The CF3 + NO2 rate constant measured between 1.5 and 110 Torr and between 251 and 295 K by time resolved infrared emission

Article abstract of DOI:10.1039/b004709k

The rate constant k₁ for reaction of CF3 with NO2 was investigated over the pressure range 1.5-110 Torr Ar or N2 and at temperatures between 251 and 295 K. Time resolved emission from vibrationally excited CF2O and FNO products was used as a detection method for CF3 removal. The rate constant at room temperature shows no marked variation with pressure, with an average value of k₁ = (1.75 +/- 0.26) x 10^-11 cm³ molecule^-1 s^-1, where the errors are 2?, but a small reduction at the lowest pressure indicates the possible contribution from a third body recombination process. We conclude that this process is minor, and if it is treated as a separate reaction channel it results in a limiting high pressure branching ratio of 13 +/- 7 percent, which drops to half that value at 10 Torr. As the limiting high and low pressure rate constants are within the 2? error bars of k₁ we recommend the average value as applying over the full pressure range. Combination with previous results shows that of the possible bimolecular channels, formation of CF2O + FNO dominates (ca. 95 percent). No temperature dependence is seen, within experimental error. Pitfalls in the extraction of rate constants from the analysis of time resolved product emission are discussed, and a method is described to assign rate constants unambiguously to reactive formation and collisional quenching. CF3 radicals were formed from the 248 nm photolysis of CF3I, and observations of emission from the radical show previously unobserved excitation in the asymmetric C-F stretching mode ν₃.

Full text of DOI:10.1039/b004709k

View Article Online / Journal Homepage / Table of Contents for this issue
The CF + NO rate constant measured between 1.5 and 110 Torr
3
2
and between 251 and 295 K by time resolved infrared emission
Ciara Breheny, Gus Hancock* and Claire Morrell
Physical and T heoretical Chemistry L aboratory, Oxford University, South Parks Road, Oxford,
UK OX1 3QX
Received 13th June 2000, Accepted 11th September 2000
First published as an Advance Article on the web 26th October 2000
The rate constant k for reaction of CF with NO was investigated over the pressure range 1.5È110 Torr Ar
1
3
2
or N and at temperatures between 251 and 295 K. Time resolved emission from vibrationally excited CF O
2
2
and FNO products was used as a detection method for CF removal. The rate constant at room temperature
3
shows no marked variation with pressure, with an average value of k \ (1.75 ^ 0.26) ] 10~11 cm3
1
molecule~1 s~1, where the errors are 2p, but a small reduction at the lowest pressure indicates the possible
contribution from a third body recombination process. We conclude that this process is minor, and if it is
treated as a separate reaction channel it results in a limiting high pressure branching ratio of 13 ^ 7%, which
drops to half that value at 10 Torr. As the limiting high and low pressure rate constants are within the 2p error
bars of k , we recommend the average value as applying over the full pressure range. Combination with
1
previous results shows that of the possible bimolecular channels, formation of CF O ] FNO dominates (ca.
2
95%). No temperature dependence is seen, within experimental error. Pitfalls in the extraction of rate constants
from the analysis of time resolved product emission are discussed, and a method is described to assign rate
constants unambiguously to reactive formation and collisional quenching. CF radicals were formed from the
3
248 nm photolysis of CF I, and observations of emission from the radical show previously unobserved
3
excitation in the asymmetric CÈF stretching mode m .
3
pitfalls in using kinetic observations of products for rate con-
stant measurements.
Introduction
Atmospheric degradation of a number of Ñuorocarbon species
Although the CF ] NO reaction is only of minor atmo-
3
2
leads to the formation of the CF radical,1 and its subsequent
spheric importance, its rate constant has received a great deal
of experimental attention, and the results of eight previous
studies are shown in Table 1.3h10 At Ðrst sight, these data
seem to indicate a trend of increasing rate constant with
increase in total pressure, but closer inspection shows that this
interpretation depends crucially on the accuracy of the two
lowest pressure studies. The very low pressure photolysis mea-
surement of Rossi et al.3 was carried out under conditions
which have been suggested as leading to an overestimation of
3
chemistry has been the subject of a number of experimental
investigations. The radicalÏs major atmospheric fate is recom-
bination with molecular oxygen, and we have recently investi-
gated the rate constant for this reaction by using time resolved
infrared (IR) emission as a marker for the decay of the radical
in the presence of O and NO .2 The emission results from
2
2
vibrationally excited products formed from the fast reaction of
CF with NO , and one of the requirements of the study of
3
2
the competition between removal by NO and O has been a
thorough understanding of the CF ] NO reaction. Here we
describe measurements which provide values of the CF
] NO rate constant between 1.5 and 110 Torr total pressure
and between 251 and 295 K and which illustrate potential
the low pressure of NO ,4 and hence to a low value of the
2
2
2
rate constant. Bevilacqua et al.6 reported only a brief investi-
3
2
gation of the reaction in the course of a study in which the
major aim was to observe the kinetics of CF O and CF O
3
2
3
2
3
species, and their value has correspondingly large error bars.
Table 1 Measurements of k
1
Study
Pressure range/Torr
Rate constant/10~11 cm3 molecule~1 s~1
Rossi et al.3
0.005
0.27 ^ 0.05
1.0 ^ 0.7
1.6 ^ 0.3
1.5 ^ 0.2
2.5 ^ 0.3
2.4 ^ 0.5
2.5 ^ 0.3
3.2 ^ 0.7
Bevilacqua et al.6
Sehested et al.8a
Pagsberg et al.10
Francisco and Li5
Oum and Hancock7
Sugawara et al.4
Vahktin9
0.8È2 (He)
2È10 (Ar/N )
2
3È17 (Ar)
4È5 (He/Ar)
5 (Ar)
4È20 (Ar)
300È600 (He)
760 (SF )
Sehested et al.8b
This work
2.10`0.7
6
~0.3
1.5È110 (Ar/N )
1.75 ^ 0.26
2
a IR Ñuorescence. b Pulse radiolysis.
DOI: 10.1039/b004709k
Phys. Chem. Chem. Phys., 2000, 2, 5105È5112
This journal is ( The Owner Societies 2000
5105

View Article Online
Furthermore, a pressure dependent reaction which from the
data of Table 1 would appear to have reached its limiting high
pressure value at around 5 Torr would need an unrealistically
high value of its third body recombination rate constant to
reproduce the apparent fall-o† behaviour. A series of consis-
tent measurements over a wider pressure range than provided
by the individual studies in Table 1 would therefore appear
warranted. Furthermore, no measurements have been report-
ed at temperatures other than close to 295 K.
at least a factor of Ðve faster than reaction, and so that the fall
of the emission signal corresponds to process (1). In this way
we were able to measure the rate constant over the pressure
range 1.5È110 Torr and between 251 and 295 K, and conclude
that within experimental error a single value of the rate con-
stant is appropriate under all conditions.
Experimental
Several product channels have been suggested for the CF
CF radicals were formed by the 248 nm pulsed photolysis of
3
3
] NO reaction:
CF I. For experiments at room temperature the radiation
2
3
from an excimer laser (Lambda Physik Compex 102, 250 mJ
CF ] NO ] CF O ] FNO
per pulse, 10 Hz) was multipassed (13 times) into a stainless
steel reaction vessel. IR radiation was collected with Welsh
optics and directed either into an FTIR spectrometer (Bruker
IFS/66) or through an interference Ðlter and observed with
either an InSb (Graseby IS2) or HgCdTe (Kolmar D317)
detector. The interferometer was operated in step scan mode12
and the time resolution of the detectors and associated elec-
tronics was set to D2 ls. Signals were digitised at 1 ls
channel width and averaged for a pre-set number of laser
shots. A reaction vessel without multipass or Welsh collection
optics was used for experiments at lower temperatures.
2
2
2
*H¡ \ [270 kJ mol~1 (1a)
298
] CF O ] F ] NO
2
*H¡ \ [31 kJ mol~1 (1b)
298
] CF O ] NO
3
*H¡ \ [226 kJ mol~1 (1c)
298
] CF NO (CF ONO)
3
2
3
*H¡ \ [247([314) kJ mol~1 (1d)
NO (BOC, [99.5%) was puriÐed by reaction with excess
298
2
O
with the O then being removed by freezeÈthaw cycling.
and the data of Table 1 have been obtained by observations of
2
2
Ar (BOC, [99.999%), N O (BOC, [99.997%), N (BOC,
the kinetic behaviour of both reactants,3,4,6,8h10 and pro-
2
2
[99.998%) and CF I (Fluorochem, [99%) were used
ducts.4,5,7,8 Time resolved observations of CF O and FNO
3
2
without further puriÐcation. A slow Ñow of gases was used to
have shown them to be directly formed by reaction (1),
replenish photolysed molecules and to remove products, with
reagent partial pressures calculated from measured total pres-
sures and reagent Ñows.
4,5,7,8,10 and product analyses have shown that this channel
dominates.6,8h11 A branching ratio of k /k B 3 has been
1a 1d
inferred by the long time observation of stable addition pro-
ducts following the reaction at high pressures ([300 Torr),9,11
but quantitative observations of CF loss and CF O pro-
Results
3
2
duction by time resolved diode laser absorption at low pres-
sures (\17 Torr) have concluded that 95 ^ 4% of the reaction
proceeds through channel (1a), with no stable adduct observ-
(a) Emission from 248 nm irradiation of separate samples of
CF I and NO
3
2
able on the time-scale of the CF decay.10 As the formation of
3
The 248 nm irradiation of CF I results in prompt IR emission
from CF , and a delayed feature which is from the parent
molecules formed vibrationally excited by energy transfer. The
strongest prompt feature is between 1200 and 1270 cm~1 and
is assigned to *m \ [1 transitions in CF , shifting with time
to higher wavenumbers as vibrational relaxation takes place.
Fig. 1 shows a contour plot of the spectral intensities around
1200 cm~1 as a function of time. At higher wavenumbers (and
at a factor 20 lower intensity) similar behaviour was seen for
bands near 2300 and 2500 cm~1, assigned as the combination
band *m \ *m \ [1 and the overtone band *m \ [2,
respectively. Previous experimental studies at this wavelength
have either identiÐed13 or assumed14 excitation in the sym-
metric modes of CF (the umbrella m and the stretch m ), with
CF NO may be a†ected by long time heterogeneous pro-
3
3
2
cesses, a study over a larger pressure range than employed in
the individual measurements shown in Table 1 will help to
clarify the importance of the recombination channel (1d).
The interpretation of reactant loss data in terms of the
overall rate constant for process (1) is straightforward, but
that of product formation can be ambiguous if the observed
product itself is also subject to removal processes. This is the
case in the present experiments (and in previous studies5,7,8),
3
3
3
where time resolved IR emissions from the CF O and FNO
products are observed. For a reaction sequence (1a) producing
2
3
1
3
vibrationally excited CF Ot followed by the quenching
2
process
3
2
1
CF Ot ] M ] CF O ] M
(2)
the time-of-Ñight measurements of Felder showing the clearest
evidence of m excitation.13 In contrast to the behaviour of the
asymmetric m mode, we did not detect emission on the
2
2
the time dependent concentration of CF Ot (and hence the
2
2
time dependence of the observed signal) will take the form
3
*m \ [2 overtone near 1400 cm~1, presumably because of
2
low Einstein A factors for this transition. Our data are not
A
k [CF ] [NO ]
3 0
B
1
2
[CF Ot] \
(e~k2*M+t [ e~k1*NO2+t) (3)
sufficient to determine the extent of vibrational partitioning
2
k [NO ] [ k [M]
1
2
2
into the observed modes, but our observations of the *m \
3
leading to a double exponential rise and fall behaviour where
the rising rate will always be the faster of the two rates
k [NO ] and k [M]. Additional evidence therefore needs to
[2 overtone implies excitation of m at least equal to 2, and
3
suggests that the vibrational distribution in the CF fragment
3
is more complex than previously considered. Finally in this
1
2
2
be found to decide which of the rising or falling rates should
be associated with reaction, and in only one of the previous
IR emission studies7 has this been explicitly provided. We
note that expression (3) clearly cannot apply when the rates of
processes (1) and (2) are equal, and we discuss the practical
problems associated with Ðtting the data when this condition
is approached.
section, a band at 1185 cm~1 was seen to peak at a delay from
the prompt CF emission, and was removed by a cell contain-
3
ing CF I as a cold gas Ðlter (1.8 Torr, 8 cm pathlength). The
3
band is identiÐed as the antisymmetric CF stretching mode,
*m \ [1, in CF I formed by energy transfer from the
4
3
nascent CF . A sample of NO also showed emission features
3
2
when irradiated at 248 nm, and these will be described in a
In this study, we used time resolved IR emission to measure
the rate of process (1) under conditions such that quenching is
separate publication.2 Fortunately for this study, the absorp-
tion of NO is at a minimum at 248 nm (with an absorption
2
5106
Phys. Chem. Chem. Phys., 2000, 2, 5105È5112

View Article Online
Fig. 2 Plots of the emission spectra between 1100 and 2100 cm~1 as
Fig. 1 Contour plot of the emission from 248 nm irradiated CF I
a function of time after the irradiation of a mixture of CF I (9 mTorr),
3
3
NO (63 mTorr) and Ar (1.8 Torr). Spectra are summed in 20 ls inter-
2
(51 mTorr in 1.8 Torr Ar) as a function of time and wavenumber. The
feature near 1250 cm~1 appears promptly and shifts to higher wave-
vals between 20 and 200 ls at a resolution of 6 cm~1. Features near
number with time, and is assigned to the *m \ [1 transition in the
1900 cm~1 are assigned as the *m \ [1 bands of CF O (near 1900
3
CF radical. The band at 1185 cm~1 reaches its maximum at later
3
1
2
cm~1 and shifting to higher wavenumbers with time) and the *m \
1
times, and is assigned to the *m \ [1 transition in the parent CF I,
[1 bands of FNO (showing the double peaked P and R branches
4
3
formed by energy transfer from CF . Resolution 4 cm~1.
centred at 1844 cm~1). Between 1170 and 1280 cm~1 the emission
3
seen in the absence of NO (Fig. 1) is augmented by emission from the
2
*m \ [1 bands of CF O near 1250 cm~1.
4
2
cross-section a factor of 20 lower than that of CF I), but
3
bands corresponding to fundamental and Ðrst overtone tran-
sitions in NO(v) produced by photolysis are clearly seen for
FNO. In principle, time resolved data at speciÐc wavenum-
bers, characteristic of particular species, can be extracted from
the spectra of Fig. 2 and 3. Fig. 4(a) shows decay curves for
two di†erent wavenumber regions, and illustrates a potential
pitfall in the observations of unresolved IR emission. Emission
summed between 1850 and 1910 cm~1 peaks at earlier times
than that at higher wavenumbers, 1920È1970 cm~1, and this
levels v \ 1È8 between 1700 and 1900 cm~1 and between 3400
and 3700 cm~1, respectively. Emission from NO (0,0,1È0,0,0)
2
at 1600 cm~1 results from excited molecules formed by energy
transfer processes, and at higher pressures and lower tem-
peratures emission from NO produced by photolysis of the
2
N O dimer can be seen at 1450 and 2750 cm~1, and is
2
4
is because the emission from CF O shifts to higher wavenum-
assigned to highly vibrationally excited NO emitting in the
2
2
ber with time because of vibrational cascade from the nascent
*m \ [1 and *m \ *m \ [1 bands, respectively.
3
3
1
vibrationally excited levels. Careful inspection of the time
resolved spectra shows that in the region 1700È2100 cm~1 in
Fig. 3 the FNO emission does not show a high wavenumber
shift, and thus for kinetics to reÑect the true time dependence
of the nascent products of process (1) emission should be
viewed in a region that encompasses FNO or the highest
Although, as will be shown, the bands from NO irradiation
2
are weak in comparison with those from the products of the
CF ] NO reaction, an understanding of their identiÐcation
3
2
and behaviour is crucial as they interfere kinetically with the
data.
vibrational levels (i.e., the lowest wavenumber) of CF O. Fig.
3 shows the transmission of an interference Ðlter in such a
2
(b) Emission from CF + NO
3
2
region, 1850È1910 cm~1, and Fig. 4(b) illustrates the time
Fig. 2 shows time resolved spectra in the region 1100È2100
cm~1 arising from the photolysis of a mixture of CF I and
3
NO . Additional features are seen, particularly near 1900
2
cm~1, which are absent in the irradiation of the separate
species, and are identical with those seen in our previous
study of the reaction when the CF reagent was formed from
3
infrared multiple photon dissociation of CF I.7 Fig. 3 shows a
3
time integrated spectrum between 1700 and 2100 cm~1, where
the emission is identiÐed as CF O emitting in the *m \ [1
2
1
band (the CÈO stretch) near 1930 cm~1 and from FNO emit-
ting in the *m \ [1 band, showing the double peak expected
1
at this resolution from the P and R branches of the (1,0,0 ] 0,
0,0) band centred at 1844 cm~1. The superior resolution and
wavenumber accuracy of the present instrument compared
with our previous laboratory-made step scan interferometer is
illustrated in the spectra observed in Fig. 3 and shown in Fig.
4 of our previous work.7 In the latter study, a small wavenum-
ber shift had led to our assignment of the two peaks near 1844
cm~1 as arising from FNO and another reaction product, and
the isomeric FON was suggested as a possibility. This now
seems not to be the case; the present spectrum is entirely con-
sistent with the double peaked carrier near 1844 cm~1 being
Fig. 3 Time integrated spectrum from the data in Fig. 2 between 0
and 400 ls. Marked on the spectrum are the positions of the absorp-
tion bands of CF O and FNO, where long time emission from the
2
products of reaction would be expected. Also shown as a dashed line
is the transmission curve of the Ðlter used in kinetic experiments.
Phys. Chem. Chem. Phys., 2000, 2, 5105È5112
5107

View Article Online
Fig. 4 (a) Time dependences of the emission summed between 1920
and 1970 cm~1 (upper trace) and 1850 and 1910 cm~1 (lower trace)
taken from the data in Fig. 2. It can be seen that the emission at the
higher wavenumber peaks at a later time, indicative of vibrational
cascade. (b) Time dependences of the emission following irradiation of
a mixture of CF I (47 mTorr), NO (150 mTorr) and Ar (6.8 Torr).
The lower trace shows emission measured through the interference
Ðlter whose spectrum is shown in Fig. 3. The upper trace is from time
resolved spectra of the kind shown in Fig. 2, summed over the wave-
number range corresponding to the Ðlter transmission. The two traces
have been separated in vertical scale, as they are indistinguishable,
apart from the clear improvement in S/N ratio in the Ðlter data.
Fig. 5 Results of Ðtting the double exponential expression (3) to
emission data taken through the interference Ðlter and corrected for
the small amount of emission from NO as explained in the text. The
upper panel shows the rising rates and the lower panel the falling
3
2
rates, each as a function of NO pressure for a range of total pressures
2
of Ar. P(Ar) \ (+) 2.8, (=) 7.8, (]) 12.8, ())17.8, (>) 21.8 and (L)
25.3 Torr.
tion a low pressure bimolecular component [processes (1aÈc)]
of ca. 2 ] 10~12 cm3 molecule~1 s~1, slower than previously
observed. The rising rate constant is for quenching of the
resolved behaviour in this region obtained from the interfer-
ometer traces compared with that from the Ðlter. The time
dependences are identical, but the enhanced S/N ratio of the
Ðlter data allows a far more comprehensive evaluation of the
kinetics, and was used in all further kinetic measurements.
As described above, there is an additional emitting species
in the region of the Ðlter, namely that from vibrationally
emitting species by NO and experimental error is invoked to
2
excited NO formed from the 248 nm photolysis of NO .
2
Although there is very little nascent NO formed in the levels
viewed by the Ðlter (v \ 1, 2 and 3),2,15 vibrational cascade
will lead to these levels being populated at longer times and
thus interfere with the extraction of k . The small e†ect can be
1
seen at long times in the low noise Ðlter trace of Fig. 4(b),
where the signal does not fall exponentially to zero at the rate
expected from the earlier time behaviour. Traces were taken
with and without CF I present and subtracted to eliminate
3
this e†ect. This procedure is valid only if CF I has no e†ect on
3
the NO(v) kinetics, and this was proved from measurements of
the behaviour of the NO(1È0) emission with and without
CF I present. A cold gas Ðlter was placed in front of the
3
detector under conditions such that emission from NO(1È0)
was absorbed, but non-resonant emission (from other species
and from higher levels of NO) was una†ected. The kinetic
behaviour of the NO(1È0) emission was identical in the two
cases, showing that CF I at the pressures used had a negligi-
3
ble e†ect on the vibrational cascade of NO.
Data such as those of Fig. 4(b), corrected for NO emission,
were found to Ðt the double exponential expression (3), and
rising and falling rates were linear functions of [NO ] as
shown in Fig. 5. The extracted rising and falling rate constants
Fig. 6 Slopes of the plots of rising rates (upper panel) and falling
2
rates (lower panel) against NO pressure shown in Fig. 5 as a function
2
of total pressure of Ar. The data appear to be consistent with a pres-
are shown as a function of Ar pressure in Fig. 6. At Ðrst sight
the data could be interpreted in the following way. The falling
rate constant corresponds to process (1), as it shows a depen-
dence on total pressure which can only be explained by a
third body recombination e†ect [process (1d)], and which
would not be expected for simple quenching. There is in addi-
sure dependent falling rate constant, which could be interpreted as
that for reaction (1) having a dominant third body component,
process (1d). This explanation, as explained in the text, is erroneous,
and illustrates the pitfalls of attempting to Ðt expression (3) to data
even as good as the example shown in Fig. 4(b) when rising and
falling rates are similar.
5108
Phys. Chem. Chem. Phys., 2000, 2, 5105È5112

View Article Online
account for its variation with Ar pressure. There are, however,
several problems with this interpretation. First, the marked
quencher, which does not react with CF , so that quenching
is always the faster process and the falling rate then becomes
3
intercept of the falling rate at zero NO pressure (Fig. 5)
remains unexplained, and second, the noticeable downward
the true rate of reaction. Fig. 7 shows the e†ect of adding 1
2
Torr N O as the chosen quencher. The rising rate is now
2
trend in rising rate constant with increasing total pressure
(Fig. 6) has been ignored. The interpretation turns out to be
incorrect, and is illustrated here to emphasise the pitfalls of
analysing data even of such high quality as shown in Fig. 4(b).
The problem arises through the relative magnitudes of the
rising and falling rates. At low Ar pressure, reaction is faster
than quenching, but the reverse is true at high Ar pressure.
This means that identiÐcation of each of the rising and falling
rate constants of Fig. 6 with a single process is incorrect
except in the limits of high and low Ar pressures. The trends
in Fig. 6 illustrate this qualitatively, in that similar rate con-
stants are returned from the analysis of for example the rising
rates at low [Ar] and the falling rates at high [Ar], which now
would be associated dominantly with the reactive component.
Simulations of reaction (1) and quenching (2) under these con-
ditions were Ðtted by the double exponential expression (3)
(taking great care over the Ðtting when the rates approached
equality), and the extracted rate constants showed the same
behaviour as seen in Fig. 6. At some combination of Ar and
markedly increased, but at the expense of a reduced signal,
and one of the consequences of this is that the emission from
NO becomes proportionally larger. Fig. 8 shows the emission
signals with and without CF I present together with the sub-
3
tracted data which now can be analysed to extract k . In con-
1
trast to the data shown in Fig. 4(a), spectra taken with N O
2
present now showed no marked di†erence in the time behav-
iour summed over the same spectral windows as before, illus-
trating the added advantage of removal of the e†ect of
vibrational cascade by addition of N O. Furthermore, obser-
2
vation of the e†ect of N O on CF emission showed that the
2
3
vibrationally excited reactant was quenched on a time scale
faster than that of reaction, thus removing the potential
problem of CF relaxation.
3
Emission traces in the presence of N O were analysed by
2
the FACSIMILE model including the reactions in Table 2,
and an example of the Ðt is shown in Fig. 8. The returned rate
constants k are shown as a function of total pressure (both
1
Ar and N ) between 1.5 and 37 Torr in Fig. 9. The straight
2
NO concentrations this means that the true reaction and
dashed line is the unweighted average of the values of k
2
1
quenching rates will become equal, and these processes would
returned from 67 measurements at a variety of NO pressures
2
produce a time dependent behaviour that cannot be analysed
by eqn. (3). We still Ðnd, however, very good Ðts to the experi-
mental data but which return di†erent rates under conditions
where we would expect similarity.
We believe that there are two reasons for this. We have
made no allowance for the behaviour of vibrationally excited
CF , undoubtedly produced but in an unknown fraction by
3
the photolysis of CF I. Vibrational cascade and/or changed
3
reactivity with vibrational excitation could a†ect the returned
rates. In addition, there is secondary chemistry occurring in
the system, and the time dependent behaviour will not be pre-
cisely described by a simple reaction scheme which incorpor-
ates only processes (1) and (2). O atoms are formed from NO
photolysis and, although their main fate will be the reaction
2
with NO , they can regenerate CF radicals by reaction with
Fig. 7 Kinetic behaviour observed through the Ðlter with and
2
3
CF I, thus leading to an underestimate of the rate of process
without N O (1 Torr) present. The upper curve shows the emission
3
2
(1). In addition, F atoms, formed both as a minor product in
accompanying irradiation of CF I (19 mTorr), NO (110 mTorr) and
3
2
Ar (4.8 Torr), and the lower curve illustrates the reduction of emission
reaction (1b) (estimated as 1.5%7) and from the O ] CF
3
and the increase in its rising rate when N O is added.
reaction, react rapidly with CF I to regenerate CF . A
2
3
3
FACSIMILE16 Ðt to the data including the reactions listed in
Table 2 was carried out, and produced values of k which
1
showed a less marked trend with total pressure than the
falling rate constants shown in Fig. 6, but which exhibited
non-systematic variations, which were unexpected given the
high quality of the raw data.
We conclude that when rates of formation and quenching
are of similar magnitudes, when secondary chemistry occurs
and when the inÑuence of vibrationally excited CF is
3
neglected, it is difficult to extract meaningful data from the
emission traces. We chose to solve these problems by increas-
ing the ratio of falling and rising rates by adding an efficient
Table 2 Rate constants used in the FACSIMILE Ðts
Rate constant/
cm3 molecule~1 s~1
Process
Ref.
Fig. 8 Emission signals observed through the interference Ðlter with
CF ] NO ] CF Ot ] FNO
0.98 k
3
2
2
1
and without CF I (9 mTorr) for irradiation of NO (104 mTorr), Ar
CF ] NO ] CF Ot ] F ] NO
0.02 k
3
2
3
2
2
1
(5.1 Torr) and N O (1.4 Torr). With CF I absent the emission shows a
CF Ot ] M ] CF O ] M
k
2
3
2
2
2
sharp rise, a slow increase and slower decrease resulting from vibra-
O ] NO ] O ] NO
(9.7 ^ 1.0) ] 10~12
17
18
19
20
21
2
2
tionally excited NO formed from photolysis of NO and relaxing
O ] CF I ] CF ] IO
(3.6 `1.5) ] 10~12
2
3
3
~1.0
through the spectral window of the Ðlter. With CF I present addi-
O ] CF ] CF O ] F
(3.1 ^ 0.8) ] 10~11
3
tional emission from CF O and FNO is seen. The di†erence is shown
2
3
2
F ] CF I ] CF ] IF
(1.2 ^ 0.6) ] 10~10
(3.9 ^ 1.3) ] 10~12
3
3
in the inset together with a kinetic Ðt (virtually indistinguishable from
CF ] CF ] C F
2 6
3
3
the data) as described in the text.
Phys. Chem. Chem. Phys., 2000, 2, 5105È5112
5109

View Article Online
Fig. 9 Rate constants k obtained from the Ðts to data such as in
Fig. 10 Plot of the falling rate as extracted from double exponential
1
Fig. 8 as a function of Ar (…) and N (L). The horizontal straight line
Ðts to data such as shown in Fig. 8 as a function of NO pressure for
2
2
total pressures of Ar or N between 1.5 and 110 Torr. The line now
2
shows the weighted average of all the data, yielding a value of
(1.75 ^ 0.26) ] 10~11 cm3 molecule~1 s~1 for reaction (1). The dotted
line shows a Ðt to the data assuming that the high pressure limit is
reached at 40 Torr, with fall-o† parameters identical with those
extracted from Vahktin.9 As can be seen, the dotted line does not Ðt
the data well, and predicts an unrealistic limiting low pressure rate
constant as explained in the text. The solid curve is a Ðt to the data
including a contribution from a third body dependent association
reaction, as explained in the text.
passes sensibly through zero (in contrast to the data in Fig. 5 taken
without N O present), and the slope has
a
value of
2
(1.52 ^ 0.06) ] 10~11 cm3 molecule~1 s~1 (2p error limits), ca. 14%
lower than the value obtained from the full Ðts. The data, however,
show that there is no marked di†erence in rate constant as a function
of pressure over the range 1.5È110 Torr.
between 37 and 110 Torr there is no signiÐcant change in the
value of k .
1
Temperature dependent measurements of the reaction were
(25È160 mTorr) and total pressures (1.5È37 Torr), k \ (1.75
1
carried out in a single pass cell, and Fig. 11 shows examples of
data taken at 288 and 251 K compared with the traces at
room temperature. Although the lower temperature S/N ratio
is far worse in comparison with the data taken under condi-
tions of multipass irradiation and Welsh collection optics, it
can be seen that the falling rates are very similar in all cases.
Data were taken at two Ar pressures, 5 and 19 Torr, with
again no measurable e†ect of third body concentration.
Analysis of the decays was carried out in the same way as
before, with rate constants in the simulation modiÐed from
their values in Table 2 to account for the lower temperatures.
As N O is necessarily present in the NO sample and its
proportion increases as the temperature is lowered, its pho-
tolysis behaviour needs to be considered. We observe signals
from vibrationally excited NO formed from N O , either by
direct photolysis or by secondary dissociation of the dominant
NO product during the laser pulse.2 At lower temperatures
the e†ect is to increase the NO background signal seen
^ 0.26) ] 10~11 cm3 molecule~1 s~1, where the errors are
2p. The percentage errors of the individual Ðts to the data
such as that shown in Fig. 8 were far lower than the 2p error
limits quoted for the mean value, and we attribute the latter
predominantly to random errors in the measured concentra-
tions of the Ñowing reagents.
Although it is clear that there is no marked variation in k
1
with total pressure or identity of bu†er gas (Ar or N ), the
2
data show hints of a small reduction in value below about 6
Torr. We return to this point later. Extensive checks were
made on the data. First, the e†ect of secondary chemistry
involving F atoms was assessed by adding H to the system so
2
4
2
2
that they were removed dominantly by the F ] H reaction.
2
FACSIMILE Ðtting with and without H produced the same
2
values of k . Laser Ñuences were varied to alter the extent of
1
2
4
radicalÈradical reactions, and again produced no change in
the returned values of k . A sensitivity analysis of the Ðtting
2
1
procedures was carried out by altering values of the rate con-
stants in Table 2 by their error limits, by changing the branch-
ing ratio of eqn. (1b) between 0 and 5% and by altering other
parameters such as absorption cross-sections by 10%. Per-
centage changes in the returned k values were considerably
1
less than the statistical errors quoted above. Variation of gas
residence time between shots showed that the e†ect of reagent
depletion was negligible, and CF I precursor variation
3
between 3 and 20 mTorr produced the same value of k .
1
Fig. 10 shows the falling rate obtained from Ðts of the data
to eqn. (3) with N O present as a function of NO concentra-
2
2
tion, i.e., without the inclusion of secondary chemistry in order
to illustrate the magnitude of the e†ect. The slope of the line
yields a rate constant of 1.5 ] 10~11 cm3 molecule~1 s~1,
14% lower than from the FACSIMILE Ðts because of the
e†ect of neglecting the secondary chemistry which can re-form
CF . The falling rate is independent of Ar or N pressure and,
3
2
in contrast to the data in Fig. 5, now passes sensibly through
zero. Also shown in Fig. 10 are data for N pressures up to
2
110 Torr. At pressures above 37 Torr the reduction in the
overall signal made adequate Ðts from the FACSIMILE mod-
elling difficult because of the rapid signal rise compared with
the associated detector and ampliÐer response. Double expo-
nential rise and fall Ðts, however, were still possible, and the
Fig. 11 Comparison of data taken at three temperatures for the
same reactant concentrations, showing no variation of the time
dependence within experimental error. The clear advantage of data
taken at room temperature with the multipass excitation and collec-
tion optics is evident. Rate constants extracted from lower tem-
perature data such as these are listed in Table 3.
invariance of the falling rate with N pressure up to this value,
2
as shown in Fig. 10, indicates that we can conclude that
5110
Phys. Chem. Chem. Phys., 2000, 2, 5105È5112

View Article Online
through the interference Ðlter, and this can be allowed for by
subtraction in the same way as before. Additional O atoms
are formed which could a†ect the secondary chemistry, but
Ðtting routines in which the extremes of a quantum yield of
unity or zero for O atoms produced from N O yielded no
s~1, there is an indication that the lowest pressure measure-
ments (\6 Torr) cluster below the horizontal straight dashed
line shown in Fig. 9, and a possible explanation is that there is
a small third body e†ect underlying a dominant bimolecular
component. We Ðrst note that if this is so, any extrapolation
from our present data to reach the value of 3.2 ] 10~11 cm3
molecule~1 s~1 at 300 Torr as reported by Vakhtin9 is impos-
sible. Vakhtin measured the branching ratio (1d)/(1a) by end
product analysis of CF O and CF NO as 30È35% at three
2
4
di†erence in the returned values of k within experimental
1
error as shown in Table 3. The rate constants at four di†erent
temperatures between 215 and 295 K are shown in Table 3.
Although the error bars at the lower temperatures are con-
siderably larger than at room temperature (reÑecting the
poorer quality of the low temperature data as shown in Fig.
11), there is seen to be no noticeable change in the rate con-
stant in this pressure range 295È251 K.
2
3
2
pressures, 150, 300 and 600 Torr He, dropping slightly to 25%
at 4 Torr, and from these data he modelled a fall-o† curve. If
we model a similar branching ratio, but constrain the high
pressure limiting rate constant to give a value of k \ 1.75
1
] 10~11 cm3 molecule~1 s~1, we produce the curve shown as
a dotted line in Fig. 9. As can be seen, this is an unsatisfactory
Ðt to the data, and it also results in a low pressure limiting
rate constant for the association reaction (1d) as 4.3 ] 10~28
cm6 molecule~2 s~1, which is over an order of magnitude
Discussion
We Ðrst comment on the values of the rate constant in com-
parison with other studies shown in Table 1, and consider Ðrst
the previous measurements from time resolved IR emis-
sion.5,7,8 Previous data from our group were taken at rela-
tively low Ar pressures (5 Torr) and we identiÐed the rising
rate as corresponding to reaction and the falling rate to quen-
larger than other recombination reactions involving CF with
3
NO23 and O 2,17 and CF O with NO .17
2
3
2
2
We now examine the evidence for the formation of a stable
addition product in other studies. Substantial (ca. 30%) for-
mation of CF NO has been seen in some end product
analyses9,11 whereas in others only small (\11%8) or
negligible5 amounts were observed. Real time observation of
the species would remove the uncertainty of wall reactions
which can a†ect such analyses. In the chemical ionisation
measurements of Bevilacqua et al.,6 evidence for formation
CF NO (or CF ONO) is described, but not quantiÐed, with
formation of CF O simply stated as being predominant. Most
recently, however, Pagsberg et al.10 measured the yield of
CF O from reaction (1) as 95% over the pressure range 3È17
Torr by observing both CF loss and CF O formation quanti-
tatively by IR absorption, but were unable to observe
CF NO formation with the same technique. We shall now
ching because of the e†ect of adding up to 2.5 Torr of N or
2
3
2
O . The rising rate was unchanged with N , but increased
2
2
with O , as expected for reactive removal, and the falling rate
2
increased with both gases, as expected for quenching.
Although the rising and falling rates extracted from the Ðts of
eqn. (3) appeared to be well behaved, it is also clear that under
these conditions (5 Torr Ar) they were sufficiently close to one
another to render the data treatment unreliable, particularly
given the lower S/N ratio of the earlier experiments which did
not have the advantage of Welsh cell collection optics.
Although our present data agree with the previous work
within the error bars, the agreement is probably fortuitous. In
the two other reported IR emission experiments the S/N
ratios were considerably lower than those in the present
experiments, and the problems of data analysis alluded to
above were not addressed. The rate constants obtained in
these seemingly identical studies, (2.5 ^ 0.4) ] 10~11 (ref. 5)
and (1.6 ^ 0.3) ] 10~11 cm3 molecule~1 s~1 (ref. 8), are suffi-
ciently di†erent to indicate problems with such data analysis,
and the agreement of one of them8 with the present result is
again probably fortuitous.
3
2
3
2
2
3
2
3
2
examine whether our data can be Ðtted with any conÐdence to
a model which incorporates both bimolecular and third body
processes, and we Ðrst address the problem with reactions for-
mulated in eqn. (1), namely with two separate and indepen-
dent channels leading to bimolecular [reactions (1aÈc)] and
recombination [reaction (1d)] products. To do this we would
have several adjustable parameters, namely the true bimolecu-
lar component of reaction (1), incorporating processes (1aÈc),
and the fall-o† behaviour of process (1d), so that many poten-
tial Ðts would be possible. As we can conclude from Fig. 9
that bimolecular kinetics dominate, we simplify the Ðtting by
Ðxing the value for the low pressure limiting rate constant for
reaction (1d). We choose a low pressure value which is identi-
cal with that for the CF ] O recombination reaction,
Rates of removal of reactants should in principle provide
unambiguous kinetic data, and as Table 1 shows our results
agree well with the most recent determinations from loss of
CF 6,10 and NO ,8 but are outside the error limits of an
3
2
earlier CF loss measurement.4 Our results, however, are
3
almost a factor of two lower than the higher pressure (300 and
3
2
600 Torr He) measurements of Vakhtin,9 which were carried
out by relative rate constant measurements of process (1) com-
pared with the recombination of CF with NO, with the rate
namely (2.5 ^ 0.3) ] 10~29 cm6 molecule~2 s~1.2 This value
is similar to those reported for CF recombination with NO23
3
and CF O with NO .17 We use the Troe expression for the
3
3
2
constant for the latter process taken from the literature.22 As
fall-o† behaviour,24 with the broadening factor F \ 0.6 as
c
these high pressure rate constants are considerably faster than
those in the present study and as this study9 was one of
two9,11 in which CF NO was detected as a signiÐcant end
found for the CF ] O reaction.2 Fig. 9 shows the resultant
3
2
best Ðt to the data with two adjustable parameters, namely the
bimolecular value of k [the sum of processes (1aÈc)] and the
3
2
1
product, we need to consider carefully the total pressure
dependence of the reaction.
high pressure limiting value of process (1d). At low pressure k
1
now becomes (1.61 ^ 0.04) ] 10~11 cm3 molecule~1 s~1, with
Although the present data show that between 1.5 and 110
Torr the value of k is (1.75 ^ 0.26) ] 10~11 cm3 molecule~1
the limiting high pressure value of k \ (0.21 ^ 0.11) ] 10~11
1d
cm3 molecule~1 s~1. Our low and high pressure values of k
1
1
Table 3 Rate constant determination at low temperatures
Temperature/K
No. determinations
k /10~11 cm3 molecule~1 s~1a
1
288
251
256
6
6
12
(a) 1.75 ^ 0.34
(a) 2.00 ^ 1.0
(a) 2.00 ^ 1.0
(b) 1.74 ^ 0.34
(b) 1.80 ^ 0.90
(b) 1.80 ^ 0.60
a The data marked as (a) and (b) are for quantum yields of one and zero O atoms produced in the photolysis of N O . No di†erence is seen in
2
4
the returned values of k within experimental error.
1
Phys. Chem. Chem. Phys., 2000, 2, 5105È5112
5111

View Article Online
are thus (1.61 ^ 0.04) ] 10~11 and (1.82 ^ 0.15) ] 10~11 cm3
molecule~1 s~1, and can be seen to lie well within the error
limits of the average value quoted above.
295 K, and we recommend a value of (1.75 ^ 0.26) ] 10~11
cm3 molecule~1 s~1 as encompassing all pressure conditions
and the temperature range measured. There is evidence of fall-
o† behaviour at the lowest pressures studied, and at room
temperature if this is considered to indicate a separate reac-
tion channel, we recommend a high pressure limiting branch-
ing ratio of the CF NO (or CF ONO) product as 13 ^ 7%,
A second and intuitively more satisfactory route to the pro-
duction of a stabilised adduct would be the formation of an
energised complex which either decomposes (possibly after
rearrangement) or is collisionally stabilised, i.e., both
bimolecular and third body processes originate from a
common precursor rather than separate steps indicated by
reaction (1). If formation of the adduct is the rate determining
step then the rate constant will show no pressure dependence,
but the yield of stabilised product will rise with pressure, and
hence the emission yield of process (1a) will fall. We measured
emission intensities between 3 and 25 Torr Ar, corrected these
for the measured quenching rates and found that the resultant
emission yields showed no variation within experimental error
(^10%). This indicates that over this pressure range the for-
mation of a stabilised adduct in competition with decomposi-
tion to form products is of minor importance. The data
therefore seem to favour separation of bimolecular and ter-
molecular processes, but it should be emphasised that there is
a need for direct observation of stabilised products at higher
pressures under conditions such that heterogeneous formation
can be eliminated.
The present data provide no further evidence of channels
other than process (1a), but taken with our previous estimates
of a yield of 1.5% of channel (1b),7 and the most recent time
resolved observations of a branching ratio of NO from the
sum of processes (1b) and (1c) to be only 1%,10 we conclude
that reaction (1a) is the dominant bimolecular step, and we
concur with the estimate of Pagsberg et al.10 of a bimolecular
branching ratio of 95 ^ 4% at low pressures (\17 Torr). The
Ðtting of the data in Fig. 9 yields a limiting high pressure yield
of 13 ^ 7% for process (1d) if it is treated as a separate inde-
pendent step, the relative errors being large because of its
small contribution, and we note that it has reached 50% of its
limiting value at a pressure of 10 Torr with a Ðtted value of
1.72 ] 10~11 cm3 molecule~1 s~1. Although parameterisation
of the contributions from processes (1aÈc) and (1d) will yield
the various values of the rate constants and quantum yields,
we present a simple result which Ðts in with our error bars.
3
2
3
dropping to half this value at 10 Torr. Bimolecular reaction is
dominant, and the evidence points clearly to CF O and FNO
2
being the overwhelmingly dominant bimolecular component.
Our results highlight the difficulties in drawing kinetic conclu-
sions from time resolved product observations when rates of
formation and removal of an observed species are of the same
magnitude.
Acknowledgements
We are grateful to the EPSRC, the Royal Society and to the
National Environmental Development Organisation, Japan,
for support of this work.
References
1
T. J. Wallington, D. R. Worsnop, O. J. Nielsen, J. Sehested, W.
DeBruyn and J. A. Shorter, Environ. Sci. T echnol., 1992, 26, 1318;
O. J. Nielsen, T. Ellermann, J. Sehested, E. Bartkiewicz, T. J.
Wallington and T. J. Hurley, Int. J. Chem. Kinet., 1992, 24, 1009;
E. O. Edney and D. J. Driscoll, Int. J. Chem. Kinet., 1992, 24,
1067; E. C. Tuazon and R. Atkinson, J. Atmos. Chem., 1993, 16,
301; J. Sehested, T. Ellermann, O. J. Nielsen, T. J. Wallington
and T. J. Hurley, Int. J. Chem. Kinet., 1993, 25, 701.
C. Breheny, G. Hancock and C. Morrell, Z. Phys. Chem., in the
press.
2
3
4
M. J. Rossi, J. R. Barker and D. M. Golden, J. Phys. Chem., 1979,
71, 3722.
K. Sugawara, T. Nakanaga, H. Takeo and C. Matsumura, J.
Phys. Chem., 1989, 93, 1894.
5
6
J. C. Francisco and Z. Li, Chem. Phys. L ett., 1989, 162, 528.
T. J. Bevilacqua, D. R. Hanson and C. J. Howard, J. Phys. Chem.,
1993, 97, 3750.
7
8
K. W. Oum and G. Hancock, J. Phys. Chem., 1997, 101, 2634.
J. Sehested, O. J. Nielsen, C. A. Rinaldi, S. I. Lane and J. C.
Ferrero, Int. J. Chem. Kinet., 1996, 28, 579.
9
A. B. Vakhtin, Int. J. Chem. Kinet., 1997, 29, 203.
10 P. Pagsberg, J. T. Jodkowski, E. Ratajczak and A. Sillesen, Chem.
We recommend a value of k which applies over all pressures
1
Phys. L ett., 1998, 286, 138.
of (1.75 ^ 0.26) ] 10~11 cm3 molecule~1 s~1 (2p error limits),
11 N. OÏSullivan, J. Treacy and H. Sidebottom, paper presented at
the Joint CECÈEUROTRAC Workshop, HALIPP/LACTOZ
Working Group, Strasbourg, 1995.
12 P. Biggs, G. Hancock, D. E. Heard and R. P. Wayne, Meas. Sci.
T echnol., 1990, 1, 630.
13 P. Felder, Chem. Phys., 1990, 143, 141.
14 G. N. A. van Veen, T. Baller and A. E. de Vries, Chem. Phys.,
1985, 93, 277; M. A. Young and G. C. Pimentel, J. Phys. Chem.,
1990, 94, 4884.
15 J. McFarlane, J. C. Polanyi and J. G. Shapter, J. Photochem.
Photobiol., A, 1991, 58, 139.
16 FACSIMILE, Process and Chemical Reaction Modeller, AEA
Technology, Harwell, 1998.
17 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 No. 12, JPL Publication, No.
97-4, NASA, Pasadena, CA, 1997.
18 M. K. Gilles, A. A. Turnipseed, R. K. Talukdar, Y. Rudich, P. W.
Villalta, L. G. Huey, J. B. Burkholder and A. R. Ravishankara, J.
Phys. Chem., 1996, 100, 14005.
and recommend that if CF NO (CF ONO) is formed in a
3
2
3
separate third body association step, it has a limiting yield of
13 ^ 7% which drops to 6 ^ 3% at 10 Torr. If this is so, our
results suggests that previous end product observations of a
30% quantum yield for process (1d) at pressures above 300
Torr9,11 are overestimates, but we are in good agreement with
the more recent quantum yield data of Pagsberg et al.10 which
show the dominance of process (1a).
Finally, we consider our observed lack of a temperature
dependence. The bimolecular reaction is rapid and only small
temperature e†ects would be expected, but a decrease in tem-
perature would be expected to increase the recombination rate
constant in the fall-o† regime. Again we make a comparison
with the CF ] O recombination reaction, where the mea-
3
2
sured temperature dependence in the pressure range near 6
Torr would indicate a 60% increase in the rate constant when
the temperature decreases from 295 to 256 K.25 For a contri-
bution of \10% for process (1d) at room temperature and 6
Torr, this increase would again not be noticeable within the
large error bars associated with the lower temperature mea-
surements.
19 K. R. Ryan and I. C. Plumb, J. Phys. Chem., 1982, 86, 4678.
20 J. W. Bozzelli and M. Kaufmann, J. Phys. Chem., 1973, 77, 1748.
21 A. B. Vakhtin, Int. J. Chem. Kinet., 1996, 28, 443.
22 A. B. Vakhtin and A. K. Petrov, Chem. Phys., 1991, 149, 427; J.
Masanet, F. Caralp, L. Ley and R. Lesclaux, Chem. Phys., 1992,
160, 383.
Conclusions
23 E. W. Kaiser, T. J. Wallington and M. D. Hurley, Int. J. Chem.
Kinet., 1995, 27, 205.
The reaction between CF and NO shows no pressure or
24 J. Troe, J. Phys. Chem., 1979, 83, 114.
3
2
temperature dependence within the random errors of our
measurements between 1.5 and 110 Torr and between 251 and
25 F. Caralp, R. Lesclaux and A. M. Dognon, Chem. Phys. L ett.,
1986, 129, 433.
5112
Phys. Chem. Chem. Phys., 2000, 2, 5105È5112