Electronic excitation of tetrafluoroethylene, C2F4

Chemical Physics 297 (2004) 257–269 www.elsevier.com/locate/chemphys

Electronic excitation of tetrafluoroethylene, C2F4 S. Eden a,*, P. Lim~ ao-Vieira a,1, P.A. Kendall a, N.J. Mason a,2, J. Delwiche b, M.-J. Hubin-Franskin c,3, T. Tanaka d, M. Kitajima d, H. Tanaka d, H. Cho e, S.V. Hoffmann f a Department of Physics and Astronomy, University College London, Gower Street, London WC1E 6BT, UK Laboratoire Thermodynamique et Spectroscopie, Universit
e de Li ege, Institut de Chimie-B^ at. B6C, B-4000 Li ege, Belgium Laboratoire de Spectroscopie dÕElectrons Diffus
es, Universit
e de Li ege, Institut de Chimie-B^ at. B6C, B-4000 Li ege, Belgium d Department of Physics, Sophia University, Chyoicho 7-1, Chiyoda-ku, Tokyo 102-8854, Japan e Department of Physics, Chungnam National University, Daejon, Republic of Korea f Institute of Storage Rings, University of Aarhus, Ny Munkgade, Aarhus, Denmark

b c

Received 8 July 2003; accepted 24 October 2003

Abstract The VUV photoabsorption spectrum of tetrafluoroethylene is reported in the wavelength range; 115–320 nm (10.8–3.9 eV). The result is compared with high-energy electron energy loss spectra (EELS). Electrons of incident energy 100 and 30 eV are analysed post-interaction at scattering angles of 5°, 15° and 30° in the energy loss range; 3.0–15.7 eV. Both the photoabsorption and EELS measurements represent the highest resolution data yet reported. The spectra are found to be in good agreement with each other and generally consistent with previous experimental and theoretical work. New photoabsorption features are observed and Rydberg and vibrational assignments suggested. The He(I) photoelectron spectrum is recorded from 9.7 to 20.9 eV and compared to earlier work. New vibrational structure is observed in the first photoelectron band. Implications relevant to the production of CF2 radicals by photon and electron impact are discussed. Ó 2003 Elsevier B.V. All rights reserved. Keywords: Tetrafluoroethylene; C2 F4 ; Photoabsorption; Synchrotron radiation; Electron energy loss spectroscopy; Electronic and vibrational excitation; Rydberg series; Photoelectron spectroscopy

1. Introduction Tetrafluoroethylene, also known as TFE, tetrafluoroethylene, perfluoroethylene, perfluoroethene, and fluoroplast 4, is an industrial gas with a broad range of applications, most importantly in the field of technological plasmas. It is formed within plasma processing cells by electron and photon impact dissociation or *

Corresponding author. Tel.: +44-207-679-3436; fax: +44-207-6793460. E-mail address: [email protected] (S. Eden). 1 Also of Departamento de F
ısica, FCT – Universidade Nova de Lisboa, P-2829-516 Caparica, Portugal and Centro de F
ısica Molecular, Complexo I, IST, Av. Rovisco Pais, P-1049-001 Lisboa, Portugal. 2 Also of Department of Physics and Astronomy, Open University, Walton Hall, Milton Keynes MK7 6AA, UK. 3 Directeur de recherche FNRS. 0301-0104/$ - see front matter Ó 2003 Elsevier B.V. All rights reserved. doi:10.1016/j.chemphys.2003.10.031

thermal decomposition of c-C4 F8 , a feed gas used for oxide etching [1,2]. Furthermore, C2 F4 has attracted considerable interest as an alternative feed gas for the plasma etching of silicon dioxide as it dissociates readily to form the etching fragments; CF2 and CFþ 2 [3,4]. SiO2 etching is traditionally performed using CF4 , C2 F6 , C3 F8 , CHF3 , and c-C4 F8 . These species have high global warming potentials (GWP) as they absorb strongly in the infrared and have very long residence times in the EarthÕs atmosphere. CF4 , for example, remains for up to 50 000 years [5]. The current generation of plasma reactors release a high proportion of feed gas into the atmosphere. Tetrafluoroethylene is considered to have negligible GWP due to its high reactivity with OH radicals. Acerboni et al. [6] calculate the global and yearly averaged atmospheric lifetime of the molecule to be just 1.9 days. Analysis of the reactions between

258

S. Eden et al. / Chemical Physics 297 (2004) 257–269

tetrafluoroethylene and ozone has been published by Toby and Toby [7]. However, comparison of the moleculeÕs residence time with atmospheric transport mechanisms indicates that C2 F4 released at the EarthÕs surface cannot play a significant role in the depletion of the ozone layer. It has been proposed that by using a combination of CF3 I and C2 F4 in a plasma reactor, highly efficient and selective SiO2 etching can be achieved by independently controlling etch rate and (CF2 )n polymerization on the sample surface through ðþÞ ðþÞ the relative concentration of CF3 and CF2 fragments [8]. Due to the weak C–I bond, CF3 I is an excellent source of CF3 within a discharge [9]. CF3 I is considered to have a short atmospheric lifetime due to UV photolysis and thus low GWP [10,11]. Tetrafluoroethylene has been the subject of a number of experimental and theoretical studies. Winstead and Mckoy [12] report detailed theoretical analysis of elastic and inelastic low-energy electron interactions with C2 F4 . Yoshida et al. [13] present extensive work including measurements of partial ionisation cross sections and swarm analysis to calculate further electron impact cross sections and dissociation energetics. Photoelectron spectra for tetrafluoroethylene have been recorded on a number of occasions [14–18]. However, experimental electron scattering data is limited to relatively low resolution spectra measured by Coggiola et al. [19]. Similarly, the only published photoabsorption result comparable with the present spectrum is the apparently lower resolution spectrum of B
elanger and Sandorfy [20]. The present paper provides important new high-resolution quantitative data necessary to model the behaviour of C2 F4 in future plasma reactors.

2. Brief summary of the structure and properties of C2 F4 Tetrafluoroethylene is planar in the electronic ground state and has symmetry D2h [21]. Symmetry species available to a D2h molecule are Ag , Au , B1g , B1u , B2g , B2u , B3g and B3u [22]. The angles between the two geminal C–F bonds have been found experimentally to be 112.48° [23], close to the tetrahedral angle. The molecule has 12 modes of vibration in the neutral ground state [24]. Excitation of m1 corresponds to C@C stretching while all other modes are associated with motion around C–F bonds. The least bound molecular orbital in the neutral ground state is identified as being of C@C p bonding character [14]. The lowest unoccupied MO is the conjugate p
C@C antibonding orbital [13]. Brundle et al. [15] draw attention to the fact that substitution of hydrogen by fluorine atoms in hydrocarbons does not affect the ionisation from p orbitals but shifts those from r orbitals to higher energies. This r orbital stabilisation is known as the perfluoro-effect

and is observed clearly in the comparison of the photoelectron spectra of C2 F4 and C2 H4 .

3. Experimental 3.1. Photoelectron spectroscopy He(I) (21.22 eV) photoelectron spectra of C2 F4 were taken at the Universit
e de Liege, Belgium. The apparatus has been described in detail previously [25]. Briefly, the spectrometer consists of a 180° cylindrical electrostatic analyser with a mean radius of 5 cm. The analyser is used in constant energy pass mode and is fitted with a channeltron detector. The incident photons are produced by a DC discharge in a two-stage differentially pumped lamp. The energy scale is calibrated using xenon (2 P3=2 ¼ 12.130 eV and 2 P1=2 ¼ 13.436 eV) [26] and the resolution of the present spectrum is measured from the full width half maximum of the Xe peaks to be 22 meV, in the presence of C2 F4 . The C2 F4 sample was purchased from ABCR GmbH & Co. KG and is quoted as having a minimum purity of 99%. There is, however, clear evidence for CO2 contamination. Therefore, a He(I) photoelectron spectrum of CO2 was taken to accompany the results. The CO2 sample was purchased from LÕAir Liquide Benelux S.A. and has a minimum purity of 99.995%. 3.2. Electron energy loss spectroscopy The present EEL spectrum was recorded at Sofia University, Tokyo. The experimental arrangement and procedure are described by Tanaka et al. [27]. Monochromatic electrons intercept an effusive gas beam at right angles and are analysed post-interaction in a 180° hemispherical system. The analyser can be rotated around the scattering centre covering an angular range from )10° to 130° with respect to the incident electron beam. The overall energy resolution is determined by comparison with the elastic scattering peak of He to be 30–35 meV, and the estimated angular precision is ±1.5°. The energy scale was calibrated with respect to the 19.36 eV He resonance peak. The sample used for the EEL experiments was purchased from SynQuest Lab and has a minimum purity of 99%. The gas was introduced without further purification or treatment. 3.3. VUV photoabsorption Photoabsorption measurements were made at the ASTRID facility, Aarhus University, Denmark. The experimental apparatus is described elsewhere [28]. Synchrotron radiation is passed through a static gas sample. A photo-multiplier is used to measure the

S. Eden et al. / Chemical Physics 297 (2004) 257–269

transmitted light intensity at 0.05 nm intervals and wavelength is selected using a toroidal dispersion grating. For wavelengths below 200 nm, He is flushed through the small gap between the photomultiplier and the exit window of the gas cell to prevent any absorption by air contributing to the spectrum. The LiF entrance window cuts out higher order radiation before it can enter the cell. At longer wavelengths, absorption by the air in the gap removes all higher order radiation. The minimum and maximum wavelengths between which scans are performed, 115–320 nm (10.8–3.9 eV), are determined by the transmission windows of the gas cell and the grating energy range, respectively. The sample pressure is measured using a baratron capacitance manometer and varied to give maximum absorption whilst ensuring that the transmitted signal never falls to zero (saturation). The synchrotron beam ring current is monitored throughout the collection of each spectrum. Results are compared to a background scan recorded with the cell evacuated. Absolute photoabsorption cross sections are generated using the Beer–Lambert law: It ¼ I0 expðnrxÞ;

cause its broad nature minimises the affect of differences in energy resolution [31]. The experimental full width half maximum resolution for the present results may be taken to be 0.075 nm, corresponding to 3 meV at the midpoint of the energy range studied. The sample gas was purchased from ABCR GmbH & Co. KG and has a minimum purity of 99%. 4. Results and discussion 4.1. Ionic states 4.1.1. Photoelectron results The photoelectron intensity was measured in steps of 1 meV from 9.74 to 20.85 eV. The He(I) photoelectron spectrum of tetrafluoroethylene has been reported previously by Lake and Thompson [14], Brundle et al. [15], and by Sell and Kuppermann [16]. Bieri et al. [17] report a He(II) (40.8 eV) spectrum accompanied by many body Green function calculations. An energy resolution of 20 meV is given by the authors of each of these works apart from Brundle et al. [15] who claim 15–30 meV. Jarvis et al. [18] report a threshold photoelectron spectrum for C2 F4 ionised using a synchrotron VUV source. Although the optical resolution is high for the threshold measurement, the ionisation energies are given to the same precision as in the He(I) and He(II) works. The papers are generally in good agreement with each other in terms of the energy position and structure of the bands.

ð1Þ

where It is the radiation intensity transmitted through the gas sample, I0 is that through the evacuated cell, n the molecular number density of the sample gas, r the absolute photoabsorption cross section, and x the absorption path length (25 cm). The apparatus is calibrated precisely before each measurement. SO2 is used to calibrate the energy scale since it has clearly defined sets of sharp absorption peaks from 3.8 to 5.1 eV [29] and from 5.15 to 7.25 eV [30]. The Schuman–Runge (6.9–9.5 eV) absorption band of O2 is used to calibrate the absolute cross section beυ1

3υ1

4.1.2. The ionic ground state The first band, corresponding to the ionic electronic ground state of tetrafluoroethylene, is shown in Fig. 1. The original spectrum has been treated using the 4υ1

5υ1

6υ1

υ11 υ11 υ 11 υ11 υ11 υ11 υ2 υ2 υ2 υ2 υ2 υ2 υ6 υ6 υ6 υ6 υ6 υ6 υ6

7υ1

8υ1 Non-deconvoluted Deconvoluted

Intensity (a.u.)

υ0

2υ1

259

9. 5

10

10.5

11

11.5

12

12. 5

Ionisation Energy(eV) Fig. 1. The first photoelectron band of C2 F4 , assigned to ionisation from the least bound C@C p orbital, 2b3u .

260

S. Eden et al. / Chemical Physics 297 (2004) 257–269

de-convolution technique first developed by Van Cittert [32] and improved by Allen and Grimm [33]. The band is assigned to ionisation from the weakest bound C@C p orbital [14–18]. This orbital is labelled 2b3u following the analysis of Bieri et al. [17]. However, it should be noted that Winstead and McKoy [12] prefer 1b3u while Sell and Kuppermann [16] give the orbital as b3u . Brundle et al. [15] assign the band to ionisation from the 2b2u MO. Band I is considered to begin at 10.12 eV, consistent with the adiabatic ionisation potentials measured by Lake and Thompson [14], Brundle et al. [15], and Bieri et al. [17]. The photo-ionisation work of Buckley et al. [34] gives the first ionisation energy to be 10.114 ± 0.010 eV. The present spectrum, however, shows markedly clearer definition in the first bandÕs vibrational structure than in previous spectra, even before de-convolution is applied. Energy positions and vibrational assignments of the observed peaks are given in Table 1. Comparison with the neutral ground state vibrational energy levels shows that the dominant series of peaks

Table 1 Energy positions and vibrational analysis of features observed in the first photoelectron band of C2 F4 , assigned to ionisation from the least bound C@C p orbital, labelled 2b3u [17] Peak energy in eV

Analysis

10.12 10.17 10.22 10.27 10.33 10.38 10.43 10.48 10.54 10.59 10.64 10.69 10.75 10.80 10.85 10.90 10.95 11.01 11.06 11.10 11.16 11.22 11.27 11.31 11.37 11.43 11.57 11.77

m00 , adiabatic IP m00 þ m6 m00 þ m2 m00 þ m11 m1 m1 þ m6 m1 þ m2 m1 þ m11 2m1 , vertical IP 2m1 þ m6 2m1 þ m2 2m1 þ m11 3m1 3m1 þ m6 3m1 þ m2 3m1 þ m11 4m1 4m1 þ m6 4m1 þ m2 4m1 þ m11 5m1 5m1 þ m6 5m1 þ m2 5m1 þ m11 6m1 6m1 þ m6 7m1 8m1

The m1 (C@C stretching) and m2 (CF2 s-stretching, symmetry ag ) series are assigned in agreement with previous work [14–16,18]. The series attributed in the present work to m6 excitation (CF2 rocking) has previously been assigned to CF2 scissor motion [15,16,18]. The peaks attributed to m11 excitation (CF2 s-stretching, symmetry b3u ) are observed for the first time in the present work.

observed in Fig. 1 can only be consistent with the m1 mode, C@C stretching [24] of symmetry ag . The average energy difference observed between peaks of this series is 0.208 eV, compared to 0.232 eV for the neutral ground state. C@C stretching in this band is identified by Lake and Thompson [14], Brundle et al. [15], Sell and Kuppermann [16], and by Jarvis et al. [18]. These works also give evidence for m2 motion, CF2 stretching of symmetry ag . The energy required to excite m2 vibration in the neutral molecule is 0.096 eV [24]. We observe features following each major (m1 ) peak with an average energy difference of 0.103 eV. Despite the apparently large observed vibrational excitation energy, we also assign the structure to m2 motion. Brundle et al. [15] report that while the least bound p MO is strongly C@C bonding it is also C–F anti-bonding. Thus, vibrations involving change of the inter-nuclear distance between C and F will tend take place at higher energies in the ionic ground state. Brundle et al. [15], Sell and Kuppermann [16], and Jarvis et al. [18] report the presence of a third vibrational series corresponding to CF2 scissor motion with vibrational excitation energy 0.046, 0.050, and 0.046 eV, respectively. Shimanouchi [24] gives the neutral ground state excitation energy of the m3 mode as 0.049 eV. We observe peaks following the m1 series with an average energy difference of 0.057 eV. The explanation for higher energy excitation in the first ionic state proposed by Brundle et al. [15] would not seem to extend to scissor motion so we propose these features as corresponding to m6 excitation, symmetric CF2 rocking which occurs at 0.068 eV in the neutral ground state [24]. A fourth vibrational series is observed in the present work for the first time. The average energy difference from the peaks of the m1 series is 0.152 eV. This is proposed to be consistent with excitation of m11 , CF2 stretching of symmetry b1u . Excitation of the m11 mode occurs at 0.147 eV in the neutral ground state [24]. The high excitation energy of the mode in the ionic ground state can be rationalised following the argument given above for CF2 stretching of symmetry ag [15]. A weak feature at 10.06 eV is assigned to a hot band (m6 or m8 , CF2 rocking or wagging). Yoshida et al. [13] estimate that about 20% of C2 F4 molecules are vibrationally excited at 300 K. A further small peak observed at 11.92 eV appears inconsistent with any C2 F4 vibrational mode and cannot be due to carbon dioxide contamination [35]. The feature remains unassigned. 4.1.3. Ionic excited states The energies of each of the ionic excited bands evident in the present work are in good agreement with previous spectra. However, variation in the relative intensity of the excited ionisation bands is observed between the He(I) results of Lake and Thompson [14], Brundle et al. [15], and Sell and Kupperman [16]. The

S. Eden et al. / Chemical Physics 297 (2004) 257–269

carbon dioxide tetrafluoroethylene

Band III

Band IV

Band V

nυ6

261

Band VI

nυ2

Intensity (a.u.)

Band II

15

16

17

18

19

20

21

Ionisation energy (eV) Fig. 2. C2 F4 He(I) photoelectron spectrum for excited ionisation bands accompanied by the He(I) spectrum of CO2 which is present as a contaminant.

present spectrum is corrected for the variation in the transmission efficiency of the analyser with electron energy and shows the fourth, fifth and sixth ionisation bands to be stronger relative to the second and third than measured in the earlier works. Furthermore, we disagree with the previous assignment [14,15,18] of the vibrational structure observed in the fourth ionisation band. Fig. 2 shows the bands associated with the ionic excited states of C2 F4 . The sample contamination has a significant effect within this wavelength range as it coincides with the second, third and fourth ionisation bands of CO2 [35]. Therefore, a carbon dioxide result measured at intervals of 4 meV accompanies the tetrafluoroethylene spectrum in Fig. 2. The energy positions of the CO2 features match those reported in the high-resolution work of Baltzer et al. [36] to within the experimental errors. The intensity scales of the two plots in Fig. 2 are chosen in order that the height of the major CO2 peak at 18.08 eV approximates that above the band V continuum of the corresponding peak in the C2 F4 measurement. In a number of cases, vibrational C2 F4 features are considered to be partially hidden by the CO2 contamination. Naturally, the assignment of such features is subject to greater uncertainty, as is the energy at which they occur. The energy positions of the observed features and the corresponding assignments are summarised in Table 2. The band maxima are assigned in agreement with the analysis of Bieri et al. [17]. The diffuse peak at 15.6 eV is assigned to the adiabatic ionisation energy of band II in agreement with Brundle et al. [15]. The ionisation is associated with the

removal of an electron from the 3b3g r orbital [17]. As observed by Brundle et al. [15], the energy difference of 5.5 eV between the adiabatic IP of the first (p) ionisation band and that of the second (r) is unusually large for C2 F4 ; a good example of the perfluoro-effect. The intense third band is considered to be due to the ionisation from four MOs; 4ag , 3b2u , 1au (p), and 1b1g (p), calculated by Bieri et al. [17] to occur within a narrow energy range. The fact that changes of ordering may occur between very close lying ionisation energies dissuade us from assigning the shoulder observed at 16.4 eV beyond proposing it to be due to one of the four orbitals listed above. The same argument applies to the non-assignment of the shoulder at 16.9 eV in Table 2. Jarvis et al. [18] comment that band III may be associated with electron removal from the delocalised p levels of the F atoms. Band IV is assigned to ionisation from the 3b1u r MO [17]. The band has previously been observed to include a vibrational series attributed to C–F stretching [14,15,18]. Lake and Thompson [14] and Brundle et al. [15] report CF2 scissor motion coupled to each C–F stretching excitation. The corresponding ground state excitation energies for C–F stretching and CF2 scissor motion are 0.096 and 0.049 eV, respectively [24]. The average spacing of the peaks superimposed on the fourth ionisation band in the present work is 0.049 eV, leading to an excitation energy for a C–F stretching series of 0.098 eV. Brundle et al. [15] report that, with the exception of the orbital associated with band III, all the r MOs have a small C–F bonding overlap population. Therefore, following the argument supporting the high excitation energy of the CF2 stretching mode in band I, the m2

262

S. Eden et al. / Chemical Physics 297 (2004) 257–269

Table 2 Energy positions and analysis of features observed in the excited photoelectron bands of C2 F4 C2 F4 sample 15.6D 15.93 16.4D 16.64 16.9D 17.32 17.5 17.52 17.57 17.6 17.66 17.71 17.7 17.80 17.87 17.94 18.02 18.08 18.16 18.2 19.20 19.29 19.4 19.5 19.6 19.6D 19.7D

CO2

Previously observed in C2 F4 photoelectron studies

Generalised analysis

C2 F4 assignment

15.6Br 15.93L , 15.95Br; S , 15.9Bi

Band Band Band Band Band CO2 Band Band Band Band Band Band Band Band CO2 Band CO2 CO2 CO2 Band Band Band Band Band Band Band Band

3b3g adiabatic IP 3b3g vertical IP Unassigned 4ag , 3b2u , 1au (p), 1b1g (p) vertical IP Unassigned – 3b1u adiabatic IP, m00 m6 2m6 3b1u vertical IP, 3m6 4m6 5m6 6m6 7m6 – Unassigned – – – 1b2g (p) vertical IP 2b3g adiabatic IP, m00 m2 2b3g , 1b3u (p) vertical IP, 2m2 3m2 4m2 5m2 6m2

16.64L , 16.60Br , 16.63S , 16.6Bi 17.32 17.46

17.50Br , 17.51Bi

17.60

17.57L , 17.60Br; S , 17.6Bi , 17.57J

17.74 17.87 18.01 18.08 18.15 18.24

19.40 19.48 19.57

18.45L , 18.21Br , 18.22S , 18.2Bi 19.19Br; Bi 19.41L , 19.46Br , 19.4Bi , 19.41J

II II max. III III max. III IV + CO2 contribution IV IV IV max. + CO2 contribution IV IV IV + CO2 contribution IV IV

V max. + CO2 contribution VI VI VI max. + CO2 contribution VI + CO2 contribution VI + CO2 contribution VI VI

All energies are given in eV. D Indicates that the feature is diffuse and thus its energy position is subject to greater uncertainty. The energies of C2 F4 features proposed to coincide with CO2 peaks are also given to lower precision. Energies of C2 F4 features are compared to those noted by L Lake and Thompson [14], Br Brundle et al. [15], S Sell and Kuppermann [16], Bi Bieri et al. [17], and J Jarvis et al. [18]. Ionisation band maxima are assigned in agreement with Bieri et al. [17]. The nm2 series (CF2 stretching) is assigned to band VI in agreement with Lake and Thompson [14], Brundle et al. [15], and Jarvis et al. [18]. The band IV nm6 series (CF2 rocking) is proposed for the first time in the present work.

excitation energy should be significantly lower in the fourth ionisation band than in the neutral ground state. The band IV vibrational series is thus assigned in the present work to m6 excitation, symmetric CF2 rocking. The neutral ground state excitation energy of this mode is 0.068 eV [24]. The small peak at 17.94 eV remains unassigned. Band V is assigned to ionisation from the 1b2g (p) orbital [17] and includes no evidence for vibrational structure. The sixth band, however, shows a clear series beginning at 19.20 eV with an average energy difference of 0.088 eV between peaks. The structure is assigned to m2 excitation, C–F stretching, in agreement with Lake and Thompson [14], Brundle et al. [15], and Jarvis et al. [18]. Bieri et al. [17] propose the band maximum to correspond to the energy for vertical ionisation from the closely spaced 2b3g (r) and 1b3u (p) orbitals. The adiabatic ionisation energy, and thus the m2 series, is considered to correspond to the 2b3g MO. This assignment is supported by the fact that the energy difference between C–F stretching peaks is less than 0.096 eV, the neutral ground state excitation energy [24]. As discussed above, C–F stretching is expected to occur at lower

energies in C2 Fþ 4 ionised by promotion of an electron from a r MO than from the weakest bound p MO. 4.2. Valence and Rydberg states of C2 F4 4.2.1. Electron scattering and photoabsorption analysis Electron energy loss spectra (EELS) were recorded at incident electron energies of 100 and 30 eV, for scattering angles of 5° and 15°, respectively. The scattered electron intensity was measured in steps of 20 meV from 3.0 to 15.7 eV. A third spectrum was recorded at an incident energy of 30 eV with scattered electrons collected at 30° over the energy loss range 3.0–8.8 eV. At scattering angles larger than 30°, the scattered electron intensity was found to be very weak. Coggiola et al. [19] report electron energy loss results with a maximum resolution of 60 meV for incident electrons of energy 60 and 40 eV, for scattering angles of 0° and 60°, respectively. The high-energy electron scattering spectrum covers energy loss from 6 to 16 eV and the low-energy from 3 to 10 eV. We are not aware of any other EELS study of tetrafluoroethylene.

S. Eden et al. / Chemical Physics 297 (2004) 257–269

The absolute photoabsorption cross section is reported from 3.9 to 10.8 eV. The VUV spectrum has been measured previously by B
elanger and Sandorfy [20] from 5.6 to 10.8 eV with a resolution described as moderate. Though not given explicitly, an indication of the resolution of B
elanger and Sandorfy [20] lies in the fact that wavelength was selected using a grating of 1200 lines per mm compared to 2000 at the ASTRID facility [28]. As expected, comparison of the two spectra shows the present work to give clearer peak definition. Furthermore, a significant difference in cross section is observed in the high energy part of the spectral range. As far as we are aware, the only other comparable photoabsorption result for C2 F4 is that of Sharpe et al. [37] measured from 5.9 to 6.7 eV at intervals of 1 nm (approximately 30 meV over this range). As discussed in Section 4.1.2, the tetrafluoroethylene sample used for the present spectrum includes a small percentage of carbon dioxide contaminant. However, comparison with previous VUV absorption results for CO2 [38,39] shows that the absorption of CO2 is very weak compared to C2 F4 in the present energy range. No features observed in the present spectrum are assigned to CO2 transitions. It should be noted that the assignment of features observed from 7.5 to 10 eV is complicated by the overlap of the p ! r
band, the p ! p
, singlet ! singlet band, and the region associated with Rydberg states. 4.2.2. The p ! p
, singlet ! triplet transition The 40 eV, 60° electron scattering result of Coggiola et al. [19] shows a broad feature from 3.6 to 5.6 eV with a maximum at 4.68 eV. The band is assigned to the p ! p


singlet to triplet anti-bonding transition and the plot of DCS against scattering angle confirms that the excitation is dipole forbidden [19]. The transition is clearly visible in the 30 eV, 30° plot shown in Fig. 3. The maximum measured in the present work occurs at 4.79 eV. The single-excitation-configuration-interaction (SECI) calculation of Winstead and Mckoy [12] gives the excitation energy of the transition as 4.48 eV. As expected for a dipole forbidden excitation, the band intensity is weak for electrons analysed at 30° compared to the high scattering angle measurement of Coggiola et al. [19]. No evidence for the band is observed in the present 30 eV, 15° or 100 eV, 5° spectra. The latter result can be considered to simulate photon interaction. No photoabsorption evidence is found for the p ! p
singlet to triplet band. It could be expected to see a very weak optical contribution from the dipole forbidden transition at high pressure. However, despite scanning the energy region at a sample pressure of 1.3 mbar (the highest that the baratron can measure), no absorption is observed. The cross section can be measured to an estimated sensitivity of ±0.005 Mb so the optical transition must have an absorption cross section of less than 0.01 Mb. The low signal to noise ratio apparent for this electron scattering feature means that vibrational structure cannot be identified. Due to the anti-bonding nature of the excited state, any such superimposed features are expected to be weak. 4.2.3. The p ! r
transition Fig. 4 shows the energy range over which non-zero photoabsorption cross sections are observed. The band

(π→π *) singlet→singlet

(π→π*) singlet→triplet

263

30eV, 30 deg (30eV, 30 deg) x 15 30eV, 15 deg

(π→σ *)

100eV, 5 deg

Intensity (a.u.)

4sσ 4pλ

5sσ 3dπ 3pλ

2

4

6

8

10

12

14

Energy Loss (eV) Fig. 3. C2 F4 electron energy loss spectrum (EELS) including labelled valence excitations.

16

264

S. Eden et al. / Chemical Physics 297 (2004) 257–269

extending from 7.3 to 7.9 eV with a maximum of 3.94 Mb observed at 7.78 eV corresponds to the lowest energy valence excitation identified in the present work. The diffuse peak is also clearly visible in the 100 eV, 5° and 30 eV, 30° EEL spectra with a maximum at 7.7 eV and is associated with the p ! r
, singlet ! singlet transition [21]. Coggiola et al. [19] give the excitation energy of the band as 7.7 eV. Winstead and Mckoy [12] calculate the lowest energy dipole allowed valence transition to occur at 7.74 eV and identify dipole forbidden states labelled 13 B3u and 11 B1g at 7.17 and 7.55 eV, respectively. Yoshida et al. [13] report that states at energies between the p ! p
transitions contribute to the electron scattering cross section. These optically

forbidden states may be responsible for the slightly higher relative scattered electron intensity observed in the 30 eV, 30° result than in the 30 eV, 15° spectrum. The argument may be extended to explain the high intensity of the 6.1–7.4 eV feature as a proportion of the excitation from 8 to 10 eV observed for electron scattering compared to photoabsorption. Close inspection of the p ! r
photoabsorption band suggests the presence of a weak vibrational series. The relevant structure is shown in Fig. 5 and is considered to be consistent with excitation of the m2 mode corresponding to CF2 stretching. Table 3 gives the energy positions of features assigned to the series. The average energy difference between features is 0.101 eV, slightly

40

(π→π*) singlet→singlet nν1

35

3dδ

25

-18

2

σ (Mb, 10 cm )

30

20

Adiabatic IP

15

10 nν1

(π→σ*)

3dπ

5

3pλ 3sσ

nν 2

0 6

7

8

9

10

11

Incident Energy (eV)

Fig. 4. High-resolution photoabsorption spectrum of C2 F4 including labelled vibrational series, valence excitations and Rydberg transitions.

Absolute cross section

40

Detail ((cross section x 50) - 60)

nν2 35

(π→σ*)

(π→π*) singlet→singlet nν1 ν6

ν2

ν11

ν6

ν2

ν11

ν6

25

-18

2

σ (Mb, 10 cm )

30

20

15

10 npλ 3

npλ 3

5

ν2

nsσ 0 7.5

7.6

7.7

7.8

7.9

8

8.1

8.2

8.3

8.4

8.5

Incident Energy (eV)

Fig. 5. C2 F4 photoabsorption from 7.5 to 8.5 eV showing mixing of valence and Rydberg structure.

S. Eden et al. / Chemical Physics 297 (2004) 257–269 Table 3 Energy positions of vibrational structure associated with the p ! r
, singlet ! singlet transition Photo-absorption energy in eV D

7.57 7.677 7.778 7.877 7.978 8.077 8.178 8.277

Assignment m00 m2 2m2 band maximum 3m2 4m2 5m2 6m2 7m2

The band maximum is given by Coggiola et al. [19] to be 7.7 eV. Vibrational analysis is proposed for the first time in the present work. D Indicates that the feature is diffuse and thus its energy position is subject to greater uncertainty.

higher than 0.096 eV, the excitation energy observed in the neutral ground state [24]. Brundle et al. [15] stress that the weakest bound C@C p bonding MO has a strongly C–F antibonding overlap population. Therefore, when the orbital is vacated, we expect the carbon–fluorine interatomic distance to be reduced, causing the excitation energy for CF2 stretching to occur at higher energy. The weak series and the lack of evidence for C@C vibration suggests that the molecule may be unstable in this excited state and dissociate to form CF2 radicals. 4.2.4. The p ! p
, single ! singlet transition The photoabsorption cross section between 8 and 10 eV is shown in Fig. 6. The valence band maximum corresponding to the p ! p
singlet to singlet transition for C2 F4 is reported by B
elanger and Sandorfy [20] and by Coggiola et al. [19] to occur at 8.89 and 8.84 eV, respectively. The maximum absorption observed in the present photoabsorption spectrum is 31.72 Mb and

265

corresponds to the sharp peak at 9.017 eV. The strongest intensity of scattered electrons in the EEL measurements occurs at 8.63 eV. The difference in energy corresponding to the peak cross sections is considered to be due the relative energy resolution of the results. The SECI calculation gives 9.71 eV [12], in poor agreement with the present data. The photoabsorption spectrum shows evidence for a m1 , C@C stretching series in the p ! p
singlet to singlet band. The energy positions of assigned features are listed in Table 4 and the average energy difference between m1 peaks is 0.201 eV. Further vibrations are considered to couple with the m1 series and are assigned to the m2 (CF2 s-stretching, symmetry ag ), m6 (CF2 rocking), and m11 (CF2 s-stretching, symmetry b3u ) modes, with average energy difference between features and the previous m1 peak of 0.084, 0.069, and 0.137 eV, respectively. The corresponding excitation energies in the neutral ground state are 0.232 eV (m1 ), 0.096 eV (m2 ), 0.068 eV (m6 ), and 0.147 eV (m11 ) [24]. C@C stretching can naturally be expected to occur at a lower energy in the p
antibonding state. The low excitation energies observed for CF2 stretching of symmetry ag and b3u may be associated with a C–F anti-bonding population in the p
MO countering the effect of the vacation of the p orbital discussed in Section 4.2.3. The m11 mode is anti-symmetric and thus would be expected to show very weak cross section as a single excitation [40]. However, two quanta excitation or combination with another mode can increase the probability. It is worth noting that the observed vibrational excitation energies of the CF2 stretching modes differ from those observed in the ionic ground state (Section 4.1.2) significantly and so are not associated with Rydberg transitions converging to the first ionisation potential.

Fig. 6. C2 F4 photoabsorption from 8 to 10 eV showing mixing of valence and Rydberg structure.

266

S. Eden et al. / Chemical Physics 297 (2004) 257–269

Table 4 Energy positions of vibrational structure associated with the p ! p
, singlet ! singlet transition Photo-absorption energy in eV

Assignment

8.035 8.10D 8.114 8.16D 8.227 8.299 8.310 8.360 8.421 8.49D 8.504 8.56R 8.616 8.688 8.706 8.768 8.812 8.88D 8.901 8.958 9.02R 9.103 9.157 9.225 9.308 9.364 9.436 9.56D 9.641

m00 m00 þ m6 m00 þ m2 m00 þ m11 m1 m1 þ m6 m1 þ m2 m1 þ m11 2m1 2m1 þ m6 2m1 þ m2 2m1 þ m11 3m1 3m1 þ m6 3m1 þ m2 3m1 þ m11 4m1 4m1 þ m6 4m1 þ m2 4m1 þ m11 5m1 band maximum 5m1 þ m2 5m1 þ m11 6m1 6m1 þ m2 6m1 þ m11 7m1 7m1 þ m11 8m1

Quantum defect

Energy in eV reported by B
elanger and Sandorfy [20]

Assignment

6.401 6.574 6.779 7.009 7.22D 8.557 9.253 9.457 9.574dd 9.740dd 9.840dd 9.903dd 9.951dd

1.09 – – – – 1.05 1.04 – 1.01 1.02 1.03 1.08 1.03

6.37 – – – – 8.55 9.27 – 9.58 9.74 9.84 9.90 9.95

3, m00 m1 2m1 3m1 4m1 4 5, m00 m1 6 7 8 9 10

Table 6 Energy values and quantum defects of features assigned to a Rydberg series of npk character, converging to 10.12 eV

4.2.5. Ryberg transitions To assign observed features to Rydberg series, the standard equation is used 2

Photo-absorption energy in eV

The Rydberg series is reported in agreement with B
elanger and Sandorfy [20]. Vibrational assignments are suggested for the first time in the present work. D Indicates that the feature is diffuse and thus its energy position is subject to greater uncertainty. dd Accompanies peaks which can equally be assigned to the ndd series (Table 8).

B
elanger and Sandorfy [20] and Coggiola et al. [19] observe the band maximum to occur at 8.89 and 8.84 eV, respectively. Vibrational assignments are suggested for the first time in the present work. D Indicates that the feature is diffuse and thus its energy position is subject to greater uncertainty. Features which are considered to be energetically coincident with Rydberg transitions are marked R and are given to less precision.

En ¼ EI  ðR=ðn  dÞ Þ;

Table 5 Energy values and quantum defects of features assigned to a Rydberg series of nsr character, converging to 10.12 eV

ð2Þ

where En is the energy of the Rydberg state, EI the ionisation limit to which the series converges (this may be the ionic ground state or an ionic excited state), R the Rydberg constant (13.61 eV), n the principal quantum number, and d the relevant quantum defect [41]. B
elanger and Sandorfy [20] report three Rydberg series converging to the adiabatic ionisation energy (10.12 eV), associated with promotion from the least bound C@C p MO. The assignments are found to be in close agreement with structure observed in the present photoabsorption spectrum. Tables 5, 6 and 8 list the energy values of features assigned to the nsr npk and ndd suggested by B
elanger and Sandorfy [20] accompanied by the relevant quantum defects. Further structure is assigned for the first time to a Rydberg series of

Photo-absorption energy in eV

Quantum defect

Energy in eV reported by B
elanger and Sandorfy [20]

Assignment

8.014 8.21D 8.41D 9.017 9.436 9.66D 9.864 9.79D

0.46 – – 0.49 0.54 0.56 – 0.58

8.01 8.21 8.41 9.01 9.44 9.66 – 9.79

3, m00 m1 2m1 4 5 6, m00 m1 7

The Rydberg series is and the vibrational features at 8.21 and 8.41 eV are reported in agreement with B
elanger and Sandorfy [20]. The assignment of the peak at 9.864 eV to a vibrational excitation is suggested for the first time in the present work. D Indicates that the feature is diffuse and thus its energy position is subject to greater uncertainty.

ndp character, also converging to 10.12 eV. The features proposed to belong to the ndp series are listed in Table 7 and shown in Figs. 4 and 6. On the basis of the calculated quantum defects, the n ¼ 6–10 peaks of the nsr series could equally be assigned to n ¼ 5–9 in the ndd series (Tables 5 and 8). Vibrational structure associated with the m1 , C@C stretching mode is identified in each of the Rydberg series. The clearest assigned m1 series begins at the 6.401 eV 3sr transition and is visible both in the photoabsorption and in the electron scattering spectra. The series maximum absorption cross section of 6.357 Mb occurs at 6.574 eV. The average excitation energies of m1 features assigned to

S. Eden et al. / Chemical Physics 297 (2004) 257–269 Table 7 Energy values and quantum defects of features assigned to a Rydberg series of ndp character, converging to 10.12 eV Photo-absorption energy in eV

Quantum defect

Assignment

8.39D 8.59D 9.177 9.393 9.604 9.53D 9.71D 9.817

0.20 – 0.20 – – 0.20 0.24 0.30

3, m00 m1 4, m00 m1 2m1 5 6 7

All Rydberg and vibrational assignments are suggested for the first time in the present work. D Indicates that the feature is diffuse and thus its energy position is subject to greater uncertainty.

combine with Rydberg transitions in the series converging to 10.19 eV is 0.203 eV. As would be expected, this energy difference is very close to 0.208 eV, the average energy difference observed between intense m1 peaks in the ionic ground state (see Section 4.1.2). B
elanger and Sandorfy [20] state that vibrational fine structure dominated by C@C stretching is observed in all the fluoroethylene Rydberg series and give the energy positions of features beginning at the 3pk C2 F4 peak. However, only the first two m1 features are considered to be consistent with the present photoabsorption result. It should be noted that the peaks observed in the present photoabsorption spectrum at 8.244, 8.440, 8.469, 8.737, and 9.694 eV remain unassigned. Coggiola et al. [19] report electron scattering structure at high energies attributed to Ryberg series converging to the second ionisation potential, given as 15.93 eV. The present spectra show peaks occurring at the energies given in Table 9 with the corresponding suggested assignments. The transitions are labelled in Fig. 3. 4.2.6. Absolute photoabsorption cross section The most important error source in the present C2 F4 photoabsorption cross section is considered to be the

267

Table 8 Energy values and quantum defects of features assigned to a Rydberg series of ndd character, converging to 10.12 eV Photo-absorption energy in eV

Quantum defect

Energy in eV reported by B
elanger and Sandorfy [20]

Assignment

8.640 8.843 9.043 9.27D 9.486 9.574sr 9.740sr 9.840sr 9.903sr 9.951sr

)0.03 – – 0.00 – 0.01 0.02 0.03 0.08 0.03

8.64 – – 9.22 – 9.60 9.74 – – –

3, m00 m1 2m1 4, m00 m1 5 6 7 8 9

The Rydberg series is reported in agreement with B
elanger and Sandorfy [20]. Vibrational assignments are suggested for the first time in the present work. D Indicates that the feature is diffuse and thus its energy position is subject to greater uncertainty. sr Accompanies peaks which can equally be assigned to the nsr series (Table 5).

contribution to the sample pressure of the CO2 contamination identified in the high energy photoelectron analysis. Therefore, it is expected that the true absorption cross section is marginally higher than that recorded in the present work. Comparison with the previous spectra shows the present cross section to be of 30–50% higher than that measured by B
elanger and Sandorfy [20] and 25–30% lower than the result of Sharpe et al. [37]. The calculated oscillator strength corresponding to the presently assigned p ! r
transition is given by Winstead and Mckoy [12] as 0.0545, 9% of that calculated for the following optically allowed p ! p
transition. The present result leads us to report a maximum value of 12.5%. Therefore the present absolute cross sectional data are considered generally to be in good agreement with the recent calculations. The atmospheric destruction of C2 F4 by UV photolysis was modelled as a function of altitude using the program described by Lim~ao-Vieira et al. [42] and found

Table 9 Energy values and quantum defects of features assigned to Rydberg series converging to excited ionic states Electron energy loss in eV 100 eV, 5°

30 eV, 15°

12.93 13.67 14.79 15.13 15.35

12.93 13.69 14.73 – 15.37

100 eV, 5° Quantum defect

Energy loss in eV Coggiola et al. [19]

Assignment

0.74 0.35 0.55 1.00 0.75

13.3 – 15.0 – –

3pk ! 15.6 eV 3dp ! 15.6 eV 5sr ! 15.6 eV 4sr ! 16.6 eVVP 4pk ! 16.6 eVVP

VP Indicates that the quantum defect has been calculated using a vertical ionisation potential. The energy loss values given in the fourth column are those corresponding to the present assignment. The 3dp, 4sr and 4pk assignments are suggested for the first time in the present work. Coggiola et al. [19] also observe features at 13.9 and 15.7 eV.

268

S. Eden et al. / Chemical Physics 297 (2004) 257–269

to be insignificant in comparison with by reaction with OH radicals Acerboni et al. [6].

5. Conclusions The valence bands and the Rydberg transitions below the first ionisation energy of tetrafluoroethylene are attributed exclusively to excitation from the least bound C@C p orbital, an implication of the perfluoro-effect. Valence transitions are assigned to C@C antibonding orbitals in agreement with previous work and a Rydberg series of ndp character is proposed for the first time. Further features observed in the photoabsorption and EELS are assigned to vibrational modes coupled to the observed Rydberg series and valence transitions. Vibration is principally associated with C@C stretching at markedly lower excitation energy than in the neutral ground state. In the p
singlet excited state and in the ionic ground state, particularly strong motion is suggested by the identification of quanta of CF2 stretching of symmetry ag and b3u coupled to C@C stretching. However, the intensity of the vibrational structure associated with the p ! p
, singlet ! singlet transition is difficult to quantify due to energy mixing with Rydberg transitions. The results presented are consistent with the dominant dissociation pathway being the formation of CF2 radicals, as required for the use of C2 F4 as a feed gas for SiO2 etching. New spectroscopic structure is observed in the photoabsorption and photoelectron analysis and assignments suggested. The absolute photoabsorption cross section represents the highest resolution most reliable VUV spectrum available for this key industrial molecule.

Acknowledgements The authors wish to acknowledge the European Commission for access to the ASTRID facility and the support of technical staff at Aarhus University. PLV acknowledges the Portuguese ‘‘Fundacß~ ao para a Ci^encia e a Tecnologia’’ for a Postgraduate Scholarship. S.E., P.K. and N.J.M. acknowledge the support of UK funding councils; EPSRC, NERC and CLRC and the British Council during the course of this research. M.-J.H.F. acknowledges the FNRS of Belgium for a research position and financial support. J.D. acknowledges the Universit
e de Liege, Belgium, for financial support.

References [1] H. Hayashi, S. Morishita, T. Tatsumi, Y. Hikosaka, S. Noda, H. Nakagawa, S. Kobayashi, M. Inoue, T. Hoshino, J. Vac. Sci. Technol. A 17 (1999) 2557. [2] A.N. Goyette, J. de Urquijo, Y. Wang, L.G. Christopherou, J.K. Orloff, J. Chem. Phys. 114 (20) (2001) 8932.

[3] T.K. Minton, P. Felder, R.C. Scales, J.R. Huber, Chem. Phys. Lett. 164 (2,3) (1989) 113. [4] S. Samukawa, T. Mukai, J. Vac. Sci. Technol. A 17 (5) (1999) 2463. [5] S. Raoux, T. Tanaka, M. Bahn, H. Ponnekanti, M. Seamons, T. Deacon, L.-Q. Xia, F. Pham, D. Silvetti, D. Cheung, K. Fairbairn, J. Vac. Sci. Technol. B 71 (2) (1999) 477. [6] G. Acerboni, J.A. Buekes, N.R. Jensen, J. Hjorth, G. Myhre, C.J. Nielsen, J.K. Sundet, Atm. Environ. 35 (2001) 4113. [7] F.S. Toby, S. Toby, J. Phys. Chem. 80 (21) (1976) 2313. [8] S. Samukawa, T. Mukai, K. Noguchi, Mater. Sci. Semicond. Process. 2 (3) (1999) 203. [9] M. Kitajima, M. Okamoto, K. Sunohara, H. Tanaka, H. Cho, S. Samukawa, S. Eden, N.J. Mason, J. Phys. B 35 (2002) 3257. [10] S. Solomon, J.B. Burkholder, A.R. Ravishankara, R.R. Garcia, J. Geophys. Res. 99 (1994) 20929. [11] N.J. Mason, P. Lim~ao-Vieira, S. Eden, P. Kendall, S. Pathak, A. Dawes, J. Tennyson, P. Tegeder, M. Kitajima, M. Okamoto, K. Sunohara, H. Tanaka, H. Cho, S. Samukawa, S. Hoffmann, D. Newnham, S.M. Spyrou, Int. J. Mass Spec. 223–224 (2003) 647. [12] C. Winstead, V. Mckoy, J. Chem. Phys. 116 (4) (2002) 1380. [13] K. Yoshida, S. Goto, H. Tagashira, C. Winstead, B.V. Mckoy, W.L. Morgan, J. Appl. Phys. 91 (5) (2002) 2637. [14] R.F. Lake, H. Thompson, Proc. Roy. Soc., Ser. A 315 (1970) 323. [15] C.R. Brundle, M.B. Robin, N.A. Kuebler, H. Basch, J. Am. Chem. Soc. 94 (5) (1972) 1451. [16] J.A. Sell, A. Kuperman, J. Chem. Phys. 71 (11) (1979) 4703. [17] G. Bieri, W. Von Niessen, L. Asbrink, A. Svensson, Chem. Phys. 60 (1981) 61. [18] G.K. Jarvis, K.J. Boyle, C.A. Mayhew, R.P. Tuckett, J. Phys. Chem. A 102 (1998) 3230. [19] M.J. Coggiola, W.M. Flicker, O.A. Mosher, A. Kuppermann, J. Chem. Phys. 65 (7) (1976) 2655. [20] G. B
elanger, C. Sandorfy, J. Chem. Phys. 55 (5) (1971) 2055. [21] M.B. Robin, Higher Excited States of Polyatomic Molecules, vol. II, Academic, New York, 1975. [22] T. Shimanouchi, J. Phys. Chem. Ref. Data 3 (1) (1974) 269. [23] J.L. Carlos Jr., R.R. Karl Jr., S.H. Bauer, J. Chem. Soc. Faraday Trans. II 70 (1974) 177. [24] T. Shimanouchi, in: Tables of Molecular Vibrational Frequencies Consolidated, vol. I, National Bureau of Standards, 1972, 1-160. [25] J. Delwiche, P. Natalis, J. Momigny, J.E. Collin, J. Electron, Spectrosc. Relat. Phenom. 1 (1972) 219. [26] J.H.D. Eland, Photoelectron Spectroscopy, Butterworths, London, 1984. [27] H. Tanaka, T. Ishikawa, T. Masai, T. Sagara, L. Boesten, M. Takekawa, Y. Itikawa, M. Kimura, Phys. Rev. A 57 (1998) 1798. [28] www.isa.au.dk/SR/UV1/uv1.html. [29] A.C. Vandaele, P.C. Simon, J.M. Guilmot, M. Carleer, R. Colin, J. Geophys. Res. – Atmospheres 99 (D12) (1994) 25599. [30] D.E. Freeman, K. Yoshino, J.R. Esmond, W.H. Parkinson, Planet. Space Sci. 32 (1984) 1125. [31] K. Watanabe, Adv. Geophys. 5 (1958) 153. [32] P.H. Van Cittert, Z. Phys. 69 (1931) 298. [33] J.D. Allen, F.A. Grimm, Chem. Phys. Lett. 66 (1979) 72. [34] T.J. Buckley, R.D. Johnson, R.E. Huie, Z. Zhang, S.C. Kuo, R.B. Klemm, J. Phys. Chem. 99 (1995) 4879. [35] D.W. Turner, C. Baker, A.D. Baker, C.R. Brundle, Molecular Photoelectron Spectroscopy, Wiley Interscience, London, 1970. [36] P. Baltzer, F.T. Chau, J.H.D. Eland, L. Karlsson, M. Lundqvist, J. Rostas, K.Y. Tam, H. Veenhuizen, B. Wannberg, J. Chem. Phys. 104 (22) (1996) 8922. [37] S. Sharpe, B. Hartneet, H.S. Sethi, D.S. Sethi, J. Photochem. 38 (1987) 1. [38] J.W. Rabalais, J.M. McDonald, V. Scherr, S.P. McGlynn, Chem. Rev. 71 (1) (1971) 73.

S. Eden et al. / Chemical Physics 297 (2004) 257–269 [39] C. Cossart-Magos, M. Jungen, F. Launay, Mol. Phys. 61 (5) (1987) 1077. [40] G. Herzberg, Molecular Spectra and Molecular Structure. III. Electronic Spectra and Electronic Spectra of Polyatomic Molecules, Van Nostrand, New York, 1966.

269

[41] C. Sandorfy (Ed.), The Role of Rydberg States in Spectroscopy and Photochemistry, Kluwer Academic Publishers, Dordrecht, 1999. [42] P. Lim~ao-Vieira, S. Eden, P.A. Kendall, N.J. Mason, S.V. Hoffmann, Chem. Phys. Lett. 364 (2002) 535.