Fluorine-atom initiated oxidation of CF3CF2H (HFC-125) studied by FTIR spectroscopy: Product yields and kinetic modelling

Article abstract of DOI:10.1039/a702105d

A comprehensive study has been carried out of the oxidation of CF₃CF₂H (HFC-125). Reaction was initiated by continuous photolysis of F₂ in the near-UV. The F atoms produced abstracted a hydrogen atom from CF₃CF₂H initiating oxidation in gas mixtures containing variable amounts of O₂ and made up to a total pressure of 700 Torr with N₂. Product yields were measured as a function of time using FTIR spectroscopy. Experiments were performed at room temperature and in mixtures containing different ratios of [F₂] to [CF₃CF₂H]. The major products were COF₂, CF₃O₃CF₃ and CF₃O₃C₂F₅, consistent with C-C bond scission being the dominant loss process for CF₃CF₂O radicals and with previous studies which used chlorine atoms to initiate oxidation. Attempts to match the experimental results with predictions using the FACSIMILE chemical modelling program were moderately successful and confirmed recent results regarding the equilibrium constant for: F + O₂(+M) = FO₂(+M).

Full text of DOI:10.1039/a702105d

View Article Online / Journal Homepage / Table of Contents for this issue
Fluorine-atom initiated oxidation of CF CF H (HFC-125) studied
3
2
by FTIR spectroscopy: product yields and kinetic modelling
Alam S. Hasson, Christopher M. Moore and Ian W. M. Smith*
T he School of Chemistry, T he University of Birmingham, Edgbaston, Birmingham, UK B15 2T T
A comprehensive study has been carried out of the oxidation of CF CF H (HFC-125). Reaction was initiated by continuous
3
2
photolysis of F in the near-UV. The F atoms produced abstracted a hydrogen atom from CF CF H initiating oxidation in gas
2
3
2
mixtures containing variable amounts of O and made up to a total pressure of 700 Torr with N . Product yields were measured
as a function of time using FTIR spectroscopy. Experiments were performed at room temperature and in mixtures containing
2
2
di†erent ratios of [F ] to [CF CF H]. The major products were COF , CF O CF and CF O C F , consistent with CwC
2
3
2
2
3
3
3
3 3 2 5
bond scission being the dominant loss process for CF CF O radicals and with previous studies which used chlorine atoms to
3
2
initiate oxidation. Attempts to match the experimental results with predictions using the FACSIMILE chemical modelling pro-
gram were moderately successful and conÐrmed recent results regarding the equilibrium constant for: F ] O (]M) \ FO (]M).
2
2
Since it was realised1 and accepted2 that the release of man-
made chloroÑuorocarbons (CFCs) has a deleterious e†ect on
the EarthÏs atmosphere, there has been an intense e†ort to
Ðnd more benign replacement compounds. Of course, the very
chemical stability which originally made the CFCs so attrac-
tive for a range of industrial uses lies at the root of their
environmental unsuitability. In particular, their impervious-
ness to attack by OH radicals means that they have unusually
long lifetimes in the atmosphere.3h6 Consequently, they even-
tually reach the stratosphere, in amounts virtually
undiminished by removal in the troposphere, and there they
are broken down by UV light releasing chlorine atoms which
initiate chain processes destroying ozone. In addition to their
ability to destroy stratospheric ozone, the long atmospheric
lifetime of the CFCs, coupled with their IR absorption
spectra, means that they are particularly potent “greenhouse
gasesÏ.4h6
The search for replacements has focused on the hydro-
chloroÑuorocarbons (HCFCs) and hydroÑuorocarbons
(HFCs). These compounds have similar, desirable, physical
properties to the CFCs but the inclusion of at least one hydro-
gen atom in their structure means that they undergo attack by
OH radicals which initiates their breakdown in the tropo-
sphere and leads to a signiÐcant reduction in atmospheric life-
times relative to the CFCs. The investigation of the industrial
manufacture and suitability of these compounds has been
accompanied by a major scientiÐc e†ort to understand all
details of their chemical behaviour in the atmosphere. One
result has been that HCFCs have been abandoned as any-
thing other than temporary replacements for the CFCs since,
although their ozone-destroying potential is less than that of
the CFCs, it remains sufficiently high to cause concern chieÑy
because of the presence of chlorine atoms in these compounds.
Present attention, both industrial and scientiÐc, is strongly
focused on the HFCs.7
a replacement for CFC-12 (CF Cl ) in automotive air-
2
2
conditioning systems8 and whose chemistry we will consider
in a future publication. In the present paper we report the
results of laboratory experiments on the oxidation of
HFC-125 (CF CF H) for which the steps identiÐed above are
3
2
CF CF H ] OH ] CF CF ] H O
(1)
(2)
3
2
3
2
2
CF CF ] O ] M ] CF CF O ] M
3
2
2
3
2 2
CF CF O ] NO(RO ) ] CF CF O ] NO (RO ] O )
3
2
2
2
3
2
2
2
(3a,b)
The fate of the substituted alkoxy radicals formed at this stage
of the oxidation of a HFC is especially interesting, from both
a fundamental chemical viewpoint and from the context of
atmospheric chemistry. It depends, inter alia, on whether (a)
the radical still contains one or more H atoms, and (b) it is a
C or higher substituted hydrocarbon. If at least one H atom
1
is present, then its abstraction by O , as in the case of a non-
2
substituted alkoxy radical, will be important. For radicals
derived from other than C HFCs this process will be com-
1
petitive with CwC bond cleavage. Thus, although a speciÐc
widespread use has not yet been identiÐed for HFC-125, its
chemistry serves as a useful benchmark for other C HFCs,
2
because the substituted alkoxy radical derived from this com-
pound contains no H atom, so that only CwC bond cleavage
is likely:
CF CF O ] CF ] COF
(4)
3
2
3
2
simplifying the overall chemistry.
There appear to have been only three previous laboratory
investigations of the oxidation of HFC-125 under simulated
atmospheric conditions. Both Edney and Driscoll9 and
Tuazon and Atkinson10 have reported chlorine-initiated pho-
toxidation studies of CFCs, including experiments on HFC-
125. In both cases, experiments were conducted at room
temperature on mixtures containing a large excess (700 and
740 Torr, respectively) of dry air. As in the present work, pro-
ducts were identiÐed and their yields determined using FTIR
spectroscopy. Both groups of workers determined the yield for
Because of the increasing use and release of the HFCs there
has been an intense worldwide e†ort over the past few years
to understand their atmospheric chemistry. It is now clear7
that the Ðrst few steps of this chemistry parallels that of the
alkanes: (i) abstraction of an H-atom by OH; (ii) addition by
the resultant radical of O ; and (iii) conversion of that peroxy
radical to a substituted alkoxy radical by reaction with either
NO or another peroxy radical. At that stage, the chemistry
may start to diverge.
COF to be unity within experimental error. In addition,
2
2
Sehested et al.11 have reported the results of similar experi-
ments in which the dependence of product yields on the ratio
of [Cl ] : [CF CF H] present in the mixture was observed.
2
3
2
HFCs based on C H have attracted a good deal of atten-
The yield of COF was found to decrease from 1.0 to 0.63 as
2
6
2
tion, especially HFC134a (CF CFH ), which is being used as
the ratio [CF CF H] : [Cl ] was increased from 0.013 to 1.26.
3
2
3
2
2
J. Chem. Soc., Faraday T rans., 1997, 93(16), 2693È2699
2693

View Article Online
In the present experiments, the oxidation of the HFC-125
was initiated by F atoms, generated by continuous photolysis
duced, they were left for ca. 15 min in the dark to ensure com-
plete mixing and equilibration to the temperature of the cell.
During this period spectra were recorded to observe if any
dark reactions occurred.
It was found that mixtures that were left for extended
periods in the dark (ca. 1 h) showed no detectable loss of the
of F
2
F ] hl ] 2F
(5)
(6)
2
CF CF H ] F ] CF CF ] HF
3
2
3
2
HFC. However, small amounts of COF (@1 mTorr) did
2
The atoms therefore acted as a surrogate for OH radicals.
Changes in the concentrations of reagents and products of the
subsequent reactions were measured using FTIR spectroscopy
employing a long optical path length to increase the sensi-
tivity of the technique. Experiments were carried out at total
pressures of 700 Torr and, in contrast to what has been done
sometimes appear over the Ðrst 10 min. This was particularly
evident if the chamber was not Ñushed out after a previous
run, and it was prevented by ensuring that the cell was
cleaned thoroughly between experiments. Our experiments on
the oxidation of CF CF H were carried out at 298 K and at
3
2
three oxygen partial pressures 10, 140 and 700 Torr. The
experiments were repeated at least three times for each partial
in previous studies, in the presence of varying amounts of O .
2
In addition, the dependence of product yields on the concen-
pressure of O and the yields shown are the average of all
2
trations of F and CF CF H in the originally prepared
such experiments.
2
3
2
mixture was observed. The experimental results are compared
with the results of kinetic modelling which was performed
using the FACSIMILE program.12 The present study di†ers
from those undertaken previously9h11 on the oxidation of
HFC-125 in that reaction is initiated by F, rather than Cl,
atoms. In addition, it is somewhat more comprehensive, in
that product yields were measured as a function of time, the
[CF CF H] : [F ] ratio, and the partial pressure of O
HFC-125 was obtained from Fluorochem and was the Ðrst
component admitted to the reaction vessel after de-gassing by
successive pumpÈfreezeÈthaw cycles. Some experiments were
also performed using CF H (obtained from Air Products) and
3
it was treated in a similar manner. The F was admitted as a
2
4.8% mixture in He (Distillers), any HF present being reduced
by passage of this mixture through a trap containing NaF. O
2
(BOC Ltd.) and N (BOC Ltd., “oxygen-freeÏ) were taken
2
3
2
2
2
included in the mixture that was photolysed.
directly from cylinders. As well as these gases used in the
initial mixtures, it was necessary to calibrate the sensitivity of
the FTIR to various product molecules, especially COF
which is not available commercially.
2
Experimental method and procedures
COF was made by the method of Fawcett et al14 by bub-
2
The present experiments were carried out in an apparatus
which has been used in previous measurements on the relative
rate of F-atom reactions with alkanes and HFCs and it was
described in some detail in a paper13 reporting that work. In
brief, it consists of a FTIR spectrometer (Nicolet Magna,
series 550) coupled to a large cylindrical (1.3 m long by 15 cm
diameter) cell. Two continuously operating “blacklampsÏ
(Phillips TL40N/08, 35 W output) are mounted within the cell
providing near-UV radiation, and the cell is equipped with
White cell mirrors to provide a long pathlength for analysis
by IR absorption. In the experiments reported here, mixtures
bling phosgene (COCl , 10 g) through
a solution of
2
NEt É 3HF (32 g) and acetonitrile (CH CN, 50 g) heated to
3
3
40 ¡C and held under a mild vacuum. A cold Ðnger immersed
in solid CO prevented any phosgene reaching the liquid-air
2
trap in which the products of reaction were collected. Subse-
quently, the COF was distilled from a trap at [82 ¡C to
2
remove any CO and was transferred to a blackened storage
2
bulb on the Pyrex gas line. Prior to its use, the gas was sub-
jected to several freezeÈpumpÈthaw cycles. Only minute quan-
tities of CO and SiF were observed as impurities in the
2
4
sample.
containing between 0.75 and 1.25 Torr of F and O20 mTorr
2
of HFC-125 were usually photolysed for periods of between
25 min and 1 h, although in some cases reaction was e†ec-
tively over after only 5 min. The optical path length was
usually set at 22 m and the spectra were recorded by co-
adding 16 scans taken at a resolution of 0.25 cm~1. At this
resolution, each scan took 3 s, enabling a spectrum to be
recorded in 48 s.
The reactor is internally coated with TeÑon to minimise
wall reactions and the temperature was controlled to within
^4 K by passing water through tubing coiled around and in
contact with the wall of the chamber. Gases were handled in a
Pyrex manifold and, to facilitate rapid mixing, were admitted
to the reaction cell via a narrow drilled tube which runs the
length of the chamber. Once all the gases had been intro-
Results
A main objective of the present work was to explain quanti-
tatively the kinetics and mechanism of the oxidation of
CF CF H under a range of experimental conditions. For this
3
2
purpose, the observed changes in concentration of reagents
and products have been compared with predictions made by
the modelling program FACSIMILE.12 In the investigations
of the oxidation of HFC-125 initiated by Cl atoms that are
cited in Table 1 only Ðnal product yields were measured.
Because the absorption of F in the near-UV is weaker than
that of Cl , the photolysis rate and hence the subsequent rates
of reaction are appreciably slower in the case of F-atom initi-
2
2
Table 1 Previous studies of the yield of COF in the oxidation of HFC-125a
2
ref.
pressure/Torr
results
Edney and Driscoll9
Tuazon and Atkinson10
Sehested et al.11
700
740
760
yield of COF at 298 K \ 1.09 ^ 0.05
2
yield of COF at 298 K B 1.0
2
yields depend on [Cl ]: [HFC-125]
(see test)
2
Wallington and Neilsen7
at 298 K, p(Cl ) \ 897 mTorr and
2
p(HFC-125) \ 11.8 mTorr:
yield of COF \ 1.00
2
yield of CF O CF \ 0.20
3
3
3
yield of CF O C F \ 0.06
3
3 2 5
a Studies conducted by continuous UV photolysis of Cl to generate Cl atoms with detection of products by FTIR spectroscopy.
2
2694
J. Chem. Soc., Faraday T rans., 1997, V ol. 93

View Article Online
ated oxidation than for Cl-atom initiated oxidation, under the
same conditions of illumination. In the present experiments,
the decay of reagent and the build-up of intermediates and
products could be conveniently followed as a function of time.
Although similar kinetic measurements could be carried out
using Cl atoms to initiate the reaction if the intensity of the
COF . Based on experiments of this kind and calibrations
of absorption strength with concentration in the bands of
2
CF CF H at 3000 and between 1280 and 1335 cm~1, and of
3
2
COF between 1800 and 2000 cm~1, it was possible to calcu-
2
late the changes in concentration of CF CF H and of COF .
The variation of these absolute concentrations with time is
shown in Fig. 2.
3
2
2
light photolysing Cl was reduced, no experiments of this kind
2
appear to have been performed on HFC-125, or indeed on
Fig. 2 shows that there is a signiÐcant induction period
during which both the consumption of CF CF H and the for-
any other HFC.
3
2
However, in order to compare the experimental and model
results, it was necessary to determine the Ðrst-order rate con-
stant for photolysis of F (k ). In experiments designed to
mation of the reaction products is slower than at longer times.
The reason for this is discussed later. Product yields per
CF CF H reacted were determined from observations at
2
p, 5
3
2
estimate k , F was photolysed in the presence of HCl and
longer times where, as the lines in Fig. 3 show, the product
concentrations increased linearly with loss of CF CF H. The
p, 5
2
the decrease in the concentration of HCl and the increase in
the concentration of HF were simultaneously measured. The
rates of HCl loss and HF formation were found to be the
same and were assumed to be twice the rate of photolysis of
F yielding a value of k \ [d ln[F ]/dt of 3.3 ] 10~6 s~1.
3
2
yield of COF determined in this way is 1.05 ^ 0.06. This
2
Ðnding is consistent with the production of COF via reac-
tions (5), (6), (2), (3b) and (4) only; which overall amounts to:
2
2
p, 5
2
2F ] 2CF CF H ] O ] 2HF ] 2COF ] 2CF
(7)
The reaction of Cl atoms, released in the reaction of F atoms
3
2
2
2
3
with HCl, and F , is known to be extremely slow15 and was
Furthermore, our Ðnding of a unit yield of COF in the
2
2
assumed not to generate additional F atoms in this system.
F-atom initiated oxidation of HFC-125 is in agreement with
This conclusion was supported by modelling calculations.
In our Ðrst experiments on the F-atom initiated oxidation
previous reports of this quantity in the Cl-atom initiated oxi-
dation (see Table 1).
of HFC-125, gas mixtures containing 140 Torr of O and a
Clearly a unit yield of COF for each CF CF H lost is con-
2
2
3
2
high ratio of F (1 Torr) to CF CF H (4 mTorr) were irradi-
sistent with the mechanism given above but, in addition, it is
2
3
2
ated for between 20 and 50 min, with spectra recorded every
few minutes to observe the loss of CF CF H and the forma-
tion of reaction products, especially COF . Fig. 1 shows IR
necessary to consider the fate of the CF radicals, or actually
3
the CF O radicals to which they will be immediately con-
3
2
2
3
2
verted under our experimental conditions. In this respect it is
interesting to note that the sequence of rapid reactions:
spectra from such a mixture recorded before illumination and
after 45 min of photolysis, showing clearly the loss of
CF CF H and the parallel increase in the concentration of
2CF CF O ] 2CF CF O ] O
(3b@)
(4)
3
2
2
3
2
2
3
2
2CF CF O ] 2CF ] 2COF
3
2
3
2
2CF ] 2O ] 2CF O
(8)
3
2
3 2
ÈÈÈÈÈÈÈÈÈÈÈÈÈÈÈÈÈÈÈÈ
yields overall 2CF CF O ] O ] 2CF O ] 2COF
(9)
3
2
2
2
3
2
2
so that there is no net removal of peroxy radicals. Further-
more, as reaction proceeds, the reaction of CF O with
3
2
Fig. 1 An example of IR spectra recorded (a) before, and (b) after 45
min photolysis of a mixture containing 1 Torr F , 4 mTorr CF CF H
2
3
2
and 140 Torr O , made up to 700 Torr total pressure with N . Bands
Fig. 2 Comparison of how the concentrations of HFC-125 (L),
2
2
due to CF CF H, COF , CF O CF and CF O C F are identiÐed.
COF (>) and CF O CF (=) were observed to vary with time in a
3
2
2
3
3
3
3
3
2 5
2
3
3
3
The di†erence in absorbance scale in these two diagrams should be
noted. The small sharp features between 1400 and 1800 cm~1 are due
to water which has not been cleared entirely from the path of IR
analysing radiation outside the photochemical reactor. The sharp
typical experiment with those predicted using FACSIMILE and the
rate data listed in Table 2: HFC-125 (ÈÈÈ), COF (È È È È) and
2
CF O CF (-------). The variations shown by the upper lines, i.e.
3
3
3
HFC-125 ( È È ), COF (È-È-È) and CF O CF (È--È--È), were
calculated using k \ 2 ] 10~6 s~1 rather than k \ 3.3 ] 10~6
p, 5 p, 5
2
3
3
3
feature at 1030 cm~1 is apparently due to SiF impurity, the presence
4
of which varied from experiment to experiment.
s~1.
J. Chem. Soc., Faraday T rans., 1997, V ol. 93
2695

View Article Online
Fig. 5 Dependence of yields of COF (>), CF O CF (=) and
2
3
3
3
CF O C F (…) on partial pressure of O . The partial pressures of
Fig. 3 Comparisons between the loss of CF CF H and increases in
3
3
2 5
2
3
2
F
and CF CF H were 1 Torr and 4 mTorr, respectively, and the
the concentrations of the products COF (>), CF O CF (=) and
2
3
2
2
3
3
3
total pressure was 700 Torr.
CF O C F (…). The reaction mixture was the same as that speciÐed
3
3 2 5
in the caption to Fig. 1.
CF CF O and the reaction between two CF O will become
from the oxidation of CF CF H. In order to quantify the
3
2
2
3 2
3
2
more important and yield CF O.
yield of CF O CF , we had hoped to generate stoichiometric
3
3
3
3
In addition to COF , like others,7,11 we have identiÐed
quantities of it from the F-atom initiated oxidation of CF H
2
3
CF O CF and CF O C F among the products resulting
in our apparatus. However, it was clear that this reaction pro-
3
3
3
3
3 2 5
duced several di†erent species and was not a clean source of
bis(triÑuoromethyl) trioxide. Consequently, we observed the
two trioxides by their absorptions at 1292 and 1082 cm~1 and
their concentration were estimated using absorption cross-
sections that had been measured previously by Wallington et
al.,16 in agreement with earlier measurements17,18 on the IR
spectra of these compounds. Fig. 2 shows how the concentra-
tion of CF O CF increases with time and Fig. 3 shows plots
3
3
3
of the concentrations of CF O CF and CF O C F
3
3
3
3 3 2 5
observed at various times after the initiation of reaction vs. the
decrease in the concentration of CF CF H at the same times.
3
2
As with COF , after a short induction period, the yields of
2
both these trioxides rose linearly with time and with decrease
in the concentration of CF CF H, and yields of 0.19 ^ 0.02
3
2
for CF O CF and 0.18 ^ 0.02 for CF O C F could be
3
3
3
3 3 2 5
determined. The combination of these yields with that of
COF gives a C atom yield of 1.97 ^ 0.09 and a F atom yield
2
of 4.68 ^ 0.23, relative to each molecule of CF CF H
3
2
destroyed, suggesting that the yields of any other products
must be very small. The quoted errors are single standard
deviations and the calculation of the F atom yield does not
include the HF product which is formed in reaction (6) and for
which the F atom comes from photolysed F . In contrast to
2
Sehested et al.,11 we were unable to detect CF OH or
3
CF O CF as products of reaction. Using absorption cross-
3
2
3
sections for CF OH or CF O CF measured previously,11,16
3
3
2
3
and assuming the BeerÈLambert law, we estimate that this
means that the yields of these species must be below ca. 0.5
and 0.3% under our experimental conditions.
Following the initial experiments using a large excess of F
2
over CF CF H and 140 Torr of O , two further series of
Fig. 4 Comparison between (a) the experimental and (b) the calcu-
3
2
2
experiments were conducted: one in which the partial pressure
of was maintained at 140 Torr but the ratio of
[CF CF H] to [F ] was varied, and a second where the
lated variation in the yields of COF (>), CF O CF (=) and
2
3
3
3
CF O C F (…) with changes in the relative concentrations of
2 5
CF CF H and F . In the experiments, the concentration of
O
3
3
2
3
2
2
3
2
2
CF CF H was varied whilst the partial pressures of F and O and
partial pressure of O was varied between 10 and 700 Torr
3
2
2
2
2
the total pressure were kept constant at 1, 140 and 700 Torr, respec-
tively. The lines are polynomials Ðt to the data for each species and
are included simply to “guide the eyeÏ. The yields in (b) were calcu-
lated according to the full model with the rate constants listed in
Table 2.
using the same concentrations of F and CF CF H.
The results of the Ðrst series of experiments are displayed in
Fig. 4(a). It can be seen that as the [CF CF H] : [F ] ratio
2
3
2
3
2
2
was increased the yield of COF fell and this decrease was
2
2696
J. Chem. Soc., Faraday T rans., 1997, V ol. 93

View Article Online
compensated for by an increase in the concentrations of the
trioxides. This observation is qualitatively similar to that
observed in experiments on the Cl-atom initiated oxidation of
CF CF H by Sehested et al.7,11 The results displayed in Fig.
with CF CF H has been measured as 3.65 ] 10~13 cm3
3
2
molecule~1 s~1 13 and 3.5 ] 10~13 cm3 molecule~1 s~1 19 in
two independent studies, so it must be considered well estab-
lished. Under typical conditions in our experiments (i.e. at a
maximum HFC partial pressure of 20 mTorr) the pseudo-Ðrst-
order rate constants for reaction of the F atoms would be
k [CF CF H] B 240 s~1. The addition of O to CF CF
3
2
5 show that the yields also depend, though not very strongly,
on the partial pressure of O included in the reaction mix-
2
tures. The yields of the trioxides were higher and that of
6
3
2
2
3
2
COF lower at 10 Torr of O than at 140 Torr of O ,
would be even faster at the pressures employed in our experi-
2
2
2
although these changes were not monotonic.
ments. Thus with 140 Torr of O present, k [O ] B 107 s~1.7
2
2
2
These pseudo-Ðrst-order rate constants demonstrate that
the reaction of F atoms, released in the photolysis of F , and
the conversion to substituted peroxy radicals of any substi-
Discussion
2
Qualitatively, the production of the trioxides, CF O CF and
tuted alkyl radicals, created by H atom abstraction from the
HFC, are very fast on the timescale of our experiments, sug-
gesting that steady-state concentrations of most of the active
species should be rapidly established. However, one factor
which has received relatively little attention in the interpreta-
tion of experiments in which F atoms initiate photooxidation
3
3
3
CF O C F , and the absence of the peroxide CF O CF in
3 2 5
the oxidation of CF CF H is not difficult to explain and has a
3
3
2
3
3
2
number of interesting implications. It seems that the CF O
3
2
radicals formed in reaction (8) may either react with
CF CF O radicals to form the mixed trioxide
3
2
is the role of FO created by the three-body association
CF O ] CF CF O(]M) ] CF O C F (]M) (10)
2
3
2
3
2
3 3 2 5
F ] O ] M % FO ] M
(12, [12)
or they can be converted to CF O via reaction (3b) with
2
2
3
2
another peroxy radical (i.e. CF O or CF CF O ). Since the
lifetime of CF CF O radicals with respect to dissociation to
The rate constant for the “forwardÏ reaction (12) has been
measured several times in direct experiments with fair agree-
ment.20,21,27h29 The latest kinetic data evaluation20d accepts
the results of Campuzano-Jost et al.21 They measured rate
constants at temperatures between 100 and 420 K and total
pressures between 1 and 1000 bar and concluded that, with
3
3
2 2
3
2
CF ] COF is short (see Table 2), and hence their steady-
3
2
state concentration will be low, the latter loss process for
CF O is much more probable, a conclusion which is con-
3
2
Ðrmed by the modelling calculations (see below). Once CF O
3
radicals are formed then they react with CF O or
M \ N , the rate constant in the limit of low pressure at 300
3
2
2
CF CF O to produce the observed trioxides,
K would be k \ 5.8 ] 10~33 cm6 molecule~2 s~1. Applying
3
2 2
12
small corrections for “fall-o†Ï and for the change in tem-
CF O ] CF O ,
3
3
2
perature, we Ðnd that the second-order rate constant for reac-
tion (12) at 295 K and a total pressure of 700 Torr is
1.25 ] 10~13 cm3 molecule~1 s~1, so that, in a mixture con-
taining 140 Torr of O , k [O ][M] \ 5.7 ] 105 s~1. This
CF CF O (]M) ] CF O CF ,
3
2
2
3
3
3
CF O C F (]M) (11a,b)
3
3 2 5
2
12
2
but they do not apparently associate to form the peroxide
CF O CF . Reaction (11b) between CF O and CF CF O ,
value is much greater than the Ðrst-order rate constant for the
reaction of F atoms with CF CF H (see above).
3
2
3
3
3
2
2
3
2
as well as reaction (10) if it occurs, serve to reduce the yield of
There has been far less agreement in the values derived for
k , the rate constant for dissociation of FO [or the equi-
librium constant for reaction (12)] than for k . This uncer-
COF .
2
~12
2
12
Before discussing the e†ects on the product yields of chang-
ing the [CF CF H] : [F ] ratio and the partial pressure of
tainty appears to have been removed by the comprehensive
3
2
2
O , we consider some of the features peculiar to experiments
measurements of Campuzano-Jost et al.21 Their experiments
yield a value of the equilibrium constant associated with reac-
2
where oxidation is initiated by F atoms, rather than Cl atoms.
The second-order rate constant for the reaction of F atoms
tion (12) i.e. k /k
\ 3.9 ] 10~16 cm3 molecule~1 (412.6
12 ~12
Table 2 Reactions and rate constants used in modelling calculations
ka/cm3 molecule~1 s~1
or s~1
reaction
ref.
F ] hl ] 2F
3.3 ] 10~6
this work
13
2
F ] CF CF H ] HF ] CF CF
3.65 ] 10~13
3
2
3
2
3.5 ] 10~13
1.25 ] 10~13
280
1.1 ] 10~11
2.1 ] 10~12
2 ] 104
19
F ] O ] FO
20(c), 21
20(c), 21
Èb
11
Èc
20
23
24
Èd
25
Èe
Èe
Èe
Èf
11
2
2
2
FO ] F ] O
2
CF CF ] O ] CF CF O
3
2
2
3
2 2
2CF CF O ] 2CF CF O ] O
3
2
2
3
2
2
CF CF O ] CF ] COF
3
2
3
2
CF ] O ] CF O
8.0 ] 10~12
1.8 ] 10~12
1.2 ] 10~12
2.0 ] 10~12
4.2 ] 10~12
2.5 ] 10~11
2.5 ] 10~11
2.5 ] 10~11
1.0 ] 10~15
4.0 ] 10~5
3
2
3 2
2CF O ] 2CF O ] O
3
2
3
2
CF O ] C F O ] CF O ] C F O ] O
2 5 2 5 2
2CF O ] CF OOCF
2
2
2
3
3
3
3
CF O ] CF O ] CF O CF
3
3
2
3
3
3
CF O ] CF CF O ] CF O CF CF
3
3
2
2
3
3
2
3
3
CF CF O ] CF O ] CF O CF CF
3
2
3
2
3
3
2
CF O ] CF CF H ] CF OH ] CF CF
3
3
2
3
3
2
CF OH ] COF ] HF
3
2
a Rate constants for association and dissociation reactions are quoted for the conditions appropriate to our experiments; i.e. 700 Torr total
pressure of O ÈN . b Set equal to the limiting high-pressure rate constant for reaction between CF and O given in ref. 20(a). c This rate
2
2
3
2
constant is set equal to that for CF CFHO ] CF ] HFCO given in ref. 22. d Taken to be the mean of the rate constants for CF O ] CF O
3
3
3
2
3 2
and C F O ] C F O . e These rate constants are not known; we used values for all three reactions corresponding to that used in ref. 24 for
2 5 2 5
CF O ] CF CF O . f Set equal to the rate constant for reaction between CF O and CF CFH given in ref. 26.
2
2
3
3
2
2
3
3
2
J. Chem. Soc., Faraday T rans., 1997, V ol. 93
2697

View Article Online
Torr~1) so that, in a equilibrium mixture containing 140 Torr
of O , [F]/[FO ] \ 5.7 ] 10~4.
2
2
Because reaction (12) is much faster than that of the F
atoms with CF CF H under our experimental conditions,
3
2
during the early part of the photoinitiated reaction the great
majority of the F atoms produced by photodissociation of F
2
react with O rather than with the HFC until not only the
2
equilibrium ratio [F]/[FO ] is established (which will occur
2
very rapidly), but also the steady-state concentration of F
atoms, determined by the balance between the rates of pho-
tolysis of F and the rate of the reaction between F atoms and
2
HFC-125, is reached. As a result, an “induction periodÏ is to
be expected before the rate of loss of the HFC and the rate of
formation of any resultant product attain their steady values.
This e†ect is apparent for the F-atom initiated oxidation of
CF CF H in the experimental results displayed in Fig. 2.
3
2
To examine this e†ect and the changes in concentration of
reagents and products under various experimental conditions
in more detail, we employed the modelling program
FACSIMILE12 including in the mechanism the reactions
listed in Table 2. Fig. 2 shows the result of a comparison
between the observed concentrations of CF CF H, COF and
3
2
2
CF O CF and those calculated using the rate constants for
3
3
3
the elementary reactions which are given in Table 2. The
agreement between these experimental results and the predic-
tion of the model depends on the rate constants for reactions
Fig. 6 Calculated dependence of COF (>), CF O CF (=) and
2
3
3
3
CF O C F (…) yields on the steady-state concentration of F atoms
(6), (12), ([12) and the rate of photolysis of F . The Ðrst three
3
3 2 5
2
of these rate constants are well established. As the dashed line
in Fig. 2 shows, the agreement between the experimental and
calculated results can be improved by lowering the photolysis
rate by 40% which is within the uncertainty of our determi-
nation, given the possibility of a slow reaction between Cl
mated from our experiments on F ÈHCl mixtures (see above).
Fig. 4(b) shows the results in a plot of product yields vs.
2
[CF CF H] : [F ]. Although there are signiÐcant discrep-
3
2
2
ancies between the experimental [see Fig. 4(a)] and model
results, the model goes some way to explaining the observed
variations in product yields. The e†ect of changing the steady-
state concentration of F atoms leads, of course, to changes in
the concentrations of peroxy and alkoxy radicals and it is this
which leads to the variation in product yields. Although the
Cl-atom initiated reaction system will di†er in details, it seems
likely that the explanation for the variation in product yields
found by Sehested et al.7,11 must be very similar.
atoms and F .
2
The calculations just referred to and others described below
suggest that FO serves simply as a temporary reservoir for F
2
atoms. There appears to be no strong evidence that FO plays
2
any major role in the oxidative chemistry of HFC-125. In par-
ticular, we note that reactions like
CF CF O ] FO ] CF CF O ] FO ] O
(13)
3
2
2
2
3
2
2
are endothermic and therefore cannot assist in the conversion
of substituted peroxy radicals to substituted alkoxy radicals
We next attempted to model the variations in relative
product yields that were observed when either the
[CF CF H] : [F ] ratio or the partial pressure of O was
Finally we performed model calculations on mixtures con-
taining di†erent partial pressures of O to see whether such
2
calculations could explain the experimental results displayed
earlier in Fig. 5. Here the calculations were even less successful
in describing the variation of observed product yields with
experimental conditions. The model indicated that the relative
3
2
2
2
varied. Since the primary e†ect of both these changes must be
to alter the steady-state concentration of F atoms, we decided
to begin by carrying out modelling calculations using FAC-
SIMILE to investigate how the yields of products depended
on the concentration of F atoms. In these preliminary calcu-
lations, rather than include reactions (5), (6) (12) and ([12) in
the reaction mechanism, a steady-state concentration of F
atoms was speciÐed and then the model was run including the
remainder of the reactions given in Table 2 to calculate
product yields. The results demonstrated that the product
yields do indeed show signiÐcant, and complex, dependences
on the steady-state concentration of F atoms. As an example
of these results, Fig. 6 shows how the calculated product
yields vary with the concentration of F atoms when the initial
concentration of CF CF H was set at 1 ] 1014 molecule
yields of COF and the two trioxides do depend on the con-
2
centration of O present, but neither the form of the variation
2
nor its absolute magnitude were reproduced. A possible
reason is the occurrence of slow but signiÐcant reactions
involving F, FO and FO but in the absence of any Ðrm data
2
regarding these reactions29 we made no further attempts to
match the calculated results to those found experimentally.
Summary and Conclusions
We have carried out the Ðrst thorough examination of the
F-atom initiated oxidation of HFC-125 (CF CF H). Under
3
2
laboratory conditions, the main products are COF ,
2
3
2
CF O CF and CF O C F . It is clear that the main, and
probably only, loss process for the CF CF O radicals formed
in this oxidation is scission to COF ] CF . These CF CF O
radicals are formed in our experiments either by reaction
between two CF CF O radicals or by reaction of
cm~3 (43 mTorr) and the partial pressure of O was 140
3
3
3
3 3 2 5
2
Torr. Over the range of the curves shown in bold in Fig. 6, the
3
2
3
calculated product yields and their variation are similar to
those found experimentally as the ratio [CF CF H] : [F ]
2
3
2
3
2
2
was changed. However, it should be noted that this range of F
atom concentrations is much larger than the range of
[CF CF H] : [F ] ratios covered in our experiments.
3
2 2
CF CF O with CF O . In the atmosphere, the oxidation of
3
2
2
3 2
HFC-125 is initiated by its reaction with OH radicals for
which the rate constant is given by 4.9 ] 10~13 exp([1655/
T ),20c yielding a lifetime of approximately 20 years. Conse-
quently, any HFC-125 released into the atmosphere will
3
2
2
We then carried out full model calculations with the rate
constants for reactions (6), (12) and ([12) set to their well
established literature values and with k
given the value esti-
p, 5
J. Chem. Soc., Faraday T rans., 1997, V ol. 93
2698

View Article Online
13 C. M. Moore and I. W. M. Smith, J. Chem. Soc., Faraday T rans.,
1995, 91, 3014.
disperse widely. CF CF O radicals will then be formed by the
3
2
reactions of CF CF O radicals either with other peroxy rad-
3
2 2
14 F. S. Fawcett, C. W. Tullock and D. D. Co†men, J. Am. Chem.
Soc., 1962, 84, 4275.
icals or with NO. It seems certain that the overwhelming fate
of CF CF O radicals in the atmosphere will be to fragment to
15 C. M. Moore, I. W. M. Smith and D. W. A. Stewart, Int. J. Chem.
Kinet., 1994, 26, 813.
16 T. J. Wallington, J. Sehested, M. A. Dearth and M. D. Hurley, J.
Photochem. Photobiol. A: Chem., 1993, 70, 5.
17 L. R. Anderson and W. B. Fox, J. Am. Chem. Soc., 1967, 89, 4313.
18 P. G. Thompson, J. Am. Chem. Soc., 1967, 89, 4316.
19 T. J. Wallington, M. D. Hurley, J. Shi, M. M. Maricq, J. Sehested,
O. J. Nielsen and T. Ellermann, Int. J. Chem. Kinet., 1993, 25,
651.
3
2
COF ] CF .
2
3
Although the formation of the trioxides, CF O CF and
3
3
3
CF O C F , will be unimportant in the atmosphere, our ina-
3
3 2 5
bility to model how the observed product yields in the oxida-
tion of CF CF H depend on the [CF CF H] : [F ]. ratio
3
2
3
2
2
and on the O partial pressure is disappointing. It demon-
strates that our knowledge of some of the secondary chemistry
2
in this system is still less than complete.
20 (a) R. Atkinson, D. L. Baulch, R. A. Cox, R. F. Hampson, J. A.
Kerr and J. Troe, J. Phys. Chem. Ref. Data, 1989, 18, 881; (b) R.
Atkinson, D. L. Baulch, R. A. Cox, R. F. Hampson, J. A. Kerr
and J. Troe, J. Phys. Chem. Ref. Data, 1992, 21, 1125; (c) R.
Atkinson, D. L. Baulch, R. A. Cox, R. F. Hampson, J. A. Kerr
and J. Troe, J. Phys. Chem. Ref. Data, 1997, 26, 521; (d) W. B.
Demore, S. P. Sander, D. M. Golden, R. F. Hampson, M. J.
Kurylo, C. J. Howard, A. R. Ravishankara, C. E. Kolb and M. J.
Molina, Chemical Kinetics and Photochemical Data for Use in
Stratospheric Modeling Evaluation Number 11, NASA JPL-
Publication 94-26, 1994.
We thank SERC and NERC for support of our work under
the programme in Laboratory Measurements for Atmospheric
Chemistry. A.H. and C.M.M. are also grateful to the same
bodies for the award of research studentships. We thank Pro-
fessor J. A. Kerr for showing us the latest IUPAC kinetic data
evaluation [ref. 20(c)] prior to publication.
References
21 P. Campuzano-Jost, A. Croce de Cobos, H. Hippler, M. Siefke
and J. Troe, J. Chem. Phys., 1995, 102, 5317.
1
2
3
4
F. S. Rowland and M. J. Molina, Nature (L ondon), 1974, 249, 810.
F. S. Rowland and M. J. Molina, Chem. Eng. News, 1994, 72, 8.
D. J. Weubbles, J. Geophys. Res., 1983, 88, 1433.
ScientiÐc Assessment of Stratospheric Ozone, 1989, vol II; World
Meteorological Organization, Global Ozone Research and Mon-
itoring Project, report no. 20.
R. P. Wayne, Chemistry of Atmospheres, Clarendon Press,
Oxford, 2nd edn., 1991.
A. R. Ravishankara and E. R. Lovejoy, J. Chem. Soc., Faraday
T rans., 1994, 90, 2159.
T. J. Wallington and O. J. Nielsen, in Progress and Problems in
Atmospheric Chemistry, ed. J. R. Barker, World ScientiÐc, Singa-
pore, 1995, ch. 15.
22 M. M. Maricq and J. S. Szente, J. Phys. Chem., 1992, 96, 10 862.
23 O. J. Nielsen, T. Ellerman, J. Sehested, E. Bartkiewicz, T. J. Wall-
ington and M. D. Hurley, Int. J. Chem. Kinet., 1992, 24, 1009.
24 P. Biggs, C. E. Canosa-Mas, J.-M. Fracheboud, C. J. Percival,
R. P. Wayne and D. E. Shallcross, J. Chem. Soc., Faraday T rans.,
1997, 93, 379.
5
6
7
25 (a) R. C. Kennedy and J. B. Levy, J. Phys. Chem., 1972, 76, 3480;
(b) L. Batt and R. Walsh, Int. J. Chem. Kinet., 1982, 14, 933.
26 J. Sehested and T. J. Wallington, Environ. Sci. T echnol., 1993, 27,
146.
27 P. Pagsberg, E. Ratajczak, A. Sillesen and J. T. Jodkowski, Chem.
Phys. L ett., 1987, 141, 88.
8
9
B. Sukornick, Int. J. T hermophys., 1982, 10, 553.
E. O. Edney and D. J. Driscoll, Int. J. Chem. Kinet., 1992, 24,
1067.
28 J. L. Lyman and R. Holland, J. Phys. Chem., 1988, 92, 7232.
29 T. Ellerman, J. Sehested, O. J. Nielsen, P. Pagsberg and T. J.
Wallington, Chem. Phys. L ett., 1994, 218, 287.
10 E. C. Tuazon and R. Atkinson, J. Atmos. Chem., 1993, 16, 301.
11 J. Sehested, T. Ellerman, O. J. Nielsen, T. J. Wallington and
M. D. Hurley, Int. J. Chem. Kinet., 1993, 25, 701.
12 FACSIMILE program; UKAEA Harwell Laboratory, 1987.
Paper 7/02105D; Received 26th March, 1997
J. Chem. Soc., Faraday T rans., 1997, V ol. 93
2699