THE JOURNAL OF CHEMICAL PHYSICS 122, 144309 共2005兲 Complete valence double photoionization of SF6 R. Feifela兲 and J. H. D. Eland Physical and Theoretical Chemistry Laboratory, Oxford University, South Parks Road, Oxford OX1 3QZ, United Kingdom L. Storchi and F. Tarantelli Dipartimento di Chimica, Università di Perugia, and CNR I.S.T.M, Via Elce di Sotto 8, 06123 Perugia, Italy 共Received 14 December 2004; accepted 25 January 2005; published online 11 April 2005兲 Single photon double ionization of SF6 has been investigated at the photon energies 38.71, 40.814, and 48.372 eV by using a recently developed time-of-flight photoelectron–photoelectron coincidence spectroscopy technique which gives complete two-dimensional e− – e− spectra. The first complete single photon double ionization electron spectrum of SF6 up to a binding energy of ⬃48 eV is presented and accurately interpreted with the aid of Green’s function ADC共2兲 calculations. Spectra which reflect either mainly direct or mainly indirect 共via interatomic coulombic decay of F 2s holes兲 double ionization of SF6 are extracted from the coincidence map and discussed. A previous, very low value for the onset of double ionization of SF6 is found to energetically coincide with a peak structure related to secondary inelastic scattering events. © 2005 American Institute of Physics. 关DOI: 10.1063/1.1872837兴 I. INTRODUCTION Sulphur hexafluoride 共SF6兲 is important for its high dielectric strength in high voltage applications 共see Ref. 1, and references therein兲 and is widely used in plasma etching processes 共see Ref. 2, and references therein兲. To understand its physical and chemical behavior in various circumstances properly, knowledge of its electronic structure is essential. The electronic configuration of SF6 in its neutral ground state can be denoted as 共core兲22关共4a1g兲2共3t1u兲6共2eg兲4兴 ⫻共5a1g兲2共4t1u兲6共1t2g兲6共3eg兲4共1t2u兲6共5t1u兲6共1t1g兲6 , where the square brackets denote the inner-valence orbitals with ionization potentials in the range of outer-valence double ionization. Several investigations have been carried out accordingly, in particular using conventional photoelectron spectroscopy in both the valence and core regions, and have created much information concerning the cationic states 共see Refs. 3–7, and references therein兲. Various types of coincidence experiments have been performed in order to investigate the complex fragmentation behavior of SF6. Hitchcock et al. conducted electron impact excited electron-ion coincidence studies both in the S 2p ionization region 共160– 230 eV兲8 and in the valence shell ionization region 共15– 63 eV兲9 in order to determine photoabsorption and photofragmentation cross sections for SF6. They also studied dissocative and nondissociative double ionization and found that dissociative double ionization is an appreciable fraction of all double ionization events even close to the double ionization threshold. Furthermore, they “visually estimated” from the onsets of their oscillator strength a兲 Electronic mail: [email protected] 0021-9606/2005/122共14兲/144309/9/$22.50 curves, threshold potentials for ions produced in photofragmentation of SF6, in particular for SF2+ 4 共33± 2 eV兲 and for 共40± 1 eV兲 which were found to be about 6 eV lower SF2+ 2 than the pioneering values of Dibeler and Mohler.10 Masuoka and Samson11 reported partial cross sections for the singly + charged fragments 关SFm 共m = 0 – 5兲 + F+兴 and doubly charged 2+ fragments 关SFn 共n = 0 – 4兲兴 created upon single photon double ionization in the 75– 125 eV region using a technique related to that of Refs. 8 and 9. Frasinski et al.12 determined threshold values for formation of fragment ion pairs by making use of a photoion–photoion coincidence setup, and found the lowest threshold value of 41.1 eV for the formation of both SF+5 + F+ and SF+3 + F+; moreover, the fragment ion pair SF+3 + F+2 seems to occur at the same threshold value according to their investigations. About the same time, Joachims et al.13 measured appearance energies for formation of dications with a photoionization mass spectrometer and reported 2+ 2+ in Ref. 13 SF2+ 4 to occur at 38.5 eV, SF3 at 42.6 eV, SF2 at 2+ 43.8 eV, and SF at 54.7 eV. Eland and Treeves-Brown14 performed photoelectron–photoion–photoion coincidence experiments in order to investigate doubly charged ion dissociation patterns of SF6 in greater detail and derived kinetic energy release distributions. This work was later on elaborated by Hsieh and Eland15 applying a position sensitive detector which allowed them to measure the momentum vector of two coincident fragment ions and hence allowed to determine charge separation mechanisms, most clearly for SF+3 – F+ formation. Creasy and co-workers16 applied a threshold photoelectron–photoion coincidence technique to study “state-selectively” the fragmentation channels and branching ratios for the valence states in the binding energy range 15– 28 eV. Recently, Peterka et al.17 applied an ion 122, 144309-1 © 2005 American Institute of Physics Downloaded 14 Apr 2005 to 147.142.186.54. Redistribution subject to AIP license or copyright, see http://jcp.aip.org/jcp/copyright.jsp 144309-2 J. Chem. Phys. 122, 144309 共2005兲 Feifel et al. imaging technique in the same energy region and characterized the angular and translational energy distributions of the product ions. The electronic states of the dications have been the subject of several previous works. Griffiths and Harris18 determined, by double charge transfer spectroscopy, the lowest double ionization threshold to be at 38.9± 0.5 eV; their data suggest that at least two more electronically excited states of exist between 38.9 and 44 eV. Bancroft and SF2+ 6 co-workers19 performed a comparative study of normal and resonant Auger electron spectra for excitations both around the S 2p and the F 1s edges. They found that the normal Auger spectra are essentially grouped into two parts, one of which consists of transitions whose final states have two holes in the outer valence orbitals and a second which comprises transitions whose final states have one hole in an outer valence orbital and one in an inner-valence orbital. However, no further assignments were made for the observed double hole states in the normal Auger spectra. Sato et al.,20 building on the results of Ref. 19, performed Auger electron– photoion and photoion–photoion coincidence experiments and found that the dications having two holes in the outer valence orbitals dissociate into various fragments, while those having one hole in the outer valence and one hole in the inner valence orbitals mostly dissociate into atomic ionic fragments F+ and S+. Lange et al.21 recently presented an extensive study on the fragmentation of SFn+ 6 共n 艌 2兲 produced by impact of 2 MeV He2+ ions, which was accompanied by quantum chemical ab initio calculations of the vertical double ionization potentials in the valence region; they calculated the lowest vertical double ionisation potential of SF6 to be 39.0 eV. Furthermore, they calculated appearance energies for various fragmentation channels of doubly ionized SF6 at 0 K including zero-point vibration energy at the so-called “MP2/TZVPP” level and received similar values 2+ for the formation of SF2+ 4 and SF2 as reported before by Ref. 9, and they predicted the lowest fragmentation channel, SF+5 + F+, to occur as low as 31.33 eV. More recently, Yencha and co-workers22 recorded the first threshold photoelectron– photoelectron coincidence 共TPEsCO兲 spectrum of SF6 and suggested that inner valence SF+6 states play a major role in the formation of the SF2+ 6 ion states; interestingly, the onset of double ionization was found in their work to occur at 31.98± 0.02 eV. In the recently developed time-of-flight photoelectron– photoelectron coincidence spectroscopy technique 共TOFPEPECO兲 applied here,23 the energies of all the electrons ejected in pairs at each chosen photon energy are measured. Complete maps of intensity as a function of the two energies give both the dication spectrum and details of the photoionization process itself. We present in this work the first complete single photon double ionization electron spectrum of SF6 up to a binding energy of ⬃48 eV. Our interpretation is based on information obtainable from the coincidence map itself and on a new ab initio theoretical simulation of the spectrum performed by the Green’s function ADC共2兲 method.24 II. EXPERIMENT AND DATA ANALYSIS PROCEDURES Electron–electron coincidence data of SF6 were recorded by means of a recently developed time-of-flight electron– electron coincidence 共TOF-PEPECO兲 technique23,25,26 at the selected photon energies 38.71, 40.814, and 48.372 eV. This technique is based on a magnetic bottle type spectrometer as described in detail by Ref. 27. Briefly, wavelength selected light from a pulsed low-pressure discharge helium lamp ionizes an effusive jet of the target gas in a crossed beam configuration. Electrons from ionization are directed by the inhomogeneous magnetic field of a conically shaped permanent magnet to follow nominally on-axis the field lines inside a long solenoid to the 5.5 m distant multichannel plate detector. The helium lamp works at a typical repetition rate of 8 kHz and emits light pulses which are about 10 ns long. The timing of electron signals is referenced either to an electrical pulse generated when the lamp “fires” or to the simultaneous visible light pulse detected by a photomultiplier. Pairs of electrons arriving within 20 s of each other are recognized as true coincidences. The times of flight can be fitted with good precision to the simple form t= D − t0 , 共E + E0兲1/2 共1兲 with just two fitting parameter t0 and E0. The flight length D is fixed to 9265 ns 共eV兲−1/2. Energy calibration of the spectra is done by recording the well-known single electron spectrum of molecular oxygen before and after each run. In order to check for secondary collisions of electrons with the target gas, the pressure was varied from 6.5⫻ 10−6 to 3.3 ⫻ 10−5 mbar in the interaction region. Commercially available SF6 gas with a stated purity of ⬎99% was used for the experiments. The purity of the gas has been checked carefully by recording the single electron valence band spectrum before and after each coincidence run. III. THEORETICAL METHODS The calculations of the double ionization spectrum of SF6 reported in this paper were carried out using the secondorder Green’s function algebraic diagrammatic construction method.24 This is a powerful direct approach for the theoretical study of ionization processes, leading to a sparse symmetric eigenvalue problem in the space of the ionic configurations, represented in the basis of the Hartree–Fock orbitals of the neutral system. The eigenvalues of the system are the ionization energies, while the eigenvectors are related to the spectroscopic amplitudes. For double ionization, the ADC共2兲 configuration space comprises all the two-hole 共2h兲 configurations 共the main space兲 and their three-hole-one-particle single excitations 共3h1p兲. The matrix elements in the main space are of second order in the electron repulsion and of first order in the remaining space. Therefore, the ADC eigensolutions provide results which are accurate beyond second order with respect to the naked 2h configurations, and beyond first order for the satellite excitations. This is essentially equivalent to ionized state energies obtained by configuration interaction including also the vast space of 4h2p configurations. The ADC共2兲 method is implemented in a di- Downloaded 14 Apr 2005 to 147.142.186.54. Redistribution subject to AIP license or copyright, see http://jcp.aip.org/jcp/copyright.jsp 144309-3 Valence photoionization of SF6 FIG. 1. Double ionization electron spectra of SF6 recorded at the photon energies 38.71, 40.814, and 48.372 eV. In all three spectra, the signal rises strongly around 37.5 eV binding energy. rect integral-driven algorithm28 coupled to a block-Lanczos iterative diagonaliser29 which provides fast convergence onto the spectral envelope of the main space eigenvector projection, even when individual eigenvectors are not fully converged.30 The Hartree–Fock preliminary calculations have been performed on SF6 at the experimental geometry 共r = 1.564 Å兲31 with the GAMESS-UK program32 using a contracted Gaussian double zeta33 plus polarization34 basis set. The computed double ionization spectrum is fully converged up to 60 eV binding energy. IV. EXPERIMENTAL RESULTS In Fig. 1 we present double ionization electron spectra of SF6 obtained from the full coincidence data sets which were recorded at the photon energies 38.71, 40.814, and 48.372 eV. In all three spectra, the signal rises strongly at ⬃37.5± 0.1 eV binding energy, and the observed features remain essentially the same 共as far as they can be investigated兲 for all chosen photon energies as guided by the dashed lines. Below 37.5 eV binding energy an extended tail is visible in all the spectra with an onset at around 25 eV; it arises from false coincidence events, and its relative height depends on instrumental conditions. In order to account for possible pressure-dependent spectral features we recorded the double ionization electron spectrum of SF6 for different gas pressures in the interaction region. In Fig. 2 we show the 48.372 eV spectrum recorded at pressures p = 6.5⫻ 10−6 mbar 共lower panel兲 and p = 3.3 ⫻ 10−5 mbar 共higher panel兲, respectively. The broad background below 37.5 eV binding energy increases in intensity at the higher pressure, but in addition we observe four peaks in the high pressure spectrum starting at a binding energy of 31.33 eV 共peak value兲, separated by ⬃1.2 eV each. This energy spacing is similar to that of the first four outermost single electron valence bands of SF6 as reported by Ref. 5; hence the pressure-dependent peaks arise from secondary collision of primary electrons with the target gas leading to single ionization of a second neutral SF6 molecule. The po- J. Chem. Phys. 122, 144309 共2005兲 FIG. 2. Double ionization electron spectra of SF6 measured for different gas pressures in the interaction region. The arrow indicates the extraordinarily low value reported by Yencha et al. 共Ref. 22兲 for the onset of double ionization of SF6. sition of the first pressure-induced peak 共⬃31.5 eV兲 closely matches twice the vertical ionization energy for the lowest state of SF+6 . Such inelastic scattering features, already known from conventional photoelectron spectroscopy 共see, e.g., Ref. 35兲, can be seen in TOF-PEPECO spectra of all molecules at sufficiently high pressure.36 We note that the valley between the first and the second peak located at 31.98± 0.02 eV 共marked by the arrow in Fig. 2兲 corresponds exactly to the value for the onset of double ionization of SF6 reported by Yencha et al.,22 which suggests that this extraordinarily low value may have a similar origin. In comparing the double ionization electron spectrum recorded at 48.372 eV photon energy with the TPEsCO results of Yencha et al.,22 both similarities and dissimilarities can be encountered. In both cases strong broad structures are observed with an onset around 37.5 eV. However, whereas the TPEsCO spectrum shows essentially a single broad spectral feature centered about 40.44 eV binding energy and which falls off substantially around 43 eV, the complete electron–electron spectrum shown in Figs. 1 and 2 consists of two broad parts with a distinct minimum around 39.66 eV and which falls off less rapidly toward higher binding energies. In order to give an interpretation of their TPEsCO spectrum, Yencha et al.22 compared it with the corresponding inner valence part of both the ordinary and the threshold photoelectron spectrum. They found in particular that the centroid of the TPEsCO spectrum at 40.44 eV is very close to the centroid of the partially unresolved structures between ⬃37 and ⬃45 eV binding energy in the conventional photoelectron spectrum which may, in a simplified picture, be associated with the formation of the 共2eg兲−1 共36– 40.5 eV region兲, the 共3t1u兲−1 共40.5– 43 eV region兲 and the 共4a1g兲−1 共43– 45.5 eV region兲 inner valence states of SF+6 共cf. Ref. 5兲; already from the early calculations reported by Siegbahn et al.3 it is known that those orbitals are predominantly of F 2s character. On this basis, Yencha et al.22 suggested that double ionization in SF6 may proceed predominantly through the inner-valence ionic states followed by autoionization with Downloaded 14 Apr 2005 to 147.142.186.54. Redistribution subject to AIP license or copyright, see http://jcp.aip.org/jcp/copyright.jsp 144309-4 Feifel et al. J. Chem. Phys. 122, 144309 共2005兲 FIG. 3. Comparison of the single electron spectrum and the double ionization electron spectrum of SF6 measured simultaneously at the photon energy 48.372 eV. formation of the two lowest states of SF2+ 6 . A similar comparison is made in Fig. 3, where we display the single electron spectrum 共upper panel兲 recorded simultaneously in our experiment with the double ionization electron spectrum 共lower panel兲 of SF6 at 48.372 eV photon energy. The wellknown outer valence F 2A1g state of SF+6 at the binding energy of 26.82± 0.02 eV 共cf. Ref. 5兲 is included for reference purposes. As can be seen, the spectral regions between ⬃37 and ⬃45 eV appear to be fairly similar as they consist in both cases of broad, overlapping features; in particular the main features of the double ionization electron spectrum occur at somewhat lower binding energies compared to the features in the single electron spectrum as indicated by the dashed lines, which makes the interpretation of Yencha et al.22 interesting for further investigations. In order to get a deeper insight into single photon double ionization of SF6, it is helpful to study the coincidence map itself. In Fig. 4 we display, in isometric perspective, the complete coincidence map of SF6 recorded at 48.372 eV photon energy, where the intensity is shown as a function of the sum of the two electron energies E1 + E2 on both the kinetic and ionization energy scales 共into the page兲, and the kinetic energy of one of the electrons E1 共across the page兲. As we can see, the intensity varies quite strongly in the low kinetic energy region of E1, but it is fairly uniform in the region close FIG. 4. Complete coincidence map of SF6 recorded at 48.372 eV photon energy in isometric perspective, where the intensity is shown as a function of the sum of the two electron energies E1 + E2 relative to the photon energy 共into the page兲 and the kinetic energy of one of the electrons E1 共across the page兲. FIG. 5. Comparison of differently obtained double ionization electron spectra. 共a兲 The spectrum where all electrons were taken 共cf. Figs. 1–3兲; 共b兲 the spectrum obtained by choosing a slice of coincidence events from the low kinetic energy region of Fig. 4 共E1 from 0 to 1 eV kinetic energy at 38 eV binding energy and from proportionally narrower ranges at lower electron pair energies兲; 共c兲 a similarly created spectrum, where a much narrower E1 range of 0 – 0.1 eV kinetic energy at 38 eV binding energy was chosen; 共d兲 the spectrum created by selecting a slice of coincidence events from the fairly uniform area close to the main diagonal of Fig. 4, where the width of the slice was chosen to vary from 2 eV at 38 eV binding energy to proportionally narrower values at lower electron pair energies. to the main diagonal of this plot. Double ionization spectra of SF6 can be projected from the coincidence map in various different ways by taking either all electrons or only certain selections of them. In Fig. 5 we show a comparison of different double ionization electron spectra. Panel 共a兲 presents the spectrum where all electrons were taken 共cf. Figs. 1–3兲; panel 共b兲 shows the spectrum obtained by choosing a slice of coincidence events from the low kinetic energy region of Fig. 4 共E1 from 0 to 1 eV kinetic energy at 38 eV binding energy and from proportionally narrower ranges at lower electron pair energies兲; panel 共c兲 shows a similarly created spectrum, where a much narrower E1 range of 0 – 0.1 eV kinetic energy at 38 eV binding energy was chosen; panel 共d兲 shows the spectrum created by selecting a slice of coincidence events from the fairly uniform area close to the main diagonal of Fig. 4, where the width of the slice was chosen to vary from 2 eV at 38 eV binding energy to proportionally narrower values at lower electron pair energies. Clearly, the spectra are rather different; in particular, spectrum 共b兲 and 共c兲 show, in comparison to spectrum 共d兲, drastic changes in relative intensity of certain features as marked by the arrows and the dashed lines, and appear generally to be less structured. The differences in spectral appearance can be rationalized as follows; spectrum 共b兲 and 共c兲 are strongly dominated by coincidence events related to low kinetic energy electrons Downloaded 14 Apr 2005 to 147.142.186.54. Redistribution subject to AIP license or copyright, see http://jcp.aip.org/jcp/copyright.jsp 144309-5 Valence photoionization of SF6 J. Chem. Phys. 122, 144309 共2005兲 V. THEORETICAL ANALYSIS OF THE SPECTRUM AND DISCUSSION FIG. 6. Electron distributions related to two different parts of the dicationic spectrum as marked by the dashed lines. whose distribution is seen to be structured 共cf. Fig. 4兲 and which probably originate from molecular autoionization involving the inner valence electronic states of SF+6 as proposed by Yencha et al.22 It is noteworthy that spectrum 共c兲 is indeed very reminiscent to the TPEsCO spectrum reported in Ref. 22. In contrast, spectrum 共d兲, where electrons with nearly equal energies are taken from a region which shows no structure in the coincidence map, is probably representing a direct SF2+ 6 spectrum. By returning our attention to the double ionization spectrum where all electrons were taken into account 共cf. Figs. 1–3兲, and focusing on the gross features below and above the spectral minimum at 39.66 eV binding energy, the question arises if different double ionization mechanisms are operating in these two parts. In particular, one could think of one region reflecting primarily direct and the other one indirect, dissociative double ionization routes as observed before for other systems 共see Refs. 36–41, and references therein兲. The threshold values for formation of fragment ion pairs and the nonexistence of stable SF2+ 6 ions demonstrate that all double ionization in SF6 is dissociative. If separate double ionization mechanisms were operating, one could expect to see a difference in the electron distributions related to the formation of the relevant parts of the dication spectrum. In Fig. 6 we show the electron distributions related to two different parts of the dicationic spectrum as marked by the dashed lines, one below the minimum at 39.66 eV and another one above. As we can see, the electron distributions are very much the same in the two cases. Furthermore, the complete electron– electron coincidence map 共cf. Fig. 4兲 does not contain any features at constant kinetic energy E1 which could be related to autoionization in fragment fluorine atoms, in contrast to other molecules such as oxygen where indirect atomic autoionization is found to be dominant.39–41 Thus there is no direct evidence for involvement of dissociative indirect double ionization in SF6, but the observations cannot entirely exclude it. In order to analyze the double ionization spectrum in greater depth and provide a more theoretically profound interpretation, we have carried out a detailed ab initio simulation using the Green’s function ADC共2兲 method,24 of postHartree–Fock second-order accuracy. Such calculations have been successfully used in the past to study the double ionization of many molecular systems 共for a recent example see Ref. 42兲. Key features of the method are that it is efficient enough to permit the accurate calculation of the whole manifold of double ionization transitions and that, using a population analysis of the double ionization eigenvectors,43 we can obtain a neat and useful picture of the two-hole density distribution in the final states in terms of localized atomic components. The computed double ionization spectrum of SF6 comprises nearly 29 000 electronic states up to a binding energy of 110 eV, a number of states which attests to the complexity of such spectra and calculations. The density of states increases dramatically with the ionization energy, but it is high even at low energy: over 100 dicationic states are computed to lie in the lowest 10 eV of the spectrum. The population analysis of the eigenvectors yields a decomposition of the 2h projection 共i.e., the part expressible as ejection of two electrons from the Hartree–Fock occupied orbitals, with no further excitations; this part—as long as ground state electron correlation may be treated as a perturbation—is experimentally the most relevant part兲 in four localized components, which, using a well-established notation,43 we shall denote as F−1F⬘−1, S−1F−1, F−2, and S−2. The first component, F−1F⬘−1, measures the fraction of two-hole charge where the two holes are localized each on a different fluorine atom. On simple electron and atom counting arguments, this component may be expected to be dominant. S−1F−1 gives the twohole component with one vacancy on sulfur and one on a fluorine atom, while S−2 and F−2 express the portion of twohole density where both holes are localized on sulfur and on a single fluorine, respectively. In octahedral SF6, the pairs of fluorine atoms are not all equivalent. Consequently, the F−1F⬘−1 component may be further subdivided in two terms, one measuring the charge involving pairs of vicinal fluorine atoms and the second expressing the localization of the holes on pairs of equatorially opposed 共distal兲 fluorines. Figure 7 displays the energy distribution of the population components over the whole double ionization spectrum, obtained by Gaussian broadening of the discrete data, and immediately enlightens the nature of the spectrum, placing our experiment in context. The spectrum is, in its qualitative features, similar to those of other fluorides29,42,44 and we shall therefore give a summary description of it, referring the reader to the previous studies for more details. As expected, the F−1F⬘−1 character dominates in global terms, while, at the opposite end, the S−2 component is everywhere too small to be seen on the scale of the figure. The F−1F⬘−1 component 共and, consequently, the whole spectrum兲, exhibits three main zones of high intensity: below 50 eV, between 60 and 75 eV, and around 90 eV. As it is known that the fluorine components of the spectrum basically retain an atomic character,44 Downloaded 14 Apr 2005 to 147.142.186.54. Redistribution subject to AIP license or copyright, see http://jcp.aip.org/jcp/copyright.jsp 144309-6 Feifel et al. J. Chem. Phys. 122, 144309 共2005兲 FIG. 7. Plot of the computed 2h population components of the double ionization spectrum of SF6, obtained by Gaussian broadening of the discrete data 共FWHM = 1 eV兲. Solid line: total F−1F⬘−1 component; dotted: distal F−1F⬘−1 component; dashed: S−1F−1 component; Gray-filled: F−2 component. The S−2 component is omitted because is too small on the scale of the figure. these three zones may be easily interpreted in terms of the ionization of a pair of fluorine atoms. At low energy, we have the states where both atoms have lost an outer-valence 2p electron; in the intermediate region we see the dicationic states where one atom is again deprived of a 2p electron, while the other has lost an inner-valence 2s electron; and finally, at high energy, both atoms have undergone innervalence ionization. The relative intensity of the three bands may be easily related to the different number of available 2p and 2s electrons. The three bands are furthermore roughly equally spaced, reflecting the difference in binding energy of a fluorine 2p and 2s electron, because the repulsion between holes residing on distant atoms is approximately constant on average. In a more refined view, note that the distal F−1F⬘−1 component increases at the low-energy end of each band, exactly because its hole–hole repulsion energy is smaller than in the adjacent-pair component, owing to the larger distance between the fluorine atoms. The distal F−1F⬘−1 component is of course much less intense than the vicinal counterpart due simply to the smaller number of such fluorine pairs 共1 in 5兲. The F−2 density component, shown as a gray-filled curve in Fig. 7, also displays an essentially atomic-like three-band pattern, similar to the above. However, apart from the evident drop in intensity compared to the F−1F⬘−1 spectrum, we should note two other characterizing differences. The first is that the three F−2 zones are shifted by about 10 eV to higher energy compared to the F−1F⬘−1 ones, because of the obviously larger repulsion energy of two electron holes confined on a single atom. As a consequence, there is little overlap between the F−1F⬘−1 and F−2 spectra, with the remarkable result that the F−2 component, though globally much smaller, peaks out dominantly in at least two regions of the double ionization spectrum, just above 50 and around 80 eV. The second difference is that the intermediate 2p−12s−1 region is actually composed of two bands, one centered at about 72 eV and the second around 80 eV. This reflects the significant energy split between triplet and singlet states due to the spatial compactness of the inner-valence hole. 共The 2s−2 single-site band above 100 eV, on the other hand, obviously has no triplet component.兲 Finally, Fig. 7 emphasises that the S−1F−1 component of the two-hole charge is everywhere very small and more uniform, so that it is never dominant. In fact, it tends to be largest in the same energy regions where the F−1F⬘−1 component dominates the spectrum. Having so characterized the double ionization spectrum of SF6 with the help of the population analysis, we now see clearly that the photon energy used in the experiment is sufficient to probe almost completely the outer-valence double ionization region, corresponding essentially to the ionization of two 2p electrons on different fluorine atoms. We can now attempt to give a more detailed account of the experimental spectrum based on our theoretical results. In order to do so with quantitative accuracy, we would need to calculate the energy-dependent double photoionization cross section for the over 100 electronic states found below the photon energy, but this is an impossible task. Resorting to a simpler, more qualitative, approach, we may use the broadened 2h spectroscopic factors as a rough estimate of the spectral profile. The results are shown in Fig. 8, below the experimental spectrum of Fig. 5共d兲. The latter has been chosen for comparison since, as previously discussed, it corresponds to an even energy distribution between the two outgoing electrons and is thus more likely to reflect a direct double photoionization mechanism. We shall later return to this point. The computed spectrum has been shifted up in energy by 1.8 eV in order to obtain alignment with the experimental profile. The necessity of such shift was expected42 and it may be explained as due to a relatively deficient account of ground state electron correlation in our calculations. To realistically incorporate the electron energy dependence of the double ionization cross section, we have applied a linear filter to the theoretical intensity, ranging from 1 at the double ionization threshold to 0 at the photon energy of 48.372 eV, which corresponds to vanishing electron kinetic energy. Figure 8 shows that, in spite of some quantitative discrepancies in the relative spectral intensities, the computed spectrum reproduces without ambiguity—in fact, with remarkable detail—most experimental features. It proves beyond doubt that the whole measured spectrum does indeed Downloaded 14 Apr 2005 to 147.142.186.54. Redistribution subject to AIP license or copyright, see http://jcp.aip.org/jcp/copyright.jsp 144309-7 Valence photoionization of SF6 J. Chem. Phys. 122, 144309 共2005兲 FIG. 8. Theoretical double ionization spectrum of SF6 shown below the experimental spectrum of Fig. 5共d兲. The theoretical spectrum is obtained by Gaussian broadening of the computed 2h components of the states 共shown as vertical bars兲, with FWHM increasing linearly from 0.5 eV at low energy to 1.0 eV. The computed intensity incorporates a linear fall-off factor going from 1 at 38.0 eV to 0 at the threshold of 48.372 eV. The dotted line shows the total 2h spectrum, the solid line 共and the bar spectrum兲 omits the distal F−1F⬘−1 component. explore the outer-valence direct vertical double ionization of SF6. There are actually two theoretical profiles in the figure. The dotted line results from the convolution of the total 2h strength of the states, while the solid line 共and the associated discrete bar spectrum兲 results by eliminating the distal F−1F⬘−1 component. Although both spectra provide essentially the same qualitative description of the experiment, it is very interesting to note that deleting the distal F−1F⬘−1 component markedly improves the theoretical profile at low energy, where the dropped component is largest. This suggests, not implausibly, that dicationic states where the two holes are to a larger extent localized on more distant atomic sites are relatively more difficult to reach by direct double photoionization. Both the experimental and theoretical spectra show a first composite band below 40 eV, well separated from the rest of the spectrum, in which three principal peaks can be identified. This low-energy part of the spectrum comprises states where exclusively the fluorine p lone-pairs are ionized 共the sulfur-containing components of the states are nearly exactly vanishing here兲. The states appear to cluster in three neatly separated groups of nearly degenerate levels, as is evidenced by the bar spectrum. Note that the degree of neardegeneracy is such that, for example, for the highest energy group, comprising four dicationic states, only one line is visible. Somewhat surprisingly, such tight clustering is the nontrivial result of two different contributing factors: the type of fluorine pair involved 共whether vicinal or distal兲 and the relative orientation of their nonbonding 2p orbitals ionized, which can be parallel, skewed 共i.e., at right angles and noncoplanar兲, and, in the vicinal case, orthogonal but coplanar. A more detailed population analysis thus shows, for example, that vicinal parallel F−1F⬘−1 pairs contribute nearly exclusively to the first and third groups, while the middle group of states involves almost only vicinal coplanar orthogonal pairs and distal parallel pairs. Skewed equatorially opposed pairs contribute only to the two lower energy groups, while skewed vicinal pairs appear in all three groups. The first band is followed by an evident gap and then by the intense broad feature appearing between 40 and 43 eV. As attested by the computed bar spectrum, the number of dicationic states reached in this ionization energy range is very large and it is impossible to characterize them in terms of orbital ionization, as configuration mixing dominates. The states appear to give rise to two large bands, although the theoretical spectrum exaggerates the double structure and the relative intensities are also poorly reproduced. At higher energy the intensity declines rapidly, but both the experimental and theoretical profiles evidence a broad composite structure between 43 and 45 eV and a final band around 46 eV. Here the calculations appear to begin to overestimate slightly the relative energy of the states involved, which is not unexpected since, at high energy, correlation effects and the influence of higher excitations not included in our theoretical scheme are more strongly felt. Based on the results of our calculations, we may finally briefly comment on the origin of the different spectra obtained by sampling different kinetic energy distribution regions, as shown in Fig. 5. In particular, as we discussed earlier, when one of the two outgoing electrons has small kinetic energy, an autoionization mechanism with secondary electron emission may be thought to be playing a nonnegligible role in the double ionisation. As Fig. 3 and Ref. 5 show, singly ionized states of SF6 in the F 2s region lie indeed in the same binding energy region, and above, the lowest doubly ionized states. In Ref. 5, maxima in the innervalence photoelectron spectrum of SF6 were reported at 39.8, 40.9, and 44.1 eV and these figures are all well above the double ionization threshold. This is quite remarkable, for the possibility of decay of valence singly ionized states, even inner-valence ones, by secondary electron emission is uncommon in small molecules: due to the repulsion energy between two electron vacancies confined in a small region of space, doubly ionized states lie normally well above the whole 共valence兲 single ionisation spectrum. SF6 thus represents an interesting case because, as we have seen, the possibility of localization of the two holes on relatively distant fluorine atoms lowers the repulsion energy enough to place a large number of the outer-valence dicationic states below the single inner-valence-hole levels. As to the expected rate for Downloaded 14 Apr 2005 to 147.142.186.54. Redistribution subject to AIP license or copyright, see http://jcp.aip.org/jcp/copyright.jsp 144309-8 J. Chem. Phys. 122, 144309 共2005兲 Feifel et al. FIG. 9. Plot of the S−1F−1 component of the dicationic states computed below 45.5 eV, shown below the experimental profile of Fig. 5共c兲, and obtained by Gaussian broadening 共FWHM= 1 eV兲 of the discrete data shown as vertical bars. A fall-off filter like the one of Fig. 8 is incorporated between 38 and 45.5 eV. Triplet states have been reduced in intensity by 2 / 3 to simulate a decay process. such decay process, we recall that, in more extended chemical systems, inner-valence decay accompanied by the creation of a secondary electron vacancy in regions even very distant from the primary hole is well known as the so-called interatomic Coulombic decay 共ICD兲. It has been theoretically predicted and shown to be a very efficient process,45 before being experimentally observed.46–48 Moreover, it has been recently shown to play an important role in the Xe 4d ionization of xenon fluorides49 and has also been observed in 2s-ionized van der Waals dimers of neon.47 On these grounds, double ionization of SF6 by ICD may undoubtedly be the prevailing mechanism at low enough kinetic energies of the first electron. Going from a direct double photoionization mechanism to a decay one would of course change, possibly very significantly, in general, the spectral profile, even if the same manifold of final states is reached. As previously mentioned, the varying profiles of Fig. 5 support the hypothesis of a change of mechanism. Our present calculations, where transition rates and line broadenings are only qualitatively estimated, do not allow us to discriminate conclusively between the two mechanisms, and the issue is especially complicated by the fact that more than one decaying state is involved. We note however that the decrease in the relative intensity of the lowest energy band in Figs. 5共b兲 and 5共c兲 suggests, reasonably, a relatively smaller rate of decay from the innervalence, more bonding, states to dicationic states where the two vacancies are on widely separated nonbonding orbitals. More specifically, the role of the S−1F−1 character of the states may be enhanced in a F 2s hole decay, as has been found for the Xe 4d decay in xenon fluorides.49 Also the high-energy half of the spectrum appears to drop quite sharply in relative intensity, especially in Fig. 5共c兲, probably because of its proximity to the decay threshold. There are in general further differences to be taken into account between a decay mechanism and direct photoionization. One concerns the different activity of final singlet and triplet spin states. Another, very important one, concerns the effect of nuclear dynamics, which is profoundly different in decay processes from that involved in sudden direct transitions.50 This is known to affect in very significant ways not only line broadenings, but also the effective energy positions of the spectral bands.51 As an illustration of some of these suggestions, Fig. 9, shows a plot of the S−1F−1 population component up to 45.5 eV in comparison to the experimental spectrum of Fig. 5共c兲. VI. SUMMARY Single photon double ionization of SF6 has been investigated by using a recently developed time-of-flight photoelectron–photoelectron coincidence spectroscopy technique. The first complete single photon double ionization electron spectrum of SF6 up to a binding energy of ⬃48 eV was presented, and spectra which reflect either mainly direct or mainly indirect double ionization of SF6 were extracted from the coincidence map. The theoretical Green’s function calculations performed here have enabled us to give a complete and satisfactorily accurate description of the measured double ionization spectrum. The charge density analysis of the dication states has permitted a full characterization of the spectral features and given a useful context for discussing the indirect mechanism for producing outer valence dicationic states via decay of inner-valence F 2s holes. It is shown that such mechanism is in fact an intramolecular interatomic Coulombic decay and should therefore be very efficient. ACKNOWLEDGMENTS Professor N. Kosugi is acknowledged for valuable discussions on the subject. R.F. would like to thank the Swedish Research Council 共VR兲 and the Swedish Foundation for International Cooperation in Research and Higher Education 共STINT兲 for financial support of his stay at Oxford University. L.S. and F.T. thank the italian M.I.U.R. and C.N.R. for financial support. 1 V. N. Maller and M. S. Naidu, Advances in High Voltage Insulation and Arc Interruption in SF6 and Vacuum 共Pergamon, Oxford, 1981兲. 2 D. M. Manos and D. L. Flamm, Plasma Etching: An Introduction 共Academic, Boston, MA, 1989兲. 3 K. Siegbahn C. Nordling, G. Johansson et al., ESCA Applied to Free Downloaded 14 Apr 2005 to 147.142.186.54. Redistribution subject to AIP license or copyright, see http://jcp.aip.org/jcp/copyright.jsp 144309-9 Valence photoionization of SF6 Molecules 共North-Holland, Amsterdam, 1969兲. L. Karlsson, L. Mattsson, R. Jadrny, T. Bergmark, and K. Siegbahn, Phys. Scr. 14, 230 共1976兲. 5 D. M. P. Holland, M. A. MacDonald, M. A. Hayes, L. Karlsson, M. Lundqvist, B. Wannberg, and W. von Niessen, Chem. Phys. 192, 333 共1995兲. 6 D. M. P. Holland, M. A. MacDonald, P. Baltzer, L. Karlsson, M. Lundqvist, B. Wannberg, and P. Baltzer, J. Phys. B 29, 487 共1996兲. 7 L. Yang, H. Ågren, V. Carravetta, O. Vahtras, L. Karlsson, B. Wannberg, D. M. P. Holland, and M. A. MacDonald, J. Electron Spectrosc. Relat. Phenom. 94, 163 共1998兲. 8 A. P. Hitchcock, C. E. Brion, and M. J. Van der Wiel, J. Phys. B 11, 3245 共1978兲. 9 A. P. Hitchcock and M. J. Van der Wiel, J. Phys. B 12, 2153 共1979兲. 10 V. H. Dibeler and F. L. Mohler, J. Res. Natl. Bur. Stand. 40, 25 共1948兲. 11 T. Masuoka and A. R. Samson, J. Chem. Phys. 75, 4946 共1981兲. 12 L. J. Fransinski, M. Stankiewicz, K. J. Randall, P. A. Hatherly, and K. Codling, J. Phys. B 19, L819 共1986兲. 13 H.-W. Joachims, E. Rühl, and H. Baumgärtel, BESSY Jahresbericht, BESSY GmbH, Berlin 共1986兲. 共It is an annual report which one can request from the synchrotron radiation research facility Bessy in Germany by sending an e-mail to “[email protected]”; for further information see also “www.bessy.de” 14 J. H. D. Eland and B. J. Treeves-Brown, AIP Conf. Proc. 258, 100 共1992兲. 15 S. Hsieh and J. H. D. Eland, Rapid Commun. Mass Spectrom. 9, 1261 共1995兲. 16 J. C. Creasy, H. M. Jones, D. M. Smith, R. P. Tuckett, P. A. Hatherly, K. Codling, and I. Powis, Chem. Phys. 174, 441 共1993兲. 17 D. S. Peterka, M. Ahmed, C. Y. Ng, and A. G. Suits, Chem. Phys. Lett. 312, 108 共1999兲. 18 W. J. Griffiths and F. M. Harris, Int. J. Mass Spectrom. Ion Processes 85, 259 共1988兲. 19 G. M. Bancroft, K. H. Tan, O.-P. Sairanen, S. Aksela, and H. Aksela, Phys. Rev. A 41, 3716 共1990兲. 20 G. M. Bancroft, K. H. Tan, O.-P. Sairanen, S. Aksela, and H. Aksela, Phys. Rev. A 41, 3716 共1990兲. 21 M. Lange, O. Pfaff, U. Müller, and R. Brenn, Chem. Phys. 230, 117 共1998兲. 22 A. J. Yencha, M. C. A. Lopes, D. B. Thompson, and G. C. King, J. Phys. B 33, 945 共2000兲. 23 J. H. D. Eland, O. Vieuxmaire, T. Kinugawa, P. Lablanquie, R. I. Hall, and F. Pennet, Phys. Rev. Lett. 90, 053003 共2003兲. 24 J. Schirmer and A. Barth, Z. Phys. A 317, 267 共1984兲. 25 J. H. D. Eland, S. S. W. Ho, and H. L. Worthington, Chem. Phys. 290, 27 共2003兲. 26 J. H. D. Eland, Chem. Phys. 294, 171 共2003兲. 27 P. Kruit and F. H. Read, J. Phys. E 16, 313 共1983兲. 28 F. Tarantelli, 共unpublished兲. 29 F. O. Gottfried, F. Tarantelli, and L. S. Cederbaum, Phys. Rev. A 53, 2118 共1996兲. 30 H.-D. Meyer and S. Pal, J. Chem. Phys. 91, 6195 共1989兲. 31 G. Herzberg, Molecular Spectra and Molecular Structure, III. Electronic 4 J. Chem. Phys. 122, 144309 共2005兲 Spectra and Electronic Structure of Polyatomic Molecules 共Van Nostrand, New York, 1966兲. 32 GAMESS-UK is a package of ab initio programs written by M. F. Guest, J. H. van Lenthe, J. Kendrick, K. Schoffel, and P. Sherwood, with contributions from R. D. Amos, R. J. Buenker, H. J. J. van Dam, M. Dupuis, N. C. Handy, I. H. Hillier, P. J. Knowles, V. Bonacic-Koutecky, W. von Niessen, R. J. Harrison, A. P. Rendell, V. R. Saunders, A. J. Stone, D. J. Tozer, and A. H. de Vries. The package is derived from the original GAMESS code due to M. Dupuis, D. Spangler, and J. Wendoloski, NRCC Software Catalog, Vol. 1, Program No. QG01 共GAMESS兲, 1980. 33 T. H. Dunning, J. Chem. Phys. 53, 2823 共1970兲. 34 R. Ahlrichs and P. R. Taylor, J. Chim. Phys. Phys.-Chim. Biol. 78, 315 共1981兲. 35 P. Baltzer, L. Karlsson, M. Lundqvist, B. Wannberg, M. Larsson, M. A. Hayes, J. B. West, M. R. F. Siggel, A. C. Parr, and J. L. Dehmer, J. Phys. B 27, 4915 共1994兲. 36 J. H. D. Eland, S. S. W. Ho, and H. L. Worthington, Chem. Phys. 290, 27 共2003兲. 37 J. H. D. Eland, P. Lablanquie, M. Lavolle’e, M. Simon, R. I. Hall, M. Hochlaf, and F. Pennet, J. Phys. B 30, 2177 共1997兲. 38 D. B. Thompson, G. Dawber, N. Gulley, M. A. MacDonald, and G. C. King, J. Phys. B 30, L147 共1997兲. 39 P. Bolognesi, D. B. Thompson, L. Avaldi, M. A. MacDonald, M. C. A. Lopes, D. R. Copper, and G. C. King, Phys. Rev. Lett. 82, 2075 共1999兲. 40 R. Feifel and J. H. D. Eland, Phys. Rev. A 共in press兲. 41 R. Feifel, J. H. D. Eland, and D. Edvardsson, J. Chem. Phys. 122144308 共2005兲, preceding paper. 42 C. Villani and F. Tarantelli, J. Chem. Phys. 120, 1775 共2004兲, and references therein. 43 F. Tarantelli, A. Sgamellotti, and L. S. Cederbaum, J. Chem. Phys. 94, 523 共1991兲. 44 F. Tarantelli and L. S. Cederbaum, Phys. Rev. Lett. 71, 649 共1993兲; F. O. Gottfried, L. S. Cederbaum, and F. Tarantelli, J. Chem. Phys. 104, 9754 共1996兲. 45 L. S. Cederbaum, J. Zobeley, and F. Tarantelli, Phys. Rev. Lett. 79, 4778 共1997兲; J. Zobeley, L. S. Cederbaum, and F. Tarantelli, J. Chem. Phys. 108, 9737 共1998兲; R. Santra, J. Zobeley, and L. S. Cederbaum, Phys. Rev. B 64, 245104 共2001兲. 46 S. Marburger, O. Kugeler, U. Hergenhahn, and T. Möller, Phys. Rev. Lett. 90, 203401 共2003兲. 47 T. Jahnke, A. Czasch, M. S. Schöffler et al., Phys. Rev. Lett. 93, 163401 共2004兲. 48 G. Öhrwall, M. Tchaplyguine, M. Lundwall, R. Feifel, H. Bergersen, T. Rander, A. Lindblad, J. Schulz, S. Peredkov, S. Barth, S. Marburger, U. Hergenhahn, S. Svensson, and O. Björneholm, Phys. Rev. Lett. 93, 173401 共2004兲. 49 C. Buth, R. Santra, and L. S. Cederbaum, J. Chem. Phys. 119, 10575 共2003兲. 50 L. S. Cederbaum and F. Tarantelli, J. Chem. Phys. 98, 9691 共1993兲. 51 D. Minelli, F. Tarantelli, A. Sgamellotti, and L. Cederbaum, J. Chem. Phys. 107, 6070 共1997兲, and references therein. Downloaded 14 Apr 2005 to 147.142.186.54. Redistribution subject to AIP license or copyright, see http://jcp.aip.org/jcp/copyright.jsp
© Copyright 2026 Paperzz