turn, is larger than the initial radical
concentration. The ratios observed for [Cl]0:∆[NO]∞:[NO2]∞
are not quite the 1:-3:2 expected from the principal reactions
(1), (5), and (6) because of the additional minor reactions and
because the initial NO concentration is not in large excess of
the radical population.
Figure 3. Concentration versus time profiles obtained from fitting the
time-resolved absorbances of Figure 1 to the reference spectra of Figure
2 (see text). The solid lines display the simultaneous best fit of the
data to the reaction model of Table 1, treating k1 and the scaling of
“products” as fitting parameters.
The smooth curves in Figure 4 represent fits of the reaction
model simultaneously to the NO and NO2 data treating k1 as an
adjustable parameter. As with the UV fits described above,
this represents an overconstrained system, particularly since the
yield of NO2 and the amount of NO consumed must be
consistent with the initial radical population. The latter quantity
is measured independently by substituting ethane for acetalde-
hyde, omitting NO, and measuring the amount of C2H5O2
produced. The fact that both the NO and NO2 traces are fit
well adds confidence to the values obtained for k1. Small
apparently systematic deviations between the data and the fits
arise primarily from systematic uncertainties in one or more of
contribute primarily to the peroxynitrate yield owing to the
significant overlap of its spectrum with those of the latter
compounds.
The time-resolved spectra are fit via eq 10 to four reference
spectra: those of CH3C(O)O2, CH3O2, the CH3ONO combina-
tion, and the peroxynitrate products. Modifying the procedure
used above to combine spectral contributions, e.g., omitting NO
from the CH3ONO composite, has little effect on the concentra-
tions of CH3C(O)O2 and CH3O2 determined by the fitting
procedure, a small effect on [CH3ONO], and a medium effect
on the apparent peroxynitrate yield. Because of the uncertainties
in the secondary chemistry and those introduced by the lumping
of reference spectra, the CH3ONO and peroxynitrate yields were
summed into an overall secondary product category. Figure 3
displays the results, showing the decay of CH3C(O)O2, the rise
and fall of CH3O2, and the net product accumulation.
For each UV experiment, the three time versus concentration
profiles are fit simultaneously to the reaction model of Table
1. Nominally, the only fitting parameter is k1, the rate constant
for the CH3C(O)O2 + NO reaction under investigation. Fitting
three concentration dependences with a single free parameter
represents an over constrained system. Unfortunately, because
of systematic errors such as the optical cross sections of the
products or the manner in which they are combined, the absolute
yield of the products is suspect; however, its relative variation
with time is more reliable. Thus, a second adjustable parameter
is introduced to scale the product yield, in order that the absolute
magnitude does not give undue weight to the quality of fit, and
hence the determination of k1. The solid lines in Figure 3
provide examples of the fits that are obtained via this procedure.
The resulting rate constants are listed in Table 2 and plotted as
a function of temperature in Figure 5.
the quantities σNO, σNO , or [Cl]0, which cannot be simulta-
2
neously accommodated by a model with a single degree of
freedom. The fact that the initial rise observed for NO2 lags
predictions may be caused by reaction (1) producing NO2 with
a non-Boltzmann distribution of vibrational excitation that is
then rapidly relaxed by collisions with buffer gas.
The principal sources of error incurred in the IR determination
of k1 are associated with the fact that three reactions consume
NO and the fact that two reactions produce NO2. Thus, an
uncertainty of 20% in k5, the methylperoxy reaction with nitric
oxide, translates into an error of 14% in k1. Because the changes
in [NO] and [NO2] are considerably larger than [Cl]0, k1 is also
sensitive to the uncertainty in this quantity; an uncertainty of
5% in [Cl]0 contributes a 12% error to k1. Additionally, 5%
uncertainties in σNO and σNO contribute to k1 errors of 13%
2
and 6%, respectively. The net error, assuming the above
uncertainties to be statistically independent and including a 5%
error due to signal noise, amounts to about (25%.
IV. Discussion
Figure 5 depicts the variation of the CH3C(O)O2 + NO rate
constant with temperature. The UV and IR measurements
provide consistent values for k1, with both methods indicating