Shock Tube Study of the Oxidation of C3F6 by N2O

Article abstract of DOI:10.1021/jp991065v

The kinetics of the high-temperature oxidation of C₃F₆ by O(³P) have been studied by experiment, using a single-pulse shock tube, and by kinetic modeling. The O atoms were generated by the thermal decomposition of N₂O. Three mixtures, each diluted in argon, were studied: 0.6 mol % of C₃F₆ with 1.5 mol % N₂O; 6.2 mol % of C₃F₆ with 0.6 mol % of N₂O; and 6.3 mol % of N₂O and 0.7 mol % of C₃F₆. The temperatures were in the range 1300-1600 K, the residence times behind the reflected shock were in the range 550-850 μs, and the pressures were between 16 and 20 atm. Fluorinated products have been quantified with gas chromatography, oxidized products with Fourier transform infrared spectroscopy; identification of unknown fluorocarbons has been performed with gas chromatography-mass spectrometry. The most significant products detected were C₂F₆, C₂F₄, CF₂O, CO, CO₂, and CF₄. A detailed kinetic scheme is presented to model the experimental reactant and product yields as a function of temperature. Modeling showed that O-addition to either carbon of the double bond of C₃F₆ occurs. The rate constant for O-addition to the terminal carbon of the double bond, C₃F₆ + O(³P) → ³CF₃CF + CF₂O, was deduced to be k₇₁ = 10^12.7T^0.05 exp(-0.4 kJ mol^-1/RT) cm³ mol^-1 s^-1, and for addition to the central carbon, C₃F₆ + O(³P) → CF₃ + CF₂CF=O, k₇₂ = 10^12.5 cm³ mol^-1 s^-1. Under oxidizer-rich conditions, ignition of the C₃F₆ occurred. Rate of production analyses showed that ignition was propagated by an F atom chain involving the CF₂ + O and unimolecular CFO decomposition reactions. Under C₃F₆-rich conditions, single- and double-bond pyrolysis were the important destruction routes.

Full text of DOI:10.1021/jp991065v

J. Phys. Chem. A 1999, 103, 5967-5977
5967
Shock Tube Study of the Oxidation of C3F6 by N2O
Robert G. Hynes,*,† John C. Mackie,*,† and Assaad R. Masri‡
School of Chemistry and Department of Mechanical and Mechatronic Engineering, UniVersity of Sydney,
NSW 2006, Australia
ReceiVed: March 26, 1999; In Final Form: May 19, 1999
The kinetics of the high-temperature oxidation of C3F6 by O(3P) have been studied by experiment, using a
single-pulse shock tube, and by kinetic modeling. The O atoms were generated by the thermal decomposition
of N2O. Three mixtures, each diluted in argon, were studied: 0.6 mol % of C3F6 with 1.5 mol % N2O; 6.2
mol % of C3F6 with 0.6 mol % of N2O; and 6.3 mol % of N2O and 0.7 mol % of C3F6. The temperatures were
in the range 1300-1600 K, the residence times behind the reflected shock were in the range 550-850 µs,
and the pressures were between 16 and 20 atm. Fluorinated products have been quantified with gas
chromatography, oxidized products with Fourier transform infrared spectroscopy; identification of unknown
fluorocarbons has been performed with gas chromatography-mass spectrometry. The most significant products
detected were C2F6, C2F4, CF2O, CO, CO2, and CF4. A detailed kinetic scheme is presented to model the
experimental reactant and product yields as a function of temperature. Modeling showed that O-addition to
either carbon of the double bond of C3F6 occurs. The rate constant for O-addition to the terminal carbon of
3
the double bond, C3F6 + O(3P) f CF3CF + CF2O, was deduced to be k71 ) 1012.7T0.05 exp(-0.4 kJ mol-1/
RT) cm3 mol-1 s-1, and for addition to the central carbon, C3F6 + O(3P) f CF3 + CF2CFdO, k72 ) 1012.5
cm3 mol-1 s-1. Under oxidizer-rich conditions, ignition of the C3F6 occurred. Rate of production analyses
showed that ignition was propagated by an F atom chain involving the CF2 + O and unimolecular CFO
decomposition reactions. Under C3F6-rich conditions, single- and double-bond pyrolysis were the important
destruction routes.
Introduction
In the case of C2F4, there is also a lack of high-temperature
(i.e., above 500 K) kinetic data for O atom addition. A recent
Hydrofluorocarbons such as 2-H heptafluoropropane, CF3-
CHFCF3, have become viable alternatives to ozone-depleting
halons for flame inhibition. Recent flame1 and shock tube2
studies on 2-H heptafluoropropane show that an important
product of inhibitor decomposition is C3F6, produced by the
1,2 elimination of HF. Since the flame chemistry of C3F6 is
uncertain at present, it is of interest to study in detail the high-
temperature oxidation of C3F6. For this purpose, the thermal
decomposition of N2O is used to produce O atoms. The
advantage of using N2O is that it begins to decompose and
liberate O atoms at temperatures well below those required for
pyrolysis of C3F6. Thus, at least for temperatures below the onset
of pyrolysis, N2O should be a useful precursor of O atoms to
study the reactions between these atoms and C3F6.
To date, little is known of the O atom addition kinetics to
unsaturated fluorocarbons at high temperatures. Heicklen and
Saunders3 previously studied addition of O to C3F6 at room
temperature, using the mercury-sensitized decomposition of N2O
as a source of O atoms. An important observation was that CF2O
was a major product and the CF3CFO yield was substantially
less. The rate constant for the reaction of O with C3F6 producing
CF2O and CF3C¨ F, measured in competition with C2H4, was
found to be 2.1 × 1010 cm3 mol-1 s-1 at 297 K. The other
O-addition channel producing CF3CFO and CF2 in an assumed
triplet state was found to be 6.5 times slower than that which
produces CF2O.
C2F4-O2 flame study4 utilizes an O-atom addition rate constant
of 8.13 × 1012 exp(-5.3 kJ mol-1/RT) cm3 mol-1 s-1. However,
this rate constant cannot be used as an analogy for the C3F6 +
O(3P) reaction because the latter reaction would be expected to
have two channels: addition to the central carbon atom, which
might be expected to form CF3CFO and triplet CF2, and addition
to the terminal carbon, which should lead to triplet CF3C¨ F
carbene (initially) and CF2O.
Therefore, the purpose of this work is to study the oxidation
of C3F6 in the presence of N2O. By measuring stable product
yields as a function of initial concentration of reactants and
temperature and comparing these yields to numerically modeled
profiles, the rate constants for the addition of O atoms to C3F6
can be deduced. The rate constants for other secondary processes
will also be determined.
Experimental Section
Quantification of the reactants and the fluorinated products
is carried out with a Hewlett-Packard 6890 gas chromatograph
equipped with a 25 m SGE-BP1 column (5 µm film thickness)
operated isothermally at -20 °C. Detection was by FID. N2O,
O2, and N2 were quantified using the same GC equipped with
a HP PLOT Molecular Sieve 5A column, operated with a
temperature program of -10 °C isothermal for 7 min, then
ramped up to 200 °C to elute the N2O. Detection was by TCD.
In this work, the oxidation of hexafluoropropene was studied
diluted in argon in three concentration ranges. First, the
decomposition of a near-stoichiometric mixture of 0.6 mol %
C3F6 and 1.5 mol % N2O was studied. In the second study, a
* Corresponding authors. E-mail for RGH: hynes_r@chem.usyd.edu.au.
E-mail for JCM: j.mackie@chem.usyd.edu.au. Fax: 61-2-9351-3329.
† School of Chemistry.
‡ Department of Mechanical and Mechatronic Engineering.
10.1021/jp991065v CCC: $18.00 © 1999 American Chemical Society
Published on Web 07/09/1999

5968 J. Phys. Chem. A, Vol. 103, No. 30, 1999
Hynes et al.
large excess of N2O was used, 6.3 mol % N2O with 0.7% C3F6.
Third, an excess amount of C3F6, 6.2 mol %, was reacted with
0.6 mol % of N2O. The assumed stoichiometry of this reaction
(to be discussed later) is
C3F6 + 3N2O f CF2O + CF4 + CO2 + 3 N2
The diluent gas was argon (British Oxygen Corporation Gases
99.995%; minimum purity). Pure hexafluoropropene (Mathe-
sons; 99+ % purity) was analyzed by GC and showed no
impurities; it was used without purification. The nitrous oxide
(British Oxygen Corporation Gases; 99.5% mimimum purity)
was similarly analyzed. An atmosphere of the N2O showed small
traces of N2 (0.25%) and O2 (0.066%). When used diluted, these
two impurities were undetectable by GC. A preshock GC
analysis (using both thermal conductivity and flame ionization
detection) of the C3F6/N2O/Ar reactant mixture did not reveal
the presence of any significant impurities.
The pressure and temperature behind the reflected shock were
calculated from the measured incident and reflected shock
velocities. Residence times were obtained from the pressure
profiles measured using Kistler pressure transducers. Signals
were then collected by a Tektronix TDS 430A two-channel
digitizing oscilloscope. Experiments with excess N2O were
performed over the temperature range 1300-1550 K, while
experiments with excess C3F6 were performed over the range
1400-1650 K. Typical residence times behind the reflected
shocks were 550-850 µs, with pressures between 16 and 20
atm. The excess C3F6 runs had residence times that were shorter
than the excess N2O runs by ∼100 µs.
Figure 1. Variation with temperature of the experimental C3F6, C2F4,
and C2F6 yields. The respective experimental points are represented
by the symbols (b, O, 1); model profiles are indicated by the lines:
(s, - - -, - - -). Initial mixture: 0.6 mol % C3F6 and 1.5 mol % N2O.
Quantification of fluorocarbons was performed by measuring
the FID response for C2F6 (Sigma-Aldrich; 98+% purity) and
C3F6. The FID response of C2F4, for which calibration samples
were not available, was assumed to be 0.667 of the response of
C3F6 on a molar basis. FTIR analysis was performed by taking
∼100 Torr of the reaction mixture and diluting it with argon to
1 atm. Sample CO and CO2 concentrations were determined
by comparing infrared peak areas with those of a gas mixture
(Scotty Specialty Gases, Mix 216) containing 1.00 ( 0.02% of
the said gases in nitrogen. Where sample concentrations were
less than 1%, a second mixture (British Oxygen Corp. Gases,
Special Beta Mixture) containing ∼420 ppm of CO and CO2
in N2 was used. Sample CF4 concentrations were determined
by comparing sample IR peak areas with those samples of the
pure compound (Mathesons 99.9% purity) diluted in argon.
CF2O was quantified by comparing sample peak areas with those
from a standard spectrum obtained at the same resolution and
same total pressure. All spectra were measured at a total pressure
of 1 atm with an instrument resolution of 0.5 cm-1. FTIR
calibration for CF3CFO (Fluorochem. Ltd.; 96.3% purity;
impurities due to CF3CClO and traces of acetic acid) was
performed using small concentrations of the gas in argon.
Identification of heavier products was performed with a Hewlett-
Packard 5890 II GC interfaced with a Hewlett-Packard 5898A
MS Engine, equipped with an SGE-BP1 column and operated
isothermally at -60 °C.
Figure 2. Variation with temperature of the experimental CO, CF2O,
and CO2 yields. The respective experimental points are represented by
the symbols (b, O, 1); model profiles are indicated by the lines: (s,
- - -, - - -). Initial mixture: 0.6 mol % C3F6 and 1.5 mol % N2O.
the formation of higher molecular weight fluorocarbons, includ-
ing polymers. However, polymers were not detected in the gas
samples.
Figures 1-3 show profiles of product yields versus the frozen
reflected shocked gas temperature. Yields for nitrogen-contain-
ing species are based on the initial N2O concentration, whereas
yields for fluorocarbon and partially and fully oxidized species
are based on the initial C3F6 concentration. Figure 1 shows that
significant C3F6 decomposition occurs at temperatures above
1400 K. To verify that this was not due to pyrolysis, a small
number of runs was performed with 0.6 mol % C3F6 in argon.
These experiments showed that pyrolysis is significant only
above 1600 K. This result is in accord with the C3F6 pyrolysis
study of Bauer et al.5 The yields of C2F4 and C2F6 were similar
at temperatures up to 1500 K. The maximum yield of both
species at 1500 K was 0.25. The most significant of the oxidized
products was CO. At 1500 K, the yield of CO was 0.24
compared with the CF2O yield of 0.12 and the CO2 yield of
only 0.03. An important observation was that CF3CFO was not
detected in any of the runs. This suggests that the room-
temperature oxidation route proposed by Saunders and Heicklen3
is inappropriate at high temperatures. Small quantities of NO
were formed. However, quantification of this species was
difficult due to its low absorption coefficient and hence low
signal-to-noise ratio. Subtracting C3F6, CF2O, and N2O signals
Results and Discussion
1. Mixtures of 0.6 mol % C3F6 and 1.5 mol % N2O (Near
Stoichiometric). Elemental carbon, oxygen, fluorine, and
nitrogen mass balances of unconsumed reactants and recovered
products in all sets were generally of the order of 100 ( 10%
for temperatures of e1500 K. At temperatures above 1560 K,
the fluorine and carbon recovery dropped to 80% suggesting

Shock Tube Study of the Oxidation of C6F6 by N2O
J. Phys. Chem. A, Vol. 103, No. 30, 1999 5969
Figure 4. Variation with time of the experimental pressure profiles.
For the vertical scale, 10 mV ) 1 PSI (0.07 atm). The lines (s, - - -)
refer to a pre-ignition run (1440 K) and a post-ignition run (1540 K),
respectively. Initial mixture: 0.7 mol % C3F6 and 6.3 mol % N2O.
Figure 3. Variation with temperature of the experimental N2O and
N2 yields. The respective experimental points are represented by the
symbols (b, O); model profiles are indicated by the lines: (s, - - -).
Initial mixture: 0.6 mol % C3F6 and 1.5 mol % N2O.
from the product FTIR spectra did not reveal the presence of
any other significant species. Other species such as CF3NO,6
CF3OF,7 and FNO8 were searched for but not detected in the
infrared spectra. This is also true of the other two sets of
experiments to be discussed below.
Arrhenius plots of the rate of disappearance of N2O and C3F6
and the rate of appearance of CF2O showed that the activation
energy for N2O and C3F6 decomposition and CF2O formation
were very similar. This indicates that the rate-determining step
in this series was the decomposition of N2O to form O atoms.
Figure 3 shows a plot of N2O and N2 yields versus
temperature. These were the only significant nitrogenous
compounds detected in the product mixtures. At 1500 K, only
20% of the initial N2O has decomposed, with the principal
nitrogenous product being N2. NO2 was not detected in any of
the infrared spectra, and the NO was found in insignificant
concentrations.
2. Mixtures of 0.7 mol % C3F6 and 6.3 mol % N2O (Excess
Oxidant). The oxidation process in this series was found to
differ considerably from that described in the first series. Over
the temperature range 1300-1400 K, the behavior of the two
series is similar. Above 1410 K, ignition of the C3F6 was
observed along with the production of significant quantities of
CF4 and O2. NO and NO2 were observed; however, because of
the presence of O2 in the products, secondary oxidation occurred
within the infrared cell, making the quantification of these
species difficult. A good indication that ignition had indeed
occurred is given by comparing shock tube pressure traces at
temperatures above and below the ignition temperature. The
variation of pressure with time is shown in Figure 4. The post-
ignition run reaches a higher pressure, ∼20.5 atm, compared
with the pre-ignition run which reached 16.5 atm. The oscilla-
tions in the post-ignition run may be due to detonation effects
or acoustic waves travelling in the body of the shock tube.
Kinetic modeling (discussed below) did not support the pos-
sibility of oscillating kinetics. The ∆t in the figure refers to the
ignition delay, which is measured to be 240 ( 50 µs where the
initial shocked gas temperature is 1540 K.
Figure 5. The variation with temperature of the experimental C3F6,
C2F4, and C2F6 yields. The respective experimental points are repre-
sented by the symbols (b, O, 1); model profiles are indicated by the
lines: (s, - - -, - - -). Initial mixture: 0.7 mol % C3F6 and 6.3 mol %
N2O.
presumably more CF3 becomes available for recombination.
Above an initial temperature of 1450 K, both the C2F4 and C2F6
profiles drop to zero. Figure 6 shows the profiles of CF2O, CO,
and CO2. Above 1450 K, a CF2O yield of 0.9 was obtained;
CO and CO2 yields were 0.5 and 0.3, respectively. This is
different from the previous series where the major partially
oxidized product was CO. Product yields show a large increase
above the temperature for ignition. Figure 7 shows the profile
of CF4 (max. yield of ∼1.0) and O2. The appearance of CF4 at
temperatures above 1450 K is related to the appearance of the
spike in the C2F6 profile. This is again due to the large increase
in concentration at 1450 K of CF3 radicals, which recombine
to yield C2F6. The appearance of CF4 suggests the presence of
a large concentration of F atoms which are terminated by
recombination with CF3. The kinetics of the ignition process
will be discussed in the numerical modeling section. Figure 8
shows the yield profiles for N2O and N2. This plot shows that
above 1450 K the N2O has almost entirely decomposed to N2.
As mentioned above, NO and NO2 were observed in the product
mixture, with maximum measured yields of ∼0.1 each. Second-
ary reactions involving NO and O2 within the infrared cell
prevented accurate concentration measurements, and these are
not reported in this work.
Figure 5 shows yield versus temperature plots for C3F6, C2F4,
and C2F6. Unlike the first series, the C3F6 is completely
decomposed when the initial reflected shocked gas temperature
is above 1450 K. The C2F4 and C2F6 profiles show a steady
increase up until 1430 K, after which, the C2F4 yield drops to
zero at 1450 K. However, the C2F6 shows a sharp increase, as

5970 J. Phys. Chem. A, Vol. 103, No. 30, 1999
Hynes et al.
Figure 9. Variation with temperature of the experimental C3F6, C2F4,
and C2F6 yields. The respective experimental points are represented
by the symbols (b, O, 1); model profiles are indicated by the lines:
(s, - - -, - - -). Note that the C2F4 yields have been multiplied by 2
for clarity. Initial mixture: 6.2 mol % C3F6 and 0.6 mol % N2O.
Figure 6. Variation with temperature of the experimental CO, CF2O,
and CO2 yields. The respective experimental points are represented by
the symbols (b, O, 1); model profiles are indicated by the lines: (s,
- - -, - - -). Initial mixture: 0.7 mol % C3F6 and 6.3 mol % N2O.
for F atoms at equilibrium. The stoichiometry predicts a yield
of 1.0 for CO2; experimentally, CO and CO2 are both observed
with a combined yield of 0.8. This indicates that, at 1600 K,
partial, rather than complete, oxidation to CO2 is occurring.
3. Mixtures of 6.2 mol % C3F6 and 0.6 mol % N2O (Fuel
Rich). This series also shows somewhat different decomposition
characteristics from the previous two series. As there is a large
excess of C3F6 in the mixture, it is expected that some pyrolysis
will occur. Figure 9 shows the variation with temperature of
the C3F6, C2F4, and C2F6 yields. At a temperature of 1640 K,
only 20-25% of the C3F6 has decomposed. The major products
of decomposition in this series are C2F4 (maximum yield at 1640
K of ∼0.09) and C2F6 (0.035). Note in Figure 9 that the C2F4
yields have been multiplied by two for clarity. GC-MS was used
to detect heavier fluorocarbon compounds in this series, and
the two most important were iso-C4F8 (CF2dCF(CF3)2) and
2-C4F8 (CF3CF2dCF2CF3). The maximum yields of each of
these species at 1640 K were 0.025 and 0.011, respectively.
Other heavier fluorocarbons include octafluoropropane, C4F6
(identified as hexafluoro-2-butyne), perfluorinated cyclobutenes,
hexafluorobenzene, pentafluorocyclopentadiene, and pentafluo-
ro(trifluoromethyl)benzene. Figure 10 shows the CF2O, CO, and
CO2 product yields. Note that both the CF2O and CO2 yields
have been multiplied by two for clarity. The yields of these
species are quite low with the maximum CF2O yield of 0.018
at 1640 K. The yields of CO and CO2 at this temperature were
0.017 and 0.0015, respectively. The yields of the two C4F8
isomers are consistently low in all three sets of experiments.
Figure 11 shows the N2O and N2 profiles where essentially all
of the nitrogen in the N2O is converted to N2. Small yields
(<0.01) of NO and NO2 were detected in the infrared spectra.
In the presence of a large excess of O2, NO may be oxidized to
NO2 in the infrared cell and the appearance of the NO2 may be
an artifact. This oxidation of NO is supported by the loss of O2
at temperatures where more O2 is expected and by the low levels
of NO recorded. Thus, the NO and NO2 measurements are not
reliable and are not reported here. Also, no CF3CFO was
detected in any of the product gas samples.
Figure 7. Variation with temperature of the experimental CF4 and O2
yields. The respective experimental points are represented by the
symbols (b, O); model profiles are indicated by the lines: (s, - - -).
Initial mixture: 0.7 mol % C3F6 and 6.3 mol % N2O.
Figure 8. Variation with temperature of the experimental N2O and
N2 yields. The respective experimental points are represented by the
symbols (b, O); model profiles are indicated by the lines: (s, - - -).
Initial mixture: 0.7 mol % C3F6 and 6.3 mol % N2O.
The Reaction Model. A 77-step reaction model with 38
species was assembled in this work to model the oxidation of
C3F6, as summarized in Table 1. It should be noted that the
reverse Arrhenius parameters are fitted with a T0 dependence
in order to verify the accuracy of the thermochemical data. In
In terms of the stoichiometry proposed above for the C3F6
oxidation, the measured yields of ∼1.0 for both CF4 and CF2O
are consistent. Further, these two products are the major sinks

Shock Tube Study of the Oxidation of C6F6 by N2O
J. Phys. Chem. A, Vol. 103, No. 30, 1999 5971
The self-reaction, N2O + N2O ) N2 + O + N2O, is also a
sensitive reaction whose rate constant of 4 × 1018T-0.73 exp-
(-266 kJ mol-1/RT) cm3 mol-1 s-1 was obtained in this work.
The A-factor obtained for this reaction agrees well with that of
Vandooren et al.,10 while the barrier is 3 kJ/mol higher than
theirs. Also important is the reaction N2O + CO ) CO2 + N2.
The A-factor of 8 × 1012 s-1 and the barrier of 151 kJ/mol
were derived in this work. The resulting rate constant showed
good agreement with the available literature values,11 although
the scatter in the literature is large.
There are only three reactions linking the fluorine and
nitrogen chemistry in this mechanism. The first is the termo-
lecular reaction of F atoms with NO forming FNO. For this
reaction, the rate data used in the model are based on a fit of
the available literature values; however, the data were measured
at room temperature and therefore some uncertainty is associated
with the rate constant. FNO was not detected in the second set
of experiments, although the model does predict its presence.
This may be attributed to its known high reactivity with metal
surfaces.12 The second reaction is CF3 + NO ) CF3NO whose
rate constant is based on the room-temperature data of Masanet
et al.13 and uncertainties would also be expected with this rate
constant. CF3 can also react with FNO, yielding CF4 and NO.
The chosen rate constant is similar to that for the CF3 + ClNO
reaction at 298 K.14
Figure 10. Variation with temperature of the experimental CO, CF2O,
and CO2 yields. The respective experimental points are represented by
the symbols (b, O, 1); model profiles are indicated by the lines: (s,
- - -, - - -). Note that the CF2O and CO2 yields have been multiplied
by 2 for clarity. Initial mixture: 6.2 mol % C3F6 and 0.6 mol % N2O.
The most important reactions involving C3F6 are the addition
of O atoms across the double bond. The addition is complicated
by the possibility that O atoms can attack either carbon of the
double bond; therefore, rate constants for attack at both these
centers were included in the model. When the O atom adds to
the terminal carbon, a triplet biradical is assumed to form, CF3-
CF•CF2O•. This biradical is estimated to lie ∼130 kJ/mol below
the energy of the reactants. This estimate is based on preliminary
G2-MP2 calculations.15 An error of at least 40 kJ/mol is
assumed. The biradical can undergo three possible reactions.
First, 1,2 F atom shift to form CF3CF2 + CFO is possible. A
barrier of 155 kJ/mol was estimated on the basis of the
calculated barrier15 for 1,2 F shift in the singlet CF3C¨ F f C2F4
reaction. The products lie ∼60 kJ/mol below the biradical.
Figure 11. Variation with temperature of the experimental N2O and
N2 yields. The respective experimental points are represented by the
symbols (b, O); model profiles are indicated by the lines: (s, - - -).
Initial mixture: 6.2 mol % C3F6 and 0.6 mol % N2O.
CF3CF•CF2O• f CF3CF2 + CFO
∆rH ) -60 kJ/mol
the actual numerical modeling, the reverse rate parameters are
computed from the equilibrium constant at each temperature.
The temperature dependences of the A-factors are included.
Many of the rate constants used for the fluorine chemistry
were obtained from Burgess et al.9 The nitrogen chemistry was
obtained from Vandooren et al.10 Some rate constants were
obtained by fitting the known experimental data from a kinetic
database.11 Where these sources were insufficient, other refer-
ences were used or estimates were made in this work.
The thermochemical data used in the modeling have been
described in ref 2 except for that of CF3CF, which has been
modified in this work (vide infra).
Since it is the decomposition of N2O that produces the O(3P)
atoms, the kinetics of O atom formation are first discussed. The
most important route for the destruction of N2O is the unimo-
lecular decomposition, N2O ) N2 + O. The barrier of 265 kJ/
mol and an A-factor of 2.7 × 1011 s-1 derived in this work
yield a rate constant which, over the temperature range of 1000-
2000 K, is in good agreement with current literature values.11
It should be noted that under the experimental conditions
(pressures ∼ 20 atm) this reaction should be at, or close to, the
high-pressure limit. It is also a highly sensitive reaction, and
the A-factor has thus been deduced with high precision.
Second, C-C rupture may occur to form the triplet CF3C¨ F and
CF2O. These products also lie ∼60 kJ/mol below the energy of
the biradical.
CF3CF•CF2O• f 3CF3C¨ F + CF2O
∆rH ) -60 kJ/mol
Third, F loss may occur to form CF3CF•CFdO + F (both
doublets). The enthalpy of formation of CF3CF•CFdO (∆fH°298
) -1045 kJ/mol) was estimated using the enthalpy of the
aldehyde, CF3CF2CFdO (∆fH°298 ) -1465 kJ/mol; calculated
by group additivities) and a C-F bond enthalpy of ∼500 kJ/
mol. The products of this channel were estimated to lie ∼42
kJ/mol above the triplet biradical.
CF3CF•CF2O• f CF3CF•CFdO + F
∆rH ) 42 kJ/mol
A QRRK calculation16 was performed considering all three
channels. The high-pressure addition rate constant of 3.2 × 1012
cm3 mol-1 s-1 was assumed, equivalent to that for O addition
to C2F4 at 1500 K.9 For the dissociation of the biradical back
to reactants, an A-factor and barrier of 5.0 × 1013 s-1 and 130
kJ/mol were used. The A-factor and barrier for dissociation of

5972 J. Phys. Chem. A, Vol. 103, No. 30, 1999
Hynes et al.
TABLE 1: Reaction Model for C3F6 Oxidation in N2Oa
forward reaction reverse reaction
forward reaction reverse reaction
ref;
note
ref;
note
reaction
log A
n
E
log A n
E
reaction
log A
n
E
log A n
E
1 2O + M ) O2 + M
2 N2O ) O + N2
13.3
11.4
0
0
-7.5 14.7
0
0
0
0
0
0
0
488.6 33
108.0 PW
100.3 PW
265.5 10
453.7 11; fit 44 2CFO ) CO + CF2O
318.1 10
40 CFO + F ) CF2O
12.0
0
0
0.0 14.6
13.4 14.3
0
0
497.2 PW
356.3 11; fit
265.7 9.8
41 CFO + F2 ) CF2O + F 12.8
3 2N2O ) N2 + O + N2O 18.6 -0.73 265.7 14.3
42 CFO ) CO + F
43 CFO + O ) CO2 + F
25.0 -4.46 141.0 8.5 0 -43.5 PW
4 N2O + O ) 2NO
5 N2O + O ) N2 + O2
6 NO + N ) N2 + O
13.8
13.9
12.5 0.3
0
0
107.2 12.2
114.5 13.6
0.0 14.2
13.9
13.3
13.5
13.3
0
0
0
0
0.0 15.8
1.3 15.4
0.0 17.3
0.0
0
0
0
390.5 9; 3A
365.9 41
381.8 42
0.0 9
45 2CF3 ) CF3-CF3
46 CF3-CF2 + O f
CF3 + CFO + F
7 NO + M ) N + O + M 15.1
0
621.0 14.4
-8.7 10
8 NO + O ) N + O2
9.6
1
173.1 13.8
0
51.1 10
47 CF3-CF2 + O )
CF2O + CF3
13.0
0
0.0 13.1
0
438.5 9
9 N + N2O ) N2 + NO
10 NO2 + O ) NO + O2
13.0
12.6
0
0
83.1 12.0
-1.0 12.1
0
0
556.0 10
194.8 34
48 CF3-CF2 ) CF3 + CF2 15.6
0
0
235.4 13.3
0.0 12.9
0
0
20.9 43
49 CF3-CF2 + F ) 2CF3
13.5
16.7
12.2
9.3
135.3 44;
C2H5+H
11 NO2 + M ) NO +
O + M
15.3
12.9
15.7
9.1
0
0
0
0
0
0
292.9 12.4
150.6 13.6
0.0 17.1
0
0
0
0
0
-7.4 11; fit 50 CF2CF2 ) 2CF2
516.8 PW 51 CF3CF ) CF2CF2
0
0
1
292.6 13.5
150.6 13.0
0.0 12.0
0
0
0
0
0
30.5 45
288.9 14
402.6 9
532.5 9
246.0 9
12 N2O + CO )
CO2 + N2
13 2N + M ) N2 + M
944.4 11; fit 52 CF2CF2 + O )
CF2 + CF2O
108.9 35
14 2NO + O2 ) 2NO2
4.4 11.5
53 CF2CF + O2 )
CF2O + CFO
26.7 -4.55 22.9 10.9
15 NO + N2O )
N2 + NO2
16 2NO ) N2 + O2
17 O + CO + M )
CO2 + M
18 O2 + CO ) O + CO2
19 2F + M ) F2 + M
14.4
209.2 14.6
352.6 11; fit 54 CF2CF + O )
CF2CO + F
13.5
0
0.0 14.3
13.2
19.0 -1.5
254.8 14.5
20.9 16.0
0
0
435.8 11; fit 55 2CF ) C2F2
13.7
7.0
0
2
0.0 17.3
7.9 12.8
0
0
441.7 estd
155.6 9
526.6 36
56 C2F2 + O )
CFCO + F
57 CF2CF ) C2F2 + F
58 CF2CO + O )
CF2O + CO
12.4
0
200.0 13.4
0
227.0 33
13.3
13.0
0
0
334.7 13.1
33.5 13.6
0
0
26.2 estd
713.5 9
10.1 0.77 117.2 11.7 0 -28.1 2
20 F + NO + M )
FNO + M
16.2
0
2.5 17.4
0
229.7 11; fit 59 CFCO + O )
CFO + CO
132.6 12
307.2 13
14.0
0
0.0 14.5
0
613.0 9
22 CF3 + NO ) CF3NO
21 CF3 + FNO )
CF4 + NO
13.0
11.5
0
0
0.0 15.4
0.0 13.5
0
0
60 CFCO + F ) CF2 + CO 13.5
0
0
0.0 14.2
0.0 12.3
0
0
457.2 9; A/3
87.7 9; 2A
61 CF2CF2 + F )
CF3 + CF2
13.8
23 CF3 + F ) CF4
13.0
0
0.0 16.2
0
534.4 15
62 CF2CF2 + F )
CF3-CF2
13.8
0
32.5 14.7
0
0
334.7 2
24 CF3 + F2 ) CF4 + F
25 CF3 + M ) CF2 +
F + M
12.4
16.0
0
0
10.5 14.4
348.9 14.2
0
0
390.7 37
-0.9 PW
63 CF2CF + F ) 2CF2
64 cyclo-C3F6 f
CF2CF2 + CF2
13.3
13.0
0
0
0.0 13.2
182.0
249.7 9
0.0 2
26 CF3 + O2 ) CF3O + O 9.4 1.14 90.0 13.6
0
0
5.5 9
65 CF2CF2 + CF2 f
11.5 0.5
35.6
0.0 2
cyclo-C3F6
27 CF3 + O ) CF2O + F
13.0
0
0.0 13.6
303.2 38;
A/2
66 2CF3CF ) 2-C4F8
11.7
0
12.6 14.2
0
550.9 2; A/10
28 CF2O + F ) CF3O
29 CF3O + F ) CF3OF
12.4
13.0
0
0
33.9 13.3
0.0 14.5
0
0
129.1 39
206.7 40;
A/3
67 2CF2CF ) C4F6
68 C3F6 + CF2 ) i-C4F8
12.5
12.0
0
0
0.0 17.5
133.9 15.9
0
0
551.5 2
402.2 2; A/40
30 CF2 + O2 ) CF2O + O 13.3
0
0
0
0
0
110.9 14.3
110.9 14.0
4.2 13.4
0
0
0
0
0
267.8 25
69 C3F6 + F )
CF3-CF2 + CF2
70 C3F6 + F )
CF3CF + CF3
13.4
13.5
0
0
0.0 12.8
50.2 13.0
0.4 12.7
0.0 0.0
250.0
0
0
0
0
44.7 PW
4.3 PW
31 CF3CF + O2 )
CF3COF + O
13.3
13.6
13.0
13.0
370.6 estd
32 CF2 + O ) CFO + F
160.1 9; A/2 71 C3F6 + O )
CF3CF + CF2O
72 C3F6 + O f
CF3 + CF2CFdO
73 CF2CFdO f
12.7 0.05
257.6 PW;QRRK
fit
0.0 PW
33 CF3CF + O )
CFO + CF3
34 CF3CF + F )
CF2 + CF3
0.0 12.1
381.8 PW
230.1 PW
12.5
15.3
0
0
4.2 12.3
CFO + CF2
74 C3F6 ) CF3 + CF2CF
75 C3F6 ) CF3CF + CF2
76 CF3-CF2 + CF2 )
n-C3F7
35 CF3CF ) CF3 + CF
36 CF + F ) CF2
37 CF + O2 ) CFO + O
15.0
13.8
13.3
0
0
0
288.7 12.5
0.0 15.6
7.5 13.5
0
0
0
14.7 PW
500.0 9
167.3 9
16.7
15.7
12.6
0
0
0
441.8 13.9
401.7 13.4
8.4 14.8
0
0
0
22.4 PW
6.0 PW
205.6 2
38 CF + O ) CO + F
39 F + CFO ) F2 + CO
13.9
11.1
0
0
4.2 15.0
0
0
527.4 9; A*2 77 n-C3F7 )
14.7
10.5
0
0
190.0 13.3
0.0 11.4
0
0
40.4 2; 5A
CF2CF2 + CF3
33.5 11.7
55.1 9
78 CF3O + CO )
CO2 + CF3
124.7 estd
a Units for A, cm3 mol-1 s-1 or s-1 as appropriate. Units for E, kJ/mol. PW indicates rate constant evaluated in the present work. Estd. indicates
rate constant was estimated in the present work; fit indicates that the database of ref 11 was used to provide a fit to currently known rate data; 2A
and A/2 indicates A-factor multiplied and divided by 2 for this work. The carbene CF3CF produced in reaction 71 is assumed to be initially a triplet,
but at ∼20 atm, rapid spin relaxation to the singlet state is expected to occur (see text).
the biradical via the first channel are 1.6 × 1012 s-1 and 155
kJ/mol; the parameters for the second channel were 2.0 × 1015
s-1 and 13 kJ/mol. The parameters for the third channel were
A ) 2 × 1014 s-1 and 55 kJ/mol (reaction endothermicity plus
a 13 kJ/mol barrier). The calculation showed that the second
channel, with its negligible barrier, proceeds with a rate that is
several orders of magnitude faster than the first channel and at
least a factor of 100 faster than the third channel. Therefore,
the overall rate constant for O-addition to form triplet CF3C¨ F
and CF2O is 3.2 × 1012 cm3 mol-1 s-1 at 1500 K (note that the
QRRK fit is used as the rate constant for reaction 71 in Table
1).

Shock Tube Study of the Oxidation of C6F6 by N2O
J. Phys. Chem. A, Vol. 103, No. 30, 1999 5973
Under the high-pressure conditions of these experiments, it
is assumed that the triplet CF3C¨ F will undergo rapid collision-
induced spin relaxation to form the singlet carbene. This
assumption is justified in part by the quenching rate constant17
of 3CF2 by O2 of 2.5 × 1012 cm3 mol-1 s-1. The 3CF2 quenching
experiments of Young et al.18 show that NO and O2 are
significantly better quenchers than Ar. The predicted concentra-
tions of NO in the first and third sets of experiments in this
paper are ∼1 × 10-9 mol cm-3; in the second set, [NO] ≈ 1 ×
10-6 mol cm-3. The quenching rate constant determined by
•CF2CFdO.
CF2 CFO•CF3 f CF3CFdOCF2 + F
•
•
∆rH ) 110 kJ/mol
Because of the endothermicity of these last two channels, it is
assumed that only the first channel is important.
The most likely fate of the •CF2CFO radicals is C-C rupture
to form singlet CF2 and CFO radicals. Although the bond
enthalpy is ∼250 kJ/mol, the high A-factor (∼1015 s-1) would
ensure that this is a rapid process.
3
Koda19 for CF2 and NO of 2.9 × 1013 cm3 mol-1 s-1 gives a
lifetime of ∼25 µs for sets I and III; for set II, the lifetime is
only 0.03 µs. Another indication is given by the rate constant
for triplet SO2 relaxation20 with argon as the collider species.
•CF2CFO f 1C¨ F2 + CFO
∆rH ) 250 kJ/mol
3
The rate constant for SO2 spin relaxation is 3.3 × 1010 cm3
mol-1 s-1, and this gives the triplet state lifetime of 0.2 µs at
In the kinetic model, the O-addition to the central carbon is
treated as a two-step process. The reaction is assumed to rapidly
20 atm and 1500 K.
•
form CF2CFO + CF3, with a rate constant of 3.2 × 1012 cm3
3
Although the rate constant for CF3C¨ F is not known, the
mol-1 s-1. The second step is the irreversible decomposition
3
3
studies on CF2 and SO2 suggest that the upper-limit lifetime
•
of the CF2CFO radicals to 1C¨ F2 + CFO. The A-factor is
3
of CF3C¨ F should be fractions of microseconds. The shock-
assumed to be 2 × 1015 s-1 with a barrier of 250 kJ/mol.
Both of the above additions could also proceed along singlet
surfaces to form a stable epoxide, C3F6O, where the addition
reaction would have to proceed over a barrier associated with
triplet-singlet curve-crossing. However, epoxides were not
detected in any of the experiments, further indicating that both
reactions probably follow triplet surfaces.
tube experiments of this work have reaction times of hundreds
of microseconds. Second, NO, O2, and Ar are present in the
reaction mixture, and these should adequately quench the
3
3CF3CF¨. Therefore, the CF3C¨ F should not contribute signifi-
cantly to the overall chemistry of the system.
The singlet-triplet energy gap for the CF3C¨ F has been
determined in recent difference-dedicated configuration interac-
tion calculations by Garcia et al.21 and CBS-Q calculations by
van Stekelenborg.22 A value of 72 kJ/mol was found, with the
singlet being lower in energy. This energy gap further suggests
that rapid triplet-singlet conversion should occur, especially
when the singlet is thermodynamically more stable.
When O atoms add to the central carbon of C3F6, traces of
CF3CFO were observed in the room-temperature experiments
of Saunders and Heicklen.3 Because the formation of CF3CFO
3
and CF2 from the triplet biradical is endothermic by 19 kJ/
mol the CF3CFO could only form by way of a triplet to singlet
curve-crossing. As their experiments involved photosensitization
by the heavy atom Hg*, it is possible that triplet-singlet
crossing could take place via spin-orbit perturbations through
a short-lived complex incorporating the Hg.
When O-addition to the central carbon occurs, it is suggested
•
that the triplet biradical CF2 CFO•CF3 forms. Its energy is
assumed to lie ∼130 kJ/mol below the reactants. This triplet
biradical can undergo C-C bond scission to form CF3 and •CF2-
CFdO radicals (both doublets). The latter radical has an
estimated ∆fH°298 ) -613 kJ/mol based on the enthalpy of
formation of CF3CFdO (∆fH°298 ) -1024 kJ/mol; calculated
by isodesmic reactions at the G2-MP2 level15) and a C-F bond
enthalpy of ∼500 kJ/mol. The energy of the products lies 75
kJ/mol below that of the triplet biradical; the rupture should be
rapid because of the high A-factor (∼1015 s-1) and large
exothermicity.
The pyrolysis reactions for C3F6 were important in the
modeling of the third reaction set (excess C3F6). The A-factors
for both the unimolecular single- and double-bond rupture were
optimized to give the correct product distributions (residual C3F6
and C2F4). Rupture of the double bond leads to the singlet
carbenes CF3CF¨ and CF2. In previous work, the enthalpy of
formation for CF3C¨ F at 298 K was estimated by Hynes et al.2
to be -586 kJ/mol. However, recent G2-MP2 isodesmic reaction
calculations15 now give the heat of formation at 298 K to be
-520 kJ/mol. This value is in excellent agreement with a value
of -523 kJ/mol computed at the CBS-Q level by van Steke-
lenborg.22 The ab initio barrier and A-factor for the isomeriza-
tion15 of the carbene into C2F4 result in a higher A value and a
lower activation energy compared with that found in previous
work.2 These ab initio values are incorporated into the modeling
in this work.
The addition of F atoms to C3F6 was also complicated by
the possibility of addition to either carbon of the double bond,
and two reactions were included to describe these. Rate constants
for both of these reactions were estimated based on the C2F4 +
F rate constant of Burgess et al.9
Other important secondary reactions include reactions of F
atoms with radicals and other molecules. These include the
termination of CF3 with F forming CF4, a product that is only
detected in runs where ignition has occurred. The bimolecular
rate constant of Plumb and Ryan24 measured at 298 K was used.
At ∼20 atm, this reaction is likely to be close to the high-
pressure limit. The rate constant for the reaction of CF2 and O
yielding CFO and F is based on that of Burgess et al.,9 with the
CF2 CFO•CF3 f •CF2CFO + CF3
∆rH ) -75 kJ/mol
•
This biradical could also undergo C-C scission to form the
triplet CF2 and singlet CF3CFO. To determine the energy of
the products, the enthalpy of formation of the triplet CF2 (47.4
kJ/mol) was required. This enthalpy was deduced by adding
the calculated singlet-triplet energy gap (234 kJ/mol 17) to the
enthalpy of formation of the singlet (-186.6 kJ/mol9). The
energy of the products was found to lie ∼90 kJ/mol above that
•
of the CF3 + CF2CFdO.
CF2 CFO•CF3 f CF3CFdO + 3CF2
•
∆rH ) 19 kJ/mol
A third possible channel is the elimination of F atoms from
•
the triplet biradical to form CF3CdOCF2 (estimated ∆fH°298
) -975 kJ/mol based on ∆fH°298 ) -1398 kJ/mol for CF3Cd
OCF323 and a C-F bond enthalpy of ∼500 kJ/mol). The reaction
is endothermic and lies ∼185 kJ/mol above that of the CF3 +

5974 J. Phys. Chem. A, Vol. 103, No. 30, 1999
Hynes et al.
A-factor reduced by a factor of 2 in this work. The rate constant
for the reaction of CF3C¨ F and O was estimated on the basis of
the aforementioned reaction except that the barrier is assumed
to be negligible and the A-factor slightly lower. In this work, it
is assumed that the reactants singlet CF3C¨ F and O(3P) correlate
with the products CFO and CF3 on a triplet surface, rather than
with CF3CFO on a singlet surface (which would include a barrier
for association and curve-crossing). The rate constant estimated
for O2 + CF3C¨ F is assumed to be equal to that derived by
Keating and Matula25 for CF2 and O2.
A sensitive reaction in this study is the unimolecular
decomposition of the FCO radical. Under the high-pressure
conditions of this work (∼20 atm), the termolecular association
rate constant F + CO + M ) FCO + M derived by Burgess et
al.9 (at 1 atm) was found to be inappropriate under our
experimental conditions. Therefore, an RRKM calculation was
performed for the unimolecular decomposition, FCO ) F +
CO, with the UNIMOL26 program using the transition state
geometry of Francisco27 calculated at the MP2 level and the
Figure 12. Plot of CF4 model mole fractions (s) versus time for an
experiment which exhibits ignition. The initial temperature was 1540
K. Initial mixture: 0.7 mol % C3F6 and 6.3 mol % N2O.
The ignition delay mentioned above has been simulated using
the chemical kinetic model. To achieve this, the model CF4 mole
fraction was calculated with a constant-pressure model and
plotted against time as shown in Figure 12. This species was
chosen because it is present in significant quantities only after
ignition. The time required for the CF4 to reach 50% of its
equilibrium mole fraction is about 220 µs. This delay time is
in good agreement with the observed ignition delay time of 240
( 50 µs. Note that a constant volume calculation predicts a
delay time that is 50 µs shorter.
Figures 9-11 show that the model has provided good fits to
all the major species for the third set of experimental data. The
two isomers of C4F8 were also detected in low yields in the
experiments. Although these species were accounted for in the
kinetic model, their yields are too small to model accurately
and their yield profiles were thus omitted in this work.
Sensitivity and Rate of Production Analyses. Reaction flux
analyses for all sets were performed at a short reaction time of
100 µs to avoid thermal equilibration. Also, the calculated
temperature, assuming a constant-pressure system, increases
rapidly after 100 µs in the set II analyses due to ignition and
decreases in set III due to reaction cooling behind the reflected
shock front.
critical energies of Knyazev et al.28 The value of ∆E
)
down
350 cm-1 was used in the calculation. At 20 atm, a rate constant
for the unimolecular decomposition reaction of 1025.8T-4.46 exp-
(-141 kJ mol-1/RT) was derived; however, the A-factor was
reduced by a factor of 5 to ensure correct modeling of the
products. This RRKM calculation for the FCO decomposition
showed good agreement with the 1 atm value of Burgess et al.9
(the reverse rate constant is calculated from the equilibrium
constant at 1500 K) and the correct falloff behavior was
predicted when compared with the low-pressure experimental
results of Knyazev et al.28 However, the high-pressure limit for
the FCO decomposition was not reached at 1000 atm where
the calculated kuni/k∞ = 8 × 10-2 at 1500 K. This suggests that
the rate constant used in this work is reasonable in light of the
uncertainties in the RRKM calculations and the absence of a
measured high-pressure limit value for the decomposition.
Kinetic Modeling. Kinetic modeling with the mechanism
given in Table 1 was performed with the Sandia code
CHEMKIN29 together with a shock-tube code30 modified to
include the effects of quenching by the reflected rarefaction
wave and the LSODE ordinary differential equation solver.31
The code SENKIN32 was used for sensitivity analyses.
Set I. In terms of the rate of production of O atoms, the
reaction with the most flux is reaction 3, the N2O self-reaction,
which accounts for 100% O atom production. O atom addition
to both the terminal and central carbons of the double bond
(reactions 71 and 72, respectively) is important in the destruction
of the C3F6. As the temperature rises from 1390 to 1495 K, the
addition of F atoms to the central carbon (reaction 69) becomes
the dominant C3F6 destruction route with 40% of the flux. The
addition of O atoms to either of the double-bonded carbons of
C3F6 plays a lesser role (30% by reaction 71, 15% by reaction
72). F atoms are produced predominantly by reaction 42 (CFO
) CO + F; 33%) and reaction 32 (CF2 + O ) CFO + F; 16%).
The major consumption route of F atoms is reaction 69. FCO
is an important radical in this system and is generated by reaction
73 (26%) and by reaction 32 (23%); reaction 42 is the primary
consumption channel. An interesting feature of the rate of
production analysis is that a radical chain mechanism operates,
enabling F-atom attack to proceed more efficiently than O-atom
attack. The F atoms produced by reactions 42 and 32 are
consumed in the F attack on C3F6 (reaction 69). The resulting
fluoroethyl radicals undergo C-C rupture producing CF3 (the
primary production route for this radical) and CF2. The CF2
produced by this reaction and reaction 50 (C2F4 pyrolysis) is
Figures 1-3 show the results of modeling the first set of
experimental data. These figures show that all major stable
product and unconsumed reactant yields have been modeled
successfully over the temperature range.
Figures 6-9 show the results of modeling the second data
set in which ignition takes place. In Figure 6, the model tends
to overpredict the rate of C3F6 consumption somewhat. The peak
C2F6 yield is predicted to be ∼40% higher than that observed
experimentally; however, this could also be attributed to
difficulties in obtaining experimental yield data at the precise
temperature where ignition occurs. The C2F4 yield is also slightly
overpredicted by the model. Figure 7 shows that the model
reproduces the ‘spike’ in the CO profile, and the CO2 and CF2O
profiles have also been successfully modeled. Figure 8 shows
that good agreement was obtained between model and experi-
mental CF4 and O2 yield profiles.
As stated above, small yields of NO and NO2 were detected
in the products, the latter products possibly arising from NO
oxidation by O2 in the infrared cell. The kinetic model predicts
that NO2 should be essentially undetectable under these shock
conditions, further supporting the view that the measured NO2
is indeed an artifact.

Shock Tube Study of the Oxidation of C6F6 by N2O
J. Phys. Chem. A, Vol. 103, No. 30, 1999 5975
Figure 14. Reaction fluxes for C3F6 decomposition and product
formation. The largest arrows indicate the channels with the greatest
flux. The conditions are 0.6% N2O and 6.3% C3F6 at a temperature of
1600 K.
Figure 13. Reaction fluxes for C3F6 decomposition and product
formation. The largest arrows indicate the channels with the greatest
flux. The conditions are 1.5% N2O and 0.6% C3F6 at a temperature of
1500 K.
CF3 (reaction 61). The CF2 reacts with O atoms (reaction 32)
to produce F and CFO. In this way, a high concentration of F
atoms build up, propagated by reactions 61 and 69. The F atom
attack reactions are all exothermic as written (∆rH ) -46 and
-80 kJ/mol for reactions 69 and 61, respectively) and the O
atom reactions are even more exothermic (∆rH ) -160, -255,
and -203 kJ/mol for reactions 32, 71, and 72, respectively).
Ignition in this set occurs due to the large concentration of O
atoms and the large flux through reactions 71 and 72. Reactions
32 and 42 also have a greater flux than in set I. This creates a
pool of F atoms that are consumed through reactions 69 and
61. The constant pressure sensitivity calculations show an
increase in temperature of ∼250 K at the end of the reaction
time.
Set III. Analysis of this set shows the expected pyrolytic
behavior since a large excess of C3F6 was used. The dominant
C3F6 decomposition pathways were reactions 75 (double bond
rupture) and 74 (single bond rupture). Reaction 75 dominated
over 74 as a result of its lower barrier. The flux through the
O-addition channels is less than in the previous reaction sets
with only 3% consumption of C3F6 taking place at 1600 K
through reaction 71. Reactions 75 and 71 produce the carbene
CF3C¨ F and reaction 74 produces the fluorovinyl radical CF2d
CF. The primary decomposition routes for CF3CF are reactions
-75 and the 1,2 F-shift isomerization reaction, 51, producing
C2F4. Reaction -74, producing C3F6, and the recombination
into C4F6 consume the fluorovinyl radicals (reaction 67).
Since the O atom is in lower concentrations in this set, less
CFO is produced and hence less F. The F attack on C3F6
(reaction 69) is less important in this series and consumes only
6.5% of the C3F6 at 1600 K. Reactions producing F and CFO
are as described in the previous two sets. The flux diagram in
Figure 14 shows the production and consumption routes of the
major products. The formation and destruction of C2F6 and C2F4
are as described in sets I and II. CF2O is essentially formed by
reaction 71, CO entirely by reaction 42, and CO2 by reactions
12 (94%) and 43 (6%).
successively oxidized by O atoms to CFO and then CO, with F
production accompanying each reaction. The F so produced
reacts with C3F6 to continue the chain.
The formation and destruction channels of the important
stable products are shown schematically in Figure 13. CF2O,
under these conditions, is particularly stable and has no
significant decomposition channel.
The sensitive reactions involving C3F6 are the O-atom
addition reactions 71 and 72. Also important is the CF2 + O
reaction 32. The most sensitive reaction for all products is N2O
decomposition (reaction 2). Also sensitive are reactions 12, 32,
and 43.
Set II. At temperatures below the point of ignition (∼1450
K), the oxidation proceeds as described in set I. An analysis of
the flux data at 1550 K shows that decomposition of C3F6
becomes driven by F addition reaction (reaction 69). This is a
result of significantly greater flux through reactions 32 and 42.
The formation pathways for CF2O have also changed with
reaction 71 decreasing in flux and reaction 44 (CFO recombina-
tion; 42%) becoming dominant. Reaction 27 (O + CF3) accounts
for 20% of CF2O formation. In terms of CO and CO2
production, the pathways are similar to those described in the
previous set except that greater flux passes through reaction 12
in the case of CO2 formation. C2F6 and C2F4 formation routes
are as described in Figure 13.
Two species observed in set II and not observed in set I are
CF4 and O2. The formation of the former was through reaction
23 (99%), although the consumption of F by this channel was
6% and CF3 consumption was less than 1%. O2 formation was
caused by reaction 5 where N2O consumes O.
Reaction rate sensitivities and fluxes are also used to explain
the basis of the ignition process. Sensitivities for this set are
studied at an initial temperature of 1550 K and 20 atm. At short
reaction times, the oxidation of C3F6 is dominated by O-addition
reactions 71 and 72. From reactions 32 and 73 CFO is produced.
This subsequently breaks down to F and CO (reaction 42). At
longer times, the F attacks the C3F6 through reaction 69
producing CF2 radicals. F also attacks C2F4 producing CF2 and
At 1600 K and low residence times, the C3F6 mole fraction
displays sensitivity to reactions 75, 51, 68, and 67. Significantly,

5976 J. Phys. Chem. A, Vol. 103, No. 30, 1999
Hynes et al.
reaction 75 has greater sensitivity than reaction 2. The C2F4
also displays sensitivity to reactions 75 and 51. The remaining
products are sensitive to reactions already described in the
previous two sets.
forms from the recombination of F and CF3 radicals. Despite
the large concentration of O atoms in set II, partial, rather than
complete, oxidation takes place, with CF2O and CO being the
major partially oxidized products. The rate constant for O2
oxidation of CO (reaction 18) has a high (200 kJ/mol) barrier,
and the O2 concentration is relatively low except under the
conditions of high temperatures and large O excess. The rate
constant for O atom oxidation of CO (reaction 17) is termo-
lecular and therefore slow. The dominant route for CO oxidation
to CO2 is by N2O oxidation (reaction 12) by virtue of the large
N2O concentrations used.
Role of C3F6 in C3 Fluorocarbon Flame Inhibition. In the
context of flame inhibition by CF3CHFCF3, C3F6 can play an
important role. In lean, atmospheric pressure flames,1 the thermal
decomposition route forming C3F6 accounts for ∼40% of the
calculated CF3CHFCF3 decomposition channels. In these lean,
atmospheric pressure flames, C3F6 could not be detected in the
postflame gases. Because flame radicals such as H, O, and OH
are in superequilibrium concentrations in the reaction zone of
the flame, rapid exothermic destruction of the C3F6 is expected
on the basis of the results of this work. The ignition behavior
of C3F6 as described in set II could, in part, account for the
large temperature increases observed when CF3CHFCF3 is added
to a flame. The ignition process however depends on the relative
O and C3F6 concentrations. A near-stoichiometric mixture did
not ignite, but a 10-fold excess of O atoms did cause ignition.
Despite a lack of knowledge of certain rate data for C3F6
destruction in a flame, especially channels involving OH, our
reaction flux analysis of the CF3CHFCF3-inhibited flame data1
suggests that OH addition to C3F6 would still be the dominant
destruction route, although O + C3F6 reactions are nonetheless
important channels. Because inhibitor molecules such as CF3-
CHFCF3 decompose exothermically in a flame, the extra heat
released through C3F6 ignition (if the C3F6 were in an O-atom-
rich environment) could lead to increased flame propagation,
rather than inhibition. This may contribute to the relative
inefficiency of CF3CHFCF3 as an inhibitor compared with CF3-
Br.
Under C3F6-rich conditions, pyrolysis was observed with the
rupture of both single and double bonds being the principal
decomposition reactions.
Acknowledgment. We thank Dr. G. B. Bacskay and Mr.
M. Smith for performing G2-MP2 calculations.
References and Notes
(1) Hynes, R. G.; Mackie, J. C.; Masri, A. R. Combust. Flame 1998,
113, 554.
(2) Hynes, R. G.; Mackie, J. C.; Masri, A. R. J. Phys. Chem. A 1999,
103, 54.
(3) Saunders, D.; Heicklen, J. J. Am. Chem. Soc. 1965, 87, 4062.
(4) Douglass, C. H.; Ladouceur, H. D.; Shamamian, V. A.; McDonald,
J. R. Combust. Flame 1995, 100, 529.
(5) Bauer, S. H.; Hou, K. C.; Resler, E. L., Jr. Phys. Fluids Suppl.
1969, 1, 125.
(6) Shurvell, H. F.; Dass, S. C.; Gordon, R. D. Can. J. Chem. 1974,
52, 3150.
(7) Lagemann, R. T.; Jones, E. A.; Woltz, P. J. H. J. Chem. Phys.
1952, 20, 1768.
(8) Chen, J.; Zhu, T.; Niki, H. J. Phys. Chem. 1992, 92, 6115.
(9) Burgess, D. R., Jr.; Zachariah, M. R.; Tsang, W.; Westmoreland,
P. R. Prog. Energy Combust. Sci. 1996, 21, 453.
(10) Vandooren, J.; Van Tiggelen, P. J.; Pauwels, J.-F. Combust. Flame
1997, 109, 647.
(11) Westley, F.; Herron, J. T.; Frizzell, D.; Hampson, R. F.; Mallard,
W. G. NIST Chemical Kinetics Database 17-2Q98; NIST Standard
Reference Data; NIST: Gaithersburg, MD, 1998.
(12) Greenwood, N. N.; Earnshaw, A. Chemistry of the Elements;
Pergamon Press: Oxford, U.K., 1993; p 507.
(13) Masanet, J.; Caralp, F.; Ley, L.; Lesclaux, R. Chem. Phys. 1992,
160, 383.
The behaviors of perfluoroalkenes and perfluoroalkanes are
markedly different. As shown in this paper, the initiation step
for the oxidation of C3F6 is essentially a barrierless process. In
contrast, the routes for perfluoroalkane (e.g., C2F6) decomposi-
tion are C-C scission (barrier ∼406 kJ/mol), C-F loss (515
kJ/mol), and F abstraction by O (300 kJ/mol). Because of these
high-energy destruction routes, perfluorinated alkanes tend to
behave thermally (via heat absorption and dilution) rather than
chemically in a flame environment. Perfluoroalkenes tend be
more fuel-like.4
(14) Rossi, M. J.; Barker, J. R.; Golden, D. M. J. Chem. Phys. 1979,
71, 3722.
(15) Smith, M. B. Sc. (Hons) Thesis, School of Chemistry, University
of Sydney, 1998.
(16) Dean, A. M.; Westmoreland, P. R. Int. J. Chem. Kinet. 1987, 19,
207.
Conclusion
An examination of the product yields and reaction path
analysis of the reaction between C3F6 and N2O shows that C3F6
decomposes primarily by O addition to both carbons of the
double bond. The analysis also shows that, when CFO radicals
form, they readily undergo unimolecular F loss. These F atoms
are then able to attack the C3F6 and cause a radical chain that
also involves CF2 carbenes. Significantly, no CF3CFO or the
epoxide, C3F6O, was observed in any of the experiments,
suggesting that O-addition proceeds along a triplet surface. In
the presence of excess N2O, ignition was observed to occur at
temperatures above ∼1410 K, owing to the large excess of O
atoms and the resulting large F radical production. The reactions
that give rise to ignition are initiated by O addition to both
carbons of the double bond, reactions 71 and 72. The flux
through these two reactions is greater than in set I owing to the
large excess of O atoms. The subsequent exothermic oxidation
of CF2 (reaction 32) and unimolecular decomposition of CFO
radicals (reaction 42) produce F atoms. The large pool of F
atoms so produced reacts with C3F6 and C2F4 via reactions 61
and 69 respectively producing CF2 and CF3 radicals, ultimately.
The large F pool is evidenced by the observation of CF4, which
(17) Koda, S. Chem. Phys. 1982, 66, 383.
(18) Young, R. A.; Blauer, J.; Bower, R.; Lin, C. L. J. Chem. Phys.
1988, 88, 4834.
(19) Koda, S. J. Phys. Chem. 1979, 83, 2065.
(20) Briggs, J. P.; Caton, R. B.; Smith, M. J. Can. J. Chem. 1975, 53,
2133.
(21) Garcia, V. M.; Castell, O.; Reguero, M.; Caballol, R. Mol. Phys.
1996, 87, 1395.
(22) Van Stekelenborg, J. Private communication, January 1999.
(23) Gordon, A. S. Int. J. Chem. Kinet. 1972, 4, 541.
(24) Plumb, I. C.; Ryan, K. R. Plasma Chem. Plasma Process. 1986, 6,
11.
(25) Keating, E. L.; Matula, R. A. J. Chem. Phys. 1977, 66, 1237.
(26) Gilbert, R. G.; Smith, S. C.; Jordan, M. J. T. UNIMOL Program
Suite for the Calculation of Falloff CurVes for Unimolecular and Recom-
bination Reactions; School of Chemistry, University of Sydney: Sydney,
1993.
(27) Francisco, J. S.; Goldstein, A. N.; Williams, I. H. J. Chem. Phys.
1988, 89, 3044.
(28) Knyazev, V. D.; Bencsura, A.; Slagle, I. R. J. Phys. Chem. 1989,
93, 1916.
(29) Kee, R. J.; Miller, J. A.; Jefferson, T. H. CHEMKIN, A General
Purpose, Problem Independent, Transportable FORTRAN Chemical Kinetics
Code Package; Report SAN80-003; Sandia Laboratories: Albuquerque,
NM, March 1980.

Shock Tube Study of the Oxidation of C6F6 by N2O
J. Phys. Chem. A, Vol. 103, No. 30, 1999 5977
(30) Mitchell, R. E.; Kee, R. J. A General Purpose Computer Code for
Predicting Chemical Kinetic BehaVior Behind Incident and Reflected Shocks;
Report SAN82-8205; Sandia Laboratories: Albuquerque, NM, March 1982.
(31) Hindmarsh, A. C. LSODE and LSODI, Two New Initial Value
Differential Equation SolVers; ACM Signum Newsletter 1980, 15, 4.
(32) Lutz, A. E.; Kee, R. J.; Miller, J. A. SENKIN: A FORTRAN
Program Predicting Homogeneous Gas-Phase Chemical Kinetics with
SensitiVity Analysis; Report SAND87-8248: Sandia Laboratories: Albu-
querque, NM, Feb 1988.
(36) Wagner, H. Gg.; Zabel, F. Ber. Bunsen-Ges. Phys. Chem. 1974,
72, 705.
(37) Teitel’boim, M. A.; Vedeneev, V. I. Kinet. Catal. 1986, 26, 1119.
(38) Tsai, C.; Belanger, S. M.; Kim, J. T.; Lord, J. R.; McFadden, D.
L. J. Phys. Chem. 1989, 93, 1916.
(39) Kennedy, R. C.; Levy, J. B. J. Phys. Chem. 1972, 76, 3480.
(40) Heicklen, J. AdV. Photochem. 1988, 14, 177.
(41) Maricq, M. M.; Szente, J. J.; Khitrov, G. A.; Francisco, J. S. Chem.
Phys. Lett. 1992, 199, 71.
(42) Glanzer, K.; Maier, M.; Troe, J. J. Phys. Chem. 1980, 84, 1681.
(43) Ainagos, A. F. Kinet. Catal. 1991, 32, 720.
(33) Tsang, W.; Hampson, R. F., Jr. J. Phys. Chem. Ref. Data 1986,
15, 1087.
(34) Tsang, W.; Herron, J. T. J. Phys. Chem. Ref. Data 1991, 20, 609.
(35) Atkinson, R.; Baulch, D. L.; Cox, R. A.; Hampson, R. F., Jr.; Kerr,
J. A.; Rossi, M. J.; Troe, J. J. Phys. Chem. Ref. Data 1997, 26, 1329.
(44) Warnatz, J. In Combustion Chemistry; Gardiner, W. C., Jr., Ed.;
Springer-Verlag: New York, 1984.
(45) Bauer, S. H.; Javanovic, S. Int. J. Chem. Kinet. 1998, 30, 171.