Difluorocarbene studied with threshold photoelectron spectroscopy (TPES): Measurement of the first adiabatic ionization energy (AIE) of CF2

Article abstract of DOI:10.1002/chem.200801699

The first photoelectron band of difluorocarbene CF₂, has been studied by threshold photoelectron (TPE) spectroscopy. CF₂ was prepared by microwave discharge of a flowing mixture of hexafluoropropene, C₃F₆, and argon. A vibrationally resolved band was observed in which at least twenty-two components were observed. In the first PE band of CF₂, the adiabatic ionization energy differs significantly from the vertical ionization energy because, for the ionization CF ₂⁺ (X²A₁)+e^- ← CF₂ (X¹A₁), there is an increase in the FCF bond angle (by ≈20°) and a decrease in the C-F bond length (by ≈ 0.7 A). The adiabatic component was not observed in the experimental TPE spectrum. However, on comparing this spectrum with an ab initio/Franck-Condon simulation of this band, using results from high-level ab initio calculations, the structure associated with the vibrational components could be assigned. This led to alignment of the experimental TPE spectrum and the computed Franck-Condon envelope, and a determination of the first adiabatic ionization energy of CF₂ as (11.362± 0.005) eV. From the assignment of the vibrational structure, values were obtained for the harmonic and fundamental frequencies of the symmetric stretching mode (v₁′) and symmetric bending mode (v₂′) in CF₂⁺ (X²A₁).

Full text of DOI:10.1002/chem.200801699

DOI: 10.1002/chem.200801699
Difluorocarbene Studied with Threshold Photoelectron Spectroscopy
(TPES): Measurement of the First Adiabatic Ionization Energy (AIE) of CF2
Fabrizio Innocenti,[a] Marie Eypper,[a] Edmond P. F. Lee,[a] Stefano Stranges,[b]
Daniel K. W. Mok,[d] Foo-tim Chau,[d] George C. King,[c] and John M. Dyke*[a]
Abstract: The first photoelectron band
of difluorocarbene CF2, has been stud-
ied by threshold photoelectron (TPE)
spectroscopy. CF2 was prepared by mi-
crowave discharge of a flowing mixture
of hexafluoropropene, C3F6, and argon.
A vibrationally resolved band was ob-
served in which at least twenty-two
components were observed. In the first
PE band of CF2, the adiabatic ioniza-
tion energy differs significantly from
the vertical ionization energy because,
FCF bond angle (by ꢁ208) and a de-
vibrational components could be as-
signed. This led to alignment of the ex-
perimental TPE spectrum and the com-
puted Franck–Condon envelope, and a
determination of the first adiabatic ion-
ization energy of CF2 as (11.362ꢂ
0.005) eV. From the assignment of the
vibrational structure, values were ob-
tained for the harmonic and fundamen-
tal frequencies of the symmetric
stretching mode (n1’) and symmetric
ꢀ
crease in the C F bond length (by
ꢁ0.7 ꢀ). The adiabatic component was
not observed in the experimental TPE
spectrum. However, on comparing this
spectrum with an ab initio/Franck–
Condon simulation of this band, using
results from high-level ab initio calcula-
tions, the structure associated with the
Keywords: carbenes
· photoelec-
+
2
˜
bending mode (n2’) in CF2 (X A1).
tron spectroscopy · reactive inter-
mediates · vibrational spectroscopy
!
+
2
ꢀ
˜
for the ionization CF2 (X A1)+e
1
˜
CF2 (X A1), there is an increase in the
Introduction
ters of the ground state, X1A1, and a number of excited
states have been derived from electronic emission[4] and ab-
sorption spectroscopy,[2,3] and single vibronic level (SVL)
fluorescence.[1] Several microwave studies have been report-
ed which provide precise rotational constants for the X1A1
state.[5,6]
Difluorocarbene, CF2, is important in plasma processing and
is produced in the stratosphere by photodissociation of
chlorofluorocarbons by radiation from the sun. It has been
studied in some detail spectroscopically in the ultraviolet,[1–4]
microwave[5,6] and infrared regions.[7–10] Structural parame-
Small reactive intermediates, such as CF2, play important
roles in determining the etching rate, selectivity and aniso-
tropy of plasma-etching processes. The mechanisms of CF2
production and destruction in fluorocarbon reactive ion-
etching processes have been investigated using techniques
such as laser induced fluorescence,[11] broad-band UV ab-
sorption spectroscopy,[12,13] infrared diode laser spectrosco-
py[14,15] and cavity-ring down spectroscopy.[16] It has been
found that the primary mechanism for CFx production
(where x=1, 2 and 3) is neutralisation and fragmentation of
[a] Dr. F. Innocenti, M. Eypper, Dr. E. P. F. Lee, Prof. Dr. J. M. Dyke
School of Chemistry, University of Southampton
Highfield, Southampton SO171BJ (UK)
E-mail: jmdyke@soton.ac.uk
[b] Dr. S. Stranges
Department of Chemistry and INSTM Unit
“La Sapienza”, University of Rome (Italy)
ISNM-CNR sez. Roma La Sapienza and Laboratoria
TASC-INFM-CNR
+
CFx (x=1–4) ions incident on the powered electrode and
34012 Trieste (Italy)
not direct electron-impact induced fragmentation of the
feedstock gas.[17] In order to understand, model and ulti-
mately control plasmas, spectroscopic methods such as those
listed above are needed to monitor reactive intermediates in
plasmas, and rate-constant measurements and reaction en-
[c] Prof. G. C. King
School of Physics and Astronomy, Manchester University
Manchester M139PL (UK)
[d] Dr. D. K. W. Mok, Prof. F.-t. Chau
Department of Applied Biology and Chemical Technology
The Hong Kong Polytechnic University
Hung Hom (Hong Kong)
thalpy determinations are required for relevant reactions in-
+
volving CF2 and CF2
.
11452
ꢁ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2008, 14, 11452 – 11460

FULL PAPER
2
Ab initio calculations: For the X A1 state of CF2+, geometry optimization
and vibrational frequency calculations were carried out using the restrict-
ed spin CCSD(T) method (RCCSD(T)) with augmented correlation-con-
sistent polarized valence (aug-cc-pVXZ or AVXZ) and core-valence
(aug-cc-pCVXZ or ACVXZ) basis sets of up to the quintuple-zeta (X=
˜
The first adiabatic ionization energy (AIE) of CF2 is im-
portant in contributing to the determination of the enthalpy
of ion–molecule reactions involving CF2 in plasmas. Previ-
+
ous determinations of the first AIE of CF2 have included a
study by vacuum ultraviolet photoelectron spectroscopy
(PES),[18] which gave the first AIE as (11.42ꢂ0.01) eV, a
study with photoionization mass spectrometry (PIMS),[19]
using radiation derived from a synchrotron source, which
gave the first AIE as (11.445ꢂ0.025) eV, and a study by
electron impact mass spectrometry, which gave a value of
(11.5ꢂ0.4) eV.[20] The PES study[18] shows the first band of
CF2 to consist of regular structure, with at least fifteen com-
ponents. It was interpreted[18] as a regular series in the defor-
mation mode in the ionic state. The vertical ionization
energy (VIE) was measured as (12.240ꢂ0.005) eV. In the
present work, this band has been re-investigated at higher
resolution using threshold photoelectron spectroscopy
(TPES). For this a spectrometer which has been specifically
designed to study reactive intermediates with the PE and
angle-resolved constant-ionic-state (CIS) methods was
modified to allow TPES measurements to be performed.[21]
The objective is to obtain a higher resolution spectrum of
the first band of CF2. The spectrum obtained, supported by
appropriate ab initio/Franck–Condon factor calculations,
should allow the first AIE to be determined more reliably
than previously and enable the vibrational structure in the
first band to be analysed more thoroughly.
Q or 5) quality. With the core-valence basis sets, all electrons were corre-
lated. For the X A1 state of CF2, ab initio total energies at different bond
1
˜
lengths and angles, the RCCSD(T)/aug-cc-pV5Z potential energy func-
tion (PEF) and anharmonic vibrational wavefunctions have been taken
from our previous study on CF2,[26] except for RCCSD(T)/aug-cc-pCV5Z
results, which have been obtained in the present study. The largest aug-
cc-pCV5Z calculations have 543 contracted basis functions.
For the evaluation of the best theoretical geometrical parameters (re and
qe) and adiabatic ionization energy (AIE), the 1/X3 formula[27] was used
to extrapolate the computed RCCSD(T)/ACVQZ and RCCSD(T)/
ACV5Z values to the complete basis set (CBS) limit. Since the
RCCSD(T)/ACVXZ (X=Q or 5) values were used in the extrapolation,
core correlation contributions have already be accounted for. For the cor-
rection for zero-point vibrational energies (DZPE) in order to give AIE0,
1
˜
available experimental fundamental vibrational frequencies for the X A1
2
state of CF2 were used.[28] For the X A1 state of CF2+, the best theoretical
˜
ZPE was used, and it was estimated using the CBS value {extrapolation
employing the 1/X3 formula, with the ZPEs evaluated using the comput-
ed RCCSD(T)/AVQZ and RCCSD(T)/AV5Z harmonic vibrational fre-
quencies} plus core correlation correction {the difference between the
RCCSD(T)/ACVQZ and RCCSD(T)/AVQZ values}.
Potential energy functions, anharmonic vibrational wavefunctions and
2
+
˜
Franck–Condon factor calculations: The PEF of the X A1 state of CF2
was fitted to 106 computed CASSCF/MRCI+D/AV5Z energies in the
ranges of 0.9ꢃr(CF)ꢃ1.95 ꢀ and 70.0ꢃq(FCF)ꢃ160.08. The multi-ref-
G
erence CASSCF/MRCI method (including the Davidson correction, with
a full valence active space) has been used in energy scans for the fitting
2
+
˜
of the PEF of the X A1 state of CF2 because multi-reference character
becomes non-negligible in the region with rꢄ1.6 ꢀ. Nevertheless, com-
puted CI coefficients of the major electronic configuration obtained from
Experimental Section
2
+
˜
the MRCI calculation of the X A1 state of CF2 in this region have
values larger than 0.778, and the sums of the squares of the computed CI
coefficients of all reference configurations, ꢀACTHNUTRGNEUNG
(Cref)2, have values larger
The experiments reported here were undertaken on the Circularly Polar-
ized Beamline (4.2R, Polar) at the Elettra synchrotron radiation source
(Trieste). A photoelectron spectrometer has been used which was specifi-
cally designed to study reactive intermediates with PE and CIS spectros-
copy.[22–24] This spectrometer has recently been modified to allow TPE
spectra to be obtained.[21] In order to record TPE spectra, the photoelec-
tron spectrometer was tuned to detect near-zero energy (threshold) pho-
than 0.959, indicating that the computed MRCI wavefunctions describe
the electronic state studied adequately. The root-mean-square (r.m.s) de-
2
+
˜
viations of the fitted PEFs of the X A1 state of CF2 from computed ab
initio energies is 12.9 cmꢀ1
.
The details of the coordinates and polynomial employed for the PEFs,
the rovibrational Hamiltonian[29] and anharmonic vibrational wavefunc-
tions used in the variational calculations, and the Franck–Condon (FC)
factor calculations which include Duschinsky rotation and anharmonicity
have been described previously[26, 30] and hence will not be repeated here.
toelectrons. The detection of threshold electrons was optimised using the
!
Ar+(2P3/2,2P1/2
)
Ar
N
tion obtained was typically about 5 meV as estimated from the full-width
!
at half maximum of the main (3p)ꢀ1 Ar
G
)
Ar
U
Nevertheless, some details of the harmonic basis functions used in the
tional photoelectron (PE) spectra were also recorded as described in ear-
lier work[25] and the same procedures were used to normalize the spectra
for photon flux and the transmission function of the spectrometer.
2
˜
calculation of the anharmonic vibrational wavefunctions of the X A1
+
state of CF2 are given. The vibrational quantum numbers of the har-
monic basis functions of the symmetric stretching and bending modes
employed in the calculation of anharmonic wavefunctions have values of
up to v1’=12, v2’=25 with the restriction of (v1’+v2’)ꢃ25.
CF2 was produced by a microwave discharge of flowing hexafluoropro-
pene, C3F6, diluted with argon. Preliminary experiments were carried out
in Southampton in order to determine the optimum pressures which max-
imise the intensity of the first CF2 band in the PE spectra. The optimum
2
+
˜
The best computed geometry of the X A1 state of CF2 obtained at the
1
partial pressures were: DpACHTNUGRTNEUNG
(C3F6)=5ꢂ10ꢀ6 and Dp(Ar)=1ꢂ10ꢀ7 mbar.
˜
CBS limit in the present study, the experimental geometry of the X A1
state of CF2 (re =1.2975 ꢀ and qe =104.818, derived from experimentally
derived Ae and Be values)[31] and the experimental AIE0 value of
11.362 eV (see later) obtained from the TPES spectrum in the present
study were used in the FC factor calculations. In addition, FC factors
These partial pressures were measured using an ionization gauge con-
nected to the main vacuum chamber and are with respect to the back-
ground pressure in the vacuum chamber (3ꢂ10ꢀ7 mbar).
were calculated at Boltzmann vibrational temperatures of 0 K and 600 K.
!
+
2
ꢀ
1
˜
˜
The CF2 (X A1)+e
CF2 (X A1) photoelectron band has been simulat-
Computational Details
ed with these computed FC factors. The method includes allowance for
Duschinsky rotation and anharmonicity. Gaussian linewidth of
A
0.005 eV full-width-at-half-maximum (FWHM) was used for each vibra-
tional component in the spectral simulation, as the experimental thresh-
old photoelectron spectrum (TPES) has a resolution of ca. 0.005 meV
FWHM.
In order to compute potential energy functions for the ground neutral
and ionic states so that Franck–Condon factors could be calculated for
the first PE band of CF2, ab initio calculations were performed. These
can be described as follows:
Chem. Eur. J. 2008, 14, 11452 – 11460
ꢁ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
11453

J. M. Dyke et al.
The MOLPRO suite of programs[32] was employed for all ab initio calcu-
autoionization.[39,40] This is primarily due to the high density
of Rydberg states, which are parts of series which converge
on higher rovibronic ionic levels. These can autoionize and
produce electrons of low kinetic energy that are detected in
TPES. As a result, autoionization processes can lead to non-
Franck–Condon distributions as well as the observation of
ionic vibrational levels which are not observed in the
normal Franck–Condon distribution. Also, these autoioniza-
tion processes may lead to observation of ionic states that
cannot be observed by direct ionization from the ground
state in conventional PES. The TPE spectrum in Figure 2
shows a better resolution and a very different vibrational
profile with respect to the Franck–Condon distribution of
the PE spectrum.
2
˜
lations carried out in the present investigation. The PEF of the X A1
+
+
2
˜
state of CF2 and the full list of computed FC factors of the CF2 (X A1)
!
+ eꢀ CF2 (X A1) photoionization process are available, upon request,
1
˜
from the authors.
Results and Discussion
Experimental spectra: The UV photoelectron spectrum of
C3F6 shows a structured band in the region 10.4–11.9 eV and
five broad bands in the ionization energy region 14.0–
17.0 eV.[33,34] Microwave discharge of flowing C3F6/Ar mix-
tures gives rise to complete destruction of C3F6 with produc-
tion of CF2 and C2F4. It is also known from separate experi-
ments that microwave discharge of flowing C2F4/Ar mixtures
gives CF2. In this work, the discharge conditions and partial
pressures of C3F6 and Ar were optimised to give the maxi-
mum yield of CF2. A photoelectron spectrum obtained at
hn=21.0 eV from discharged C3F6/Ar in the ionization
[35,36]
region 10.0–13.2 eV showing the first bands of C2F4
CF2[18] is shown in Figure 1.
and
Figure 2. TPE (upper) and PE (lower) spectra recorded for a flowing CF2
and Ar mixture in the 11.3–13.2 eV photon energy region.
In the TPE spectrum the lower vibrational bands are
more intense and more clearly resolved than in the PE spec-
trum where the lowest vibrational component that is clearly
observed is at about 11.6 eV (two additional lower-energy
components were observed in the original PES study).[18] In
the earlier PES study of CF2,[18] the adiabatic and vertical
ionization energies were determined as 11.42ꢂ0.01 eV and
12.240ꢂ0.005 eV respectively. In the present study it is pos-
sible to measure the VIE from the PE spectrum as 12.258ꢂ
0.002 eV and the lowest vibrational band observed in the
TPE spectrum as 11.438ꢂ0.004 eV. In both cases the shift
with respect to the previous study[18] is 18 meV higher. From
the experimental TPE spectrum, it is not possible to estab-
lish if the component observed at 11.438 eV is really the
lowest vibrational component in the PE spectrum of CF2.
This is because below 11.4 eV there is a contribution from
the highest observed vibrational component of the first PE
band of C2F4 as can be seen in Figure 2 and the intensity of
a lower vibrational band of CF2, if present, would be very
small compared to the noise level. Indeed in the TPE spec-
trum in Figure 2 it is possible to observe some structure
below 11.4 eV but the signal-to-noise is not sufficient to
allow unambiguous identification of weak C2F4 contributions
and possible weak vibrational components of CF2. For this
Figure 1. UV photoelectron spectrum recorded at a photon energy of
21.0 eV of a microwave discharge of a flowing mixture of hexafluoropro-
pene, C3F6, in Ar. Spectra obtained under these conditions show that all
the C3F6 is destroyed and converted into CF2 and C2F4.
In order to record TPE spectra, the penetrating-field ana-
lyzer was tuned to detect near-zero energy (threshold) pho-
toelectrons. The detection of threshold electrons was opti-
!
(1S0) (3pꢀ1) TPE spec-
) ArACHTNUTGRNUEGN
mised using the Ar+(2P3/2,2P1/2
trum.[21,37,38] The spectral resolution obtained is typically
about 5 meV as estimated from the full-width at half maxi-
!
CTUHGNTRENNUNG CAHTUNGTRENNUGN
mum of the main (3p)ꢀ1Ar+A(2P3/2) Ar(1S0) band.
Figure 2 shows the TPE spectrum of CF2 recorded in the
photon region from 11.3 to 13.2 eV together with a PE spec-
trum recorded at hn=21.0 eV. Whilst the observed relative
intensities of vibrational components in a PE band in a con-
ventional PE spectrum are almost always governed by
Franck–Condon factors between the initial neutral and final
ionic state, those observed by TPES are often dominated by
11454
www.chemeurj.org
ꢁ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2008, 14, 11452 – 11460

Difluorocarbene
FULL PAPER
reason the experimental TPE spectrum was compared with
an ab initio/Franck–Condon computed vibrational envelope
to see if the AIE of CF2 could be established.
Figure 3 shows the experimental TPE spectrum compared
!
+
with the simulated CF2 (X2A1) CF2 (X1A1) photoelectron
spectrum at 0 K. The spectral simulation used a Gaussian
function of width 5 meV for each vibrational component
that is comparable with the experimental resolution. The
calculated adiabatic component is placed at 11.367 eV. This
was done by positioning the computed envelope so that the
structure in the higher components in the TPE spectrum
matches well with the computed structure, notably in the
third, fourth, sixth and ninth components. The 11.36 eV
Figure 4. Expanded TPE spectrum of CF2 in the 11.4–12.2 eV photon
!
(X1A1) photoelec-
CF2ACHTGNURTENNUNG
energy region (upper). Simulated CF2
N
tron spectrum for 0 K Boltzmann vibrational temperature (lower).
and the same is true for the corresponding experimental vi-
brational components which appear to be a single feature.
The exception is the vibrational component at 11.45 eV in
Figure 4 which is a single vibrational component in the cal-
culated spectrum while it appears to have two components
in the experimental spectrum.
The simulated spectrum in Figure 4 was computed with a
Boltzmann vibrational temperature of 0 K. The calculation
has been repeated using a Boltzmann distribution at a tem-
perature of 600 K and the result is shown in Figures 5 and 6.
Figure 3. TPE spectrum recorded for a flowing CF2 and Ar mixture in
+
the 11.3–13.2 eV photon energy region (upper). Simulated CF2
!
(X2A1) CF (X1A1) photoelectron spectrum for 0 K Boltzmann vibra-
2ACHTUNGTRENNUNG
tional temperature (lower).
component has a very small relative intensity and it is below
the noise level of the experimental TPE spectrum. The posi-
tions of all the observed vibrational components are in ex-
cellent agreement with the computed components although
the TPE vibrational envelope is different from the comput-
ed envelope. This is because of the non-Franck–Condon be-
haviour of threshold photoionization. The shape of each ex-
perimental vibrational component closely matches the shape
of the corresponding calculated vibrational component as
can be seen from the comparison of the lowest vibrational
components shown in Figure 4. In the third, sixth and elev-
enth experimental vibrational components (counting the
11.36 eV component, not shown in Figure 4, as the first), it
is possible to observe two features with the one at lower
ionization energy more intense; the same pattern is ob-
served in the calculated vibrational components. The fourth,
seventh and ninth experimental vibrational components
each show two features with the one at higher ionization
energy more intense; the same pattern is observed in the
calculated fourth and seventh vibrational components while
in the ninth component the two contributing features are
comparable in intensity. In the other calculated vibrational
components in Figure 4 the 5 meV band-width used is not
good enough to allow separation of different components
Figure 5. TPE spectrum recorded for a flowing CF2 and Ar mixture in
the 11.3–13.2 eV photon energy region (upper). Simulated CF2 CHTUNGTRENNUNG
+A(X2A1)
!
CF
temperatureACHTUNGTRENNUNG
(X1A1) photoelectron spectrum for 600 K Boltzmann vibrational
(lower).
The main differences with respect to the simulation at 0 K
are at ionization energies above 12.8 eV and for the vibra-
tional component in Figure 4 at ꢁ11.45 eV. This band, in
the calculation at 600 K, has a “hot band” on the higher ion-
ization energy side as can be seen in Figure 6. The shape of
the 11.45 eV vibrational component in the experimental
TPE spectrum in Figure 4 is therefore explained considering
the contribution from the “hot” band. The simulated spec-
trum at 600 K offers an even better match with the experi-
Chem. Eur. J. 2008, 14, 11452 – 11460
ꢁ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
11455

J. M. Dyke et al.
1
˜
CF2 X A1 are shown in Table 2. Previous MRSDCI/DZP re-
sults of Cai[41] (see references therein for earlier, lower level,
2
+
˜
ab initio results) on the X A1 state of CF2 are also included
in Table 1. Nevertheless, since the ab initio calculations car-
ried out in the present work are of significantly higher level
and also are more systematic than previous work, we will
focus only on our results in the following discussion. From
Table 1, the best computed geometrical parameters of the
2
+
˜
X A1 state of CF2 obtained at the CBS limit (including
core correlation as mentioned above) are re(CF)=1.2131ꢂ
0.0006 ꢀ and qe
G
tainties have been estimated by the differences between the
CBS and RCCSD(T)/ACV5Z values. These geometrical pa-
rameters are currently the most reliable values, as no experi-
mentally derived values are available at present.
Figure 6. Expanded TPE spectrum of CF2 in the 11.4–12.2 eV photon
!
(X1A1) photoelec-
CF2ACHTGNURTENNUNG
energy region (upper). Simulated CF2
N
tron spectrum for 600 K Boltzmann vibrational temperature (lower).
mental spectrum with respect Table 1. Optimized geometrical parameter (re in ꢀ and qe in degrees) and computed vibrational frequencies
2
{w1(a1), w2(a1) and w3(b2) in cmꢀ1} of, and computed adiabatic ionization energies (IE in eV) to, the X A1 state
˜
to the simulated spectrum at
0 K. Indeed, on comparing
of CF2+, obtained at different levels of calculations.
Method
re
qe
w
IE
Figure 4 with Figure 6 it is pos-
sible to notice an improvement
of the agreement with the cal-
culated spectrum at 600 K in
particular for the fourth, sixth
and ninth vibrational compo-
nents (counting the component
at 11.36 eV as the first).
The vibrational components
above 12.8 eV in the TPE ex-
perimental spectrum show con-
siderable intensity and a clear
deviation from the computed
Franck–Condon relative inten-
sities. This is probably due to
strongly autoionizing Rydberg
states converging to higher
ionic limits of CF2. The simula-
tion at 600 K shows some weak
extra structure above 12.8 eV,
due to the contributions from
“hot” bands, with respect to the
RCCSD(T)/aug-cc-pVQZ
RCCSD(T)/aug-cc-pV5Z
RCCSD(T)/aug-cc-pCVQZ[a]
RCCSD(T)/aug-cc-pCV5Z[a]
CAS/MRCI/aug-cc-pV5Z
1.2168
1.2160
1.2143
1.2137
1.2185
1.2185
124.63
124.58
124.67
124.63
124.63
124.52
1374.6, 648.6, 1698.9
1379.1, 649.7, 1701.7
1379.1, 651.2, 1704.6
11.3865
11.4010
11.3807
11.3937
A
1370.0, 653.2, –
1356.3, 651.7, –
AHCTUNGTRENNUNG
1.2131
1.222
124.59
124.5
11.412ꢂ0.018
+0.046
11.458ꢂ0.020
DACHTUNGTRENNUNG
[d]
best theoretical IE0
MRSDCI/DZP[e]
photoion. MS[f]
1259, 656, 1599
11.445ꢂ0.025
11.42ꢂ0.01
He I photoelectron spectrum[g]
TPES
–, 650ꢂ40, –
11.362ꢂ0.005
[a] All electrons correlated; see text. [b] Extrapolation to the complete basis limit (CBS), employing the 1/X3
formula and the RCCSD(T) values calculated using the aug-cc-pCVQZ (ACVQZ) and aug-cc-pCV5Z
(ACV5Z) basis sets. The estimated uncertainty for IE is the difference between the best theoretical value and
the ACV5Z value. [c] Contribution of zero-point energies (ZPE) of the two states involved: For CF2, the ex-
perimental fundamental vibrational frequencies, 1225.08, 666.25 and 1114.44 cmꢀ1 (gas phase values from
NIST WebBook),[28] were used, giving a ZPE value of 0.1863 eV. For CF2+, the CBS(1/X3; AVQZ, AV5Z)
ZPE value of 0.2315 eV and core(AVQZ, ACVQZ) correction of +0.0008 eV give the best theoretical ZPE
value of 0.2323 eV and a best estimated D
toionization mass spectrometry, from ref. [19]. [g] Ref. [18].
ACHTUNGTREN(NUNG ZPE) value of 0.045 eV. [d] CBS+DZPE. [e] From ref. [41]. [f] Pho-
simulation at 0 K (compare Figure 3 with Figure 5).
Table 2. Computed geometrical parameters and harmonic vibrational fre-
1
quencies (fundamental values in cmꢀ1) of the X A1 of CF2 obtained at
˜
The ionization energies of the calculated vibrational com-
ponents from the ab initio/Franck–Condon factor calcula-
tions are on average 5 meV higher than the more intense ex-
perimental vibrational components in the 11.6–12.6 eV
region. Hence, by moving the computed bands in Figure 3
by 5 meV to match the more intense experimental TPE vi-
brational component positions in the photon energy region
11.6–12.6 eV, we determine the AIE to be 5 meV lower than
the value obtained from the simulation. This gives a value
of 11.362ꢂ0.005 eV for the AIE.
the RCCSD(T) level of calculation using different basis sets. The
RCCSD(T)/aug-cc-pV5Z potential energy function (PEF) and anhar-
monic vibrational wavefunctions used have been taken from our previous
study on CF2.[26]
RCCSD(T)
Re [ꢀ] qe [8]
w1
w2
w3
Aug-cc-pVQZ
Aug-cc-pV5Z
1.3008 104.786 1242 [1231] 670 [668]
1.2953 104.907 1247 [1234] 675 [672]
Aug-cc-pCVQZ 1.2981 104.846
exptl.[42]
1.2975 104.81 [1225.0793] [666.24922] [1114.4435]
Ab initio results: The computed ab initio results obtained in
the present study for the X A1 state of CF2 are summar-
ized in Table 1, and the corresponding results obtained for
Regarding computed harmonic vibrational frequencies of
the X A1 state of CF2 obtained in the present study from
RCCSD(T) calculations, the largest difference between
2
+
2
+
˜
˜
11456
www.chemeurj.org
ꢁ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2008, 14, 11452 – 11460

Difluorocarbene
FULL PAPER
values obtained using different basis sets is less than 6 cmꢀ1.
The difference between the computed symmetric stretching
harmonic frequencies obtained from RCCSD(T)/AV5Z cal-
culations and the CASSCF/MRCI/AV5Z PEF is about
9 cmꢀ1, which may be considered as the largest theoretical
uncertainty associated with the computed vibrational fre-
quencies reported in this article. Before the vibrational as-
signment of the TPES spectrum is considered, it should be
noted that the only available experimental vibrational fre-
2
+
˜
quency of the X A1 state of CF2 is that of the bending
mode, 650ꢂ40 cmꢀ1, measured in the previously recorded
UV photoelectron spectrum.[18] This experimental n2’ value
agrees very well with the corresponding computed value of
651.7 cmꢀ1 obtained from the MRCI PEF of the present
study. However, as will be shown below in the comparison
between the simulated PE spectrum and the experimental
TPE spectrum, the observed vibrational structure in the UV
PE spectrum[18] is not due to a single n2’ progression, but a
number of combination bands. Nevertheless, the excellent
agreement between the simulated PE spectrum and the ob-
served TPE spectrum, in terms of position and shape of the
individual vibrational components, indicates that the com-
puted vibrational frequencies reported here are very close
to the true values.
The best computed AIE obtained at the CBS limit in the
present study is 11.412ꢂ0.018 eV. Correction of zero point
energies of the two states involved yields the best computed
AIE0 value of 11.46ꢂ0.02 eV. This value agrees very well
with the earlier experimental values of 11.445ꢂ0.025 eV
and 11.42ꢂ0.01 eV obtained from photoionization mass
spectrometry[19] and UV photoelectron spectroscopy, respec-
tively[18] but less well with the AIE derived in this work of
11.362 eV.
+
2
˜
Figure 7. Computed Franck–Condon factors of the CF2 (X A1)
+
!
eꢀ CF2 (X A1) photoionization process obtained with Boltzmann vibra-
tional temperatures of 0 K (top trace) and 600 K (bottom trace), and
those of “hot” bands arising from ionizations from the (0,1,0), (1,0,0),
1
˜
1
˜
(0,2,0) and (1,1,0) levels of the X A1 state of CF2 (middle trace).
mann vibrational temperature of 0 K, which contribute to
!
+
2
ꢀ
1
˜
˜
the CF2 (X A1) + e
CF2 (X A1) ionization process, are
shown in Figure 8. In practice, vibrational components with
v1’ and/or v2’ having values of up to at least 8 for both vibra-
tional quantum numbers have non-negligible contributions.
It should be noted, however, that computed anharmonic
wavefunctions of vibrational levels of the X A1 state of
CF2 with relatively high quantum numbers show contribu-
tions from several different basis functions. Consequently,
vibrational assignments of levels with v1’ and/or v2’, having
values larger than 7, are tentative. Nevertheless, it is clear
Computed Franck–Condon factors and simulated photoelec-
tron spectrum: The computed FC factors (FCFs) of the
!
+
2
ꢀ
1
2
˜
˜
˜
CF2 (X A1)+e
CF2 (X A1) photoionization process ob-
+
tained with Boltzmann vibrational temperatures of 0 and
600 K are shown in Figure 7 (upper and lower spectra in this
Figure, respectively; the computed FCF of the strongest vi-
brational component is set to 100 arbitrary units in all parts
of this figure). The computed FCFs of vibrational “hot”
!
that the (3,v2’,0)
(0,0,0) series is the strongest vibrational
ACTHNUGTENR(NUG 0,0,0) vibrational component
has the largest computed FC factor (see Figure 8).
!
bands arising from ionizations from the (0,1,0), (1,0,0),
progression and the (3,3,0)
1
˜
(0,2,0) and (1,1,0) levels of the X A1 state of CF2 obtained
at 600 K are also shown in Figure 7 (middle spectrum). It
can be seen from this Figure that “hot” band contributions
to the whole simulated spectrum are small. The most notice-
able contributions of “hot” bands are probably in the low
In Figure 9, computed FCFs in the 11.5–12.9 eV IE region
at a Boltzmann vibrational temperature of 600 K and vibra-
tional assignments of some of the relatively strong vibration-
1
˜
al components arising from the (0,0,0) level of the X A1
IE region, relative to ionizations from the (0,0,0) level of
the X A1 state, as has already been discussed in the compar-
state of CF2 are shown. In this IE region, some partially re-
solved components with a doublet structure were observed
1
˜
ison between the simulated PE spectrum and the experi-
mental TPE spectrum.
in the TPE spectrum. Computed FCFs in this region show
!
that these doublet structures are due to (v1’,v2’+2,0)
N
!
Although the vibrational structure of the previously re-
ported UV PE spectrum[18] appears to consist of only one vi-
brational series, our computed FCFs show that a large
number of combination bands are involved. The computed
FCFs of eight major vibrational series obtained at a Boltz-
and (v1’+1,v2’,0)
N
nitude close to n1’.
In Figure 10, computed FC factors and assignments of
some vibrational components in the adiabatic ionization
region are shown. It can been seen from the Figure that
Chem. Eur. J. 2008, 14, 11452 – 11460
ꢁ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
11457

J. M. Dyke et al.
Figure 8. Computed Franck–Condon factors of eight major vibrational
+
Figure 9. Computed
(X A1)+e
Franck–Condon
factors
of
the
CF2
!
+
2
ꢀ
1
˜
˜
progressions at 0 K of the CF2 (X A1)+e
CF2 (X A1) photoionization
!
2
ꢀ
1
˜
˜
CF2 (X A1) photoionization process in the 11.5–12.9 eV IE
process and their vibrational assignments.
region, and vibrational assignments of some vibrational components,
which contribute to some of the doublet structures in the observed
threshold photoelectron spectrum (see text).
there are a number of “hot” band components, namely the
!
!
!
(0,1,0)
(0,1,0), (1,0,0)
G
ACTHUNGTERNNU(G 1,0,0) compo-
nents at computed ionization energies of 11.359, 11.363 and
11.370 eV, respectively, which are very close to the (0,0,0)
(0,0,0) component at 11.362 eV. However, in order to re-
solve the vibrational structure in this congested region, a
better experimental resolution than 5 meV FWHM is re-
quired.
!
An overview of the assignment of the main structure in
the experimental TPE spectrum in the photon energy region
11.3–12.2 eV is shown in Figure 11. As can be seen, the
!
main structure shown corresponds to (v1’,v2’,0)
N
bination bands with v’=0, 1, 2 and 3. The doublet structure
observed in some of these bands arises because 2n2’ is ap-
proximately equal to n1’. On comparing Figure 11 with the
experimental TPE spectrum in the photon energy region
11.4–12.2 eV (Figure 6), and using the information provided
in the simulations in Figures 9 and 10, the doublet structure
observed in the experimental vibrational components can be
assigned. For example, in the first two observed components
at 11.45 and 11.52 eV the doublet structure can be assigned
+
Figure 10. Computed
(X A1)+e
some vibrational components in the AIE region.
Franck–Condon
factors
of
the
CF2
!
2
ꢀ
1
˜
˜
CF2 (X A1) photoionization process and assignments of
Once the main assignments of the structure in the experi-
mental TPE spectrum have been established, values of the
vibrational constants n1’ (sym. stretching mode) and n2’
(sym. bending mode) in the ionic state can be derived. For
n1’, we and wexe were obtained as (1370ꢂ20) and (7ꢂ10)
!
!
to the (0,1,0)
G
ACTHNGUTERNN(UG 0,1,0) ionizations, and
!
!
ACTHNGUTER(NNUG 1,0,0), respectively. Also,
the (0,2,0)
N
the doublet structure in the band at 11.77 eV can be as-
! !
AHCTUNGTREG(NNUN 0,0,0) and (2,1,0) 0,0,0).
signed to the ionizations (1,3,0)
11458
www.chemeurj.org
ꢁ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2008, 14, 11452 – 11460

Difluorocarbene
FULL PAPER
an ab initio/Franck–Condon simulation has allowed the
upper state vibrational quantum numbers associated with
each observed vibrational component to be determined. The
position of the adiabatic component could then be deter-
mined. Also, the harmonic and fundamental vibrational fre-
+
2
˜
quencies for n1’ and n2’ for the ionic state, CF2 (X A1),
have been derived from the experimental TPE spectrum.
These compare well with those determined from the poten-
+
2
˜
tial energy function for CF2 (X A1) computed in this work
with high level electronic structure calculations.
The adiabatic ionization energy of CF2 is recommended
as 11.362ꢂ0.005 eV.
Acknowledgements
Figure 11. A simulated spectrum showing assignment of the main struc-
ture in the low energy region (11.3–12.2 eV) in the first photoelectron
band of CF2. The doublet structure observed in some of the bands arises
because 2n2’ is approximately equal to n1’. More detail of the structure in
The authors are grateful to EPSRC for supporting this work (grant no.
EP/C515706/1), and for the provision of computational resources from
the National Service for Computational Chemistry. This work was also
supported by a grant from the Research Grants Council (RGC) of the
Hong Kong Special Administrative Region (competitive earmarked re-
search grant (CERG) 2007–2008 grant no. 500207). We also thank Dr. N.
Zema and the technical staff of the Polar beamline (4.2R) at Elettra,
Trieste for help and advice. M.E. thanks the EU Early Stage Research
Training Network (SEARCHERS) for financial support.
the band at 11.36 eV is shown in Figure 10. The main structure shown
!
corresponds to (v1’,v2’,0)
N
3.
cmꢀ1. The value for we, compares very well with the value of
1370.0 cmꢀ1 obtained from the CAS/MRCI/aug-cc-pV5Z
PEF while the fundamental separation (1356ꢂ20) cmꢀ1 also
compares very well with the value of 1356.3 cmꢀ1 derived
from the CAS/MRCI/aug-cc-pV5Z PEF (see values listed in
Table 1). The agreement is not so good for n2’. The harmonic
and fundamental values expected from the CAS/MRCI/aug-
cc-pV5Z PEF are 653.2 and 651.7 cmꢀ1 respectively. The cor-
responding we and wexe values obtained from the experi-
mental TPE spectrum are (635ꢂ10) cmꢀ1 and (0ꢂ5) cmꢀ1,
giving the harmonic and fundamental values as (635ꢂ10)
cmꢀ1.
[1] D. S. King, P. K. Schenck, J. C. Stephenson, J. Mol. Spectrosc. 1979,
78, 1.
[2] C. W. Mathews, J. Chem. Phys. 1966, 45, 1068.
[3] C. W. Mathews, J. Chem. Phys. 1967, 46, 2355.
[4] P. Venkateswarlu, Phys. Rev. 1950, 77, 676.
[5] A. Charo, F. C. De Lucia, J. Mol. Spectrosc. 1982, 94, 363.
[6] W. H. Kirchoff, D. R. Lide, Jr., F. X. Powell, J. Mol. Spectrosc. 1973,
47, 491.
[7] H. B. Qian, P. B. Davies, J. Mol. Spectrosc. 1995, 169, 201.
[8] J. B. Burkholder, C. J. Howard, P. A. Hamilton, J. Mol. Spectrosc.
1988, 127, 362.
[9] P. B. Davies, P. A. Hamilton, J. M. Elliott, M. J. Rice, J. Mol. Spec-
trosc. 1983, 102, 193.
This work indicates that studies with TPE spectroscopy,
supported by ab initio/Franck–Condon calculations, of other
tri- and tetra-atomic reactive intermediates, in which vibra-
tional structure in more than one vibrational mode in the
first and higher PE bands is expected, is possible and this is
planned for future work. Possible reactive intermediates
that could be studied in this way include HO2,[43] HNO,[44]
HCO,[45] CH2Cl and CHCl2.[46] In each case, the results
should lead to a more reliable determination of the AIE, as-
signment of the observed vibrational structure and hence
determination of vibrational frequencies in the ionic state.
[10] P. B. Davies, W. Lewis-Bevan, D. K. Russell, J. Chem. Phys. 1981,
75, 5602.
[11] J. P. Booth, G. Cunge, P. Chabert, N. Sadeghi, J. Appl. Phys. 1999,
85, 3097.
[12] B. A. Cruden, K. K. Gleason, H. H. Sawin, J. Appl. Phys. 2001, 89,
915.
[13] J. P. Booth, G. Cunge, F. Neuilly, N. Sadeghi, Plasma Sources Sci.
Technol. 1998, 7, 423.
[14] M. Haverlag, E. Stoffels, W. W. Stoffels, G. M. W. Kroesen, F. J.
de Hoog, J. Vac. Sci. Technol. A 1996, 14, 384.
[15] J. Rçpcke, G. Lombardi, A. Rousseau, P. B. Davies, Plasma Sources
Sci. Technol. 2006, 15, S148.
[16] a) I. M. P. Aarts, B. Hoex, A. H. M. Smets, R. Engeln, W. M. M.
Kesssels, M. C. M. van de Sanden, Appl. Phys. Lett. 2004, 84, 3079;
b) D. Romanini, A. A. Kachanov, N. Sadeghi, F. Stoeckel, Chem.
Phys. Lett. 1997, 264, 316.
[17] J. P. Booth, Plasma Sources Sci. Technol. 1999, 8, 249.
[18] J. M. Dyke, L. Golob, N. Jonathan, A. Morris, M. Okuda, J. C. S.
Faraday 1974, 70, 1828.
Conclusion
In this work, the first photoelectron band of difluorocar-
bene, CF2, has been recorded with threshold photoelectron
(TPE) spectroscopy. Extensive vibrational structure was ob-
tained for this band but the adiabatic component was not
observed. Even though the TPE vibrational envelope is
non-Franck–Condon in nature, comparison of the structure
seen in the vibrational components with that expected from
[19] T. J. Buckley, R. D. Johnson, R. E. Huie, Z. Zhang, S. C. Kuo, R. B.
Klemm, J. Phys. Chem. 1995, 99, 4879.
[20] V. Tarnovsky, K. Becker, J. Chem. Phys. 1993, 98, 7868.
[21] F. Innocenti, M. Eypper, S. Beccacei, A. Morris, S. Stranges, J. B.
West, G. C. King, J. M. Dyke, J. Phys. Chem. 2008, 112, 6939.
[22] F. Innocenti, L. Zuin, M. L. Costa, A. A. Dias, A. Morris, S.
Stranges, J. M. Dyke, J. Chem. Phys. 2007, 126, 154310.
Chem. Eur. J. 2008, 14, 11452 – 11460
ꢁ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
11459

J. M. Dyke et al.
[23] J. M. Dyke, S. D. Gamblin, A. Morris, T. G. Wright, A. E. Wright,
J. B. West, J. Electron Spectrosc. Relat. Phenom. 1998, 97, 5.
[24] J. M. Dyke, S. D. Gamblin, A. Morris, T. G. Wright, A. E. Wright,
J. B. West, J. Phys. B 1999, 32, 2763.
[25] F. Innocenti, L. Zuin, M. L. Costa, A. A. Dias, M. Goubet, A.
Morris, R. I. Oleriu, S. Stranges, J. M. Dyke, Mol. Phys. 2007, 105,
771.
[33] G. K. Jarvis, K. J. Boyle, C. A. Mayhew, R. P. Tuckett, J. Phys. Chem.
A 1998, 102, 3230.
[34] L. Golob, PhD thesis, University of Southampton (UK), 1975.
[35] S. Eden, P. Limao-Vieira, P. A. Kendall, N. J. Mason, J. Delwiche,
M. J. Hubin-Franskin, T. Tanaka, M. Kitajima, H. Tanaka, H. Cho,
S. V. Hoffman, Chem. Phys. 2004, 297, 257.
[36] R. F. Lake, H. Thompson, Proc. R. Soc. Lond. Ser. A 1970, 315, 323.
[37] G. C. King, M. Zubek, P. M. Rutter, F. H. Read, J. Phys. E 1987, 20,
440.
[26] F. T. Chau, D. K. W. Mok, E. P. F. Lee, J. M. Dyke, ChemPhysChem
2005, 6, 2037.
[27] A. Halkier, T. Helgaker, W. Klopper, P. Jorgensen, A. G. Csaszar,
Chem. Phys. Lett. 1999, 310, 385.
[38] R. I. Hall, A. McConkey, K. Ellis, G. Dawber, L. Avaldi, M. A. Mac-
Donald, G. C. King, Meas. Sci. Technol. 1992, 3, 316.
[39] P. M. Guyon, R. Spohr, W. A. Chupka, J. Berkowitz, J. Chem. Phys.
1976, 65, 1650.
[40] T. Baer, P. M. Guyon, J. Chem. Phys. 1986, 85, 4765.
[41] Z. L. Cai, Theor. Chim. Acta 1993, 86, 249.
[42] K. Sendt, G. B. Bacskay, J. Chem. Phys. 2000, 112, 2227.
[43] J. M. Dyke, A. Morris, N. Jonathan, M. J. Winter, Mol. Phys. 1981,
44, 1059.
[44] J. Baker, V. Butcher, J. M. Dyke, A. Morris, J. C. S. Faraday Trans.
1990, 86, 3843.
[45] J. M. Dyke, A. Morris, N. Jonathan, M. J. Winter, Mol. Phys. 1981,
44, 1059.
[28] J. Wormhoudt, K. E. McCurdy, J. B. Burkholder, Chem. Phys. Lett.
1989, 158, 480; C. W. Mathews, Can. J. Phys. 1967, 45, 2355; O. Suto,
J. Steinfeld, Chem. Phys. Lett. 1990, 168, 181 (Data from NIST Stan-
dard Reference Database 69, June 2005 Release: NIST Chemistry
WebBook; http://webbook.nist.gov/chemistry/).
[29] J. K. G. Watson, Mol. Phys. 1968, 15, 479.
[30] D. K. W. Mok, E. P. F. Lee, F. T. Chau, J. M. Dyke, J. Chem. Phys.
2004, 120, 1292.
[31] L. Margules, J. Demaison, J. E. Boggs, J. Phys. Chem. A 1999, 103,
7632.
[32] MOLPRO is a package of ab initio programs written by H.-J.
Werner, P. J. Knowles, R. Lindh, F. R. Manby, M. Schꢄtz, P. Celani,
T. Korona, G. Rauhut, R. D. Amos, A. Bernhardsson, A. Berning,
D. L. Cooper, M. J. O. Deegan, A. J. Dobbyn, F. Eckert, C. Hampel,
G. Hetzer, A. W. Lloyd, S. J. McNicholas, W. Meyer, M. E. Mura, A.
Nicklaß, P. Palmieri, R. Pitzer, U. Schumann, H. Stoll, A. J. Stone,
R. Tarroni, T. Thorsteinsson, 2008.
[46] L. Andrews, J. M. Dyke, N. Jonathan, N. Keddar, A. Morris, J. Am.
Chem. Soc. 1984, 106, 299.
Received: August 15, 2008
Published online: November 12, 2008
11460
www.chemeurj.org
ꢁ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2008, 14, 11452 – 11460