Elementary reaction of CF2(X) with O3: Primary products

Article abstract of DOI:10.1524/zpch.2000.214.1.095

The reaction of CF₂(X) with O₃ has been investigated in a discharge flow reactor at room temperature and low pressure (p = 2.5 mbar). CF₂(X) was produced by a microwave discharge of a mixture of CF₂Br₂/He. The reactants and products were detected by mass spectrometers (MS), which were connected to the flow systems via a continuous molecular beam sampling system. The following primary products were observed (CF₂O₃ (m/z = 98); CF₂O (m/z = 47, 66); O₂ (m/z = 32)). The reaction mechanism: CF₂(X) + O₃ ? CF₂O₃^# → CF₂O₃ → CF₂O+O₂ (1) explains the observed primary products. The molecule with the empirical formula CF₂O₃ has been observed directly by MS; the structure is discussed. by Oldenbourg Wissenschaftsverlag, Muenchen.

Full text of DOI:10.1524/zpch.2000.214.1.095

Zeitschrift für Physikalische Chemie, Bd. 214, H. 1, S. 95Ϫ112 (2000)
by Oldenbourg Wissenschaftsverlag, München
˜
Elementary Reaction of CF2(X) with O3 :
Primary Products
By Th. Beiderhase1, K. Hoyermann1 and W. Hack2
1
Institut für Physikalische Chemie, Tammannstraße 6, D-37077 Göttingen
Max-Planck-Institut für Strömungsforschung, Bunsenstraße 10, D-37073 Göttingen
2
(Received January 6, 1999)
˜
Elementary reactions / CF2(X) / O3 / Mass spectrometry
˜
The reaction of CF2(X) with O3 has been investigated in a discharge flow reactor at room
˜
temperature and low pressure (p ϭ 2.5 mbar). CF2(X) was produced by a microwave
discharge of a mixture of CF2Br2/He. The reactants and products were detected by mass
spectrometers (MS), which were connected to the flow systems via a continuous molecu-
lar beam sampling system.
The following primary products were observed (CF2O3 (m/z ϭ 98); CF2O (m/z ϭ
47, 66); O2 (m/z ϭ 32)). The reaction mechanism:
ն
˜
CF2(X) ϩ O3 { CF2O3 → CF2O3
→ CF2OϩO2
(1)
explains the observed primary products. The molecule with the empirical formula CF2O3
has been observed directly by MS; the structure is discussed.
1. Introduction
1
˜
Difluorcarbene (CF2(X A1)) is a radical, albeit not a very reactive one, and
therefore it may play an important role as a fluorine-containing intermediate
in the degradation of fluorinated compounds in the upper atmosphere and
˜
stratosphere. Also in plasma etching systems CF2(X) can be regarded as a
fluorine-containing reservoir.
In the plasma etching processes of Si and SiO2 surfaces, used in the
fabrication of micron-size features in semiconductor devices, CF4 dis-
charges are applied. In models to understand these processes, CFx radicals
and F atoms are involved [1Ϫ3]. The addition of a small amount of O2 to
the CF4 plasma is known to increase significantly the etching rate. The main
etching species are assumed to be F atoms, the concentration of which is
increased by O atom reactions [4]. Stable O3 molecules, which can be added
Brought to you by | New York University Bobst Library Technical Services
Authenticated
Download Date | 7/31/15 9:33 AM

96
Th. Beiderhase, K. Hoyermann and W. Hack
to the system in large quantities compared to the concentration at labile O
atoms available, might also cause an increase in the F atom concentration
˜
in the C-F-O system. One possible pathway is the reaction of CF2(X) with
ozone.
˜
CF2(X) is formed in the atmospheric chemistry by photolysis of CFCls
such as CF2Cl2, which accumulate in the earth’s atmosphere and strato-
˜
sphere [5Ϫ8]. Since the reactivity of CF2(X) in its electronic ground state
with atmospheric gases like O2, N2, CO2 is very low, the concentration of
˜
CF2(X) could reach higher values than that of other reactive intermediates.
˜
The reaction of O3 with CF2(X)
1
˜
˜
CF2(X) ϩ O3(X A1) → products
(1)
is therefore of great interest. Under atmospheric conditions, reaction (1) is
not very fast. At room temperature approximately one million CF2-O3 colli-
sions are needed before reaction occurs with an activation energy of 25 kJ/
mol [9]. The products of the gas phase reaction (1) have not previously
been measured directly. In an argon matrix a CF2-O3 complex was observed
by IR-spectroscopy [10, 11].
˜
The oxidation reaction of CF2(X) with NO2, which leads to CF2O and
NO (1), has been studied directly [12]. The oxidation reaction with O(3P)
atoms
3
˜
CF2(X) ϩ O( P) → FCO ϩ F
(5)
which is a fast process [13] leads to the formation of FCO radicals and F
atoms [14, 15].
The aim of this work was to measure directly the products of the reac-
tion (1). As the reaction is very slow under experimental conditions, only
very low product concentrations are to be expected. The ideal method for
the sensitive detection of all products is a flow reactor combined with a
mass spectrometer.
2. Experimental
The experimental arrangements, as described in detail elsewhere [12, 16],
consisted of an isothermal discharge flow reactor at low pressure and a
molecular beam sampling device connecting the flow reactor to a magnetic
deflection type mass spectrometer with electron impact ionisation. This ex-
perimental arrangement is shown in Fig. 1. The second detection device
used was a time of flight mass spectrometer which had a choice of electron
impact ionisation or photoionisation.
The experimental study of the primary products of the reaction (1), CF2
ϩ O3, is complicated by the fact that reaction (1) is very slow (i.e. low
product concentrations) compared to possible other reactions like secondary
reactions and heterogeneous reactions of O3, products, and impurities. The
Brought to you by | New York University Bobst Library Technical Services
Authenticated
Download Date | 7/31/15 9:33 AM

˜
Elementary Reaction of CF2(X) with O3 : Primary Products
97
Fig. 1. Experimental arrangement: Flow reactor connected with a magnetic deflection
type mass spectrometer via a molecular beam sampling system; by choice: analog ion
current or synchronous single ion counting.
aging of the reactor and the accumulation/desorption of products at/from
the reactor wall were investigated using the time of flight mass spec-
trometer. Time resolved mass spectra (m/z ϭ 2Ϫ300) are obtained to assess
the temporal behaviour of the reactor. The low signal-to-noise ratio of the
spectra had to be accepted for the sake of good time resolution. No interfer-
ing accumulation of products was observed between the normal cleaning
intervals. Therefore quantitative measurements were performed by coupling
the flow reactor with the magnetic deflection mass spectrometer, along with
analogous amplification or digital single ion counting of the signal. This
arrangement brought about a substantial enhancement of the signal-to-noise
ratio and the extension to the low energy impact ionisation near threshold.
The CF2 radicals were produced by the dissociation of CF2Br2 by a
microwave discharge in the side arm of a fast flow reactor, operating at a
˜
pressure in the range 0.5 Յ p/mbar Յ 10. The CF2(X) yield from the dis-
sociation of CF2Br2 depends on the concentration of CF2Br2 as shown in
Fig. 2. The dissociation (followed at m/z ϭ 129 and 131, respectively) is
Ϫ3
complete (Ն 99%) if a mole fraction of x(CF2Br2) ϭ 2.6 · 10 was used in
the experiments (see Fig. 2).
Brought to you by | New York University Bobst Library Technical Services
Authenticated
Download Date | 7/31/15 9:33 AM

98
Th. Beiderhase, K. Hoyermann and W. Hack
˜
Fig. 2. CF2(X) yield versus mol fraction of CF2Br2 in discharge: γ ϭ {[CF2Br2]ϪEϪ
[CF2Br2]ϩE}/[CF2Br2]ϪE [CF2Br2]ϮE ϭ concentration of CF2Br2 with discharge on and off,
ϩ
followed as CF2Br on m/z ϭ 129 and 131 respectively; IE ϭ 100 eV; p ϭ 2.5 mbar.
v¯ ϭ 26.6 m/s.
˜
The CF2(X) source is characterized by the time of flight mass spectrum
taken at an ionisation energy of IE ϭ 100 eV in the mass range 60 Յ m/z
Յ 206 as shown in Fig. 3. The depletion of CF2Br2, which is detected as
ϩ
the CF2Br ion (m/z ϭ 129, 131), is obvious. The increase at m/z ϭ 158,
ϩ
ϩ
ϩ
160 and 162 is due to the formation of Br2(79,79Br2 , 79,81Br2 and 81,81Br2 ).
ϩ
The formation of m/z ϭ 79, 81 (Br ) is independent of the discharge situa-
tion, i.e. the fragmentation of CF2Br2 or of Br2 forming Br ions is equal at
ϩ
the high ionisation energy. The peak at m/z ϭ 69 is due to the CF3 radical
from the combination reaction F ϩ CF2 –M→ CF2 (2). The formation of F
atoms in the CF2Br2 discharge was found and characterized in [17]. The F
atom concentration is of minor importance (0.01 Յ [F]/[CF2] Յ 0.001). At
the masses m/z ϭ 179 and 181 only marginal increases were found when
the discharge was switched on and off, respectively. From the absence of
Brought to you by | New York University Bobst Library Technical Services
Authenticated
Download Date | 7/31/15 9:33 AM

˜
Elementary Reaction of CF2(X) with O3 : Primary Products
99
Fig. 3. Time of flight mass spectra in the mass range 28 Յ m/z Յ 131; IE ϭ 100 eV, O3
¯
absent. v (flow reactor) ϭ 20.3 m/s, p ϭ 1.7 mbar, ∆t ϭ 8.6 ms (transfer time from
discharge to the tip of the expansion nozzle in the flow reactor). a) discharge “on”; b)
discharge “off”.
ϩ
C2F4Br it can be concluded that CF2Br radicals, which would combine to
form C2F4Br2 in a fast reaction (k ϭ 1.0 · 1013 cm3/mol s [18], were not
ϩ
present in the discharge system, since C2F4Br (m/z ϭ 179, 181) is the
ϩ
main fragment ion of C2F4Br2. No signal was found at m/z ϭ 100 (C2F4 )
indicating that the combination reaction CF2 ϩ CF2 –M→ C2F4 (3) is too slow
under the experimental conditions given in Fig. 3 (k3 ϭ 2.6 · 1010 cm3/mol s
[19].
The reactant O3 was injected into the main flow by a movable inlet
probe. The different positions of the probe correspond to different reaction
times. The O3 concentration was obtained by the titration reaction:
O3 ϩ F → OF ϩ O2
with a known F atom concentration.
(6)
The samples were withdrawn continuously by a molecular beam sam-
pling device. The ions are produced by electron impact with variable elec-
tron impact ionisation energies (10Ϫ30 eV, 70 eV, 100 eV) and analysed by
the mass spectrometer with high resolution. High ionisation energies were
applied to obtain high sensitivity whereas at low ionisation energies ion
fragmentation was reduced or suppressed. Sensitive detection was ac-
complished by synchronous single ion counting combined with chemical
Brought to you by | New York University Bobst Library Technical Services
Authenticated
Download Date | 7/31/15 9:33 AM

100
Th. Beiderhase, K. Hoyermann and W. Hack
modulation of the reaction by switching the microwave discharge on/off
and gating the two-channel counter to the conditions “CF2 ϩ O3’’/“·CF2Br2
ϩ O3’’ in the reactor.
Gases of the highest commercially available purity were used without
further purification (He 99.996%, Messer Griesheim, CF2Br2 (Ն98%)
Merck-Schuchardt). Ozone was produced from O2 (Ն99.996% UCAR) with
a commercially available ozonizer (Argentox GI 10). O3 was adsorbed on
silicagel at the temperature of CO2(s) in two traps and thus separated from
the oxygen. The ozone was desorbed from the silicagel and transfered by
He into the reactor.
3. Results
The mass spectra of the products of the reaction:
˜
˜
CF2(X) ϩ O3(X) → products
(1)
were taken at high (70 eV) and low ionisation energy. Three overview spec-
tra in the mass ranges 15 Յ m/z Յ 50, 40 Յ m/z Յ 50 and 60 Յ m/z Յ
100 are shown in Figs. 4, 5, and 6. The interpretation of the mass spectra
was performed with the data summarized in Table 1.
˜
The depletion of CF2(X) due to reaction (1) is seen in Fig. 4 (m/z ϭ 50,
ϩ
CF2 ). The increase of the peaks m/z ϭ 44 and 47 cannot be assigned to the
products CO2 and COF since they are fragment ion peaks (see below). The
increase at m/z ϭ 48 is simply caused by the added O3.
From the depletion of CF2 observed at m/z ϭ 50 as a function of reac-
tion time a rate constant k1 could be obtained from the simple expression:
ln{[CF2](tR)/[CF2]0} ϭ k [O3] tR ,
(I)
if the CF2 depletion in the system was only due to reaction (1). The sensi-
tivity of the CF2 detection, however, is not sufficient to use such small
initial CF2 concentrations that parallel reactions are suppressed:
CF2 ϩ Br2 → products
(7)
and
CF2 ϩ CF2 –M→ C2F4
(3)
For the determination of k1 the reactions (7) and (3) have to be taken into
account (k7 ϭ 1.6 · 109 cm3/mol s [27] k3 ϭ 2.6 · 1010 cm3/mol s [19]). The
rate for the removal at the wall:
CF2 –w–a→ll products
(8)
Ϫ1
was determined separately to be kW ϭ 10 s under the experimental con-
ditions. From the CF2 depletion in the flow reactor in the presence of O3 an
upper limit for k1 was obtained, which is higher than that determined di-
Brought to you by | New York University Bobst Library Technical Services
Authenticated
Download Date | 7/31/15 9:33 AM

˜
Elementary Reaction of CF2(X) with O3 : Primary Products
101
Fig. 4. Mass spectrum for the reaction CF2 ϩ O3 → product; IE ϭ 70 eV, [CF2]0/[O3]0 Х
1/2O. a) in the absence of O3; b) in the presence of O3.
rectly at very low CF2 concentrations; where reactions (3) and (7) can be
neglected [9]. As the derived value in (a) is an upper limit only, it does not
contradict the results given in (a). A computer simulation of the full mech-
Brought to you by | New York University Bobst Library Technical Services
Authenticated
Download Date | 7/31/15 9:33 AM

102
Th. Beiderhase, K. Hoyermann and W. Hack
Fig. 5. Mass spectrum for the reaction CF2 ϩ O3 → product; IE ϭ 17 eV. a) discharge
“on”; b) discharge “off”.
anism resulted the total depletion of CF2 obtained under our experimental
conditions due to reaction (1). Expressed in molefraction this reads as:
Ϫ5
∆ x (CF2) Х 4 ·10 ).
The mass spectrum at low masses in the range 16 Յ m/z Յ 50 and at
an ionisation energy of IE ϭ 17 eV is shown in Fig. 5. The peak m/z ϭ 48
ϩ
(O3) decreased due to reaction (1). The formation of O2 (m/z ϭ 32, O2 )
ϩ
was observed. The mass peak m/z ϭ 32 (O2 ) is a fragment peak of O3, too,
and thus should decrease by reaction (1); the observed increase indicates,
however, that O2 is a reaction product. The discrimination between the two
electronic states of the oxygen molecule O2 (O2 (X3Σg ) and O2 (a1∆g)] was
Ϫ
ϩ
not obtained (see Discussion). The mass peak m/z ϭ 16 (O ) is a fragment
of O3 as well as of O2 and the signal height was found to be unchanged. The
signals with m/z ϭ 32 and 50 increased when the discharge was switched on
and the signals at m/z ϭ 48 decreased. The peaks m/z ϭ 48 and 50 are due
ϩ
to the reactants O3 and CF2, respectively; CF2 was observed only if the
discharge was on and therefore is not a fragment ion from CF2Br2.
The product spectrum for the mass range 60 Յ m/z Յ 102 taken with
the magnetic deflection type mass spectrometer at IE ϭ 17 eV is shown in
Brought to you by | New York University Bobst Library Technical Services
Authenticated
Download Date | 7/31/15 9:33 AM

˜
Elementary Reaction of CF2(X) with O3 : Primary Products
103
Fig. 6. Reaction product spectrum of reaction (1); IE ϭ 17 eV. a) absence of O3, b) pres-
Ϫ11
Ϫ10
ence of O3. [CF2]0 ϭ 1.4 · 10 mol/cm3, [O3] ϭ 9 · 10 mol/cm3; v ϭ 26.6 m/s; t ϭ
¯
R
5.9 ms, p ϭ 2.5 mbar.
Fig. 6. For the registration of the background mass spectrum, the O3 flow
was substituted by an equivalent He flow in order to keep the flow con-
ditions unchanged.
Signal increases were observed for the peaks at m/z ϭ 95, 96, 97 and
ϩ
98. The mass peaks m/z ϭ 95 and 97 can be attributed to the ion BrO .
The ratio of the peak heights is 1:1 due to the 79Br and 81Br isotope abun-
dance ratio. The increases at m/z ϭ 96 and 98 are due to HOBr (79Br and
81Br). The increase of the signal of the peak m/z ϭ 98 was larger than
the increase of m/z ϭ 96 even after subtraction of the mass spectrometric
background at m/z ϭ 98. Thus it is concluded that the mass peak m/z ϭ
ϩ
ϩ
98 is due not only to HO81Br but also the ion CF2O3 from the initial
complex CF2O3 formed in reaction (1):
˜
˜
CF2(X) ϩ O3(X) ϩ M → CF2O3 ϩ M.
(1)
The decrease of the peaks with m/z ϭ 79 and 81 belonging to Br atoms
and the observation of BrO can be interpreted by the reaction of Br atoms
ϩ
with O3 forming BrO:
Br ϩ O3 → BrO ϩ O2 .
(9)
ϩ
The peak at m/z ϭ 100, which results from C2F4 , is more or less unchanged.
Brought to you by | New York University Bobst Library Technical Services
Authenticated
Download Date | 7/31/15 9:33 AM

104
Th. Beiderhase, K. Hoyermann and W. Hack
Table 1. Mass spectra, electron impact ionisation around 70 eV, if not stated otherwise
and thermodynamic data of some compounds. I ¥ intensities relative to the main peak
(I ϭ 1).
CF2Br2; M ϭ 208/210/212, H0f, 298 ϭ Ϫ380.7 kJ/mol [20]
m/z
I
129 131
79
31
81
50
191
12
[21]
[21]
[21]
[21]
1.0 0.973 0.261 0.258 0.251 0.188 0.67
0.59
(CF2Br)2; M ϭ 258/260/262, H0f, 298 ϭ Ϫ790.8 kJ/mol [20]
m/z
I
179 181
129
131
100
31
260
50
1.00 0.974 0.340 0.331 0.170 0.134 0.124 0.75
F2CO; M ϭ 66, H0f, 298 ϭ Ϫ607.9 kJ/mol [22]
m/z
I
47
66
28
31
50
0.24
16
0.10
19
0.09
48
0.09
1.00 0.548 0.136 0.44
C2F4; M ϭ 100, H0f,298 ϭ Ϫ661.1 kJ/mol [20]
m/z
I
31
81
100
50
12
69
0.36
24
0.24
19
0.14
1.00 0.731 0.431 0.308 0.77
O3 ; M ϭ 48, H0f,298 ϭ ϩ132.3 kJ/mol [20]
m/z
I
16
48
32
1.00 0.53 0.37 (17 eV)
(this study)
(this study)
m/z
I
48 32
1.00 0.69
(14 eV)
FCO; M ϭ 47, H0f;298 ϭ Ϫ152.1 kJ/mol [22]
0
˜
CF2(X); M ϭ 50, Hf;298 ϭ Ϫ205.0 kJ/mol [23, 24, 25]
m/z
I
31
50
19
[23, 24, 25]
1.00 0.87 0.59 (70 eV)
m/z
I
50
31
(17 eV)
1.00 0.31
m/z
I
31
50
(70 eV)
(approximate, this study)
1.00 0.7
0
˜
CF3(X); M ϭ 69, Hf,298 ϭ Ϫ460.2 kJ/mol [24, 25]
m/z
I
m/z
I
50
31
69
19
1.00 0.89 0.50 0.46 (70 eV)
69
1.00
[23, 24, 25]
50
Ϫ
31
Ϫ
19
Ϫ
(17 eV)
FO2;
M ϭ 51,
H0f,298 ϭ ϩ12.6 kJ/mol
[20]
CF2Br; M ϭ 129/131, H0f,298 ϭ Ϫ220 kJ/mol
(estimated)
[20]
BrO;
M ϭ 95/97,
M ϭ 96/89,
H0f,298 ϭ ϩ125.9 kJ/mol
H0f,298 ϭ Ϫ37.6 kJ/mol
HOBr
[26]
Brought to you by | New York University Bobst Library Technical Services
Authenticated
Download Date | 7/31/15 9:33 AM

˜
Elementary Reaction of CF2(X) with O3 : Primary Products
105
This indicates that at the high CF2 concentrations, which are needed in order
to observe a reaction effect, the combination reaction (3) contributes to the
CF2 radical consumption.
The mass peaks m/z ϭ 80 and 82 are due to HBr, which is formed in the
heterogeneous reaction of Br atoms from the CF2Br2 discharge and traces of
adsorbed water.
The mass spectral range around m/z ϭ 66 was examined in detail as
inserted in Fig. 6. A small increase was found at m/z ϭ 66 (the mass peak
ϩ
of CF2O ) at low ionisation energy (17 eV). This indicates that the CF2O3
complex that was formed initially decomposes via:
CF2Oն3 → CF2O ϩ O2
leading to the formation of CF2O (m/z ϭ 66, CF2O ) and O2 (m/z ϭ 32,
(1b)
ϩ
ϩ
ϩ
O2 ) as observed in the spectrum (Fig. 5). The signal at m/z ϭ 69 (CF3 ) did
not change by the addition of O3. Thus the CF3 reaction with O3 was un-
important in the system studied.
The assignment of the primary products of reaction (1) (CF2O3, CF2O,
O2) was deduced from the overview mass spectra recorded as the analogous
ion currents with the inherent limits of mass spectrometric detectability. In
order to enhance the confidence level of the ion current measurements and
to further unravel the fragment ion formation of the products CF2O3, CF2O,
O2 and of byproducts use has been made of the synchronous single ion
counting technique together with a systematic variation of the ionisation
energy (10 Յ IE/eV Յ 70 eV).
The results obtained at IE ϭ 17 eV are shown in Fig. 7. The peaks were
measured by single ion counting and corrected for the mass spectrometric
background. In this spectrum the formation of m/z ϭ 32 (O2) was observed.
ϩ
The formation of CF2O was detected (m/z ϭ 66, CF2O ) directly by the
ϩ
parent ion and by the fragment m/z ϭ 47 (FCO ). The formation of the
ϩ
initial complex CF2O3 was detected directly by its CF2O3 ion (m/z ϭ 98).
ϩ
The contribution of HO 81Br isotope to the peak at m/z ϭ 98 (CF2O3 ϩ
ϩ
HO 81Br ) can be subtracted via the signal of HO 79Br isotope (m/z ϭ 96),
taking into account the known natural isotope abundance (79Br/81Br ϭ
1.028). The formation of BrO (m/z ϭ 95, 97) and the heterogeneous forma-
tion of HBr (m/z ϭ 80, 82), as well as the consumption of Br/Br2 (m/z ϭ
ϩ
79, 81; Br ) do not influence the deduction of the mechanism of reaction
(1). In this more sensitive experiment a consumption of CF3 (m/z ϭ 69,
ϩ
CF3 ) was also observed. A significant increase was observed at m/z ϭ 50
ϩ
(CF2 ) where a decrease due to the consumption of CF2 would be expected.
This signal at m/z ϭ 50 is explained as a fragment ion of CF2O3 at the low
ionisation energy (IE ϭ 17 eV). At high ionisation energy (IE ϭ 70 eV) a
ϩ
decrease was observed (Fig. 4) due to the dominant CF2 ion formation from
˜
the CF2(X) radical.
Brought to you by | New York University Bobst Library Technical Services
Authenticated
Download Date | 7/31/15 9:33 AM

106
Th. Beiderhase, K. Hoyermann and W. Hack
Fig. 7. Product spectrum of the reaction (1); IE ϭ 17 eV, p ϭ 1.9 mbar, T ϭ 298 K,
Ϫ11
Ϫ9
v ϭ 21.6 m/s, t ϭ 6.5 ms, [CF ] ϭ 9.6 · 10 mol/cm3, [O3]0 ϭ 1.1 · 10 mol/cm3.
¯
R
2 0
In order to prove that CF2O is a direct product of reaction (1) at the low
pressures applied in the present experiments (i.e. that m/z ϭ 47 and 66 is
not only a fragment ion of CF2O3) the products were compared with the
products of the reaction:
CHF2 ϩ O3 → CHF2Oն ϩ O2
(10)
CHF2Oն → CF2O ϩ H.
This is a fast reaction k10(298 K) ϭ 1.4 · 1012 cm3/mol s in which CF2O is
formed as the main product [28]. The results obtained for four different
ionisation energies are shown in Fig. 8. The peaks m/z ϭ 66 and 47 are
compared. For the reaction (10) it is known that m/z ϭ 66 is the parent
peak of CF2O and m/z ϭ 47 the fragment peak of CF2O. In the CF2/O3
system both could be fragment peaks of CF2O3. The appearance potential
ϩ
of FCO from CF2O was determined to be 14.736 eV [22]. At low ionisa-
tion energy (i.e. 17 eV) the fragment patterns obtained in the reaction CF2
ϩ O3 and CHF2 ϩ O3 were identical. At this ionisation energy the formation
ϩ
of the fragment ion CFO (m/z ϭ 47) from CF2O3 is of no importance. For
Brought to you by | New York University Bobst Library Technical Services
Authenticated
Download Date | 7/31/15 9:33 AM

˜
Elementary Reaction of CF2(X) with O3 : Primary Products
107
Fig. 8. The products of reaction (1) and the reaction CHF2 ϩ O3 → CF2O ϩ HO2 at
different ionisation energy. ᭿, CF2 ϩ O3; ᮀ, CHF2 ϩ O3 under otherwise identical
experimental conditions. Magnetic deflection type mass spectrometer ion signals obtained
by single ion counting.
ϩ
higher ionisation energies the fragmentation forming CFO is increased and
thus the number of counts for m/z ϭ 47 increased relative to m/z ϭ 66.
Thus, it is concluded that in the reaction (1) CF2O and CF2O3 are formed
both as primary products.
Experimental results of the systematic variation of the ionisation energy
ϩ
for four examples are also shown for m/z ϭ 44 (CO2 ) in Fig. 8 as observed
in the CF2/O3 system. As the signal at m/z ϭ 44 depends strongly on the
ionisation energy and decreases relative to the signal of the primary prod-
ucts CF2O3, CF2O, and O2, it can be concluded that CO2 is not a primary
eϪ
ϩ
reaction product, but a fragment ion formed via CF2O3 –→ CO2 ϩ F2O.
Brought to you by | New York University Bobst Library Technical Services
Authenticated
Download Date | 7/31/15 9:33 AM

108
Th. Beiderhase, K. Hoyermann and W. Hack
Therefore, the following mechanism is proposed:
M
ն
˜
˜
CF2(X)ϩO3(X) { CF2O3 → CF2O3
→ CF2O ϩ O2
(1)
i.e. CF2Oն3 isomerizes or decomposes prior to collision stabilisation under
the experimental conditions at a pressure of p ϭ 2.5 mbar and forms CF2O
and O2 molecules.
The depletion of CF2 occurs not only via the very slow reaction (1) but
also via the fast combination reaction (3). With this mechanism all observed
mass peaks and fragment ion formation can be explained.
4. Discussion
The rate of reaction (1) has been determined in a flow reactor with LIF
detection of CF2 under pseudo first order conditions ([CF2]0 Ӷ [O3]0). The
rate constant was determined with the assumption that the stoichiometry of
the reaction is unity [9]. The products had not previously been determined
directly. With the determination of the products in the present work the
stoichiometry can be derived to be unity since none of the reaction products
˜
further react fast with CF2(X) under the conditions of the investigation of
the rate coefficient. Thus the rate coefficient:
Ϫ1
k1(T) ϭ 1.9 · 1012 exp(Ϫ25 kJ/mol /RT) cm3/mol s
given in (9) evaluated from the disappearance rate of the CF2 radicals can
be associated with the reaction:
˜
CF2(X) ϩ O3 → products.
(1)
(1a)
(1b)
In the first step a complex CF2Oն3 is formed:
ն
˜
˜
CF2(X) ϩ O3(X) { CF2O3 .
The complex CF2Oն3 contains sufficient energy to decompose
ն
3 Ϫ
˜
CF2O3 → CF2O(X) ϩ O2(X Σg )
in a highly exothermic reaction (∆RH ϭ Ϫ545.2 kJ/mol).
The experimental findings of this work on the formation of the primary
products of the reaction CF2 ϩ O3 → {CF2O3; CF2O ϩ O3} (compound of
the empirical molecular formula CF2O3) together with the following infor-
mation related to the CF2 ϩ O3 reaction make a detailed picture of the
dynamics of the reaction feasible beyond the above given considerations
resulting from the energetics. These informations are:
(i) Ab initio calculations of the CF2-O3 complex and matrix isolation IR
spectroscopy [10, 11] led to the identification of a weakly bound complex
Brought to you by | New York University Bobst Library Technical Services
Authenticated
Download Date | 7/31/15 9:33 AM

˜
Elementary Reaction of CF2(X) with O3 : Primary Products
109
Fig. 9. Energy diagram for the reaction CF2 ϩ O3. Primary complex structure as given
in (10) (kass, krediss, kdiss, kiso and kstab see text).
CF2-O3 with various structures (Fig. 9). Structure (1): Electrostatic inter-
action of the two terminal oxygen atoms equally with the carbon atom,
giving rise to a binding energy of E01 ϭ 16.2 kJ/mol. Structure (3): one
terminal oxygen atom interacts with the C atom with a binding energy of
E03 ϭ 9.7 kJ/mol.
(ii) A long lived species FC(O) OOF, different from the described CF2-
O3 complex, has been identified by NMR spectroscopy and was described
to be stable for more than 8 h at 100 mbar and room temperature [29].
(iii) The rate coefficient of the reaction CF2 ϩ O3, k(T), exhibits a high
temperature dependence, represented by an activation energy of Ea ϭ 25 kJ/
mol [9].
The dynamics of the reaction CF2 ϩ O3 is described as a complex form-
ing bimolecular reaction with a loose entrance and rigid exit channel [30]
Brought to you by | New York University Bobst Library Technical Services
Authenticated
Download Date | 7/31/15 9:33 AM

110
Th. Beiderhase, K. Hoyermann and W. Hack
for which the energetics are displayed in Fig. 9. Assuming independent
channels for the association (rate constant kass), redissociation of the com-
plex (specific rate constant, k(E) ϭ krediss), stabilization by collisions with
M (rate constant kstab), isomerisation of the complex (kiso), and dissociation
of the complex to the products CF2O ϩ O2 (kdiss), a semiquantitative calcu-
lation of the complex life time τ(E) is possible following the procedures of
Troe [31]. The difference of the thermal energy (H300ϪH0) of the reaction
is given by the levels CF2 ϩ O3 (300 K) and CF2 ϩ O3 (0 K). The binding
energies and the frequencies are taken from (10), the total energy E ϭ E0 ϩ
(H300ϪH0), (E0 ϭ 16.2 and 9.7 kJ/mol, respectively). The frequencies used
Ϫ1
are: 1221, 1102, 668, 1080, 1078, 707, 1224, 1109, 15 (4ϫ) cm ). The
specific rate constant k(E) for the redissociation of the complex is found to
Ϫ1
be the fastest process (k(E) Х 2 ·1010 s ). The value is thought to be re-
liable within a factor of 5, considering the uncertainties in the transition
state. At the experimental conditions of this work (a pressure of 1.3 mbar,
main carrier gas: He) the rate constant for the stabilization kstab ϭ ZLJ · [M]
(ZLJ ϭ Lenard-Jones collision frequency) is of the order of kstab
Х
Ϫ1
2.4 · 107 s . Thus in a step ladder model of redissociation/deactivation by
Ϫ1
an average transfered energy down of Ϫ͗∆E͘ Ϸ 200 cm , complete redis-
sociation and no stabilisation to the structures (1), (3) are to be expected.
The observed species of empirical formula (CF2O3) may be formed by an
independent isomerisation channel of the complex CF2O3 over a barrier to
a structure like FC(O)OOF, different from the noted species (1), (3) (Fig. 9).
In the presented flow reactor experiments also a heterogeneous process
might be responsible for the FC(O)OOF formation. But it is noteworthy to
state that the time resolved experiments with the arrangement flow reactor/
time of flight mass spectrometer the product signals followed immediately
the reaction conditions “on’’/“off’’ and did not point to a slow desorption
process for (CF2O3) formation. Thus it can be concluded that no indication
to a heterogeneous contribution to the CF2O3 formation was observed.
The small amount of the primary dissociation products (CF2O ϩ O2) is
explained by the high barrier (represented by the measured activation en-
ergy) leading to kdiss Ӷ krediss, making the redissociation process (krediss) the
major process. A direct abstraction route (CF2O3 → CF2O ϩ O2) is thought
to be negligible, as the complex (CF2O3) is substantiated by the ab initio
calculations and the experimentally proven existence of a species (CF2O3).
From the energetics it can be discussed in which electronic state O2 is
formed. The dissociation of the complex CF2O3 (1b) is highly exothermic
(∆RH ϭ Ϫ450 kJ/mol) even if O2 is formed in its first electronically excited
singlet state O2 (1∆g), as one would expect from spin conservation rules. In
the dissociation pathway (1b) the CF2O has sufficient internal energy, to
decompose further via:
˜
CF2O(X) → FCO ϩ F
∆RH ϭ 535 kJ/mol.
Brought to you by | New York University Bobst Library Technical Services
Authenticated
Download Date | 7/31/15 9:33 AM

˜
Elementary Reaction of CF2(X) with O3 : Primary Products
111
This pathway is, however, energetically accessible only if O2 is formed in
the spin forbidden triplet electronic ground state (X3 Ϫ). No indications
͚
were found that either FCO or F atoms are formed in thegsystem, this might
point to the formation of O2 in the metastable exited singlet state.
Other reaction channels can be excluded from the energetics and from
the observed products. The formation of F atoms via the reaction:
CF2 ϩ O3 → CO2 ϩ O ϩ 2 F
is endothermic by ∆RH ϭ 64.8 kJ/mol. Moreover this reaction would also
be in disagreement with the observed formation of CF2O and O2. A direct
formation of CO2 via:
CF2 ϩ O3 → CO2 ϩ F2O ∆RH ϭ Ϫ322.5 kJ/mol
is, however, highly exothermic, and could be followed by the slightly exo-
thermic subsequent reaction:
F2O ϩ O3 → 2 F ϩ 2 O2
∆RH ϭ Ϫ3.2 kJ/mol
in which the F atoms and O2 molecules are formed. This reaction sequence,
which would require a significant amount of rearrangement in the CF2O3
complex, does not, however, explain the observed formation of CF2O.
A formation of difluordioxirane:
1
3
˜
˜
˜
CF2(X) ϩ O3(X) → CF2O2(X A1) ϩ O( P)
is very endothermic by approximately ∆RH ϭ ϩ317 kJ/mol [32], even if
the O atoms are formed in the spin forbidden triplet electronic ground state.
This reaction and in particular the spin allowed reaction forming O(1D) are
of no importance in the thermal system CF2/O3.
Thus it can be concluded that the dynamics sketched in Fig. 9 represent
the mechanism of reaction (1).
The reactions of odd oxygen O and O3 lead to very different products.
In the case of O(3P) two reactive F atoms are formed
O ϩ CF2 → FCO ϩ F.
O ϩ FCO → CO2 ϩ F.
(5)
(11)
whereas in the reaction (1) two stable molecules, CF2O and O2, together
with CF2O3 result.
From the presented results O3 is unlikely to be considered as an additive
in the etching process since no actived F atoms are formed, needed to accel-
erate the process.
Acknowledgement
We are greatly indebted to Prof. H. Gg. Wagner for stimulating interest.
Financial support of the DFG (SFB 357 Molekulare Mechanismen unimole-
kularer Reaktionen) and the Fonds der Chemischen Industrie is gratefully
acknowledged.
Brought to you by | New York University Bobst Library Technical Services
Authenticated
Download Date | 7/31/15 9:33 AM

112
Th. Beiderhase, K. Hoyermann and W. Hack
References
1. I. C. Plumb and K. R. Ryan, Plasma Chem. and Plasma Proc. 6 (1986) 205.
2. K. R. Ryan and I. C. Plumb, Plasma Chem. and Plasma Proc. 6 (1986) 231.
3. J. R. Gilbert, I. R. Slagle, R. E. Graham and D. Gutman, J. Chem. Phys. 80 (1976)
14.
4. K. R. Ryan and I. C. Plumb, Plasma Chem. and Plasma Proc. 4 (1984) 271.
5. L. M. Arin and P. Warneck, J. Phys. Chem. 76 (1972) 1514.
6. R. E. Rebbert and P. J. Ausloos, J. Photochem. 4 (1975) 419.
7. F. S. Rowland and M. J. Molina, Rev. Geophys. Space Phys. 13 (1975) 1.
8. M. J. Molina and F. S. Rowland, Nature 249 (1974) 810.
9. W. Hack, M. Wagner and K. Hoyermann, J. Phys. Chem. 99 (1995) 10847.
10. S. R. Davis and L. Liu, J. Chem. Phys. 97 (1993) 3690.
11. L. Liu and S. R. Davis, J. Phys. Chem. 96 (1992) 9719.
12. J. Edelbüttel-Einhaus, W. Hack, K. Hoyermann, G. Rohde and H. Gg. Wagner, Ber.
Bunsenges. Phys. Chem. 93 (1989) 1413.
13. G. Hancock, P. D. Harrison and A. J. MacRobert, J. Chem. Soc., Faraday Trans, 2
(1986), 82, 647.
14. G. Hancock and D. E. Heard, Chem. Phys. Lett. 158 (1989) 167.
15. G. Hancock and D. E. Heard, J. Chem. Soc. Faraday Trans. 8 (1991) 1039, 1045.
16. G. Dornhöfer, W. Hack, K. Hoyermann and G. Rohde, Ber. Bunsenges. Phys. Chem.
94 (1990) 468.
17. G. Rohde, University Göttingen, PhD thesis 1991.
18. J. J. Ritter, J. Am. Chem. Soc. 100 (1978) 2441.
19. S. Sharpe, B. Harnett, H. S. Sethi and D. S. Sethi, J. Photochem. 38 (1987) 1.
20. NIST Standard Reference Database 25, Structures and Properties, Stephen E. Stein
(Ed.) Version 2.0, 1994. Spectral Database, Version 3.01, 1990.
21. NIST Standard Reference Database Series 1a, The NIST/EPA/MSDC Mass Spectral
Database, Version 3.01, 1990.
22. T. J. Buckley, R. D. Johnson, R. E. Huie, Z. Zhang, S. C. Kuo and R. B. Klemm, J.
Phys. Chem. 99 (1995) 4879.
23. V. Tarnovsky, P. Kurunczi, D. Rogozhnikov and K. Becker, Int. J. Mass Spec. Ion.
Proc. 128 (1993) 181.
24. V. Tarnovsky and K. Becker, J. Chem. Phys. 98 (1993) 7868.
25. K. H. Becker, in Electron Collisions with Molecules, Clusters and Surfaces, H.
Ehrhardt and L. A. Morgan eds., Plenum Press N.Y. (1994), pp. 127.
26. P. S. Monks, L. J. Stief, M. Krauss, S. C. Kuo and R. B. Klemm, J. Chem. Phys.
100 (1994) 1902; Chem. Phys. 211 (1993) 416.
27. K. Sugawa, R. Nakanaga, H. Takeo and Ch. Matsumura, Chem. Phys. Lett. 134
(1987) 347.
28. Th. Beiderhase, University Göttingen, PhD thesis 1995.
29. G. A. Argüello, G. Balzer-Jöllenbeck, B. Jülicher and H. Willner, Inorg. Chem. 34
(1995) 603.
30. J. Troe, J. Chem. Soc. Faraday Trans. 90 (1994) 2303.
31. J. Troe, J. Phys. Chem. 83 (1979) 114.
32. M. Rahman, M. L. McKee, P. B. Shevlin and R. Sztyrbicka, J. Am. Chem. Soc. 110
(1988) 4002.
Brought to you by | New York University Bobst Library Technical Services
Authenticated
Download Date | 7/31/15 9:33 AM