Shock wave studies of the pyrolysis of fluorocarbon oxygenates. I. The thermal dissociation of C3F6O and CF3COF

Article abstract of DOI:10.1039/c6cp06816b

The thermal decomposition of hexafluoropropylene oxide, C₃F₆O, to perfluoroacetyl fluoride, CF₃COF, and CF₂ has been studied in shock waves highly diluted in Ar between 630 and 1000 K. The measured rate constant k₁ = 1.1 × 10¹⁴exp(-162(±4) kJ mol^-1/RT) s^-1 agrees well with literature data and modelling results. Using the reaction as a precursor, equimolar mixtures of CF₃COF and CF₂ were further heated. Combining experimental observations with theoretical modelling (on the CBS-QB3 and G4MP2 ab initio composite levels), CF₃COF is shown to dissociate on two channels, either leading to CF₂ + COF₂ or to CF₃ + FCO. By monitoring the CF₂ signals, the branching ratio was determined between 1400 and 1900 K. The high pressure rate constants for the two channels were obtained from theoretical modelling as k_5,∞(CF₃COF → CF₂ + COF₂) = 7.1 × 10¹⁴exp(-320 kJ mol^-1/RT) s^-1 and k_6,∞(CF₃COF → CF₃ + FCO) = 3.9 × 10¹⁵exp(-355 kJ mol^-1/RT) s^-1. The experimental results obtained at [Ar] ≈ 5 × 10^-6 mol cm^-3 were consistent with modelling results, showing that the reaction is in the falloff range of the unimolecular dissociation. The mechanism of secondary reactions following CF₃COF dissociation has been analysed as well.

Full text of DOI:10.1039/c6cp06816b

PCCP
View Article Online
View Journal
PAPER
Shock wave studies of the pyrolysis of
fluorocarbon oxygenates. I. The thermal
dissociation of C3F6O and CF3COF
Cite this:DOI: 10.1039/c6cp06816b
C. J. Cobos,a K. Hintzer,b L. Solter,c E. Tellbach,c A. Thalerb and J. Troe*cd
¨
The thermal decomposition of hexafluoropropylene oxide, C3F6O, to perfluoroacetyl fluoride, CF3COF,
and CF2 has been studied in shock waves highly diluted in Ar between 630 and 1000 K. The measured
rate constant k1 = 1.1 ꢀ 1014 exp(ꢁ162(ꢂ4) kJ molꢁ1/RT) sꢁ1 agrees well with literature data and
modelling results. Using the reaction as a precursor, equimolar mixtures of CF3COF and CF2 were
further heated. Combining experimental observations with theoretical modelling (on the CBS-QB3 and
G4MP2 ab initio composite levels), CF3COF is shown to dissociate on two channels, either leading to
CF2 + COF2 or to CF3 + FCO. By monitoring the CF2 signals, the branching ratio was determined
between 1400 and 1900 K. The high pressure rate constants for the two channels were obtained from
theoretical modelling as k5,N(CF3COF - CF2 + COF2) = 7.1 ꢀ 1014 exp(ꢁ320 kJ molꢁ1/RT) sꢁ1 and
Received 5th October 2016,
Accepted 19th December 2016
k
6,N(CF3COF - CF3 + FCO) = 3.9 ꢀ 1015 exp(ꢁ355 kJ molꢁ1/RT) sꢁ1. The experimental results obtained
at [Ar] E 5 ꢀ 10ꢁ6 mol cmꢁ3 were consistent with modelling results, showing that the reaction is in the
falloff range of the unimolecular dissociation. The mechanism of secondary reactions following CF3COF
dissociation has been analysed as well.
DOI: 10.1039/c6cp06816b
www.rsc.org/pccp
In this way, reaction (1) behaves similar to the dissociation of
perfluorocyclopropane,8 c-C3F6, which is also a convenient
1. Introduction
While the pyrolysis of fluorocarbons, CxFy, and hydrofluoro- source for CF2 through the reaction
carbons, CxFyHz, has been studied extensively (see e.g. ref. 1),
only little is known on the thermal dissociation of fluorocarbon
c-C3F6(+M) - C2F4 + CF2(+M)
(2)
oxygenates, CxFyOz. The present series of investigations start to Reactions (1) and (2) both are characterized by limiting high
fill this gap. In the former reactions, complex elimination pressure rate constants with low energy barriers and small
reactions with low energy barriers and comparably small pre- preexponential factors. At the same time, they may be followed
exponential factors of the high pressure rate constants were by the dimerization of CF2 through8–11
encountered in some cases. In other cases, the reactions were
CF2 + CF2(+M) - C2F4(+M)
(3)
dominated by simple bond scissions with high energy barriers
and large preexponential factors. It appears of interest to At higher temperatures, C2F4 from reaction (2) dissociates by
investigate whether the introduction of oxygen leads to similar the simple bond scission11
observations.
C2F4(+M) - CF2 + CF2(+M)
(4)
We start the present work with a study of the decomposition
of hexafluoropropylene oxide, C3F6O, which is known2–7 to be a with a much larger energy barrier and a larger preexponential
direct source of difluorocarbene, CF2, by the reaction
factor. Not much is known on the corresponding reaction of
perfluoroacetyl fluoride, CF3COF, formed by reaction (1).
IR multiphoton dissociation12 suggested the primary process
to be CF3COF - C2F4 + O. However, in view of the high
endothermicity of this reaction channel and the indirect
character of the experiments, this conclusion appears to be
uncertain. Other pathways like
C3F6O(+M) - CF3COF + CF2(+M)
(1)
a INIFTA, Facultad de Ciencias Exactas, Universidad Nacional de La Plata, Argentina
b Dyneon GmbH, Gendorf, D-84508 Burgkirchen, Germany
c
¨
¨
¨
Institut fur Physikalische Chemie, Universitat Gottingen, Tammannstrasse 6,
¨
D-37077 Gottingen, Germany. E-mail: jtroe@gwdg.de
CF3COF(+M) - CF2 + COF2(+M)
- CF3 + FCO(+M)
(5)
(6)
d
¨
Max-Planck-Institut fur Biophysikalische Chemie, Am Fassberg 11,
¨
D-37077 Gottingen, Germany
This journal is ©the Owner Societies 2017
Phys. Chem. Chem. Phys.

View Article Online
Paper
PCCP
- CF3CO + F(+M)
(7)
on the basis of energy arguments should be more probable.
The present work investigated the dissociations of C3F6O
and CF3COF in shock waves by monitoring the UV absorption
of CF2. Because of the strong and well characterized properties
of this absorption,11 observations could be made over largely
varying reaction conditions. At the same time, primary and
secondary reactions could be distinguished. Like in our previous
investigations of fluorocarbons and hydrofluorocarbons,8,11,13–15 the
experiments were accompanied by quantum-chemical calculations
of the reaction pathways and theoretical modelling of the rate
constants. This combination of experiment and theory is proved
to be most advantageous again. Subsequent to the present work, a
study of the decomposition of C4F8O will be reported.16
Fig. 1 Absorption–time profile of CF2 at 248 nm in dissociation C3F6O -
CF3COF + CF2 followed by dimerization 2CF2 - C2F4 behind a reflected
shock wave (T = 801 K, [Ar] = 7.8 ꢀ 10ꢁ5 mol cmꢁ3, relative reactant
concentration [C3F6O]t=0/[Ar] = 5.1 ꢀ 10ꢁ4; optical density OD p [CF2],
see the text; Schlieren peaks indicating the arrival of the incident and
reflected shock wave).
2. Experimental technique and results
The thermal decomposition of C3F6O was studied under high
dilution by the bath gas M = Ar behind incident and reflected
shock waves. The progress of the reaction was followed exclusively
by the strong and quantitatively characterized UV absorption of
observed in multiphoton dissociation17 of C3F6O, were not
noticed (as deduced from the observation of one CF2 formed
per one C3F6O decomposed under low temperature conditions).
Since reaction (1) has a much stronger temperature dependence
than reaction (3), by increasing the temperature one can accelerate
reaction (1) to such an extent that clean second order absorption–
time profiles for CF2 dimerization become observable. This is
illustrated in Fig. 2 for an experiment at 1100 K behind the
reflected shock. On the other hand, by lowering the reactant
concentration the second order dimerization can be slowed down
to such an extent that the absorption–time profile of CF2 is
dominated by a clean first-order profile from reaction (1). This is
shown in Fig. 3 by an experiment with 35 ppm of C3F6O in Ar,
instead of about 500 ppm of Fig. 1 and 2. Signals like Fig. 3 were
represented by
CF2 at 248 nm (with absorption coefficients11 e(T)/cm2 molꢁ1
=
3.35 ꢀ 106 + 4.6 ꢀ 107 exp(ꢁ[(T + 1457 K)/1272 K]2) ꢁ 245(T/K),
base e for e). Details of our technique have been described
before8,11,13–15 and are only briefly summarized here. We used a
shock tube of 9.4 cm inner diameter with a length of the test
section of 4.15 m and of the driver section of 2.80 m. Windows for
absorption measurements were placed 9 cm before the reflecting
end plate. The light source was a Xe high pressure arc lamp
(Osram XBO 150 W/4). A Zeiss MQ3 prism monochromator served
for wavelength selection and signals were recorded using a multi-
plier (Hamamatsu E2420) – transient recorder arrangement. C3F6O
(from abcr, 97% purity) was mixed with Ar (from Air Liquide,
99.9999% purity) in large mixing vessels before the experiments.
Mixtures with about 500 and about 50 ppm in Ar were prepared.
The concentration of C3F6O in the mixtures was controlled by the
amount of CF2 formed behind the shock waves. For the 500 ppm
mixtures, the concentration of CF2 in the experiments on reaction
(1) was equal to the nominal value from mixture preparation. For
the 50 ppm mixtures, some wall adsorption occurred in the vessel
and in the shock tube before the experiments. This loss of C3F6O,
however, could easily be accounted for, again from the CF2 yield in
low temperature experiments where one CF2 is formed per one
C3F6O decomposed. The accurate determination of the starting
concentration of C3F6O at various places of the analysis was
essential, see below.
[CF2] = [C3F6O]t=0{1 ꢁ exp(ꢁk1t)}
and signals like Fig. 2 by
(8)
2.1 Dissociation of C3F6O and dimerization of CF2
The formation and consumption of CF2 in reactions (1) and (3)
at a comparably low temperature of 800 K is illustrated in Fig. 1.
While the formation of CF2 by reaction (1) corresponds to a first
order process, CF2 disappears in the second order process of
reaction (3), resulting in absorption–time profiles such as
shown in the figure. Contributions from other reactions like
C3F6O - C2F4 + COF2 or C3F6O - CFCF3 + COF2, such as
Fig. 2 As Fig. 1, showing dominant dimerization 2CF2
-
C2F4 at
higher temperature (T = 1103 K, [Ar] = 5.5 ꢀ 10ꢁ5 mol cmꢁ3, [C3F6O]t=0
/
[Ar] = 5.3 ꢀ 10ꢁ4).
Phys. Chem. Chem. Phys.
This journal is ©the Owner Societies 2017

View Article Online
PCCP
Paper
Fig. 3 As Fig. 1, showing dominant dissociation C3F6O - CF3COF + CF2
at lower temperature and lower reactant concentration (T = 728 K, [Ar] =
8.7 ꢀ 10ꢁ5 mol cmꢁ3, [C3F6O]t=0/[Ar] = 3.5 ꢀ 10ꢁ5).
Fig. 4 Rate constants k1 for C3F6O dissociation (from present work; J:
[Ar] = (1–2) ꢀ 10ꢁ5 mol cmꢁ3, K: [Ar] = (5–10) ꢀ 10ꢁ5 mol cmꢁ3; the line
corresponds to a linear fit of the points).
1/[CF2] ꢁ 1/[C3F6O]t=0 = 2k3t
(9)
Numerical fitting of signals like Fig. 1 led to k1 and k3
simultaneously. Table 1 summarizes the resulting k1 together
with the applied experimental conditions while Fig. 4 illustrates
the temperature dependence of k1. We note that the reactant
concentrations were varied by about a factor of 15. By measuring k1
behind incident and reflected waves, the bath gas concentration
[Ar] was varied by about a factor of 20. Within this variation no
dependence of k1 on [Ar] was noticed such that reaction (1)
was found to be very close to the high pressure limit of the
unimolecular dissociation. This is in accord with our modelling
results (see below). The temperature dependence of k1, as
illustrated by Fig. 4, is well represented by
Table 2 Rate constants k3 for dimerization 2CF2 - C2F4 (rate law d[CF2]/
dt = ꢁ2k3[CF2]2, experiments in Ar with [C3F6O]t=0/[Ar] E 5 ꢀ 10ꢁ4
)
T/K
[Ar]/10ꢁ5 mol cmꢁ3
k3/1010 cm3 molꢁ1 sꢁ1
739
753
764
897
940
978
1109
8.5
8.0
8.1
7.0
6.5
5.9
1.0
7.2
7.2
7.2
8.6
8.1
6.8
4.1
The modelling results of the present work and the previous
results from ref. 5–7 are compared with eqn (10). The general
agreement was found to be most satisfactory such that k1 is well
established now between 410 K (from ref. 5) and about 1000 K
from the present work. (We note that k1 was measured in pure
C3F6O in ref. 5–7 while the present experiments were performed
under high dilution of C3F6O in Ar.)
In contrast to k1, k3 depends both on the temperature and
the bath gas concentration [Ar]. Table 2 summarizes the
derived k3 as a function of T and [Ar]. The data clearly indicate
that reaction (3) under the present conditions is in the inter-
mediate range between a low pressure third order and a high
pressure second order combination reaction. This is consistent
with the earlier results for the dimerization from ref. 8–10 and
the results for the reverse thermal dissociation (4) from ref. 11.
Fig. 5 demonstrates this behaviour by including the present
results into the representation of k3 (T, [Ar]) from ref. 8 at two
representative values of [Ar].
k1 = 1.1 ꢀ 1014 exp(ꢁ162(ꢂ4)kJ molꢁ1/RT) sꢁ1 (10)
Table 1 Rate constants k1 for dissociation C3F6O - CF2 + CF3COF
(experiments in Ar with [C3F6O]t=0/[Ar] E 5 ꢀ 10ꢁ4
)
T/K
[Ar]/10ꢁ5 mol cmꢁ3
k1/sꢁ1
628
632
659
670
677
692
694
696
715
739
744
751
753
763
764
783
801
805
824
861
888
918
945
9.7
10.0
9.9
9.1
2.4
9.2
8.9
8.6
2.3
8.5
2.1
1.9
8.0
2.0
8.1
6.9
7.8
1.8
1.7
1.5
1.5
1.4
1.3
4.3 ꢀ 100
3.5 ꢀ 100
1.9 ꢀ 101
2.9 ꢀ 101
3.5 ꢀ 101
7.0 ꢀ 101
8.9 ꢀ 101
9.1 ꢀ 101
1.8 ꢀ 102
4.1 ꢀ 102
5.1 ꢀ 102
6.5 ꢀ 102
7.9 ꢀ 102
9.7 ꢀ 102
1.0 ꢀ 102
1.3 ꢀ 103
2.1 ꢀ 103
2.8 ꢀ 103
2.2 Dissociation of CF3COF
4.7 ꢀ 103 After the dissociation of C3F6O by reaction (1) is complete, one
2.0 ꢀ 104
may study the subsequent dissociation of the reaction product
3.1 ꢀ 104
5.1 ꢀ 104
CF3COF. In the present case, equimolar mixtures of CF3COF
9.2 ꢀ 104 and CF2 formed by reaction (1) are studied at temperatures
This journal is ©the Owner Societies 2017
Phys. Chem. Chem. Phys.

View Article Online
Paper
PCCP
Fig. 5 Rate constants k3 for CF2 dimerization (see Fig. 6 of ref. 8; present
work: ,: [Ar] E 5 ꢀ 10ꢁ5 mol cmꢁ3, .: [Ar] = 1 ꢀ 10ꢁ5 mol cmꢁ3; results from
Fig. 6 Concentration–time profile of CF2 at 248 nm, showing CF2
formation from CF3COF dissociation behind incident shock (T2 = 1836 K
and [Ar]2 = 4.1 ꢀ 10ꢁ6 mol cmꢁ3) and CF2 dissociation behind reflected
shock (T5 = 3930 K, [Ar]5 = 9.5 ꢀ 10ꢁ6 mol cmꢁ3) (the vertical axis
corresponds to the CF2 yield Y(CF2) = [CF2]/[C2F6O]t=0) for the reaction
behind the incident shock; the time scale for the reaction behind the
incident shock is compressed by a factor of 3.39, see the text).
C2F4 dissociation:11 n: [Ar] E 5 ꢀ 10ꢁ5 mol cmꢁ3; &: [Ar] E 1 ꢀ 10ꢁ5 mol cmꢁ3
;
results from ref. 8: J: [Ar] E 5 ꢀ 10ꢁ5 mol cmꢁ3; results from ref. 10: K: in 1 bar
of Ar; results from ref. 10: ’: in 50 Torr of N2; full and dashed lines: high pressure
limit of k3 from ref. 8 and 11; dotted line: k3 at 5 ꢀ 10ꢁ5 mol cmꢁ3 and
dash-dotted line: k3 at 1 ꢀ 10ꢁ5 mol cmꢁ3 from ref. 8).
The observed time dependence of Y(CF2) behind the incident
wave in this way leads to the sum of k5 and k6 while the reached
level of Y(CF2) corresponds to the ratio k5/(k5 + k6). Experiments
as shown in Fig. 6 could be done at temperatures behind
the incident shock between 1400 and 1900 K at [Ar] E 5 ꢀ
10ꢁ6 mol cmꢁ3. The results, e.g. k5 E 5 ꢀ 102, 9.4 ꢀ 103, and
6.4 ꢀ 104 sꢁ1 for T = 1410, 1590, and 1840 respectively, were
found to be about a factor of 2 to 5 smaller than modelled high
pressure values of k5 given below. We attribute this observation
to falloff effects which should become visible for the higher
temperatures and a smaller molecular size of CF3COF compared
to the C3F6O experiments. In this way, the two dissociation
channels of CF3COF should behave similar to the dissociation
of C2F6 for which falloff effects were analysed in ref. 19. Falloff
curves for reactions (5) and (6) were characterized with the data
modelled below and represented in the Appendix.
considerably higher than those employed in the experiments
described in Section 2.1. The shock wave experiment now is
conducted in three stages. Behind the incident wave, first,
reaction (1) occurs so rapidly that, at the arrival of the shock
front, there is an absorption step arising from the primarily
formed CF2. The CF2 yield, Y(CF2) (defined by Y(CF2) = [CF2]/
[C3F6O]t=0), in this step increases from 0 to unity. Second,
during a subsequent dissociation of CF3COF by reactions (5)
and (6), Y(CF2) further increases; the rate of this increase and
the level of the reached Y(CF2) provide access to k5 and k6.
Finally, when the reaction mixture is further heated and
compressed in the reflected shock wave, CF3 formed in reaction
(6) first rapidly decomposes to CF2 and F, before all CF2 from
the various reactions decomposes at the now very high final
temperature. Fig. 6 shows an example for the time dependence
of the CF2 absorption signal such as recorded during this three-
stage experiment. In order to convert the CF2 absorption into
concentrations, one has to account for the changes in gas
2.3 High temperature reaction mechanism
density and temperature (and, from the latter, the CF2 absorp- Fig. 6 corresponds to an experiment with very high temperatures
tion coefficient) in the incident and reflected wave. For the behind the reflected shock such that CF2 dissociates thermally.
most relevant incident shock wave here, in Fig. 6 the CF2 When lower temperatures were generated in the reflected shock,
absorptions have been converted to the yields Y(CF2) and given the situation changed because then a mechanism of unimolecular
by the vertical axis. On the other hand, the horizontal axis and bimolecular reactions was found to influence the observed
corresponds to the laboratory time. For the stationary gas CF2 absorption–time profiles. Varying the reactant concentration
behind the reflected shock, like in Fig. 1–3, also in Fig. 6 this allows one to distinguish between first and second order
corresponds to the time scale of the reacting species. In Fig. 6 processes. First, we look at low-concentration experiments with
the decay of the CF2 signal at a temperature of 3930 K of the about 40 ppm of C3F6O in Ar. Fig. 7 illustrates the formation of
reflected shock then follows exactly the expected result for CF2 CF2 by reaction (1) at 770 K behind the incident wave. At the
decomposition from ref. 18. Evaluating the kinetics behind the temperature of the reflected wave of 1380 K, that part of
incident shock, one has to account for the fact that the shock the initial C3F6O which has not yet reacted at the arrival of
heated gas here flows towards the reflecting end plate of the the reflected wave decomposes almost instantaneously to
shock tube. As is well known, the apparent ‘‘compression of the CF3COF and CF2. For the conditions of Fig. 7, reaction (3) is
laboratory time’’ in comparison to the reaction time then too slow to occur. Systematically exploring the absorption steps
corresponds to the density ratio across the incident shock front at the arrival of the reflected wave, however, reveals that more
which, in the experiment of Fig. 6, amounts to a factor of 3.39. CF2 is present than corresponds to unity yield Y = 1 from
Phys. Chem. Chem. Phys.
This journal is ©the Owner Societies 2017

View Article Online
PCCP
Paper
the reflected wave with a concentration-independent component,
determined by reactions (5) and (6), and concentration-dependent
subsequent kinetics. These observations will be further rationalized
below on the basis of the calculated energetics of various pathways.
The fact that the increase and decrease of CF2 during the first 100 ms
behind the reflected shock in Fig. 8 were not observed in low-
concentration experiments like Fig. 7 indicates that bimolecular
reactions must have been involved, see below.
3. Calculation of energy profiles and
rate constants
Like in our previous work,8,11,13–15 we performed quantum-
chemical calculations at the CBS-QB320 and the G4MP221
ab initio composite levels. The G4MP2 calculations used B3LYP/
6-31G(2df,p) optimized geometries and harmonic vibrational
frequencies (scaled by 0.9854). In order to study the influence
of the basis set on these properties, additional calculations with
the larger basis set B3LYP/6-311+G(3df) were also carried out.
The Gaussian 09 software22 was employed for all calculations.
Conventional transition state theory was used to calculate
limiting high pressure rate constants for rigid activated complexes.
In the case of simple bond scissions with loose activated complexes,
statistical adiabatic channel model calculations with a standard
parameter a/b E 0.5 were employed.23 Limiting low pressure rate
constants were estimated according to the method of ref. 24. Falloff
curves were modelled following the method obtained in ref. 24–26.
First, the calculations for reaction (1) are described. The
relevant decomposition pathway C3F6O - CF3COF + CF2 was
confirmed to involve a rigid transition state of the CF3C(O)FCF2
structure such as illustrated in Fig. 9. After this structure has
been reached, the system decomposes directly to CF3COF + CF2.
The enthalpies (at 0 K, in kJ molꢁ1) of the transition state and of
the final products at different levels of theory were found to be
(161.1, 156.1, 156.1) and (123.0, 108.4, 107.9), resp., for calculations
with model a = CBS-QB3, model b = G4MP2, and model c = G4MP2//
B3LYP/6-311+G(3df) methods. The molecular parameters at the
three levels of theory are summarized in the Appendix. The
corresponding limiting high pressure rate constants in the form
of k1,N = A1,N exp(ꢁEa,1/RT) sꢁ1 were found to be represented by
A1,N = (2.8 ꢀ 1014, 2.8 ꢀ 1014, 3.0 ꢀ 1014) sꢁ1, and Ea,1 = (167.8,
162.8, 162.8) kJ molꢁ1 for the models (a, b, c). Previous work gave
experimental values of A1,N = (1.6 ꢀ 1013, 6.3 ꢀ 1013, 1.3 ꢀ 1014) sꢁ1
and Ea,1 = (151.9, 153.1, 161.9) kJ molꢁ1 in ref. 5, 6, 7 respectively.
The agreement between the previous experimental results and
the present calculations, e.g. is illustrated by k1(750 K) = (4.2 ꢀ
102, 1.4 ꢀ 103, 8.5 ꢀ 102, 5.7 ꢀ 102, 1.3 ꢀ 103, 1.4 ꢀ 103) sꢁ1 for
ref. 5, 6, 7, model a, model b, and model c. The experimental
result from the present work, such as given by eqn (10) was
k1(750 K) = 5.6(ꢂ1) ꢀ 102 sꢁ1 which is well consistent with the
previous experiments and the model calculations. Modelling of
the falloff curves of reaction (1) indicated that, at the low
temperatures of the present experiments and with a molecule
of a size as large as C3F6O, no falloff corrections to k1 were
required, i.e. k1/k1,N Z 0.9. The measured k1 thus corresponds
Fig. 7 Absorption–time profile of CF2 from C3F6O dissociation behind
incident shock wave (T2 = 770 K, [Ar]2 = 2.1 ꢀ 10ꢁ5 mol cmꢁ3) and
further heating CF3COF + CF2 by a reflected shock wave (T5 = 1382 K,
[Ar]5
=
4.5
ꢀ
10ꢁ5 mol cmꢁ3
) in a low-concentration experiment
([C3F6O]t=0/[Ar] = 4.0 ꢀ 10ꢁ5).
reaction (1). The CF2 yield directly behind the reflected shock
wave between 1000 K and 2000 K increases with temperature
until a maximum of about Y E 1.7 is approached such as also
demonstrated in Fig. 6.
Compared to the low concentration situation of Fig. 7, the
situation changed when higher reactant concentrations like
530 ppm C3F6O in Ar were investigated. Fig. 8 shows an
example. In this case, CF2 formation by reaction (1) is followed
by dimerization (3) behind the incident shock wave. C2F4 from
dimerization (3) decomposes nearly instantaneously by reaction
(2) at a temperature of 1736 K of the reflected wave. After this,
one observes a further increase of CF2 on a time scale of 10 ms
before some CF2 is consumed and a final much slower further
increase of CF2 sets in. We note that no other absorption than
that of CF2 is observable under our conditions. We also note that
a CF2 yield of unity in Fig. 8 corresponds to OD = eꢃxꢃ[CF2] = 0.4. Like
in Fig. 6 and 7 there is hence additional CF2 formation behind
Fig. 8 As Fig. 6, in a high-concentration experiment ([C3F6O]t=0/[Ar] = 5.3
ꢀ 10ꢁ4). The CF2 signal behind the incident wave (T2 = 918 K, [Ar]2 = 1.4 ꢀ
10ꢁ5 mol cmꢁ3) shows C3F6O decomposition and CF2 dimerization; the
CF2 signal behind reflected wave (T5 = 1736 K, [Ar]5 = 2.9 ꢀ 10ꢁ5 mol cmꢁ3
)
shows the influence of reaction (6) followed by reactions (13)–(17).
This journal is ©the Owner Societies 2017
Phys. Chem. Chem. Phys.

View Article Online
Paper
PCCP
Table 3 Calculated energy barriers (on the G4MP2 ab initio composite
level with B3LYP/6-31G(2df,p) optimized geometries, harmonic vibrational
frequencies scaled by 0.9854, and single-point post-Hartree–Fock ab initio
calculations, an average absolute energy uncertainty of 4.4 kJ molꢁ1
)
ꢀ
ꢀ
ꢄa
Reaction
DH0ꢄ kJ molꢁ1 DH0 kJ molꢁ1
C3F6O - CF3COF + CF2
CF3COF - COF2 + CF2
CF3COF - FCO + CF3
CF3COF - CF3CO + F
CF3COF - C2F4 + O
CF3COF - CF3CF + O
CF3COF + CF2 - CF3CO + CF3
FCO - CO + F
(1) 108.4
156.1
293.7
360.2
—
(5) 212.5
(6) 360.2
(7) 479.5
591.2
748.1
128.9
(14) 138.5
(17) 350.6
—
—
—
138.5
350.6
CF3 - CF2 + F
C2F6 in ref. 11 and 19, resp. The present calculations also strongly
suggested that CF3COF decomposition proceeds on the two
pathways (5) and (6) while other dissociations should be of only
very minor importance. More details of the falloff curves of
reactions (5) and (6) are given in the Appendix.
4. Discussion
In contrast to the dissociation of c-C3F6, where an intermediate
1,3 biradical CF2CF2CF2 (the ‘‘bond-stretched invertomer’’ of
ref. 27) with an energy minimum is passed before the rate-
determining energy barrier is reached, we did not find such an
energy minimum near to the transition state structure CF3COFCF2
of Fig. 9 for C3F6O dissociation. In any case, c-C3F6 and C3F6O
decomposition both are convenient CF2 sources releasing CF2 with
comparably small energy requirements. Accidentally the rate
constants are very similar, with preexponential factors of
Fig. 9 Structures of C3F6O (top) and the transition state on the way to
CF3COF + CF2 (bottom): calculations at the B3LYP/6-31G(2df,p) level, (see
the text), bond lengths in Å.
to the high pressure rate constant k1,N. This result is perfectly in
line with the results for the decomposition of c-C3F6 such as
investigated in ref. 8.
A1,N E 2 ꢀ 1014
s .
ꢁ1 and activation energies of Ea,1 E 170 kJ molꢁ1
In both cases the primary decomposition may be followed by CF2
dimerization leading to C2F4 and the values for k3, obtained in the
present work and in ref. 8, were found to agree. The corresponding
results were discussed before8,11 and need not to be considered
here, again.
Next we inspect the thermal decomposition of CF3COF, on
which nothing was known from previous work. As our experiments
only provided limited information, theoretical modelling of the
reaction and its pathways proved most helpful. For this reaction
we again did G4MP2 calculations such as characterized above. The
derived energy parameters are summarized in Table 3. Obviously,
pathways (5) and (6) appear most favourable, while pathway (7) as
well as the pathways leading to O atoms can be ruled out on
energetic reasons. Calculated molecular parameters are given
again in the Appendix. In this case, modelling of the kinetics
between 1400 and 1900 K led to
While our low temperature experiments led to mixtures of
CF3COF + (CF2, 1/2C2F4) of 1 : 1 composition, the further
heating of this mixture in reflected shock waves led to a more
complicated picture. Consistent with our calculations, part of
CF3COF decomposes to CF2 + COF2 on a time scale not much
slower than that of C2F4 dissociation.11 COF2 then is thermally
much more stable.28 Reaction (5), therefore, could clearly be made
responsible for an increase of the CF2 yield in low-concentration
experiments like Fig. 7 as well as in the time-resolved experiments
like Fig. 6. The high-concentration experiments of Fig. 8, however,
revealed a more complicated mechanism of secondary reactions.
We interpret the additional CF2 signals by a reaction sequence
which is initiated by the dissociation channel (6) of CF3COF, i.e.
k5,N = 7.1 ꢀ 1014 exp(ꢁ320 kJ molꢁ1/RT) sꢁ1 (11)
k6,N = 3.9 ꢀ 1015 exp(ꢁ355 kJ molꢁ1/RT) sꢁ1 (12)
The difference in the preexponential factors reflects the fact
that reaction (5) has a more rigid transition state than reaction
(6). Because CF3COF is smaller than C3F6O and the temperatures
of interest are higher than for C3F6O, one has to expect some
falloff reductions of k5 and k6 below the high pressure values.
This was observed in our experiments as illustrated in Section
2.2. This behaviour is analogous to the dissociations of C2F4 and
CF3COF - CF3 + FCO
which is then followed by
CF3 + CF3 - CF2 + CF4
(6)
(13)
Phys. Chem. Chem. Phys.
This journal is ©the Owner Societies 2017

View Article Online
PCCP
Paper
Rotational constants for the transition state (in cmꢁ1):
0.0674, 0.0350, 0.0305.
FCO - CO + F
F + CF2 - CF3
F + CF3 - CF4
CF3 - CF2 + F
(14)
(15)
(16)
(17)
2. For molecular parameters of CF3COF (calculated on the
G4MP2//B3LYP/6-31+G(2df,p) level), see the text.
Vibrational frequencies (in cmꢁ1): 45, 219, 229, 377, 417,
508, 583, 683, 758, 801, 1108, 1196, 1251, 1318, 1928.
Vibrational frequencies for the transition state of reaction
(5) (in cmꢁ1): 305i, 83, 145, 146, 347, 421, 527, 580, 673, 723,
766, 1001, 1344, 1377, 1822, (the torsion is treated as a free
rotor).
Further reactions like CF3COF - CF3CO + F, CF3CO - CF3 + CO
according to Table 3 should be of only minor importance. We
note that FCO and CF3CO are thermally very unstable,29,30
while CF4 decomposition requires much higher temperatures
to occur.31 The rate constants of reactions (15) and (16) are
known.19,31 On their basis one concludes that these reactions
influence the partial balance between CF3 formation and
consumption in a subtle way. Our observation of a marked
concentration dependence of the CF2 signals in Fig. 7 and 8
calls for a bimolecular CF2-forming reaction which we suggest
to be reaction (13). So far nothing has been known on this
disproportionation reaction. However, it is able to explain the
additional CF2 formation shortly behind the reflected wave in
Fig. 8 which then should be followed by CF2 consumption
through reaction (15) as long as F atoms are available. All fine
details of our CF2 signals in Fig. 8 could be reproduced by the
described mechanism. However, because of the complexity of
the mixture it did not appear warranted to further deduce rate
constants from this analysis. Experiments with more specific
precursors would appear more suitable for this purpose.
In summary, the present work provided experimental and
theoretical insights into the CF2-forming decomposition of
C3F6O. Using this reaction as a precursor for CF3COF formation,
we were also able to shed light on the decomposition of CF3COF.
While the former is a simple single-channel process, the latter
was shown to be dominated by the two channels (5) and (6).
Although C3F6O and c-C3F6 decomposition have very similar rate
constants for CF2 formation, their intermediates are suggested to
be of different character, of CF3C(O)FCF2 ketene-form in the
former and of 1,3 biradical-form CF2CF2CF2 in the latter case.
Like in fluorocarbons and hydrofluorocarbons, for the present
fluorocarbon oxygenates we observed rigid activated complex
elimination processes (reactions (1) and (5)) and bond scission
processes with looser activated complexes (reaction (6)). Unravelling
these processes is made possible by the combination of kinetics
experiments and rate theories.
Rotational constants (in cmꢁ1): 0.128, 0.0831, 0.0687.
Rotational constants for the transition state of reaction (5)
(in cmꢁ1): 0.132, 0.0767, 0.0640.
3. Modelling of falloff curves for the dissociation reactions
CF3COF - CF2 + COF2
(5)
and
CF3COF - CF3 + FCO
(6)
Calculations with modelled high pressure rate constants of
eqn (11) and (12) and low pressure rate constants
k5,0 = [Ar]9.0 ꢀ 1021(T/1600 K)ꢁ13.6 exp(ꢁ339 kJ molꢁ1/RT)
cm3 molꢁ1 sꢁ1
k6,0 = [Ar]7.2 ꢀ 1022(T/1600 K)ꢁ13.6 exp(ꢁ394 kJ molꢁ1/RT)
cm3 molꢁ1 sꢁ1
The modelling assumed a temperature-independent average
energy transferred per collision of hDEi/hc = ꢁ100 cmꢁ1 which
led to low pressure collision efficiencies bc of 0.046, 0.039, and
0.029 for reaction (5) and 0.051, 0.044, and 0.033 for reaction (6)
at T = 1400, 1550, and 1850 K, respectively. The falloff curves
were constructed following the method of ref. 24–26. The falloff
curves obtained are illustrated in Fig. 10. They can be represented
Appendix
1. For molecular parameters of C3F6O (calculated on the
G4MP2//B3LYP/6-311+G(3df,3pd) level), see the text.
Vibrational frequencies (in cmꢁ1): 47, 133, 158, 249, 265,
306, 348, 425, 504, 535, 565, 577, 612, 722, 771, 807, 1015, 1131,
1154, 1181, 1203, 1256, 1340, 1538.
Vibrational frequencies for the transition state (in cmꢁ1):
297i, 50, 70, 112, 172, 227, 243, 299, 377, 384, 454, 537, 587, 665,
678, 733, 809, 1035, 1185, 1194, 1263, 1291, 1304, 1673.
Rotational constants (in cmꢁ1): 0.0737, 0.0364, 0.0310.
Fig. 10 Modelled falloff curves for CF3C(O)F - COF2 + CF2 (in blue) and
CF3C(O)F - CF3 + FCO (in red). Calculations for 1400, 1550, 1700 and 1850 K
(from bottom to top), using rate coefficients described in the Appendix.
This journal is ©the Owner Societies 2017
Phys. Chem. Chem. Phys.

View Article Online
Paper
PCCP
¨
13 C. J. Cobos, L. Solter, E. Tellbach and J. Troe, J. Phys. Chem.
in the format of ref. 26 with center broadening factors of Fcent
=
A, 2014, 118, 4880–4888.
0.064, 0.065, and 0.070 for reaction (5) and 0.055, 0.053, and
¨
14 C. J. Cobos, K. Hintzer, L. Solter, E. Tellbach, A. Thaler and
0.053 for reaction (6) at T = 1400, 1550, and 1850 K, respectively.
J. Troe, Phys. Chem. Chem. Phys., 2015, 17, 32219–32224.
¨
15 C. J. Cobos, L. Solter, E. Tellbach and J. Troe, Phys. Chem.
Acknowledgements
Chem. Phys., 2014, 16, 9797–9807.
¨
16 C. J. Cobos, K. Hintzer, L. Solter, E. Tellbach, A. Thaler and J. Troe,
Financial support by the Deutsche Forschungsgemeinschaft
(Projekt TR 69/20-1) is gratefully acknowledged.
Phys. Chem. Chem. Phys., 2016, DOI: 10.1039/C6CP06822G.
17 V. A. Dunyakhin, O. V. Kuricheva, V. V. Timofeev and Yu.
N. Zhitnev, Russ. Chem. Bull., 1995, 11, 2079–2086.
18 C. J. Cobos, A. E. Croce, K. Luther and J. Troe, J. Phys. Chem.
A, 2010, 114, 4755–4761.
19 C. J. Cobos, A. E. Croce, K. Luther and J. Troe, J. Phys. Chem.
A, 2010, 114, 4748–4754.
20 J. A. Montgomery, M. J. Frisch, J. W. Ochterski and
G. A. Petersson, J. Chem. Phys., 2007, 127, 124105.
21 L. A. Curtiss, P. C. Redfern and K. Raghavachari, J. Chem.
Phys., 2007, 127, 124105.
References
1 D. R. Burgess, M. R. Zachariah, W. Tsang and P. R.
Westmoreland, Prog. Energy Combust. Sci., 1996, 21, 453–529.
2 P. B. Sargeant, J. Org. Chem., 1970, 3, 678–682.
3 P. B. Sargeant and C. G. Krespan, J. Am. Chem. Soc., 1969, 91,
415–419.
4 W. Mahler and P. R. Resnick, J. Fluorine Chem., 1973/74, 3,
451–452.
22 M. J. Frisch, et al., Gaussian 09, Revision A.02-SMP, Gaussian,
Inc., Wallingford, CT, 2009.
23 C. J. Cobos and J. Troe, J. Chem. Phys., 1985, 3, 1010–1015.
24 J. Troe, J. Phys. Chem., 1979, 83, 114–126.
25 J. Troe, Ber. Bunsenges. Phys. Chem., 1983, 87, 161–169.
26 J. Troe and V. G. Ushakov, Z. Phys. Chem., 2014, 1, 1–10.
27 H. Wei, D. A. Hrovat, W. R. Dolbier, B. E. Smart and
W. T. Borden, Angew. Chem., Int. Ed., 2007, 46, 2666–2668.
28 A. P. Modica, J. Phys. Chem., 1970, 74, 1194–1204.
29 V. D. Knyazev, A. Bencsura and I. R. Slagle, J. Phys. Chem. A,
1997, 101, 849–852.
5 R. C. Kennedy and J. B. Levy, J. Fluorine Chem., 1976, 7, 101–114.
6 Yu. N. Zhitnev, V. V. Timofeev and N. Yu. Ignat’eva, High
Energy Chem., 1994, 28, 484–491.
7 P. J. Krusic, D. C. Roe and B. E. Smart, Isr. J. Chem., 1999, 39,
117–123.
¨
8 C. J. Cobos, L. Solter, E. Tellbach and J. Troe, J. Phys. Chem.
A, 2014, 118, 4873–4879.
9 W. J. R. Tyerman, J. Chem. Soc., Faraday Trans., 1969, 65,
1188–1198.
10 F. Battin-Leclerc, A. P. Smith, G. D. Hayman and T. P. Murrells,
J. Chem. Soc., Faraday Trans., 1996, 92, 3305–3313.
¨
30 A. Tomas, F. Caralp and R. Lesclaux, Z. Phys. Chem., 2000,
214, 1349–1365.
11 C. J. Cobos, A. E. Croce, K. Luther, L. Solter, E. Tellbach and
J. Troe, J. Phys. Chem. A, 2013, 117, 11420–11429.
12 E. B. Aslanidi, V. T. Zarubin and Yu. S. Turishchev, Laser
Chem., 1986, 6, 373–380.
¨
31 G. Knight, L. Solter, E. Tellbach and J. Troe, Phys. Chem.
Chem. Phys., 2016, 26, 17592–17596.
Phys. Chem. Chem. Phys.
This journal is ©the Owner Societies 2017