Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Electronic excitation of polyatomic molecules by fast electron impact Sze, Kong Hung 1988

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
831-UBC_1989_A1 S97.pdf [ 17.56MB ]
Metadata
JSON: 831-1.0060286.json
JSON-LD: 831-1.0060286-ld.json
RDF/XML (Pretty): 831-1.0060286-rdf.xml
RDF/JSON: 831-1.0060286-rdf.json
Turtle: 831-1.0060286-turtle.txt
N-Triples: 831-1.0060286-rdf-ntriples.txt
Original Record: 831-1.0060286-source.json
Full Text
831-1.0060286-fulltext.txt
Citation
831-1.0060286.ris

Full Text

ELECTRONIC EXCITATION OF POLYATOMIC MOLECULES BY FAST ELECTRON IMPACT by KONG HUNG SZE B.Sc, The Chinese University of Hong Kong, 1981 A THESIS SUBMITTED IN PARTIAL FULFILMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY in THE FACULTY OF GRADUATE STUDIES Chemistry We accept this thesis as conforming to the required standard THE UNIVERSITY OF BRITISH COLUMBIA December, 1988 ® Kong Hung Sze, .1988 In presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. 1 further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission. The University of British Columbia Vancouver, Canada Department DE-6 (2/88) ii ABSTRACT High resolution electron energy loss spectroscopy has been used to examine the inner-shell and valence-shell electronic excitation of a number of polyatomic molecules, including SFg, SeF6> TeFg, C\Fy C ^ F , C ^ C l , C ^ B r , C ^ I , Ni(CO)4> (CH^SO and SO2. The inner-shell and valence-shell electron energy loss spectra (ISEELS and VSEELS) were measured under low momentum transfer conditions with impact energy in the range of 1-3.7 keV and 0° scattering angle. Under these conditions the spectra are dominated by electric dipole-allowed transitions. The ISEELS spectra include all accessible core excitations of these molecules below 1000 eV equivalent photon energy. A number of specific investigations have been performed in order to extend present understanding of the physical nature of electronic excitation phenomena, in particular those involving inner-shell electrons. In addition the present work illustrates new applications of ISEELS to the study of chemical phenomena. In the investigation of (Coulombic) potential barrier effects in the "cage" molecules SFg, SeFg, TeFg and CIF^, f-type continuum shape resonances are observed for the first time in the spectra of TeFg and they show very different spectral behavior from the d-type continuum shape resonances observed in the spectra of SFg and SeFg. Consideration of both the ISEELS and VSEELS spectra indicates that there is a weakening of the potential barrier in going through the series from SFg to SeFg to TeFg. The Coulombic potential barrier model provides an extremely satisfactory understanding of (a) the co-existence of intense continuum shape resonances and intense Rydberg transitions plus direct ionization continuum; and (b) the number and symmetry of continuum shape resonances observed in the ISEELS spectra of CIF^. The physical significance of Coulombic potential barrier effects is further convincingly demonstrated by a comparison of the "central atom" inner-shell spectra of SFg, CIF^ and HCl. In contrast to earlier work, the present comparative study of the He(I) and He(II) photoelectron spectra and the VSEELS excitation spectra of the monohaloethylenes (i.e. G,H,X; X = F, Cl, Br and I) suggests that the HOMO orbital in C-H,I is predominantly iii of iodine 5pj_ (out-of-plane) character rather than of ir character whereas the reverse situation applies to the HOMO orbitals of C 2 H 3 F , C 2H 3C1 and C ^ B r . Based on a term value correlation analysis, substitutional effects are found to be most prominent in the a -type orbitals, while the n and Rydberg orbitals are less influenced. Linear correlations between the C - X bond strength and the term values for both inner-shell and valence-shell transitions to the a (CX) orbital are also observed. The ISEELS and VSEELS spectra of Ni(CO)4 are compared with the corresponding spectra in free CO. The C and O Is spectra of these two molecules show some notable similarities despite the very different manifold of final states available. In • * particular the inner-shell spectra of both molecules exhibit intense ls-> ir and Is -»a transitions. High resolution ISEELS has been used to obtain vibrational resolved C Is spectra of Ni(CO)4 and free CO. The implications and possibilities of studying d7r -»p7r back-bonding in transition metal carbonyl complexes by high resolution ISEELS spectroscopy are discussed. The inner-shell (S 2s, 2p, C Is) excitations of (CH^SO (DMSO) measured by ISEELS are compared with synchrotron radiation studies of the S Is photoabsorption spectrum. The pre-edge regions of these spectra are interpreted as excitations to common manifolds of virtual valence and Rydberg orbitals. A linear correlation between the S-C bond lengths and the term values of S 2 p 3 / 2 -» a (S-C) transitions is demonstrated for DMSO and a number of other sulfur compounds. The S 2p, 2s and O Is ISEELS spectra of S0 2 as well as the S Is photoabsorption spectrum are compared with and assigned according to the results of multichannel quantum defect theory calculations. The calculated energies and oscillator strengths of spectral features in these spectral regions are generally in good quantitative agreement with the measurements. iv TABLE OF CONTENTS Abstract ii Table of Content iv List of Tables .-. vi List of Figures xii Acknowledgments xvi Chapter 1 Introduction 1 1.1 General Considerations 1 1.2 Principles of Electron Energy Loss Spectroscopy using Fast Electrons 4 1.3 Comparison between Electron Energy Loss Spectroscopy and Photoabsorption 11 Chapter 2 Analysis of Electron Energy Loss Spectra 17 2.1 Features in the Electron Energy Loss Spectra 18 2.1.1 Transitions to Rydberg Orbitals 18 2.1.2 Transitions to Virtual Valence Orbitals 21 2.1.3 Delayed Onsets of Ionization Continua 23 2.1.4 Extended Energy Loss Fine Structures 24 2.1.5 Multiple Excitations and Ionizations 24 2.2 Comparison between Inner-Shell and Valence-Shell Electron Energy Loss Spectra 26 2.3 Potential Barrier Effects 28 2.4 Processes Related to Electronic Excitation 33 2.4.1 Ionization 33 2.4.2 Core Hole Decay Processes : X-ray fluorescence, Auger Decay, Autoionization and Dissociation 36 Chapter 3 Experimental Methods 41 3.1 Spectrometer, Operation and Spectral Acquisition 41 3.2 Energy Calibration 45 3.3 Sample Handling 45 V Chapter 4 Investigation of Potential Barrier Effects in "Cage" Molecules : 48 4.1 Inner-Shell and Valence-Shell Electronic Excitation of SFg, SeFg and TeFg 48 4.II Inner-Shell and Valence-Shell Electronic Excitation of CIF^ 105 Chapter 5 An Investigation of Substitutional Effects in Monohaloethylenes (C 2 H 3 X, X=F, Cl, Br and I) : 131 5.1 Inner-Shell Excitation by Electron Energy Loss Spectroscopy 131 5.II Photoionization by He(I) and He(II) Photoelectron Spectroscopy and Valence-Shell Excitation by Electron Energy Loss Spectroscopy 175 Chapter 6 A Case Study in Co-ordination Chemistry : Inner-Shell and Valence-Shell Electronic Excitation of Ni(CO)4 by High-Resolution Electron Energy Loss Spectroscopy 210 Chapter 7 Electronic Excitation of Sulfur Compounds 238 7.1 Inner-Shell and Valence-Shell Electronic Excitation of Dimethyl Sulfoxide 238 7.II High Resolution Inner-Shell Election Energy Loss Spectra of Sulfur Dioxide : A Comparison with Multichannel Quantum Defect Theory 274 Chapter 8 Concluding Remarks 307 References 309 vi LIST OF TABLES Table Page 3.1 Reference energies for inner-shell electron energy loss spectroscopy 46 3.2 Source and purity of samples 46 4.1.1 Symmetry of initial and final orbitals for dipole allowed transitions in MFg molecules of O n symmetry 52 4.1.2 Energies, term values and possible assignments for features in the S 2p spectrum of SFg 55 4.1.3 Energies, term values and possible assignments for features in the S 2s spectrum of SFg 56 4.1.4 Energies, term values and possible assignments for features in the F Is spectrum of SFg 56 4.1.5 Energies, term values and possible assignments for features in the F Is spectrum of SeFg 65 4.1.6 Energies, term values and possible assignments for features in the Se 3s spectrum of SeFg 65 4.1.7 Energies, term values and possible assignments for features in the Se 3p spectrum of SeFg 66 4.1.8 Energies, term values and possible assignments for features in the Se 3d spectrum of SeFg 67 4.1.9 Energies, term values and possible assignments for features in the F Is spectrum of TeFg 74 4.1.10 Energies, term values and possible assignments for features in the Te 3p spectrum of TeFg 75 4.1.11 Energies, term values and possible assignments for features in the Te 3d spectrum of TeFg 76 vii 4.1.12 Energies, term values and possible assignments for features in the Te 4s spectrum of TeFg „ 77 4.1.13 Energies, term values and possible assignments for features in the Te 4p spectrum of TeFg „ 77 4.1.14 Energies, term values and possible assignments for features in the-Te 4d spectrum of TeFg ; _ _ 78 4.1.15 Energies, term values and possible assignments for features in the VSEELS spectrum of SFg _ _ 95 4.1.16 Energies, term values and possible assignments for features in the VSEELS spectrum of SeFg „ 96 4.1.17 Energies, term values and possible assignments, for features in the VSEELS spectrum of TeFg 97 4.1.18 Summary of term values for- assignments•-of-the- ISEELS and VSEELS: spectra of SFg „. 100 4.1.19 Summary of term values for assignments of the ISEELS and VSEELS spectra of SeFg _ . 100 4.1.20 Summary of term values for assignments of the ISEELS and VSEELS •' spectra of TeFg 101 4.11.1 Calculated energies and local characters of the valence-shell molecular orbitals of C1F3 __ 109 4.11.2 Energies, term values and possible assignments for features in the CI 2p spectrum of CIF^ I l l 4.11.3 Energies, term values and possible assignments for features in the CI 2s spectrum of CIF^ , 116 4.11.4 Energies, term values and possible assignments for features in the F . l s spectrum of CIF^ 119 viii 4.11.5 Energies, term values and possible assignments for features in the VSEELS spectrum of C1F3 122 4.11.6 Summary of term values for assignments of the ISEELS and VSEELS spectra of C1F3 123 5.1.1 Energies, term values and possible assignments for features in the F Is spectrum of C 2 H 3 F 136 5.1.2 Energies, term values and possible assignments for features in the Cl 2p, 2s spectrum of C2H3C1 139 5.1.3 Energies, term values and possible assignments for features in the Br 3d spectrum of C 2 H 3 Br 144 5.1.4 Energies, term values and possible assignments for features in the Br 3p spectrum of C2HjBr 147 5.1.5 Energies, term values and possible assignments for features in the I 4d spectrum of ^ii-^l 150 5.1.6 Energies, term values and possible assignments for features in the C Is spectrum of C^H^F 156 5.1.7 Energies, term values and possible assignments for features in the C Is spectrum of C2H.3CI 157 5.1.8 Energies, term values and possible assignments for features in the C Is spectrum of C2H 3Br 158 5.1.9 Energies, term values and possible assignments for features in the C Is spectrum of CjH^I 159 5.1.10 C-X bond strengths and C(CX) ls-+a*(CX) term values for the monohaloethylenes 172 5.II.1 Vertical ionization potentials and associated vibrational structures in the first band of the He(I) photoelectron spectrum of C 9 H , F 181 ix 5.11.2 Vertical ionization potentials and associated vibrational structures in the first band of the He(I) photoelectron spectrum of C2H.JCI 182 5.11.3 Vertical ionization potentials and associated vibrational structures in the first band of the He(I) photoelectron spectrum of C2HjBr 183 5.11.4 Vertical ionization potentials and associated vibrational structures in the first band of the He(I) photoelectron spectrum of C2H2I 184 5.11.5 Energies, term values and possible assignments for features in the VSEELS spectrum of C 2 H 3 F 190 5.11.6 Energies, term values and possible assignments for features in the VSEELS spectrum of CjHjCl 194 5.11.7 Energies, term values and possible assignments for features in the VSEELS spectrum of C 2 H 3 Br 197 5.11.8 Energies, term values and possible assignments for features in the VSEELS spectrum of CjH^I 201 5.11.9 Summary of term values for assignments of the ISEELS and VSEELS spectra of the monohaloethylenes 204 5.11.10 C - X bond strengths and 2a" •* 0 (CX) term values for the monohaloethylenes 207 6.1 Energies, term values and possible assignments for features in the C Is spectrum of Ni(CO)4 and CO 217 6.2 Energies, term values and possible assignments for features in the O Is spectrum of Ni(CO)4 and CO 218 6.3 Parameters of the least squares fit of the C Is-* 7r transition in the C Is ISEELS spectrum of CO 221 6.4 Parameters of the least squares fit of the C Is-* it transition in the C Is ISEELS spectrum of Ni(CO)4 221 X 6.5 Averaged vibrational frequencies of selected isoelectronic transition metal carbonyl species 225 6.6 Energies, term values and possible assignments for features in the Ni 3p spectrum of Ni(CO)4 230 6.7 Energies, term values and possible assignments for features in the VSEELS spectrum of Ni(CO)4 233 6.8 Parameters of the least squares fit of the VSEELS spectrum of Ni(CO)4 ... 234 7.1.1 Calculated energies of the molecular orbitals of DMSO 244 7.1.2 Valence atomic orbital coefficients and local characters of 8a", 15a' and 16a' virtual valence molecular orbitals of DMSO 245 7.1.3 Energies, term values and possible assignments for features in the S 2p, 2s spectrum of DMSO 250 7.1.4 Energies, term values and possible assignments for features in the C Is spectrum of DMSO 254 7.1.5 Energies, term values and possible assignments for features in the O Is spectrum of DMSO 258 7.1.6 Energies, term values and possible assignments for features in the S Is spectrum of DMSO 261 7.1.7 R(S-C) bond lengths and a (S-C) energies (8) for various organo-sulfur compounds 264 7.1.8 R(S-O) bond lengths and a (S-O) energies (5) for various sulfur-oxygen compounds 266 7.1.9 Energies, term values and possible assignments for features in the VSEELS spectrum of DMSO 269 7.1.10 Summary of term values for assignments of the ISEELS and VSEELS spectra of DMSO 272 7.11.1 Transitions in S0 2 from the *Aj ground state for C 2 v symmetry 283 7.11.2 Experimental and calculated data for S 2p excitation in S0 2 285 xi 7.11.3 S 2p excitation of S0 2 at high resolution (164-167 eV) 292 7.11.4 Experimental and calculated data for S 2s excitation in S0 2 296 7.11.5 Experimental and calculated data for O Is excitation in S0 2 299 7.11.6 Summary of experimental and calculated term values for the core excitation spectra of S0 7 304 xii LIST OF FIGURES Figure Page 1.1 Schematic diagram of the electron energy loss experiment 6 1.2 Wavelength resolution (AX) plotted against excitation energy for fixed values of energy resolution (AE) 14 1.3 High resolution N Is ISEELS spectrum of N 2 15 2.1 Inner-shell and valence-shell electronic excitation and ionization 19 2.2 ISEELS spectrum of C H 3 I 25 2.3 S 2p ISEELS spectrum of SF 6 30 2.4 Schematic representation of the effective potential barrier in SFg 31 2.5 Photoionization process and photoelectron spectrum 35 2.6 Shake-up and shake-off processes 37 2.7 Core hole decay processes : X-ray fluorescence, Auger decay, autoionization and dissociation 39 3.1 High-resolution electron energy loss spectrometer 42 4.1.1 Comparison of (a) F Is; (b) S 2s; (c) S 2p ISEELS spectra of SFg 53 4.1.2 (a) Medium-resolution; (b) high-resolution S 2p ISEELS spectra of SF g showing the Rydberg structures 57 4.1.3 Comparison of (a) F Is; (b) Se 3s; (c) Se 3p; (d) Se 3d ISEELS spectra of SeF6 61 4.1.4 Correlation between F Is IPs and electronegativities of central atom for MF„ type molecules 63 4.1.5 Detailed Se 3d ISEELS spectrum of SeFfi 70 4.1.6 Comparison of (a) F Is; (b) Te 3p; (c) Te 3d; (d) Te 4s; (e) Te 4p; (0 Te 4d ISEELS spectra of TeF g 72 4.1.7 Detailed Te 4d ISEELS spectrum of TeF g 82 4.1.8 Comparison of the F Is ISEELS spectra of (a) SF,; (b) SeF,; (c) TeF, .. 84 xiii 4.1.9 Comparison of (a) S 2s spectrum of SFg; (b) Se 3s spectrum of SeFg", (c) Te 4s spectrum of TeFg 85 4.1.10 Comparison of (a) S 2p spectrum of SFg; (b) Se 3p spectrum of SeFg; (c) Te 4p and (d) Te 3p spectra of TeFg 86 4.1.11 Comparison of (a) Se 3d spectrum of SeFg; (b) Te 4d and (c) Te 3d spectra of TeFg 87 4.1.12 Long range VSEELS spectra of (a) SFg; (b) SeFg; (c) TeFg 90 4.1.13 Detailed VSEELS spectrum of SFg 91 4.1.14 Detailed VSEELS spectrum of SeFg 92 4.1.15 Detailed VSEELS spectrum of TeFg 93 4.11.1 The structure of the C1F3 molecule 106 4.11.2 (a) Low-resolution Cl 2p,2s; (b) high-resolution Cl 2p ISEELS spectra of C1F3 : 110 4.11.3 - Detailed Cl 2s ISEELS spectrum of C1F3 115 4.11.4 F Is ISEELS spectrum of C1F3 118 4.11.5 (a) Detailed; (b) long range VSEELS spectra of C1F3 121 4.11.6 Comparison of the (a) Cl 2p; (b) Cl 2s; (c) F Is ISEELS spectra of C1F3 „ 127 4.11.7 Comparison of potential barrier effects in central atom core excitation spectra of SFg (S 2p), C1F3 (Cl 2p) and HCl (Cl 2p) 129 5.1.1 F Is ISEELS spectrum of C 2 H 3 F 135 5.1.2 (a) High-resolution Cl 2p; (b) low-resolution Cl 2p, 2s ISEELS spectra of C 2H 3C1 138 5.1.3 (a) High-resolution; (b) low-resolution Br 3d ISEELS spectra of C 2 H 3 Br ... 143 5.1.4 Br 3p ISEELS spectrum of C ? H,Br 146 xiv 5.1.5 (a) High-resolution; (b) low-resolution I 4d ISEELS spectra of C 2 H 3 I 149 5.1.6 Low-resolution C Is ISEELS spectra of (a) C 2 H 3 F ; (b) C 2H 3C1; (c) C 2 H 3 Br; (d) C 2 H 3 I ; and (e) C 2 H 4 154 5.1.7 High-resolution C Is ISEELS spectra of (a) C 2 H 3 F ; (b) C 2H 3C1; (c) C 2 H 3 Br; (d) C-H3I; and (e) C 2 H 4 155 5.1.8 Correlation between term values for C Is (average) excitations and the electronegativity of the halogen substituent in the monohaloethylenes 169 5.1.9 Correlation between C - X bond strength and C(CX) Is -> 0 (CX) term values in the monohaloethylenes and methyl halides 171 5.11.1 He(I) and He(II) photoelectron spectra of C 2 H 3 F , C ^ C l , CjHjBr and C 2 H 3 I 179 5.11.2 Detailed first band He(I) photoelectron spectra of C 2 H 3 F , C 2H 3C1, C 2 H 3 Br and C 2 H 3 I 180 5.11.3 Energy levels correlation diagram for C 2 H 4 , C 2 H 3 F , C 2H 3C1, C 2 H 3 Br and C 2 H 3 I 186 5.11.4 Long range and detailed VSEELS spectra of C 2 H 3 F 189 5.11.5 Long range and detailed VSEELS spectra of C 2H 3C1 193 5.11.6 Long range and detailed VSEELS spectra of C 2 H 3 Br 196 5.11.7 Long range and detailed VSEELS spectra of C 2 H 3 I 200 * 5.11.8 Correlation between C - X bond strength and a (CX) term values in the valence excitation spectra of monohaloethylenes and methyl halides 206 6.1 Qualitative molecular orbital energy level diagram for Ni(CO)4 213 6.2 C Is ISEELS spectra of (a) CO; (b) Ni(CO)4 215 6.3 O Is ISEELS spectra of (a) CO; (b) Ni(CO)4 216 6.4 High-resolution ISEELS spectra of the C ls-» n transitions in (a) CO; (b) Ni(CO)4 220 XV 6.5 (a) Original; (b) background subtracted Ni 3p ISEELS spectra of Ni(CO)4 . 229 6.6 VSEELS spectra of (a) CO; (b) Ni(CO)4 232 7.1.1 The structure of the DMSO molecule 243 7.1.2 Comparison of S 2p, 2s, C Is and O Is ISEELS spectra with S Is photoabsorption spectrum of DMSO 247 7.1.3 High-resolution S 2p and low-resolution S 2p, 2s ISEELS spectra of DMSO 249 7.1.4 High-resolution and low-resolution C Is ISEELS spectra of DMSO 253 7.1.5 0 Is ISEELS spectra of DMSO 257 7.1.6 S Is photoabsorption (ionization current) spectrum of DMSO obtained using synchrotron radiation 260 * 7.1.7 Correlation between a (C-S) resonance position relative to the edge (6) and bond length for S 2p spectra of organo-sulfur compounds 263 7.1.8 Detailed and long range VSEELS spectra of DMSO 268 7.11.1 Calculated quantum defects for eigen channels in the representation of S0 2 '. 280 7.11.2 (a) S 2p excitation spectrum of S0 2 (163-176 eV); (b) multi-channel quantum defect theory calculation 284 7.11.3 Comparison of the (a) photoabsorption; and (b) high-resolution electron energy loss S 2p excitation spectra of SO-> in the energy range 164-167 eV 291 7.11.4 (a) S 2p, 2s excitation spectrum of S0 2 (160-240 eV); (b) multi-channel quantum defect theory calculation 295 7.11.5 (a) O Is excitation spectrum of S0 2 (525-545 eV); (b) multi-channel quantum defect theory calculation 298 7.11.6 (a) S Is excitation spectrum of S0 2 (1060-1240 eV); (b) multi-channel quantum defect theory calculation 301 7.11.7 Comparison of the S 2p, 2s and O Is ISEELS spectra of S0 2 with the S Is photoabsorption spectrum obtained using synchrotron radiation 303 xvi ACKNOWLEDGEMENTS I would like to express my sincere thanks to Dr. C.E. Brion for his interest, assistance, encouragement and supervision throughout the course of my studies. It has been a pleasure to have worked with him and with the other members of his research group. Special thanks are due to Dr. Suzannah Daviel for her earlier work in the development of the high performance electron energy loss spectrometer used in this work and for introducing me to the instrument. I would also like to acknowledge many helpful discussion with Dr. R.N.S. Sodhi and to thank him for suggesting the study of the CIF^ molecule. Much appreciation is due to Dr. A. Katrib of the University of Kuwait who provided the He(I) and He(II) photoelectron spectra of the monohaloethylenes and for his help in recording the electron energy loss spectra of these molecules during a short visit to UBC in the summer of 1986. Dr. G. Cooper is thanked for providing stimulating and helpful discussions concerning the analysis of the spectra of Ni(CO) 4. Dr. A. Hitchcock, Dr. S. Bodeur and Dr. M. Tronc are thanked for supplying the S Is photoabsorption spectrum of dimethyl sulfoxide as measured on the ACO synchrotron in Paris. Additional thanks are due to Dr. A. Hitchcock who also showed great interest and gave many helpful suggestions and useful discussions concerning other studies in this work. Professor J.M. Li and Mr. X.M. Tong of the Institute of Physics, Chinese Academy of Sciences, Beijing, are thanked for performing multichannel quantum defect calculations on the S 0 2 molecule. I should like to express my gratitude to the capable staff of the mechanical and electronic workshops for their assistance in the maintenance of the spectrometer. Financial support in the form of a University of British Columbia Graduate Fellowship is also gratefully acknowledged. The research work was also supported by operating grants from The Natural Sciences and Engineering Research Council of Canada. Finally, I wish to thank my parents and Winnie for their patience and encouragement This thesis is dedicated to them. 1 CHAPTER 1 INTRODUCTION 1.1 General Considerations Knowledge of the energies and transition probabilities for the production of electronically excited and ionized states of atoms and molecules is of fundamental importance. In addition to the general interest in extending basic knowledge, such information is needed in a large number of areas of practical interest For example, accurate experimental spectral information is essential for the development and testing of new quantum mechanical procedures which calculate electronic state energies and the associated cross sections for the processes involved [1]. Such information is also urgently needed to permit an adequate understanding and accurate modeling of the interaction of radiation with matter taking place in a wide variety of phenomena in radiation chemistry, physics and biology [2]. Specific important areas of application include environmental protection, radiation-induced decomposition, radiation therapy, space chemistry and physics, electron and X-ray microscopy, fusion processes, plasmas, high temperature chemistry, the nuclear industry, etc. However, until quite recently, an examination of the scientific literature revealed that only very limited quantitative spectral information was available, particularly for molecules, in the vacuum UV and soft X - ray regions of the spectrum, o i.e. at excitation energies above ~20 eV, that is for wavelengths shorter than ~600 A. These higher energy regions are important as they cover the range of many highly excited electronic states of valence electrons as well as all inner-shell transitions and most ionization and fragmentation processes. In the past decade, an increasing amount of such spectral information has become available [3,4], in particular for molecular inner-shell excited states [5-9]. The influx of new data is the direct result of the increasing availability and use of novel experimental techniques involving tuneable synchrotron radiation [10,11] as well as the use of electron energy loss spectroscopy which utilizes the virtual photon field of a fast electron beam as the excitation source [6-9]. It is with 2 experiments based on the latter technique that the present work is concerned. As early as 1904 electrons were directed to a metal surface and the transmitted electrons were analyzed for energy losses by a magnetic deflection technique [12]. By 1920 electron impact methods had produced accurate measurements of excitation and ionization potentials of many molecular and atomic gases. Although most early experiments utilized electron beams with quite broad Maxwell-Boltzmann energy spreads of ~1 eV full-width half-maximum (FWHM), it soon became obvious that the use of a monochromatic" electron beam would open up new avenues for experimental and theoretical advancement The rise of quantum mechanics and the development of the Bethe theory [13], which provided the initial formulation for the analysis of the interaction between high energy electrons and matter, injected increasing interest in the investigation of electron impact phenomena in the mid 1930's. During the following 20 years, the intervening war and subsequent interest in nuclear physics left the field practically dormant. However, since the late 1950's there has been a renewed and growing interest in the field of electron impact phenomena because of a heightened awareness of the important role of electron-molecule and electron-atom interactions in many practically relevant areas including radiation chemistry, upper-atmosphere phenomena and plasma physics. In addition, the development of more sophisticated experimental techniques, including advances in high vacuum systems, low energy electron optics, energy analysers, solid-state electronics and detectors such as channel electron multipliers, permitted the study of single particle collision processes and thus provided the basis for a rapid growth in all areas of electron spectroscopy. In the past decade, the Ph.D. works of Wight [14], Hitchcock [15] and Sodhi [16] in this laboratory and the work of King and Read et al. at the University of Manchester [17] have amply demonstrated that electron energy loss spectroscopy (EELS) is a very suitable technique for the study of gas phase atomic and molecular inner-shell and valence-shell electronic spectra. Comprehensive reviews on the most recent as well as early inner-shell electron energy loss spectroscopy (ISEELS) and valence-shell electron energy loss spectroscopy (VSEELS) studies are available in a number of publications [5-9,17-33]. The high resolution ISEELS technique is now sufficiently well 3 developed to permit its application to the detailed investigation of physical and chemical phenomena. The unique features of the ISEELS technique present many interesting and useful possibilities. In the present study, the ISEELS spectra (below energy loss E = ~ 1000 eV) as well as the VSEELS spectra have been measured under low momentum transfer conditions (1.0-3.7 keV impact energy and 0° scattering angle) for the following molecules : MF^ (M = S, Se, Te), CIF3, C ^ X (X = F, Cl, Br, I), Ni(CO)4, (CH^SO and S0 2 . In addition to the primary interest of extending fundamental spectroscopic knowledge for these molecules, several specific investigations have also been undertaken to study a number of chemical and physical effects. For example, in Chapter 4, the physical aspects of potential barrier effects [34-36] in the "cage" molecules MF^ (M = S, Se, Te) and CIF^ have been examined by a detailed consideration of the features in both the discrete and continuum regions of the respective ISEELS and VSEELS spectra. In addition, the chemical effect of f orbital participation in TeF^ is observed. In Chapter 5, the substitutional effects (inductive and resonance) on the occupied valence and unoccupied virtual valence orbitals of the monohaloethylenes (C^^X, X = F, Cl, Br, I) have been investigated using the combined techniques of ISEELS, VSEELS and photoelectron spectroscopy. The relationship between C - X bond strengths and the a (C-X) resonance term values is also discussed. In Chapter 6, possibilities for detailed investigation of drr -*pir back-bonding in transition metal carbonyl and other complexes by high resolution ISEELS technique are illustrated by a study of the ISEELS and VSEELS spectra of Ni(CO)4. In Chapter 7 high resolution ISEELS is applied to the study of DMSO and S0 2 . In particular the correlation between S-C bond lengths and the o (S-C) resonance term values is examined for various sulfur compounds. Finally, the powerful combination of experiment and high level quantum chemical calculations is illustrated by a comparison of the ISEELS spectra of SO 0 with multichannel quantum defect theory calculations. 4 1.2 Principles of Electron Energy Loss Spectroscopy using Fast Electrons Electron energy loss experiments are conceptually simple : a monoenergetic beam of electrons is used to excite the target species and the excitations are detected as energy losses (E) in the scattered electron beam. The process may be represented as e + A B —» A B + e » 1 1 E 0 ( E o - E > » where AB and AB are the initial and final (excited) states of the target species respectively; and e is the colliding electron which has energy E Q and (EQ - E) before and after the collision respectively. Energy losses up to the energy of the incident electron E o can in principle occur for accessible excitation channels of the target species, including transitions to both bound and continuum states. A. comparison with the corresponding (resonant) photoabsorption process hv + A B -> A B * 1-2 E indicates that the energy loss E of the incident electron is analogous to the photon energy (hv = E) required to produce the excited state of quantum energy E. In addition, while the two processes are physically different, it has been shown that there is a quantitative relationship between photoabsorption and electron energy loss spectra obtained under experimental conditions involving negligibly small momentum transfer to the target (i.e. high impact energy and small scattering angle). Detailed treatments of this by now well known topic have been given in earlier works [6-9,13,14,37-41] and therefore only a brief oudine of the fundamental concepts will be given in the following presentation. 5 1.2.1 Differential Cross-section A schematic representation of the EELS experiment is shown in Figure 1.1. When an electron beam passes through a volume of target gas, elections are scattered out of the beam anisotropicaily due to both elastic and inelastic collisions. As a result of inelastic collisions, the target species may undergo a variety of processes such as excitation, ionization, dissociation, etc. The transition probability of each process is usually measured by the scattering cross section for that process. Experimentally, the magnitude . of the scattered electron current I measured at energy loss E within a solid angle dfl about the scattering angle 6 after a beam of incident elections of intensity I transverses a distance w through the target gas of concentration c is given by [42] I = I o e x p ( - d f w c ) ' u where (dcr/dft) is the differential cross section (DCS) for excitation to a particular excited state of quantum energy E. Eq. 1.3 has a similar form to the well-known Beer's law [43] for photoabsorption in which the photoabsorption cross section, a ^, for the corresponding excitation is measured. It should be noted that for an electron impact 2 excitation to a continuum state the double differential cross section (d a/dfidE) in the energy range dE about E is measured. Scattering cross sections of the form (do/cin) will be used throughout the following presentation although the underlying ideas apply equally well for excitations to continuum states. 1.2.2 Oscillator Strengths The transition probability for a particular excitation is also often conveniently given in the form of the oscillator strength. In a classical picture, the oscillator strength is defined as the number of elections in free oscillation at a particular frequency. Therefore the total oscillator strength is equal to the number of elections in the target system. Although modern concepts reject the identification of a characteristic frequency with a Figure 1.1 Schematic diagram of the electron energy loss experiment 7 given election in a system, the concept of oscillator strength is retained in quantum theory as a most convenient measure of transition probability. In practice the total oscillator strength (i.e. for all processes) is normalized to the total number of electrons in the system. This is known as the Thomas-Reiche-Kuhn (TRK) sum rule [44]. For a dipole transition occurring in a photoabsorption experiment, an optical oscillator strength f(0) is used, which is defined in quantum mechanics as : 2 f (0) = 2E s=l 1.4 where E is the excitation energy, ¥ Q and V n are the ground and excited state wavefunctions respectively of an N electron target species and r g give the coordinates of the electrons. Foi excitation due to electron impact as give in Eq. 1.1, Bethe [13] introduced the concept of the generalized oscillator strength f(K) which is expressed as 2E f ( K ) = — K e (K) on v ' 1.5 where s=l 1.6 and K is the momentum transfer in the collision, given by 1.7 K = k - k o n i.e. K 2 = k 2 + k 2 - 2k k cos 6 o n o n The quantities k and k are the momenta of the incident and scattered election o n respectively and 6 is the mean scattering angle. 8 1.2.3 Bethe-Born Approximation Quantum mechanical treatment of the electron scattering process shows that there is a quantitative relationship between f(K) and the differential scattering cross section (da/dfl). This was first derived by Bethe [13], and later reviewed in detail by Inokuti [38,3?]. The Bethe theory for electron scattering is based on the first Born approximation and applies to collisions involving electrons with high incident energy. For sufficiently fast collisions, it can be assumed that the influence of the incident electron upon a target species is a small external perturbation which results in a sudden transfer of energy or momentum into the electrons of the target species. Therefore, the incident as well as the scattered electron waves may be regarded as negligibly distorted by the interaction and they can be described by plane waves. Within this assumption, the DCS for an inelastic scattering process as represented by Eq. 1.1 is given by [38] Hrr k I e I 2 __(9) = 4 _ _ _ _ _ _ 1.8 < - kQ K 4 Combination with Eq. 1.5 gives da 2 k n 1 -» ^ ( 6 ) = n-— f(K) 1.9 o The Bethe-Born equation shown above is valid for high-incident electron energy. In practice this means E Q > 5-10 times the excitation energy E and therefore will apply in most cases of the present work. 1.2.4 Limit Theorem Bethe [13] also showed that the generalized oscillator strength f(K) can be 2 expanded in a power series of K in which the first term is equal to the optical oscillator strength f(0) 9 f(K) = f(0) + ( e ^ e ^ K 2 + B K 4 + > 1.10 where e r £ 2 . are expressions of the form 1 N em = 77 < v n 1 2 rs I ¥ > 1.11 m m ! n j s o and B is also a function containing several £ m terms. The quantity £ m is regarded as th the m order multipole matrix element (i.e. m=l is electric dipole, m = 2 is electric quadrupole, etc.). It can be seen from Eq. 1.10 that L i m f ( K ) = f(0) . 1.12 K2->o Therefore in the limit of negligibly small momentum transfer ^ £ ( 0 ) = --rf(0) • 1.13 ^ E k K 2 o Under these conditions (i.e. the optical limit) only electric dipole-allowed transitions will be observed in the electron energy loss spectrum (i.e. EELS and photoabsorption spectra w i l l exhibit the same transitions). However it should be noted that the energy loss and photoabsorption spectra differ in relative intensities due to the electron impact kinematic 2 factor (2/E)(k o/k n)(l/K ) as given in the Bethe-Born relationship (Eq. 1.13). Based on the ideas discussed above, three general types of experiment have been carried out Firstly, Lassettre et al. [23,24,45] obtained optical oscillator strengths for a number of atomic and molecular discrete transitions by studying the scattering intensity for a particular transition as a function of the scattering angle (i.e. as a function of momentum transfer) and extrapolating to zero momentum transfer. Alternatively Hertel and Ross [46,47] made similar measurements for discrete valence transitions in alkali metals by 2 variation of the impact energy and extrapolating to K. = 0. However, the procedure of 10 extrapolation was tedious and involved considerable uncertainty in some cases ' due to the long extrapolations needed under the experimental conditions used. A much more useful approach was taken by Van der Wiel et al. [48-50] who showed that it was possible to avoid extrapolation procedures by making direct electron scattering measurements under conditions which sufficiently closely approximate the optical limit In practical terms this means using fast electrons, (i.e. high impact energy E Q of several keV) and measuring d at zero degree mean scattering angle. In this third method all terms except f(0) on the right hand side of Eq. 1.10 become negligible (typically <1%) as a result of the experimental conditions selected. Under such conditions of high impact .energy E Q and zero degree scattering angle (i.e. the momentum transfer is sufficiently small) the electron scattering differential cross section is simply related to the optical oscillator strength by [11,38,51] : d Q = « E f(0) . 1.14 where a and b are constants that depend on the impact energy and the angular acceptance of the electron energy loss spectrometer (i.e. independent of target species). In actual experimental situation, b usually has a value between 2 and 3 [11,51]. As is well known [38], the optical oscillator strength f(0) is direcdy related to the cross-section for photoabsorption a ^ by : f(0) = _ - _ a . 1.15 K e h Therefore an electron energy loss spectrum produced under conditions of negligible momentum transfer should be similar to the corresponding photoabsorption spectrum, differing only by a relative decrease in intensity which is proportional to the inverse of the energy loss cubed (or in practice a somewhat lower power, see above). This variation is very small (typically less than a 15 % variation) for inner-shell excitations over the 11 ~40 eV range typically studied in the present ISEELS work. Therefore, as far as relative intensities are concerned, the ISEELS and the corresponding photoabsorption spectra will be very similar, in particular when a short spectral region is studied. Furthermore, it should be noted that not only dipole allowed transitions are induced in atomic or molecular targets by fast electron impact at small momentum transfer, but also their absolute optical intensities can be obtained by kinematic conversion of the differential electron scattering cross-sections (intensities). Thus it is possible to carry out quantitative optical spectroscopy without any photons. Reviews of measurements of absolute oscillator strengths for photoabsorption and photoionization processes as obtained by fast electron impact have recently been published [3,4,6]. The ability of fast electrons to induce dipole allowed (i.e. optical) transitions can also be qualitatively understood in terms of the "virtual photon field" model [6,9]. In high-energy electron scattering at small scattering angles the target experiences a sharply pulsed (in the time domain) electric field. The frequency components of this field can be obtained by a Fourier transformation, and they represent a wide range of frequencies of essentially equal, intensity. The impulsive field of a high-energy electron therefore represents an ideal "continuum light source" of virtual photons and the resulting electron energy loss spectrum therefore resembles an optical absorption spectrum in the sense that dipole selection rules apply and the intensities are related to optical transition probabilities by purely kinematic factors as given in Eq. 1.13. 1.3 Comparison between Electron Energy Loss Spectroscopy and Photoabsorption The quantum states of atoms and molecules have traditionally been investigated by the methods of optical spectroscopy using photoabsorption techniques. There has, however, until quite recently been relatively little information on transitions in the vacuum UV and X-ray regions of the spectrum, which cover many highly excited states due to excitations of the valence electrons as well as all inner-shell processes. The reason for this shortage of information has been the limited availability of continuum light sources at energies 12 above 20 eV, until the comparatively recent advent of tuneable synchrotron radiation. Therefore, in the past two decades, a number of new techniques have been developed to provide optical data in this region. In particular electron energy loss spectroscopy under conditions of low momentum transfer (i.e. high impact energy and small scattering angle) has emerged as a viable and in some cases superior alternative to the use of tuneable synchrotron radiation for spectroscopy at vacuum UV and soft X-ray energies [4,6-9]. Since the first time synchrotron radiation was used for optical experiments by Hartman and Tomboulian [52] in 1953, this earlier unwanted by-product of high energy electron accelerators has been most usefully exploited in the form of storage-ring or synchrotron radiation. Such sources provide an intense smooth continuum [10,11] from the infrared to the X-ray region depending on the energy of the accelerator. However, significant difficulties still exist which limit the efficient usage of synchrotron radiation. The challenge in effectively using such intense sources has centered around the problem of efficient monochromatization of the short-wavelength radiation. The main difficulty is the extremely low reflectivity of optical component (mirrors and gratings) in the monochromator at these wavelengths, which usually results in severe attenuation of the intense synchrotron radiation. Consequently, energy resolution is often sacrificed to obtain sufficient intensity at shorter wavelengths. Additional difficulties include order overlapping, stray light corrections, absorption by optical components (particularly near the C Is edge, due to surface contamination) as well as problems associated with calibration of the photon energy scale. In addition usage of synchrotron radiation . is limited by the remote location and limited beam time available to most users. These limitations are avoided in the relatively inexpensive laboratory based electron energy loss technique. A further advantage of the EELS technique is that excitation over a wide range of energies from the extreme IR to the X-ray region can be covered using the same instrument in a single scan. In contrast optical methods require several different types of monochromator and grating to cover the same energy regions. Since electrons and photons are monochromated by different physical techniques, 13 the energy resolution behavior in EELS and optical experiments is very different In EELS the spectral resolution is usually expressed in terms of the energy resolution AE, which is constant regardless of the excitation energy E. In contrast in photoabsorption spectral resolution is represented by a constant wavelength resolution AX. The wavelength resolution AX in the photoabsorption process is related to an equivalent energy resolution AE by [14] : 2 AX AE = E - . L 1 6 This behavior is illustrated in Figure 1.2. Eq. 1.16 implies that the energy resolution becomes increasingly less favorable as E increases in optical experiments. In contrast AE is constant and independent of E in EELS experiments. As a result EELS has been used most effectively for the study of second-row (C to Ne) K shell and third-row (S to Ar) L shell excitations in atoms and molecules. These transitions occur in the energy loss range ~ 180-1000 eV where electron impact resolution is presently superior to that typically obtainable in photoabsorption. For example, Figure 1.3 shows an ISEELS * spectrum of the N Is -• ir band of N 2 measured in the present work at an o experimental resolution of 0.055 eV (i.e. AX = 0.004 A). Note that the present vibrationally resolved band structure of N 2 is the highest resolution yet obtained and performance with vibrational resolution is still not attainable by optical techniques in this spectral region. The superior resolution of ISEELS is also demonstrated in the study of the S 2p excitation of S0 2 in Chapter 7.II (see Figure 7.II.3). Such high resolution not only allows the study of vibrational structures of inner-shell excited molecules but also permits detailed study of Rydberg states as well as accurate determination of natural linewidths [17] of inner-shell excited states. It is of particular importance to note that the non-resonant property of electron impact excitation (Eq. 1.1) means that EELS studies are not subject to line saturation effects [38], which result in spurious intensities in optical spectra for transitions having natural linewidths smaller than the instrumental resolution. Such a situation is especially 0.001 0.01 0.1 1 10 100 Resolution A \ A Figure 1.2 Wavelength resolution (AX) plotted against excitation energy for fixed values of energy resolution (AE) (taken from Ref. [9]). The shaded area marks the excitation energies where the best electron impact resolution is presently superior to the best photoabsorption resolution. t CO H LU EQUIVALENT WAVELENGTH X(A) 30.95 30 .90 30.85 N I s — " 77"' . • • • • • • • i r V = 0 r "i r N E 0 = 2 0 0 0 e V A E = 0.055 eV equivAX = 0 .004A • » •• • J L -I 1 L J I L J I L J _ J L 4 0 0 . 5 401.0 401.5 ENERGY LOSS (eV) 4 0 2 . 0 Figure 1.3 High-resolution N 1s ISEELS spectrum of N2-16. likely at the shorter wavelengths corresponding to inner-shell photoabsorption. A further advantage that electron impact excitation has over photoabsorption is that under conditions of low impact energies and large scattering angle dipole forbidden and/or spin forbidden transitions can be observed in the EELS spectrum. In this respect, EELS offers a more complete probe of the quantum states of. the target species. The main limitations of the EELS technique are : 1. The relative drop in intensity with increasing "photon" energy (see Eq. 1.14). 2. Backgrounds from stray electrons. 3. EELS methods are most suitable for gaseous samples and are less easily applied to solids and surfaces than photoabsorption. 4. . Energy resolution is relatively poor at low energy loss (i.e. low photon energy) compared with photoabsorption. 17 CHAPTER 2 ANALYSIS OF ELECTRON ENERGY LOSS SPECTRA Electron energy loss spectra provide a wealth of information on the energies as well as transition probabilities for the production of the various quantum states of molecules. These quantities are closely related to1 the energies and relative spatial extents of the molecular orbitals involved. Therefore, any high quality theoretical study which offers accurate concomitant calculations of transition energies and probabilities will be most useful in the analysis and assignment of the experimental measurements. An example of the application of such modern high quality calculations is demonstrated in Chapter 7.II in which the inner-shell excitations of S0 2 are assigned with the assistance of quantum mechanical calculations based on multichannel quantum defect theory [53]. Though the value of modern high quality calculations is readily acknowledged, most are still highly extensive and correspondingly computationally expensive. Therefore, to date, such calculations are not common especially for large polyatomic molecules. From a pragmatic point of view, in order to use the EELS technique for the study of chemical problems in larger molecules, it is useful and necessary to be able to analyze the recorded spectra with the aid of simple qualitative and semiquantitative principles without involving extensive calculations. Most earlier ISEELS and many VSEELS spectra have thus far been interpreted on an empirical basis using the concepts of quantum defects and term values [54-56]. In addition to these, the use of relatively simple calculations such as GAUSSIAN 76 [360] and MS-Xo calculations, etc. can usually give a helpful qualitative understanding of the excitation processes involved and aid in the assignment In this chapter, the empirical concepts that are important in the analysis of the EELS spectra will be described. 18 2.1 Features in the Electron Energy Loss Spectra In electron energy loss spectroscopy, the primary processes associated with the excitation of electrons from the normally occupied valence-shell or inner-shell (core) molecular orbitals (MOs) are examined. These processes include transitions to the normally unoccupied Rydberg and virtual valence orbitals as well as ionizations to continuum states. These are shown graphically in Figure 2.1. Each of these excitation processes causes a loss of a characteristic amount of energy from the fast incident electrons and contributes to spectral features observed at the corresponding energy in the electron energy loss spectrum. In this section a description of the spectral features observed in the EELS spectra due to various kinds of processes is presented. 2.1.1 Transitions to Rydberg Orbitals Molecular Rydberg orbitals are large, diffuse atomic-like orbitals, which extend well beyond the bounds of ground state molecules. Below the ionization potential (IP) of a particular occupied level for a many-electron species, transitions to the Rydberg orbitals will result in a series of features converging to the IP. The term value for a particular transition, which is defined as the energy difference between the IP and transition energy (E) for the given Rydberg series, can be Fitted into the Rydberg equation R T = I P - E = 2.1 ( n - 6 , ) 2 where R is the Rydberg constant (13.605 eV) and 6^  is the quantum defect of the s, p, d etc. type of Rydberg orbital as represented by their angular momentum quantum number A. Eq. 2.1 was first used by Rydberg [44] to fit the spectra of alkali atoms on an empirical basis. The close similarity of this equation to the hydrogen atom formula originally postulated by Bohr [57] and predicted by quantum mechanics suggests that as in hydrogen, the transitions are characteristic of a one-electron system in which the excited electron is at so large a distance from the positively charged ion core that the 19 / / / / / / / / / / / IONIZATION CONTINUA ' RYDBERG VIRTUAL VALENCE VALENCE S CORE UNOCCUPIED LEVELS GROUND STATE OCCUPIED LEVELS * - Valence-shell excitation or ionization +- Inner-shell excitation or ionization Figure 2.1 Inner-shell and valence-shell electronic excitation and ionization 20 core appears as a point charge. It is this feature of relative size, i.e. a radius much larger than that of the core or valence orbitals, which qualitatively distinguishes a Rydberg orbital in an atom or molecule. In this treatment (Eq. 2.1), the quantum defect can be considered as a measure of the deviation from simple hydrogen-like behavior and it reflects the amount of penetration of the particular Rydberg orbital into the core, which results in an increase in the term value. Based on a survey of a vast body of electronic absorption and energy loss spectra (mostly molecules containing second and third row elements) Robin [54] has estimated empirical limits for the term values of the lowest members of Rydberg series. The typical magnitudes are in the range 2.8-5.0 eV for the lowest ns member, 2.0-2.8 eV for the lowest np member and 1.5-1.8 eV for the lowest nd member (i.e. corresponding quantum defects ^ are 0.8-1.3 for s Rydberg series, 0.4-0.8 for p Rydberg and 0.0-0.2 for d Rydberg series). It should be noted that the lowest ns Rydberg term value has the largest variation because of its penetrating nature and the possibility of mixing with virtual valence orbitals lying in the same energy region. In principle, if the term value of the lowest member of a particular Rydberg series is known (and hence ^ is known), it is possible to estimate the term values of higher members in the series by applying the Rydberg formula (Eq. 2.1). However, it becomes more difficult to identify higher n Rydberg states because of the increased density of states and the fact that the intensity of transitions to higher n Rydberg orbitals falls off rapidly as n [44]. Nevertheless, the Rydberg formula, as well as the _3 n intensity fall-off, provides an important guide for the identification of Rydberg series in the spectra. An additional useful intensity criterion for identifying the atomic-like Rydberg excitations is that transitions to levels which would be formally dipole-forbidden in the purely atomic case (e.g. s ? s, p / p, d / d, s f d, etc) generally exhibit less intensity in the molecular spectra because of the atomic contribution to the overlap. Use of this so called "pseudo-atomic dipole selection rule" is particularly helpful in analyzing ISEELS spectra since the initial orbital involved is a highly localized core orbital, which is essentially atomic-like in character. 21 Since the Rydberg orbitals are large, diffuse and atomic-like, they are essentially non-bonding in character. In contrast, the virtual valence orbitals are more confined and extend over the framework of the molecule or often over several atoms in the molecule. These (normally unoccupied) virtual valence orbitals belong to the valence-shell MOs and thus are highly characteristic of the molecule. In most molecular systems, the virtual .valence orbitals correspond to the antibonding counterparts of bonding (occupied) valence, orbitals. Therefore, . Rydberg features are generally sharper than those features corresponding to transitions terminated at the virtual valence orbitals. Furthermore, the intensity distribution of vibrational components of a Rydberg state will approach the corresponding distribution for the ionic state to which these Rydberg series converge. This can be appreciated by noting that the excited electron in the Rydberg orbital is far from the core and the vibrational motion of the excited molecule closely approaches that of the ionic state. Since the intensity distribution of vibrational levels corresponding to a given ionic state can be measured by photoelectron spectroscopy, this information can be utilized in the analysis of Rydberg features in the electron energy loss spectra. 2.1.2 Transitions to Virtual Valence Orbitals In contrast to Rydberg states which always appears below the onset of the ionization continuum towards which they converge, transitions from a particular occupied orbital to the virtual valence orbitals can occur in the discrete (T>0) or continuum (T<0) regions of the excitation spectrum of the given orbital. Although term values for virtual valence final states do not have a well-defined physical interpretation as do the term values of the Rydberg states which follow the Rydberg formula (Eq. 2.1), the term values of virtual valence states are still an important parameter for the identification of spectral features with virtual valence • final orbitals. For example, a feature which has a term value >5 eV is evidently due to a transition to a virtual valence final orbital since a term value of 5 eV is the upper limit [54] for any possible Rydberg state (see section 2.1.1 above). Since virtual valence orbitals are usually the antibonding counterparts of the bonding orbitals (which are responsible for the chemical bonds in a molecule), the 22 term value of a transition to a given virtual valence final orbital is strongly characteristic of the relevant bond represented by the localized character of that virtual valence orbital. Sette et al. [58] have recently demonstrated that the term values of a states are linearly correlated with the bond lengths of the relevant a bond in a large variety of molecules. Furthermore, Robin [56,59,60] has shown that term values of o states are also linearly correlated with the bond strength of the relevant a bond. Although further experimental and theoretical studies are required to map out the trends of the term values for the virtual valence bands, correlation of the term values of valence bands in similar species is evidently a powerful tool in analysing the excitation spectra. This approach has been used extensively in the interpretation of the ISEELS and VSEELS spectra of the monohaloethylenes (Chapter 5) and DMSO (Chapter 7). Based on spatial overlap considerations, transitions to the virtual valence orbitals are expected to be stronger than those to the Rydberg orbitals. In addition, transition to virtual valence orbitals may be resonantly enhanced at the expense of the intensity of the Rydberg states and ionization continuum because of potential barrier effects (see section 2.3 below). Robin [54] has placed an upper limit of oscillator strength of 0.08 per degree of degeneracy for Rydberg excitations as a result of a survey of existing experimental and theoretical studies. In contrast an oscillator strength of 0.3-1.0 can be readily achieved for dipole allowed virtual valence bands even in small molecules [54]. However, since virtual valence orbitals are usually of antibonding character, the valence bands are frequently vibrationally broadened and may therefore appear to be less prominent in the spectra than the sharper Rydberg features. Furthermore, it should be noted that the above argument does not imply that every dipole allowed transition to virtual valence orbital should necessarily appear as a distinctive feature in the excitation spectra. Firstly, virtual valence states having term values similar to those of the Rydberg manifolds can often be totally mixed into the surrounding "sea" of Rydberg states of the same symmetry [56] and thus can cease to exist as distinctive features. Secondly, virtual valence states above the ionization threshold (i.e. T<0), can autoionize rapidly into the underlying continuum and they may become too broad to be visible. Lastly, the extent of 23 spatial overlap between an initial orbital and a virtual valence final orbital depends on the localized character of the orbitals involved. Therefore, the intensities of transitions from different initial orbitals to a given virtual valence final orbital are expected to vary. It is possible that a dipole allowed transition between initial and final orbitals of very different localized characters may turn out to be too weak to be observable. It should be pointed out that a knowledge of the localized character of the virtual valence orbitals (for example by performing a GAUSSIAN 76 [360] calculation which is able to give the AO coefficients of each atom in the molecule for all the orbitals included in the MO scheme) and their symmetry can direct the assignment of valence bands by matching their expected intensities with the observed intensities from various initial orbitals. Conversely, analysing the variation of the intensities of valence bands from different initial orbitals can provide valuable knowledge of the localized character and symmetry of the virtual valence orbitals involved. Based on this idea, Hitchcock et al. [61] have used the oscillator strengths of the Is -+7r features in the various K-shell spectra of H C O N ^ , HCOOH and HCOF to estimate the spatial distribution of the n (C=0) orbitals in these three species. 2.1.3 Delayed Onsets of Ionization Continua For the hydrogen atom, there is an abrupt rise of the ionization cross section at the ionization threshold which is followed by a monotonic exponential decline to higher energies [44]. However, for multi-electron atoms or molecules, the ionization cross sections of various outgoing channels may be maximal either at threshold or at an energy considerably removed from threshold. The latter case is referred to as a delayed onset of the ionization continuum and is usually explained in terms of a centrifugal barrier associated with the excited electron ionizing into a channel of high angular momentum [44]. Such a centrifugal barrier limits the approach of the low-energy electron near threshold to the molecular core and thereby reduces the overlap with the originating orbital (i.e. reduces the spectral intensity). However, an ionized electron of higher energy can penetrate the centrifugal barrier and the transition moment rises with increase in 24 electron energy until the out-of-phase loop of the wavefunction of the ionized electron starts to diminish the overlap with the core orbital. Acting in combination, these effects lead to a maximum in the ionization cross section which is either close to or removed from the threshold depending on the angular momentum of the out-going electron. Figure 2.2 shows an ISEELS spectrum of CH^I, in which the 4d -»e f ionization continuum maximize more than ~28 eV beyond the I 4d ionization limit [62]. 2.1.4 Extended Energy Loss Fine Structure (EXELFS) Inner-shell ionization continua can also be modulated by the EXELFS phenomenon which is analogous to the extended X-ray absorption fine structure (EXAFS) observed in many X-ray photoabsorption spectra. When an electron is ejected from an inner-shell of a particular atomic center in a molecule, the primary electron wave can interfere with the back-scattered waves from other surrounding atomic centers. This interference causes a damped sine-wave modulation of the ionization cross section. The physical description of this process is a type of internal electron diffraction, where the initial source of incident electrons can be chosen by selecting the energy loss E (or hp in photoabsorption) such that incident electrons of various kinetic energies (above the energy of the given inner-shell) can be ejected. The period of the oscillations should therefore bear some relation to the internuclear distances of atoms surrounding the primary source from which the electrons are ejected. The series of usually weak and broad EXELFS peaks and troughs, which typically extend over many hundred of eV, can be analyzed by Fourier transformation techniques to provide important information about the local geometry of the excited atomic center. 2.1.5 Multiple Excitations and Ionizations Although the excitation spectra are generally dominated by single-electron processes, double excitation or ionization processes can sometimes be observed. The shake- up/shake- off transition (see also section 2.4.1 below) is a type of two-electron >-(I) LJ L L I > r-< LJ I 4d \ : 4d->ef : , delayed onset \ of ionization continuum I4p C 1s \ 14s \ V. ., x64 100 200 ENERGY LOSS (eV) Figure 2.2 ISEELS spectrum of C H 3 I (taken from Ref. [62]). 300 t o 26 process which arises when the "sudden" perturbation of one electron leaving the molecule (ionized) with high energy induces monopole excitation/ionization in one of the remaining electrons of the positive ioa These two processes are manifested as satellites of the main ionization line in photoelectron spectra but they appear as additional onsets of continua above the main ionization threshold in electronic excitation spectra (i.e. electron energy loss or photoabsorption spectra). Therefore, a detailed knowledge of the XPS satellite spectrum can assist in the interpretation of continuum structures in electronic excitation spectra. Another possible type of two-electron process is produced by electron correlation effects and can be described by the configuration interaction model. In this case, the so-called double excitation is described by a two-electron excited configuration mixed with a single-excited configuration, and the two-electron excitation then appears by virtue of its partial one-electron character. A clear example of double excitation can be seen in the C Is spectrum of CO (see Chapter 6 Figure 6.2 feature 5) which exhibits a vibrational^ structured band 3-6 eV beyond the C Is ionization edge [63]. 2.2 Comparison between ISEELS and VSEELS Spectra Electron energy loss spectroscopy is usually classified as either inner-shell electron energy loss spectroscopy (ISEELS) or valence-shell electron energy loss spectroscopy (VSEELS) depending on whether inner-shell (core) or valence-shell electronic excitations are studied. ISEELS spectra are in general easier to interpret due to the wide energy separation of various inner-shell orbitals. This implies that spectral structures associated with the excitations from different types of inner-shell orbitals are well separated and the initial orbitals can be unequivocally assigned on energy basis. This is in sharp contrast to the complex situation in VSEELS spectra which are frequently composed of overlapping transitions from the various closely spaced valence orbitals. Therefore inner-shell excitation spectra can readily yield information about the unoccupied levels which, in turn, can often be helpful in the analysis of valence-shell excitation spectra. 27 A useful concept in spectral assignment is that of transferability of term values. In particular, it has been demonstrated [64-66] that term values for transitions terminated at a given Rydberg orbital are transferable between ISEELS and VSEELS spectra. This can be understood by noting that the excited electron in the given Rydberg orbital, which is typically of large orbital size and diffuse, will see the rest of the molecule as a positively charged centre. Thus the term values of transitions from various initial (core or valence) orbitals to the same Rydberg orbital should be largely independent of the location of the vacancy created by the excitation. In practice, the term values for core excitation and for valence excitation to a given Rydberg final state may be very slightly different due to the fact that the Rydberg orbital may penetrate the molecular core. Hence, any such variation is usually largest for the lowest member of the ns Rydberg series which is the most penetrating Rydberg orbital. Furthermore, the lowest ns Rydberg orbital may also be mixed with any virtual valence orbital of the same symmetry which occurs in the same energy region. Previous ISEELS and VSEELS studies [64-66] have shown that the maximum variation in term values for Rydberg final states is <0.5 eV. In contrast to the situation for transitions to Rydberg orbitals, term values corresponding to transitions to a given virtual valence orbital are in general not transferable between ISEELS and VSEELS spectra. This is because the excited electron in the virtual valence orbital, which is usually localized on the framework of the molecule, is subject to significant interactions with the electron hole created by the excitation. Therefore, the corresponding term value is more sensitive to the specific location of the vacancy. Nevertheless detailed studies of excitations from core orbitals to virtual valence orbitals can still be of value for the identification of the corresponding transitions in the valence excitation spectrum. This is because the loss of shielding when an electron is excited from a valence orbital is smaller than that when an electron is excited from a core orbital. Therefore the term value for a transition to a given virtual valence orbital from a core orbital is generally larger than that of the corresponding transition from a valence orbital. This implies that the term value of a core excited virtual valence state provides an upper bound for the term value of the corresponding valence excited virtual 28 valence state. In addition, in order to further assist the identification of transitions to . virtual valence orbitals in the VSEELS spectrum, it is assumed that term values corresponding to excitations to a given virtual valence orbital from different (normally occupied) valence orbitals are transferable. This assumption is reasonable considering that the valence orbitals are, in a general sense, energetically quite similar to each other and spatially concentrated on the periphery of the molecule. In this respect the valence orbitals contrast markedly with the much more highly localized and atomic-like core orbitals. Therefore, as far as transitions to a given virtual valence orbital from the various normally occupied valence orbitals are concerned, relaxation as well as correlation effects in these valence excitations would be reasonably similar and thus the term values would be expected to be of quite similar magnitude. In the absence of reliable theoretical calculations, this approach has been used to provide reasonable assignments of VSEELS spectra in earlier studies [66,67]. 2.3 Potential Barrier Effects The study of potential barrier effects in molecules has been a subject of great interest since the original postulate of this concept to account for the "anomalous" intensity distributions observed in the inner-shell electronic excitation spectra of "cage" molecules containing highly electronegative ligands such as SFg, SO^, SiF 4 > SiCl^ and BF^ [34,35]. These unusual intensity redistributions are characterized by an enhanced probability of core to virtual valence transitions, (often called shape resonances) located below and/or above the ionization edge at the expense of transitions to the Rydberg levels and ionization continuum. It was postulated [34,35] that a strong repulsive force would act on the escaping electron near the electronegative ligands and hence an effective (Coulombic) potential barrier would exist in the vicinity of the ligands. This barrier would separate the molecular field into "inner-well" and "outer-well" regions. In particular, the virtual valence orbitals trapped in the inner-well region would have a strong overlap with the initial core orbital and thus would result in enhanced spectral . features. On the other 29 hand, the difTuse Rydberg orbitals and ionization channels located in the outer-well region would be isolated from the molecular core and this would lead to a drastic reduction in spectral intensity for transitions to the Rydberg orbitals and the ionization continuum. A spectacular example of this is the S 2p spectrum of SF g as shown in Figure 2.3 while Figure 2.4 shows a schematic representation of the corresponding potential curve containing inner-well and outer-well regions separated by a Coulombic potential barrier. The potential barrier would also support inner-well discrete valence states located above the ionization edge as quasi-stationary states which are effectively decoupled from the continuum by the barrier. In fact the name "shape resonance" means simply that the resonance behavior arises from the "shape", i.e. the barrier and associated inner and outer wells,, of a local potential. These potential barrier concepts have often been coupled with the MO scheme of unoccupied virtual valence orbitals to provide simple intuitive descriptions suitable for qualitative discussions of spectral features and their transition intensities [35,68-73]. It should be noted that shape resonances are not only observed in photoabsorption or electron energy loss experiments but also in molecular photoionization in both total and partial channels, electron-molecule scattering, photoelectron branching ratios and angular distributions, Auger electron angular distributions and electron transmission spectroscopy [74]. Potential barrier effects have been identified in the spectra of a growing and diverse collection of molecules and now appear to be active somewhere in" the observable properties of most small (nonhydride) molecules. Although the Coulombic effect of electronegative ligands is apparently an important factor in creating the potential barrier in many cases, potential barrier effects have also been observed [75,76] and predicted [36,74,77,78] in the inner-shell spectra of 7r-bonded molecules such as N 2 and CO which do not have an obvious cage of electronegative ligands. Therefore, an alternative theoretical approach applicable to inner-shell excitation spectra of molecules both with and without electronegative ligands has been proposed [36,74,77,78]. In this type of approach, which typically uses the MS-Xo scattering method, the excited electron is seen as being temporarily trapped by an effective centrifugal potential barrier until it has sufficient RELATIVE INTENSITY 31 V a l e n c e , I n n e r - w e l l R y d b e r g , O u t e r - w e l l M O ' s M O ' s F l u o r i n e SPQ S u l p h u r A O ' s M O ' s A O ' s Figure 2.4 Schematic representation of the effective potential barrier in SF6 (taken from Ref. [68]). 32 kinetic energy to tunnel out by reaching an outgoing channel which penetrates the barrier. Such rapid penetration of the barrier results in an increased spectral intensity at specific energies in certain outgoing partial wave channels. The effective centrifugal barrier is considered to be a result of the competition between the attractive electrostatic (Coulombic) and the repulsive centrifugal forces which are associated with a dominant outgoing partial wave channel of high angular momentum >2). Such a centrifugal barrier usually resides on the perimeter of the molecular framework where the centrifugal forces can compete effectively with electrostatic forces. Such barriers are known for d (4 =2) and f (^ =3) outgoing waves in atomic fields [44,56] and the atomic nomenclature is usually carried over into a description of the partial wave outgoing channels of molecular systems. This approach has been applied successfully to a number of molecular core excitations [36,77-83] and provides a computational approach for the understanding of resonance phenomena not only in electronic excitations but also in molecular photoionization experiments of various types [74]. Although in this approach the centrifugal force is considered to be the primary factor causing the potential barrier, other forces such as a high concentration of negative charge near the electronegative ligands (i.e. Coulombic), repulsive exchange force, etc. [35] may also contribute. For example, in MS-Xa studies of potential barrier effects in the X-ray photoabsorption spectra of BF^ [82] and GeCl^ [83], a combination of centrifugal forces and strong electron repulsion in the neighbourhood of the ligands is considered to be responsible for the observed barrier. Furthermore, the above scattering model, involving a centrifugal barrier associated with the high angular momentum components of the out-going electron, lacks the simple physical intuitive feel provided by the Coulombic barrier/MO approach familiar to chemical spectroscopists. Thiel [84,85] and Langhoff [361] have shown convincingly that the terminating wavefunctions of molecular shape resonances can indeed be viewed as the antibonding virtual orbitals of the ground state. This connection is a natural one since shape resonances are localized within the molecular charge distribution (i.e. inner-well region) and therefore can be realistically described by a limited basis set (minimum or extended) suitable for describing the valence MOs. In light of the above discussion, the 33 Coulombic potential barrier/MO model has been used as a comparative basis for the interpretation of shape resonances in the excitation spectra obtained in the present experimental studies. In addition, a consideration of the ISEELS spectra of the "partial cage" molecule CIF^ (see Chapter 5.II) provides new evidence concerning the existence of a Coulombic potential barrier. 2.4 Processes Related to Electronic Excitation Collision of electrons with isolated atoms and molecules can not only cause excitation but also result in a large variety of other processes such as ionization, dissociation, autoionization, Auger electron emission and X-ray fluorescence. Studies of these related processes can often provide helpful information complementary to that obtained by studying the excited states with electron energy loss or photoabsorption techniques. The following sections briefly describe these processes with emphasis on those aspects which may be useful in the analysis of ISEELS and VSEELS spectra. 2.4.1 Ionization Although the ionization continua observed in EELS or photoabsorption spectra provide a direct probe of the ionization process, ionized states of atoms or molecules are generally better studied by photoionization techniques. In photoionization experiments photons of a characteristic energy hv are used to ionize the target species : hv + AB - » A B + + e 2.2 (KE) where AB and A B + are the ground and ionized states of the target species respectively and e is the photoelectron ejected with kinetic energy KE. Considering that the photoelectron is ejected from an occupied orbital with characteristic ionization potential IP, conservation of energy demands that 34 hv = IP + KE 2.3 > i.e. KE = hv - IP 2.4 Therefore, a record of the number of electrons detected as function of their kinetic energies (i.e. a photoelectron spectrum) will contain a peak at each energy, hv - D?, corresponding to the ejection of an electron from an occupied orbital of ionization potential IP as illustrated schematically in Figure 2.5. Photoelectron experiments have been traditionally divided into two branches according to the photon source used. In the first branch, which is called photoelectron spectroscopy (PES) or ultraviolet photoelectron spectroscopy (UPS), the most commonly used light source is the He(I) resonance line with an energy of 21.22 eV. This energy is only sufficient to ionize electrons from the outer-valence orbitals. Therefore, the He(II) resonance line at 40.81 eV has been used for the study of more tightly bound inner-valence orbitals. The other branch is called electron spectroscopy for chemical analysis (ESCA) or more suitably X-ray photoelectron spectroscopy (XPS). In this case, MgK<*(1252.60 eV) and AlKo(1486.70 eV) are the most frequently used light sources, both of which are able to ionize deeply bound electrons from inner-shell as well as valence-shell orbitals. Ionization potentials obtained from XPS and UPS measurements are necessary for the analysis of ISEELS and VSEELS spectra respectively because they are required to permit calculation of the term values of various excited states, which are derived from the difference between the initial orbital IPs and the excitation energies. In principle, analysis of a Rydberg series can also yield the IP (Eq. 2.1). However, in practice, the observation of a well-defined Rydberg series can easily be obscured by a number of -3 factors such as overlapping transitions, n drop in intensity, Rydberg/valence mixing, lack of resolution, etc. Therefore, the complementary photoelectron spectroscopic technique provides the IP information not only more directly but also usually more accurately. 35 FREE ELECTRON Electron energy hv-IP 1 hv-IPc • 0" BOUND ELECTRON - | P 1 - IPo hv hv PHOTOELECTRON SPECTRUM Intensity Figure 2.5 Photoionization process and photoelectron spectrum. 36 As mentioned earlier in section 2.1.1, the vibrational envelope of an electronic transition to a Rydberg level in the excitation spectrum is similar to that of the ionic state to which the Rydberg level converges. This can be understood in the sense that the potential curves of the Rydberg states approach those of the corresponding ion states because the Rydberg final orbitals are large, diffuse and non-bonding in character. Therefore, the intensity distribution of vibrational levels in the ionic states, which can be measured by high resolution photoelectron spectroscopy, can be utilized in the analysis of Rydberg states in the EELS spectra. Studying the XPS satellite spectrum can also aid in identifying structures in the electronic excitation spectrum arising from multiple excitation processes which involve the ionization of an electron with the simultaneous excitation or ionization of another electron as illustrated schematically in Figure 2.6. The former process is called "shake-up" and the latter one is called "shake-off'. These two types of process are observed as satellite peaks and continua respectively, and are situated at lower kinetic energies (i.e. higher IPs) than the -nain line in the XPS spectrum. In excitation spectra these processes produce additional ionization continuum onsets at energies above the IP equal to the separations of XPS satellites and the main line. 2.4.2 Core-hole Decay Processes: X-ray fluorescence, Auger decay, Autoionization and Highly excited states created by either ionization or excitation of a core (i.e. inner-shell) electron can undergo a number of competing decay processes to dissipate the excess energy in the system. The various major processes for immediate decay of these highly excited states can be symbolized by : for a core-ionized state A B + ® Dissociation Auger decay X-ray fluorescence Dissociation SHAKE-UP S H A K E - O F F AB + hv ^ AB +e AB+h v ->AB 2 + + 2e ' / / / / / / / / / / / / / / / / f // IONIZATION CONTINUA //1 / / / / / / / / / / / / / / / / / • • J UNOCCUPIED LEVELS VALENCE CORE / t i f t i / f t / f f IONIZATION CONTINUA m m • • Figure 2.6 Shake-up and shake-off processes. 38 for a core-exited state AB 1 AB+ AB + hv A + B + e X-ray fluorescence Dissociation Authorization In Auger decay, the core hole is filled with a less tightly bound electron with the concomitant ejection of an electron which carries the excess energy. The Auger process (see Figure 2.7) is the dominant decay mode for core holes of light atoms. For 3 example, Auger decay is ~10 times faster than X-ray fluorescence for C and O K-shell holes [44]. In general, Auger decay rates and signal intensities depend on the overlap between the core hole, the orbital containing the electron which fills the core hole, and the orbital containing the electron which is ejected. Therefore, Auger decay is dominated by the fastest processes which usually involves occupied orbitals closest in energy to the core hole energy. A special type of such a fast Auger decay is called Coster-Kronig transition. In this process, the primary vacancy is filled by a higher lying electron within the same shell (i.e. same principal quantum number, e.g. 2p into 2s). Another type of even faster Auger decay process is called Super-Coster-Kronig transition, in which the primary vacancy is filled by a higher lying electron within the same subshell (e.g. 2 p 1 / 9 The process corresponding to Auger decay in the case of a core-excited (neutral) state is called autoionization (see Figure 2.7). Similar to the situation in Auger decay, the autoionization process involves the filling of the core vacancy with the simultaneous ejection of an electron. X-ray fluorescence is more difficult to study because of the low fluorescence yield in particular for light atoms. However, the resulting spectra are easier to interpret than Auger spectra because the final orbital is known to be that associated with the core hole and the transitions are governed ~ by dipole selection rules. In this regard, peak intensity Excitation Autoionization X-ray f luorescence IONIZATION CONTINUA UNOCCUPIED LEVELS VALENCE CORE ////A IONIZATION CONTINUA //l G R Dissociation Ionization Auger decay X-ray f luorescence //////////*/// IONIZATION CONTINUA / / / / / / / > / / / / / UNOCCUPIED LEVELS VALENCE m. m IONIZATION CONTINUA //////// : / / / / / I, 'A O R Dissociation CORE Figure 2.7 Core hole decay processes : X-ray fluorescence, Auger decay, autoionization and dissociation. 40 can be used to identify the symmetry of a particular occupied level involved in a given transition. Therefore, this technique is considered to be supplementary to photoelectron spectroscopy and it is particularly useful for investigation of the ordering of occupied valence orbitals [86]. In the dissociation mode of decay, the highly excited molecule is broken into various neutral or ionic fragments either through direct dissociation or predissociation processes. In summary, the above decay processes can be classified into three general mechanisms : radiative (X-ray fluorescence), non-radiative (Auger, autoionization) and dissociative modes. In, the first case, the excess energy is released by the emission of a photon; in the second case, the excess energy is carried away by an emitted electron; while in the third case, excess energy is dissipated in the form of bond breaking and kinetic energies of the resulting fragments. For the decay of core holes with binding energies less than ~1000 eV (i.e. those of interest in this study), the non-radiative mechanism dominates [87-89]. Nevertheless, all the decay processes described above are important in the study of ionization and electronic excitation because they govern the lifetime of the hole state and thus the natural linewidth of the feature. According to the Heisenberg uncertainty principle, the natural linewidth T) is related to the lifetime r by h r = ->s 2 71 x 1 5 In practical terms T T ~7xl0 ^ eV sec. Although details of the decay mechanisms cause large variations in the lifetimes of different core hole states, the natural linewidth generally increases with increasing ionization potential within each subshell (i.e. increases with Z). Typical linewidths for carbon, nitrogen, oxygen and fluorine K. shell excited states are about 0.10, 0.12, 0.15 and 0.20 eV respectively [89,90]. 41 CHAPTER 3 EXPERIMENTAL METHODS 3.1 The Spectrometer, Operation and Spectral Acquisition The high resolution electron energy loss spectrometer used to measure the inner-shell and valence-shell electron energy loss spectra reported in this work was developed by Daviel, Hitchcock and Brion [91] in the period 1979-1983. This state-of-the art instrument provides improved performance over earlier instruments [62] in terms of resolution, sensitivity and stability. Figure 3.1 shows a schematic diagram of the spectrometer. The design and constructional details of this spectrometer have been described earlier in great detail [91] and therefore only a brief overall description will be given here. Key areas of improvement in design include: (a) Separate differential pumping of the gun, monochromator, interaction and analyser regions of the spectrometer in order to alleviate the problems of surface contamination, retuning, and frequent cleaning that occur with single chamber instruments. This ensures long term stability as well as improved sensitivity. In addition the range of compounds which can be studied is greatly enhanced since thermal stability on the hot filament is no longer an issue. (b) Advanced electron optics [92] have been incorporated in order to improve beam currents, as well as to reduce the effects of scattering of the main beam from slit edges and the surfaces of the analyzer into the detector. (c) Large hemispherical electron energy selectors (main radius, R Q = 19 cm = 7.5 in) have been chosen in order to improve transmission and resolution for a given pass energy. This also allows reasonable voltage ratios in the lenses while maintaining high resolution and the high impact energy needed for the study of dipole allowed transitions. Operational details of the spectrometer are as follows: Electrons emitted from a MONOCHROMATOR ANALYSER T U R B O P U M P 360 L/S SCALE i ' i 11 i ' 10 20 cm T U R B O P U M P 450 L/S T U R B O P U M P 360 L/S Figure 3.1 High-resolution electron energy loss spectrometer. 43 heated (thoriated) tungsten wire hairpin in the electron gun (Cliftronics CE5AH) are accelerated to the desired impact energy (1000-3700 eV in the present studies) to form a focussed electron beam (1 mm diameter). The beam is then retarded by a two-element lens (Ll) to the required pass energy (10-100 eV in the present studies) of a hemispherical electrostatic analyzer, which acts as the monochromator. A virtual slit is formed at the monochromator exit by the accelerating (ratio 1:20) lens (L2). Further acceleration (x5) is provided by lens L3, which also focuses the beam onto P4 at the entrance of the stainless steel reaction chamber. Both L2 and L3 are two-element lenses, whose adjacent elements have been tied together at an adjustable voltage so as to limit the beam angle to zero degree. The beam is transported to the collision chamber (CC) by the single lens L4 and passes out via a zoom, energy-add, lens L5. A virtual slit is formed at the analyzer entrance and the beam is retarded to the pass energy (10-100 eV in the present studies) of the analyzer by L6 and L7, which are similar in design and function to L3 and L2, respectively. A channeltron electron multiplier (Mullard B419AL) is located behind a real (1 mm) exit aperture to detect the electrons transmitted by the analyzer. Sets of x, y deflectors (Ql to Q9) are used throughout the spectrometer, in conjunction with apertures P1-P8 to align and focus the primary electron beam. Magnetic shielding of the various regions of the spectrometer is provided by hydrogen-annealed mumetal enclosures, exterior to the vacuum housing. The primary (unscattered) electron beam is used to tune up the spectrometer by directing it through the analyzer and onto the entrance cone of the channeltron where it is monitored by a vibrating reed electrometer (Cary model 401). To obtain a spectrum, a voltage equivalent to the energy loss corresponding to the inelastic scattering is added, on top of the voltages already applied to the complete analyzer system, from the central element of L5 onwards. Thus the scattered electrons can regain their energy loss and be transmitted through the analyzer system to the channeltron. By using a suitable offset voltage and scanning the energy loss region of interest an energy loss spectrum can be obtained. The pulses that result from the channeltron amplification of single scattered electrons are processed by high gain preamplifier and amplifier/discriminator units 44 ,6 (PRA models 1762 and 1763). Count rates as large as 10 Hz can be handled due to the high gain. Spectra are recorded with the aid of a PDP 11/023 computer, and/or by a Nicolet 1070 signal averager operated in a multichannel scaling mode whereby the channel address is stepped synchronously with the voltage on the analyzer and thus a complete spectrum can be signal averaged. The spectrum is monitored on-line via a VT105 graphics terminal (PDP 11/023) and/or an oscilloscope (Nicolet 1070). Several computer programs have been developed in the early stages of the present studies to plot the completed spectrum on an X - Y point plotter (Hewlett Packard 7004B), to permit measurement of energies of features in the spectrum and also to perform data manipulation such as background subtraction, spectral stripping and addition. In the present work impact energies in the range 1000-3700 eV have been used. Under the conditions of forward scattering and high impact energies (more than 5-20 times the energy loss) the energy loss spectra will be governed by dipole selection rules (see Chapter 1 section 1.2.4). The corresponding energy resolution is in the range 0.028 to 0.3 eV FWHM depending on the pass energies selected for the monochromator (E^) and analyser (E^) and the lens voltage ratios. The theoretical resolution of this particular spectrometer is simply given by [91] In general, the experimental resolution obtained is slightly better than the theoretical value calculated by the above equation, probably because the effective sizes of the virtual apertures in the monochromator and analyser [91] are smaller than predicted. The actual resolution specified in the energy loss spectra reported in the present studies was obtained by measuring the profile of the primary electron beam. The choice of the resolution to be used is determined by the intensity of features in the spectral region of interest as well as by natural linewidth considerations of the corresponding excited states (see Chapter 2 section 2.4.2). AE F W H M 3.1 45 3.2 Energy Calibration 'The absolute energy scales for all spectra measured in this work were established by calibrating a prominent feature in the sample spectrum with respect to a known feature of a reference gas. In practice, a mixture of the sample and the reference compound was leaked into the spectrometer during the energy calibration scans in order to avoid any problems due to gas-dependent energy shifts and contact potentials. The energies of .a number of selected inner-shell atomic and molecular electronic transitions have been carefully measured by Sodhi and Brion [93] for use in the calibration of ISEELS spectra. The particular calibration energies used in the present work are listed in Table 3.1. All valence-shell spectra were calibrated using the He(I) resonance line (He Is -> 2p transition at 21.218 eV [94]). In practice, the calibration corrections are very small (typically < 0.03 eV) due to the efficient differential pumping of the different regions of the spectrometer [91]. In the electron energy loss spectra shown in the present work, all structures considered to be possible distinctive features based on signal-to-noise and band profile considerations will be designated by numbered vertical lines showing their energy positions. The energies of clearly resolved peaks were obtained directly by computer analysis of the measured spectra. The positions of incompletely resolved features and shoulders were visually estimated from the spectra. 3.3 Sample Handling All the samples studied in this work were obtained commercially. Their sources and stated minimum purity are listed in Table 3.2. The samples, except for Ni(CO)4> were used without further purification. Appropriate gas regulators were put on the cylinder of each gaseous sample (S02 > C ^ F , C 2H 3C1, SFg, SeFg, TeFg, C1F3 and Ni(CO)4) before the gas was fed into the inlet system and then steadily leaked into the collision chamber through a Granville-Phillips series 203 leak valve. For liquid samples (DMSO, C 2 H 3 Br and C 2H 3I), several freeze-pump-thaw cycles were performed to remove any dissolved gases before introduction into the spectrometer via the leak valve. 46 Table 3.1 Reference energies for inner-shell energy loss spectroscopy^3) Inner-Shell Transition Transition Energy(b) (eV) SF6 S 2p 1 / 2 t2g . " 184.54(5) Ar 2p3/2 -> 4s 244.37(2) CO C Is -> n*(v=0) 287.40(2) N2 N Is —» 7t*(v=l) 401.10(2) CO 0 1S->TI* 534.21(9) SF6 F Is -» a i g 688.27(15) Ne Is -> 3p 867.13(7) (a) Reference energies are taken from Ref. [16]. (b) Uncertainties are shown in brackets e.g. 184.54(6) means 184.54 ±0.06 eV. Table 3.2 Source and purity of samples Sample Source Purity (%) SF 6 Matheson Chemical Inc. 99.8 SeF 6 Ozark-Mahoning — TeF 6 Ozark-Mahoning — C1F 3 Matheson Chemical Inc. 98.0 C 2 H 3 F Matheson Chemical Inc. 99.9 C 2 H 3 C I Matheson Chemical Inc. 99.9 C 2H 3Br Columbia Organic Chemicals 99.0 C2H3I Pfaltz and Bauer Inc. 99.0 Ni(CO)4 Matheson Chemical Inc. 99.9 (CH 3) 2S0 Aldrich Chemical Company Inc. 99.9 S0 2 MathesonChemical Inc. 99.8 47 Since Ni(CO)4 decomposes slowly into Ni metal and free CO, the sample was purified by repeated (~5 times) freeze-thaw operations on the lecture bottle using an acetone/dry ice mixture, and pumping off the residual gas which is mainly free CO. By this means the amount of free CO in the Ni(CO)4 lecture bottle could be minimized although small amount of free CO was still evident in the valence-shell spectrum of Ni(CO)4 (see Figure 6.6). Valence-shell spectra were run before and after each inner-shell data collection period in order to verify that any contribution from free CO was negligible. In practice, the ambient pressure of the spectrometer was allowed to rise from a -7 -5 base pressure of ~3xl0 ton to 5x10 torr on sample introduction. Such a pressure maximizes the signal obtainable under the single collision conditions necessary for EELS studies. No double scattering was detected. For most samples, no impurities were apparent as indicated by their ISEELS and VSEELS spectra. However, weak structures observed in the region below ~13 eV in the VSEELS spectra of SeFg (feature 6, Figure 4.1.14) and DMSO (feature 17, Figure 7.1.8) suggest that a small amount of N 2 impurity may be present in these two samples. 48 CHAPTER 4 INVESTIGATION OF POTENTIAL BARRIER EFFECTS IN "CAGE" MOLECULES : I. INNER-SHELL AND VALENCE-SHELL ELECTRONIC EXCITATION OF SF 6, SeF6 AND TeF^ 4.1.1 Introduction Previous studies of potential barrier effects have been largely restricted to molecules with second row (i.e. B, C, N, etc.) and third row (i.e. S, Si, P, etc.) central atoms. Therefore, the investigation of such effects in molecules involving central atoms of even higher atomic number is of great interest Since the inner-shell excitation spectra of SFg have provided, perhaps, the most clearest demonstration of the potential barrier effects [68,95], SeFg and TeF^ should be promising candidates for such an extended study. It is of interest that Addison et al. [96] have recently reported that there is dramatic intensity decrease in above-edge shape resonances in the Se 3p photoabsorption spectrum of SeFg which suggests that the potential barrier may be shallower in SeFg and/or that poor penetration of the accessible outgoing channel occurs. In addition, for even heavier central atoms in the fifth row such as Te the presence of low lying unoccupied 4f orbitals affords the possibility of additional outgoing MO channels (shape resonances). In order to seek a broader understanding of the potential barrier effects and of shape resonances in these molecules the inner-shell electron energy loss spectra (ISEELS) of all accessible core levels (below 1000 eV) in SeFfi and TeF g (Se 3d, 3p, 3s and F Is in SeFg-, Te 4d, 4p, 4s, 3d, 3p and F Is in TeFg) have been measured. There is no published information for inner-shell excitation of TeF g and only the Se 3p X-ray photoabsorption spectrum of SeF^ has been reported [96]. Valence-shell electron energy loss spectra (VSEELS) of SF^, SeFg and TeFg at moderately high resolution (0.03 eV FWHM) have also been measured in the same 49 spectrometer. The "VSEELS spectra for SeFg and TeFg shown in the present work are the first reported valence-shell electron impact excitation spectra for these molecules and no photoabsorption measurements exist Although, the excitation of all inner-shell [68,95,97-107] and valence-shell [108-114] of SFg have been previously studied in some detail by both photon and electron impact the S 2p, 2s and F Is ISEELS spectra as well as the VSEELS spectrum of SFg have been remeasured in the present work in order to provide direct comparison with the relevant excitation processes in SeFg and TeFg under equivalent kinematic conditions (i.e. high impact energy and zero degree scattering angle) and at a comparable energy resolution. It should be noted that the previous lower energy resolution ISEELS work for SFg [68] which was carried out at lower impact energies and larger scattering angles than the present work, exhibits appreciable contributions from non-dipole processes due to the large momentum transfers involved. 4.1.2 Results and Discussion 4.1.2.1 General considerations It should be noted that if the MO description for the MFg (M = S, Se and Te) molecules studied in the present work is limited to a minimum basis set MO scheme (i.e. F Is and M core levels + F 2s, 2p and M s and p valence levels), transitions to only two virtual valence levels of a-^ and t u^ symmetry would be expected. However, more virtual valence levels are present if the basis set . is extended for example to include the S 3d orbitals in SFg (or Se 4d orbitals in SeFg). Such a basis set extension leads to two additional virtual valence orbitals of t^ and eg symmetry. Similarly in TeFg inclusion of the low lying empty Te 4f orbitals (which are below the Te 5d [135]) leads to additional virtual valence orbitals of t u^> ^ symmetry. Using such extended basis sets the ground state electron configurations of SFg, SeFg and TeFg can be written as follows [128,129] : . 50 (a) SF 6 Core orbitals < ( l a l g ) 2 (2a l g) 2 ( l t l u ) 6 ( l e g ) 4 (3a l g) 2 ( 2 t l u ) 6 S Is F Is S 2s S 2p Valence orbitals (4a l g) 2 (3t l u) 6 (2e g) 4 (5a l g) 2 (4t l u) 6 ( I t , / (3e g) 4 (St^li,/2 ( l t l g ) 6 Unoccupied valence (Virtual^ orbitals (6alg)° (6tlu)° (2^)° (4eg)° (b) SeF 6 Core orbitals ( l a l g ) 2 (2a l g) 2 ( l t l u ) 6 (3a l g) 2 (2t' l u) 6 ( l e g ) 4 (4a l g) 2 ( 3 t l u ) 6 (2e g) 4 ( l ^ ) 6 «. v ' v Se Is Se 2s Se 2p F Is Se 3s Se 3p Se 3d Valence orbitals (5a l g) 2 (4t l u) 6 (3e g) 4 (6a l g) 2 (5t l u) 6 (2t 2 g) 6 (4e g) 4 ( 6 t l u , l t 2 u ) 1 2 ( l t l g ) 6 Unoccupied valence (virtual^) orbitals (7alg)° (7tlu)° (3^)° (5eg)° (c) TeF, 6 Core orbitals d a l g ) 2 (2a l g) 2 ( l t l u ) 6 (3a l g) 2 ( 2 t l u ) 6 (4a l g) 2 ( 3 y 6 ( l e g ) 4 Te Is Te 2s Te 2p Te 3s Te 3p F Is ( 2 e g ) 4 i V 6 ( V 2 ( 4 t i u ) 6 (3 e/(2W6 Te 3d Te 4s Te 4p Te 4d 51 Valence orbitals (6a l g) 2 (5t l u) 6 (4eg)4 (7a l g) 2 (6t l u) 6 0 % ) 6 (5eg)4 ( T ^ ) 1 2 ( l t 1 g ) 6 Unoccupied valence (virtual) orbitals (8a l g ) ° ( 8 t l u ) ° ( 9 t l u ) ° d a 2 u ) ° ( 2 ^ ) ° In general, such an extended basis set MO scheme is considered to give a better description of the properties of the molecules. For example the involvement of d-orbitals in bond formation has been found to be very important in molecules containing third-row elements [115-117]. The . ISEELS and VSEELS spectra reported in the present work are conveniently discussed in terms of transitions to common manifolds of virtual valence orbitals (shape resonances) as well as Rydberg orbitals. In addition, the highly symmetric octahedral molecular structure of these molecules severely restricts the number of possible dipole-allowed transitions from any initial orbital. Dipole-allowed transitions from initial orbitals of a given symmetry to final orbitals of various symmetries are compiled in Table 4.1.1. 4.1.2.2 Inner-shell excitation of SFg Figure 4.1.1 shows the F . Is, S 2s and 2p ISEELS spectrat of SF^ placed on common relative energy scales aligned with reference to the respective ionization potentials (IPs), as obtained by XPS measurements [118]. The S 2s spectrum shown in Figure 4.1.1b is background subtracted from a full spectrum of the type shown in Figure 4.1.1c above —b 220 eV. The background is estimated by fitting an aE curve (E is energy loss in eV, a and b are parameters obtained from the fitting) to the S 2p continua tail preceding the S 2s structures. Similar subtraction procedures have been used in the inner-shell excitation spectra of SeF^ and TeF^ so as to enhance the visibility of weak features t These include all inner-shell excitations accessible below 1000 eV which represents the effective range of the spectrometer. Table 4.1.1 Symmetry of i n i t i a l ^ and final^) orbitals for dipole-allowed transitions in MF6 molecules (O n symmetry) Initial Orbital Final Orbital a l g t m eg t l u . t2u a l u . eu> t l u . t2u t2g a 2u. eu. t l u . t2u t l u aig, eg, tig, t 2 g t 2 u a2g. eg. tig, t 2 g (a) Symmetry of initial core orbitals are a i g (M ns); ti„ (M np); t 2 g and e 8 (M nd); and aig. ti„ and e g (F Is). (b) Symmetry of final orbitals are a i g and t i u for all discrete (below edge) final states. For transitions to final states above the ionization edge, symmetry of final orbitals are t 2 g and eg for d-type outgoing continuum channels as in SF 6 and SeF6; a2 U, ti„ and t 2 u for f-type outgoing continuum channels as in TeFg. 53 (a) SR | F I s e d g e 6 Q | g 6 t | u 2 t 2 g 4e c Fls 0 = 0 ° E o = 3 0 0 0 e V A E = 0 . 2 6 0 e V x3 I 23l5\ 7 46 8 10 J_ J !_ b CO LU LU > _ l UJ cr 6 8 0 7 0 0 (b) 6 t l u r : g S 2 s e d g e 7 \ 2t2g ^e g 7 2 0 I 23 4 7 4 0 7 6 0 0 = 0 ° E o = 3 0 0 0 e V A E = 0 . 2 6 0 e V (BACKGROUND SUBTRACTED) 7 8 0 S2s J I L (c) 2 3 0 2 5 0 2 7 0 2 9 0 310 3 3 0 S 2 p e d g e s i S2p e=oc E o = 3 0 0 0 e V A E = 0 . 2 6 0 e V 160 180 2 0 0 2 2 0 ENERGY LOSS (eV) 2 4 0 2 6 0 Figure 4.1.1 Comparison of (a) F 1s; (b) S 2s; (c) S 2p ISEELS spectra of SF6-54 lying on top of the tails of preceding, more intense continua. The ISEELS spectra of SFg shown in Figure 4.1.1 are in good agreement with previous X-ray absorption spectra [95,97-107] as well as earlier ISEELS studies [68]. Furthermore, the improved resolution of the present work has resulted in better resolved peaks and has revealed new features in these spectral regions. The energies, term values and assignments of all features observed in the S 2p, 2s and F Is ISEELS spectra of SF 6 are listed in Tables 4.1.2-4 respectively. The inner-shell excitation spectra of SF^ provide a classic demonstration of potential barrier effects in cage molecules. The S 2p spectrum shows a strong absorption in the discrete region which can be assigned to a transition to the 6a^g virtual valence level. This is followed by very weak structures assigned to a mixture of overlapping Rydberg transitions and a dipole-forbidden transition to the 6t^ u level (note that both. F Is and S 2s show a strong feature corresponding to an allowed transition to this level). Two very prominent features are seen above the edge. These can be ascribed to transitions to the d-like states of L,g and e g symmetries. It should be noted that the photoabsorption S 2p spectrum of SFg is identical, within experimental error, for both gas and solid phases [95] confirming that the transitions described above are to final state levels localized in the inner-well region i.e. inner-well (virtual valence) states and not to Rydberg states, which would be localized in the outer-well region. The S 2s spectrum is dominated by a single dipole-allowed transition to the 6t^ u level. In the case of F Is excitation, transitions to all four inner-well virtual orbitals are observed since these transitions are dipole-allowed (Table 4.1.1). MO-LCAO [119] as well as MS-Xa calculations [78,81] predict similar assignments for the prominent features observed in the inner-shell excitation spectrum of SFg. A high resolution S 2p spectrum of 0.065 eV FWHM is shown in Figure 4.1.2. The solid line is the estimated underlying valence-shell background. Both 6a.^ and 2 ^ spin-orbit split partners are better resolved than in previous spectra [68,95,97-107]. Furthermore, at least two additional small shoulders (features 1 and 2) are observed on Table 4.1.2 Energies, term values and possible assignments for features in the S 2p spectrum of S F 6 Feature (a> Energy Loss(b> Term Value (eV) Possible Assignment (eV) 2 p 3 / 2 2 p 1 / 2 2 p 3 / 2 2 p 1 / 2 1 171.84 8.56 -2 172.06 8.34 • 6 a i g -3 172.28(vert) 8.12 -4 173.38 - 8.22 - 6 a 1 g 5 175.92 4.48 - 6t!u -6 176.76 - 4.84 - 611. u 7 177.44 2.96 4s -8 178.20 2.20 - 4p -9 178.64 - 2.96 - 4s 1 0 178.92 1.48 - 5s -1 1 179.44 0.96 2.16 6s 4p 1 2 179.96 - 1.64 - 5s 1 3 180.68 - 1.00 - 6s 1 4 183.44 -3.04 - 2 t 2 g -1 5 184.54 - -2.94 - 2 t 2 g 1 6 196.2<c> -15.2(9) 4 e g 1 7 205<d> -24(9) EXELFS 1 8 218<d> -37(9) EXELFS S 2 p 3 / 2 I P 180.40<e> 0 S 2 p 1 / 2 I P 181.60(0 - 0 (a) See Figures 4.1.1 and 4.1.2. (b) Estimated uncertainty is ±0.03 eV. (c) Estimated uncertainty is ±0.3 eV. (d) Estimated uncertainty is ±1 eV. (e) S 2 p 3 / 2 I P from XPS measurement [118]. (f) S 2p 1 / 2 IP estimated by assuming A(S 2p 3 / 2 - S 2p1/2)=1.2 eV, see Ref. [87]. (g) Broad structure, term value with respect to average S 2p IP (181.00 eV). Table 4.1.3 Energies, term values and possible assignments for features in the S 2s spectrum of SF6 Feature ( a ) Energy Loss^) Term Value Possible Assignment (eV) (eV) (final orbital) 1 236.9 7.8 6a l g 2 240.5 4.2 6 t i u 3 241.7 3.0 4p 4 246.7 -2.1 2 t 2 g 5 259<c> -14 4e„ S 2s IP 244.7<d> (a) See Figure 4.1.1. (b) Estimated uncertainty is ±0.03 eV. (c) Estimated uncertainty is ±1 eV. (e) S 2s IP from XPS measurement [118]. Table 4.1.4 Energies, term values and possible assignments for features in the F Is spectrum of SFg Feature*") Energy Loss*b) Term Value Possible Assignment (eV) (eV) (final orbital) 1 688.27 6.77 6a J g 2 692.55 2.49 4p 3 694.15(vert) 0.89 -i 4 694.95 0.09 , 6 t i u 5 695.44 -0.40 6 696.64 -1.60 -1 7 699.2<c) -4.2 2 t 2 g 8 712.6<c> -17.6 4e g 9 722(d) -27 EXELFS 10 75 Kd> -56 EXELFS Is IP 695.04 (a) See Figure 4.1.1. (b) Estimated uncertainty is ±0.10 eV. (c) Estimated uncertainty is ±0.3 eV. (d) Estimated uncertainty is ±1 eV. (e) F Is IP from XPS measurement [118]. S 2p, 2p, edges I • • 1 • I i i i i I — i i i 1 I ' 170 175 180 185 ENERGY LOSS (eV) Figure 4.1.2 (a) Medium-resolution; (b) high-resolution S 2p ISEELS spectra of SF6 showing the Rydberg structures. 58 the S 2pj/2~> Ga^ g band and weak shoulders are also apparent in the other broad valence bands. These shoulders could be attributed to vibronic structures but the spacing of 0.22 eV between features 1, 2 and 3 is substantially larger than that of the ground state vibration frequency of ~0.1 eV [120]. Alternatively these shoulders may be multiplet structures due to exchange interactions [121] (see below). In a previous ISEELS study [68], it was suggested that transitions to the 6t^ u orbital are located under the Rydberg structures in the 176 to 182 eV energy region. However, optical work by Nakamura et al. [102] and also by Gluskin et al. [104] reported only weak Rydberg structures in this region, without any identification of features assignable. to the (formally dipole forbidden) 6t u^ transition. In addition, there is a significant difference between the intensities and number of Rydberg structures observed in the optical works [102,104] and the earlier ISEELS work [68]. These differences were attributed to non-dipole contributions from np Rydberg states in the earlier ISEELS spectrum [68]. In order to clarify these questions, a high resolution spectrum in the 174 to 183 eV region have been measured and it is shown in Figure 4.1.2a. Under the present conditions of higher impact energy (3000 eV) and zero degree scattering angle the 2 momentum transfer (K ~0.1 au) is significantly smaller than that in the earlier work [68] o 2 (EQ = 2500 eV, t9~2°, K ~0.6 au) and therefore the non-dipole contributions should be minimized. The present spectrum, which has higher resolution and better statistics than the earlier work [68], also shows intensity below the Rydberg structure consistent with the existence of two weak bands (features 5 and 6) which can be attributed to spin-orbit components of the dipole-forbidden 6tju transition. As has also been pointed out earlier by Hitchcock and Brion [68] this same broad structure, due to transitions to the 6t u^ orbital, can in fact be seen below the Rydbergs between 175 to 179 eV in the photoabsorption spectrum [102,104]. The dashed lines in Figure 4.1.2a indicate the estimated contributions to the underlying 6t u^ transitions while the solid line represents the estimated valence-shell background. It is noteworthy that the S 2 p ^ spin-orbit component of the transition to 6t u^ is again more intense than the S Ip^^ component as observed in the transitions both to the 6a, and the 2u orbitals. The situation is in lg 2g 59 sharp contrast to the situation for the transitions to 4s Rydberg orbital where the relative intensity of the S 2p^2 and S 2 p ^ component is very close to the statistical value of 2 to 1. It is, therefore, likely that the spin-orbit intensity ratio for each of the ^ ^1/21/2 "* ^ a l g ' ^ l u ^ 2 t2g ^ansitions *s modified by the exchange interactions of the core orbital with the penetrating final orbitals, as discussed by Schwarz [121]. Similar redistributions of spin-orbit intensity have been observed in the Br 3d spectra of CH^Br [62] and C2H^Br (see Chapter 5.1) as well as in the S 2p spectrum of SO^ (see Ref. [122] and Chapter 7.II). It should be noted that, in addition to an unusual spin-orbit intensity distribution, such final state mixing results in complicated multiplet structures on the S 2p^/21/2~* ^ b l ^ands * n SC>2 [122] which have been successfully accounted for by Shklyaeva et al. [123] in a Hartree-Fock-Roothaan calculation with allowance for spin-orbit coupling. Therefore, the shoulders observed on the 6aj band in SF^ may have a similar origin. Two ns Rydberg series (n = 4 to 6, 6 =1.86) converging to the S 2 p 3 / 2 and 2pjy2 edges respectively are observed in Figure 4.1.2 in agreement with previous optical and ISEELS works [68,102,104]. Corresponding peaks in these two ns Rydberg series are separated by 1.2 eV, in excellent agreement with the spin-orbit splitting of 1.2(1) eV derived from XPS measurements [87]. In addition, the present work supports the existence of the formally dipole-forbidden S 1/2 ^ Rydberg transitions observed earlier [68] with greater relative intensity. It should be noted that the same weak and sharp structures have also been observed (but not assigned) in the high resolution (0.08 eV) synchrotron spectrum of the SFg S 2p region by Gluskin et al. [104] and are also visible in the optical work reported by Nakamura et al. [102]. i • i I In the F Is spectrum (Figure 4.1.1a), the shoulder (feature 2) on the lower energy of the 6 ^ band, first observed by Hitchcock and Brion [68], is better resolved in the present work. Based on a term value1 of 2.49 eV, this feature is ascribed to the F Is-* 4p Rydberg transition. It should be noted that much more intense peaks attributed to transitions to the first member of the p Rydberg series have been observed at similar 60 relative energy (term value) in the F Is spectra of various fluorine containing molecules [62,124]. In addition, extra shoulders (features 4-6) have been observed on the higher energy side of the 6t u^ band in the present higher resolution F Is spectrum. One possible explanation for these shoulders is the splitting between the F Is MOs (2a-, , le and ltj u). However, the energy separations of features 4-6 (0.5-1.2 eV) are much larger than the expected separation of the F Is levels (<0.1 eV [68]). In addition, more features are observed than are predicted in the above MO scheme, even if all possible dipole-allowed and forbidden transitions are taken into consideration. Therefore a more plausible explanation for the splitting is the breaking of the octahedral symmetry of the SFg molecule during the F Is excitation,. On creating an F Is hole the symmetry of the molecule is reduced to C^ v if the vacancy is localized on one of the F centers. As a result the 6t^  virtual valence (final) orbital will split into two orbitals of a^  and e symmetries respectively and in addition the F Is (initial state) MO's will also split into two a-^  and two e MO's. In this situation, a complex spectrum will result with four components (a-^  -*a ,^ e •+e) as observed (features 3-6 in Figure 4.L1, Table 4.1.4). It is noteworthy that the 6a, band in the F Is spectrum is slightly asymmetric which may be an indication of multiple transitions with the same origin as the 6t u^ band. Features 9 and 10 have earlier been assigned to EXELFS structures [68]. These EXELFS structures (analogous to EXAFS structure in photoabsorption) are caused by scattering of the outgoing ionized electrons by neighbouring atoms. Such EXELFS structures have been observed earlier [125] in the Cl 2p continua of the chloromethane series. 4.1.2.3 Inner-shell excitation of SeFg The F Is, Se 3s, 3p and 3d excitation spectra of SeFg are shown on the same relative energy scale in Figure 4.1.3 and they are aligned with respect to the lowest IP in each spectral region (eg. Se 3p 3 / 2 f ° r Se 3p spectrum). Unfortunately there is no published measurement of IP for any of the inner-shell levels in SeFg. Therefore, all the IPs shown in Figure 4.1.3 have been estimated using data from tables of XPS measurements of other M F R type molecules [118]. These estimates involve the following i 61 7aig 7t|U F ls edge 8 =0° E 0 3 7 0 0 e V AE= 0.135 eV Fls 5Gn x2 '^ ;---\v.vs-.-'':.-.i;->^ ..^ .. 4 5 J I I 1 I I l__J L H CO LU 6 8 0 7 0 0 ( b ) 7 t ' " , Se 3s edge 7 2 0 7 4 0 6 =0° E 0 3 0 0 0 e V 7 6 0 7 8 0 Se 3s 7a I 2 ' i I i I . I . I i 1 i I i AE=O.I30eV ( B A C K G R O U N D SUBTRACTED) j I i I i LU > _ J LU r r 2 3 0 (c) 2 5 0 270 2 9 0 | Se3p^ edge 6 = 0° 310 3 3 0 E n 3 7 0 0 e V 7alg7tlu stg ^ £ = 0 . 1 3 5 ^ Se 3p ig r\ 3 t2g 5e? x3 ( B A C K G R O U N D SUBTRACTED) I i i i i i I i I . I i I (d) 160 180 7t,u 2 0 0 2 2 0 2 4 0 e =0° Se3d edges E 0 3 7 0 0 eV 2 6 0 Se 3d 56Q AE= 0.135 eV x7 I 356 _L 8 ( B A C K G R O U N D SUBTRACTED) • I • I • I • I • I 6 0 8 0 100 120 ENERGY LOSS (eV) 140 160 Figure 4.1.3 Comparison of (a) F 1s; (b) Se 3s; (c) Se 3p; (d) Se 3d ISEELS spectra of SeF6-62 approximations: (a) A good linear correlation of the F Is IP with the electronegativity of M for M F n type molecules has been found for each of second-row (N-F) or third-row (P-Cl) elements M (Figure 4.1.4). Therefore, assuming a. similar correlation for the fourth-row fluorides (As-Br) gives an F Is IP of 694.1 eV for SeFg. (b) For the following third-row fluorides (MF n) : GeF^, AsF^, AsF<. and BrF^, a linear equation of M 3d level chemical shift against n is derived by least squares fitting the XPS data [118,126]. The M 3d chemical shift, say for GeF 4, is defined as the energy difference between the Ge 3d levels in the GeF 4 molecule and the Ge atom. Extrapolating to n equal to 6 gives the Se 3d chemical shift of SeFg to be 15.7(3) eV. Adding this value to the mean value of the spin-orbit components of the atomic Se 3d ionization energy [126] gives a mean value of 72.4(1.1) eV for the Se 3d IP in SeFg. (c) As discussed in the studies on dimethyl sulfoxide (Chapter 7.II) and SC^ (Chapter 7.II), the energy difference between inner-shell IPs corresponding to core orbitals on the same atom (e.g. S 2p and S 2s levels) in different molecules is largely independent of the molecular environment surrounding the atom. Therefore, assuming that the energy difference between any two Se inner-shell IPs of SeFg is the same as that for the Se atom [126], the other inner-shell IP of SeFg can be estimated directly by adding the same chemical shift of 15.7(3) eV derived above to the IPs of other atomic Se levels [126]. Similar to the situation in SFg (Figure 4.1.1), the discrete region of each inner-shell excitation spectrum of SeFg (Figure 4.1.3) is dominated by two broad features and prominent continuum features can only be located in the F Is and Se 3p spectra. It should be noted that these features also occur at about the same relative energies with respect to the corresponding IPs. Note (see below) that the Se 3p spectrum clearly exhibits spin-orbit splitting. With these considerations in mind, it is logical to interpret O 2nd-row A 3rd-row (eq) • 3rd-row (ax) + 4th-row • 5th-row i - | 1 1 1 i 1 i i 1 1 i 6 9 2 6 9 3 6 9 4 6 9 5 6 9 6 6 9 7 F l s IP(eV) Figure 4.1.4 Correlation between F 1s IPs and electronegativities of central atom for M F n type molecules. OS U J 64 the SeFg spectra in a similar manner to that used for SFg (i.e. prominent features in all of the SeFg inner-shell spectra are attributed to a common manifold of 7a^, 7t^u> 3t,g and 5e inner-well states). The intensity distribution of these transitions in each spectrum can be understood in a simple manner by consideration of the electric-dipole selection rules, as was done in the SFg spectra. Energies, term values and possible assignments of all features in the F Is, Se 3s, 3p and 3d spectra of SeFg are summarized in Tables 4.1.5-8 respectively. Excitations of the F Is level to all four virtual valence orbitals are dipole-allowed. Feature 1 in the F Is spectrum (Figure 4.1.3a) is assigned to the 7a^  band because this band is observed to be very intense in both the Se 3p and the 3d spectra as expected by the dipole-selection rule. Similarly, feature 2 can be attributed to the transition to the 7t u^ orbital since this band is intense in the Se 3s and 3d spectra but weaker in the Se 3p spectra (see discussion below). Feature 3 and the relatively weaker feature 4 are assigned to 3 ^ and 5eg continuum shape resonances respectively since the t~ state is also lower in energy and more intense than the e state in the 2g g F Is spectrum of SFg. The fact that these continuum resonances, which are of d-type in the extended basis set description, are not observed in the Se 3s spectrum and very weak in the Se 3d spectrum is again consistent with dipole selection rules. A third broad and weak structure (feature 5), observed at 734 eV on the F Is continuum, is tentatively assigned as extended energy loss fine structure (EXELFS) because of its high relative energy of 40 eV from the F Is IP. Similar weak continuum structures observed in the F Is and S 2p spectra of SFg (see features 9 and 10 in Figure 4.1.1) have also been attributed to EXELFS [68]. The Se 3s spectrum of SeFg' is dominated by the 7t^  band (feature 2) which is the only dipole-allowed transition to the virtual valence orbitals. The 7a^ band (feature 1) only appears as a shoulder on the lower energy side of the 7t u^ band. No structure due to the 3t~ and 5e orbitals can be identified in accord with expectations based on dipole selection rules. Table 4.1.5 Energies, term values and possible assignments for features in the F Is spectrum of SeFg Feature^ Energy Loss(b) Term Value Possible Assignment (eV) (eV) (final orbital) 1 685.71 8.4 7ai g 2 691.71 2.4 7 t i u 3 699.52 -5.4 3 t 2 g 4 716<c> -22 5e g 5 734<c> -40 EXELFS F Is IP 694.l<d> (a) See Figure 4.1.3. (b) Estimated uncertainty is ±0.15 eV. (c) Estimated uncertainty is ±1 eV. (d) F Is IP estimated as discussed in section 4.1.2.3. Table 4.1.6 Energies, term values and possible assignments for features in the Se 3s spectrum of SeF6 Feature ( a ) Energy Loss Term Value Possible Assignment (eV) (eV) (final orbital) 1 236<b> 1 1 7aj g 2 241.8<c> 5.4 7 t l u Se3sIP 247.2<d> (a) See Figure 4.1.3. (b) Estimated uncertainty is ±1 eV. (c) Estimated uncertainty is ±0.1 eV. (d) Se 3s IP estimated as discussed in section 4.1.2.3. Table 4.1.7 Energies, term values and possible assignments for features in the Se 3p spectrum of SeFg Feature ( a ) Energy Loss(b> Term Value (eV) Possible Assignment 3 P 3 / 2 3 p i / 2 3 p 3 / 2 3 p i / 2 1 167.23 10.4 - 7a l g 2 172.94 4.7 10.4 7ti„ 7a l g 3 181.08 -3.5 2.2 3 t 2 g 7 t l u 4 186.9<c> - -3.6 - 3 t 2 g 5 197.0<c> -16.6(e> 5eg Se 3p 3 / 2IP 177.6<d> Se 3p 1 / 2IP 183.3<d> (a) See Figure 4.1.3. (b) Estimated uncertainty is ±0.05 eV. (c) Estimated uncertainty is ±0.5 eV. (d) Se 3p3/2,i/2 IPs estimated as discussed in section 4.1.2.3. (e) Broad structure, term value with respect to average Se 3p IP (180.5 eV). Table 4.1.8 Energies, term values and possible assignments for features in the Se 3d spectrum of SeFg Feature M Energy Loss^ Term Value (eV) Possible Assignment 3 d 5 / 2 3 d 3 / 2 3 d 3 / 2 3 d 3 / 2 1 62.9 9.1 - 7a l g 2 63.7<c> - 9.1 - 7a l g 3 67.20 4.8 - 7 t l u 4 67.9(c> - 4.9 - 7 t l u 5 68.62 3.4 - 5p 6 69.48 - 3.4 - 5p 7 89<d> -17<f> 5eg 8 113<d> -4l(f> EXELFS Se 3d 5 / 2IP 72.0<e> Se 3d 3 / 2IP 72.8<e> (a) See Figures 4.1.3 and 4.1.5 (b) Estimated uncertainty is ±0.02 eV. (c) Estimated position (see text section 4.1.2.3). (d) Estimated uncertainty is ±1 eV. (e) Se 3d5/2>3/2 IPs estimated as discussed in section 4.1.2.3. (f) Broad structure, term value with respect to average Se 3d IP (72.4 eV). 68 A lower resolution (~-0.6 eV FWHM) Se 3p photoabsorption spectrum of SeFg has been reported by Addison et al. [96]. The present higher resolution (0.14 eV FWHM) ISEELS spectrum is consistent with this optical work [96] except that the energies reported by Addison et al. [96] are generally 0.1-0.5 eV lower than those reported in the present work (Table 4.1.7). Although the ISEELS spectrum is measured at a much higher resolution than the optical work, no additional features are observed. It should be noted that in contrast to the S 2p spectrum of SFg the Se 3p features are superimposed on top of the both the valence and Se 3d continua. Presumably fast Coster-Kronig type autoionization induces significant band broadening in the Se 3p features so that no further structures are resolved. Following previous assignments [96], features (1,2) and (3,4) have been attributed to spin-orbit split partners of the excitation to the 7a^g and 3t,g virtual valence orbitals respectively. Such assignments give a Se 3p spin-orbit splitting of 5.7 eV, which is reasonably close to that reported (6.3 eV) for the free Se atom [126]. The relative intensities of peaks 1 and 2 and also of peaks 3 and 4 are close to the expected statistical value of 2 for such spin-orbit pairs. This in marked contrast to the very different situation in the S 2p spectrum of SFg where the statistical ratio is severely perturbed by exchange effects (see section 4.1.2.2). It is noteworthy that peak 3 is asymmetric. A plausible explanation is that a weak Se Sp^-* 7t u^ transition may be present on the lower energy side of peak 3 since similar low intensity, dipole-forbidden t j u bands have been observed in the S 2p spectrum of SFg (Figure 4.1.2a). If this is the case then the Se 3p^2 ** 7t^  transition would be underneath the peak assigned to the Se 3py2 "*7aig transition (feature 2). The fact that peak 2 is slightly broader than peak 1 (Figure 4.1.3c) is consistent with such an interpretation. A further structure (feature 5), seen on the Se 3p continuum is also apparent at the same energy (197 eV) in the optical spectrum of SeFg [96]. This broad feature is most likely due to the transitions from the Se 3p 3 / , and 3py, levels to the 5eg orbital since such transitions are dipole-allowed and the band is located at a similar relative energy as the 5eg band (feature 4, Figure 4.1.3a) in the F Is spectrum. 69 Two broad bands (features 1 and 3 of low and high relative intensity respectively) are observed on the leading edge of the Se 3d spectrum (Figure 4.1.3d). These bands are ascribed to transitions to the 7a^g and 7t^ u orbitals respectively according to their term values as well as their relative intensities which are consistent with dipole selection rules. Each of these broad bands is considered to be composed of two transitions, from the Se 3d^/2 and 3 d ^ spin-orbit split levels. This is consistent with a broad small shoulder (feature 4) in the high resolution (0.055 eV FWHM) Se 3d spectrum shown in Figure 4.1.5. While it is not possible to obtain a definitive deconvolution of these bands, the dotted lines shown in Figure 4.1.5 indicate the estimated positions of the Se 3d^ / 2 components. The sharp features 5 and 6 are most likely spin-orbit partners of transitions to the same Rydberg level since they have an energy separation of 0.86 eV which is reasonably close to the Se 3d spin-orbit splitting of ~1 eV for the free Se atom [127]. The features 5 and 6 are best assigned as 5p Rydberg levels since transitions to the s and d Rydberg series are dipole-forbidden for Se 3d excitation. Two broad, weak features (7 and 8) can be identified in Figure 4.1.3d above the Se 3d IPs. Feature 7 is assigned as the dipole- forbidden Se 3 d ^ 3/2 5eg transition since it corresponds to the positions of 5e bands in the other inner-shell excitations. Feature 8, which is ~41 eV above Se 3d IPs, is probably an EXELFS type structure as discussed above. It should be noted that this feature occurs at approximately the same position as the EXELFS structure (feature 5, Figure 4.1.3a) in the F Is spectrum. This is consistent with the expectation that these EXELFS structures are mainly due to back-scattering of the ionization electrons between atoms linked by the Se-F bond. The observation of Rydberg transitions in the Se 3d spectrum but not in the other core excitation spectra of SeF^ could be a reflection of the relatively more diffuse nature of the Se 3d orbitals. However, an alternative explanation is that the relative increase in Rydberg intensity is a result of the weakening of the potential barrier in going from SF,, to SeF,-. Such a weakening of the potential barrier is also consistent 0 = 0 ° E 0 = 3700 eV ^ E = O.I35eV 0 = 0 ° E 0 = 2000 eV AE = 0.055 eV Se 3d3, edge I I ' ' ' I I I I I I I I I I I 1—I 1 1 1—I 1 1—L 60 64 68 72 76 80 ENERGY LOSS (eV) igure 4.1.5 Detailed Se 3d ISEELS spectrum of SeF6; (upper trace) low-resolution, (lower trace) high-resolution. 71 with the general decrease in intensity of the continuum shape resonances in the inner-shell excitation spectra of SeFg relative to the corresponding situation in SFg. Further discussion of this topic will be presented in section 4.1.2.5. 4.1.2.4 Inner-shell excitation of TeFg An overview of the various inner-shell spectra of TeFg is shown in Figure 4.1.6. There are no reported XPS measurements of IPs for any of the inner-shells in TeFg. Therefore, all the IPs shown are estimated positions, calculated in a similar fashion to those for SeFg, (see preceding section). Assuming a linear correlation between F Is IP and electronegativity of the central atom in the following fifth-row fluorides : IF^ and SbF^, gives an F Is IP of 694.1 eV for TeFg (Figure 4.1.4). The Te 3d level chemical shift (13.4 eV) is estimated from the XPS M 3d values [118] of the following fifth-row fluorides : SbF^, XeF,, XeF^ and XeFg. Similar to the situation for SFg and SeFg, the discrete regions of the inner-shell excitation spectra of TeFg are dominated by one or two prominent broad bands as expected from a consideration of the dipole selection rules. Therefore, by analogy, these features can be readily attributed to transitions to the 8a-, and 8t, virtual valence levels lg lu (see discussion below). However, the continuum regions of the TeFg spectra differ markedly from those in the corresponding spectra of SFg and SeFg. The Te 3d spectrum is particularly striking. Furthermore, while no continuum resonances can be identified in the Te 4p spectrum many prominent peaks are observed on both the Te 3d and Te 4d continua. These observations are in marked contrast to the corresponding spectra of SFg and SeF,. In fact differences in behaviour of TeF,- from SF, and SeF, have also been 6 6 6 6 noticed in other respects, such as charge distributions [128] and valence band energy levels [129]. In contrast to the general inertness of SFg and SeFg, TeFg has interesting chemical behavior since it can undergo complete hydrolysis [130] and can function as an F~ anion acceptor to form TeF^ or TeFg species [130,131]. These differences for TeFg possibly reflect the different electronic structures of the normal unoccupied virtual valence 72 00 LU \-LU > _ J L U (a) 8t,„ 8a,s F Is edge FIs l°2u 0 = 0° EQ = 3000 eV AE = 0.300 eV I 23 4 5 6 7 I 11 I I I I I • I . I • I • I • I • I (b) 6 8 0 7 0 0 gi, 7*20 I • I 7 4 0 _ l I I I L . 7 6 0 7 8 0 8 0 0 Te 3p edge | a 2 u et I 9 V i " Y 'W2 3 4 5 6 7 8 2 0 Te3p 0 = 0 ° Ec = 3000 eV AE = 0.260 eV ( B A C K G R O U N D S U B T R A C T E D ) (c) 8 2 0 I 1 I 1 I 1 I 8 4 0 8 6 0 I 1 I 1 I 8 8 0 9 0 0 l . l . i 9 2 0 9 4 0 9 6 0 ftTe 3di)/2edge Bo„8t,0 9tlu loj, 2t2u Te 3d5/2edge Te3d e = o° EG = 3000 eV AE = 0.260 eV 1" 11 rVi 11 1 l l I I I 234 5678 9 10 11 12 13 I . I . I • I . 1 . 1 L 14 5 6 0 ( d ) 5 8 0 6 0 0 8,!» ^  Te 4s edge 1 1 1 6 2 0 6 4 0 I ( B A C K G R O U N D S U B T R A C T E D ) . l' 5 . I . I . I 1 I 1 ' 6 6 0 8a,< Te4s 6 8 0 7 0 0 7 2 0 9 = 0° ED = 3000 eV AE = 0.130eV ( B A C K G R O U N D S U B T R A C T E D ) 1 l i 2, I 1 I 1 . 1 J i_ (e) I 7 0 I 9 0 2 I 0 8°ig fl 8t,„ Te 4p edges 2 3 0 2 5 0 2 7 0 2 9 0 3 I 0 •v Te4p 1 1 r 1 2 3 _ _ l . I . I , L ( B A C K G R O U N D S U B T R A C T E D ) B = 0° E c = 3000 eV AE = 0.080 eV (f) 1 0 0 1 2 0 st,L 1 6 0 180 1 • 1 . 1 • 1 2 0 0 2 2 0 2 4 0 ,Te4d5/2 ,3/2 edges 8 Q I Q / I 9 ,lu Ia2u 2,2u I I I I 1358 9 10 II Te 4d e = o° EG = 3000 eV AE = 0.080 eV 3 0 5 0 I 1 1 ( B A C K G R O U N D S U B T R A C T E D ) 1 • 1 • 1 . 1 • 1 • 1 _, L 7 0 9 0 110 1 3 0 150 1 7 0 190 ENERGY LOSS (eV) Figure 4.1.6 Comparison of (a) F 1s; (b) Te 3p; (c) Te 3d; (d) Te 4s; (e) Te 4p; (f) Te 4d ISEELS spectra of TeF6-73 levels in the case of TeF^ compared to SFg and SeF^ (see section 4.1.2.1). The continuum resonances in the inner-shell excitation spectra of SFg and SeF^ are interpreted as d-type resonances as a result of the involvement of the empty 3d and 4d AOs respectively on the central atom. However, as discussed in section 4.1.2.1 the empty 4f orbitals are lower than the empty 5d orbitals in Te. This leads to the expectation that the 4f AOs would have a larger involvement than the 5d AOs in the highly excited states of TeFg. The observation of strong continuum resonances in the Te 3d and 4d spectra (i.e. d ->f) but no continuum resonance in the 4p spectrum is consistent with the assignment of dominant f character to these resonances. With the inclusion of Te 4f AO's three extra virtual valence orbitals (9t^ u > la2 u and 21^) will be added to the minimum basis set MO scheme of TeF^ as shown in section 4.1.2.1 above. The TeF^ spectra will be tentatively assigned according to such an extended MO scheme including Te 4f orbitals. The energies, term values and possible assignments for all features in the F Is, Te 3p, 3d, 4s, 4p and 4d spectra of TeFg are summarized in Tables 4.1.9-14 respectively. Features 1 and 2 in the F Is spectrum of TeF^ (Figure 4.1.6a) can be assigned as transitions to the 8a^ and 8t^ u orbitals respectively because the intensity of these two bands in each of the various different inner-shell excitation spectra are in accord with expectations based on the dipole-selection rules. The small shoulder (feature 3) on the 8tj u band may have the same origin as similar structures observed on the 6t^ u band in the F Is spectrum of SFg (features 3-6, see Figure 4.1.1a), which are assigned as multiple transitions due to breaking of the O^ molecular symmetry during the excitation. There are five structures (features 4-8) which have been located in the F Is continuum of TeFg. Features 7 and 8 are attributed to EXELFS structures based on their high relative energies and weakness. On the other hand the features 4- 6, which are lower in energy, are attributed to the three f-type shape resonances. Since the transition to the l a 2 u orbital is dipole-forbidden from all F Is MO's, it is attributed to the weakest structure feature 5. In contrast transitions to the 9t^ u orbital are dipole-allowed from the F Is l a l g and l e ^ orbitals, these assignments are attributed to the strongest structure in Table 4.1.9 Energies, term values and possible assignments for features in the F 1s spectrum of TeF$ Feature^' Energy Loss(b> Term Value Possible Assignment (eV) (eV) (final orbital) 1 685.78 8.3 8a 1 g 2 691.22(vert) 2.9 8ti u/6p 3 692.91 1.2 8tlu 4 697.8<c> -3.7 9tlu 5 704.8(c> -10.7 1*2u 6 712.3<°> -18.2 2t2u 7 726<d> -32 EXELFS 8 758(d> -64 EXELFS SIP 694.1 <e> (a) See Figure 4.1.6. (b) Estimated uncertainty is ±0.15. (c) Estimated uncertainty is ±0.5 eV. (d) Estimated uncertainty is ±1 eV. (e) F 1s IP estimated as discussed in sections 4.1.2.3 and 4.1.2.4. Table 4.1.10 Energies, term values and possible assignments for features in the Te 3p spectrum of TeF 6 Feature'3) Energy Loss<b) Term Value Possible Assignment^) (eV) (eV) (final orbital) 1 816 16 unknown 2 824 8 8 a 1 g 3 829 3 8tiu 4 834 -2 unknown 5 836 -4 9 t i u 6 844 -12 1a2u 7 848 -16 2t2u/ed Te3p 3 / 2 IP 832.1 (c> (a) See Figure 4.1.6. (b) Estimated uncertainty is ±1 eV. (c) Te 3p 3 / 2 IP estimated as discussed in sections 4.1.2.3 and 4.1.2.4. (d) Initial state is Te 3p 3 / 2 . The Te 3p-|/2 location is outside of the spectral range of Figure 4.1.6. Table 4.1.11 Energies, term values and possible assignments for features in the Te 3d spectrum of TeF6 Featured) Energy LossC3) Term Value (eV) Possible Assignment 3ds/2 3 d 3 / 2 30*5/2 3 d 3 / 2 1 577.32 8.2 -2 581.61 3.9 -3 582.50(vert) 3.0 -4 584.90 0.6 1 1.0 5 587.59 -2.1 8.3 6 590.18 - 5.7 7 592.18(vert) - 3.7 8 593.17 - 2.7 9 595.67 -10.2 -1 0 598.56 - -2.7 1 1 603.65 -18.2 -1 2 606.14 - -10.2 1 3 612.72 - -16.8 1 4 631 (c) -40(f) 1 5 662(c) -71(0 Te 3d5 / 2IP 585.5(d) Te3d 5 / 2IP 595.9(d) ) 8a 1 g 8tiu unknown<e) 9tiu 8ai g 1a 2 u 2t2 u 8ti u 9tiu 1a2u 2t2u EXELFS EXELFS (a) See Figure 4.1.6. (b) Estimated uncertainty is ±0.15. (c) Estimated uncertainty is ±1 eV. (d) Te 3d 5 / 2 l 3 / 2 IPs estimated as discussed in sections 4.1.2.3 and 4.1.2.4. (e) Origin of this small peaks is uncertain. It is possibly a Rysberg state. (f) Broad structure, term value with respect to average Te 3d IP (590.7 eV). Table 4.1.12 Energies, term values and possible assignments for features in the Te 4s spectrum of TeF 6 Featured) Energy LossC3) Term Value (eV) Possible Assignment (eV) (eV) (final orbital) 1 164.4 8.3 8a-| g 2 179.3 , 2.4 8ti u Te4sIP 1 8 1 . 7 ( ° ) (a) See Figure 4.1.6. (b) Estimated uncertainty is ±0.5 eV. (c) Te 4s IPs estimated as discussed in sections 4.1.2.3 and 4.1.2.4. Table 4.1.13 Energies, term values and possible assignments for features in the Te 4p spectrum of TeF6 Feature^) Energy LossC3) Term Value (eV) Possible Assignment 4P3/2 4 p 1 / 2 4 p 3 / 2 4 p 1 / 2 1 113.6 6.7 - 8a-|g 2 , 116.7 3.6 - 6s/8ti u Te4p 3 / 2 IP 1 2 0 . 3 ( ° ) Te4p 1 / 2 IP 1 3 0 . 3 ( ° ) (a) See Figure 4.1.6. (b) Estimated uncertainty is ±0.5 eV. (c) Te 3d5/2l3/2 IPs estimated as discussed in sections 4.1.2.3 and 4.1.2.4. 78 Table 4.1.14 Energies, term values and possible assignments for features in the Te 4d spectrum of TeFg Feature^) Energy Loss^) Term Value (eV) Possible Assignment (eV) Unccorreted(e) Corrected^) 4 d 5 / 2 4 d 3 / 2 4 d 5 / 2 4 d 3 / 2 4 d 5 / 2 4 d 3 / 2 1 2 3 4 5 6 7 8 9 10 1 1 4 9 4(c) 50.5(c) 51.19 51.50 52.45 52.67 53.33 53.91 58(d) 66(d) 77(d) 3.3 3.5 1.5 1.2 1.6 0.0 1.3 -0.6 0.1 -5(8) -13(g) -24(8) 4.8 5.0 3.0 2.7 3.1 1.5 2.8 0.9 1.6 -3(8) -11(g) -22(g) 8a l g 8a l g 8 t i u 6p 8 t l u 7p 6p 8p 7p 9 t i „ l a 2 u 2t 2 u Te 4d IP 52.7(e) 54.0(e) 54.2(f) 55.5(f) (a) See Figures 4.1.6 and 4.1.7. (b) Estimated uncertainty is ±0.02 eV. (c) Estimated uncertainty is ±0.5 eV. (d) Estimated uncertainty is ±1 eV. (e) Te 4d5 / 2 ,3 / 2 IPs estimated as discussed in sections 4.1.2.3 and 4.1.2.4. (f) Te 4d5/2,3/2 IPs readjusted as discussed in section 4.1.2.4.6. (g) Broad structure, term value with respect to average Te 4d IP (53.4 eV for uncorrected IPs and 54.9 eV for corrected IPs). 79 this region (feature 4). Transition to the 2L , u orbital is dipole-allowed from the F Is le g orbital only, it is therefore attributed to the relatively weaker feature 6. Features 2 and 3 in the Te 3p spectrum (Figure 4.1.6b) are assigned to Te 8aig and Te ~* transitions respectively according to their term values which are very similar to those of corresponding bands in the F Is spectrum. However, the 8t u^ band is much stronger than would be expected since this band is predicted to be dipole-forbidden. Similar "unusual" strong structures are also found in the continuum region. Such an unusual distribution of intensity is not however observed in the Te 4p spectrum (Figure 4.1.6e). It should be noted that the spin-orbit splitting between Te Ip^p ^1/2 * 0 n * c s t a t e s *s 1^ ^ 0 T ^ ^ e a t o m [126] anc*> therefore a very large spin-orbit interaction will also be associated with the creation of a Te 3p core hole in TeF^. The Te 3 p ^ spectrum is expected in the region of approximately 900 eV. However, an examination of this region did not reveal any significant structure although it should be noted that this energy is at the upper limit of the spectral range accessible with the spectrometer used in the present work. Above the Te ty^/l e c ^ e ' a broad maximum (feature 7 in Figure 4.1.6b) is observed at 848 eV. The broadness of this structure (~16 eV) suggests that it is probably a delayed onset of the Te 3pj^ continuum [132]. In addition, the 2t2u shape resonance may also contribute to this feature because its term value of -16 eV is similar to that of corresponding 2t2u shape resonance in the F Is spectrum (Figure 4.1.6a, feature 6). At least three more weaker structures (features 4-6) can be identified above the Te 3p.^ 2 edge. Features 5 and 6 are ascribed to the 9t u^ and la2u f-type shape resonances respectively because their term values and the energy separations between these features are very similar to those found for the 9t^  and la2u resonances in the F Is as well as the Te 3d and 4d spectra. The origins of features 1 and 4 as well as the unusual intensities of the dipole-forbidden transitions to the 8t^  , 9t^  and 2t2u orbitals are not understood. 80 A background subtracted Te 3d excitation spectrum of TeF^ is shown in Figure 4.1.6c. It should be noted that the estimated background is a broad maximum (see insert of Figure 4.1.11c) rather than a aE b decaying tail to the underlying shells which is usually observed in inner-shell excitation spectra. Subsequent experiments showed that the large background observed is due mainly to stray scattered electrons accidentally entering the energy analyzer because of non-optimal focussing conditions of the electron lenses. A Te 3d spectrum with normal aE~ decaying background was subsequently obtained by careful retuning of the spectrometer. However, owing to the limited availability of the TeFg sample, this spectrum could not be scanned' for sufficient time to build up a good signal to noise ratio. Nevertheless, all structures were observed to be present and the overall spectral shape was consistent with the background subtracted spectrum of Figure 4.1.6c. The Te 3d excitation region is very rich in structures which can readily be fitted into two series of transitions originating from the Te 3 d ^ 2 and 3 d ^ 2 levels separated by an sph-orbit splitting of 10.4 eV estimated from the Te atom [126]. Features 1 and 5 are assigned as spin-orbit partners of the dipole-forbidden 8a^ bands because of their weak intensities and very similar term values (8.2 and 8.3 eV) to those of the 8a^g bands observed in the other spectra of TeFg. Similarly, features (2,3) and (6-8), 5 and 10, 9 and 12, 11 and 13 are ascribed as the spin-orbit components of the 8t^ u > 9t^ u > l a 2 u and 2t>u levels respectively. Feature 4 is tentatively assigned as a Rydberg structure. The multiplet structures observed on the 8t^ u bands may be due to exchange interactions since the relative intensity of those two spin-orbit partners is about 1:1 which is very different from the 3:2 theoretical statistical ratio neglecting exchange interaction [121]. Alternatively, the small shoulder observed on the 8t^ u bands may be caused by distortion of the band profile when these states autoionize into the underlying continua. Two weak and very broad maxima (features 14 and 15) are observed at 40 and 71 eV above the Te 3d average IP (590.7 eV). These high lying features are most likely EXELFS structures corresponding to those observed in the F Is spectrum (features 7 and 8 in Figure 4.1.6a) because they have very similar energy separation (32 eV in F Is and 31 81 eV in Te 3d). The higher term values of these EXELFS feature observed in the Te 3d spectrum compared with those of relevant structures in the F Is spectrum may be due to the overlap of EXELFS structures corresponding to the respective Te 3d^ / 2 and 3dj / 2 continua or due to phase shift differences [44]. Only a single broad maximum (feature 2) is clearly observable in the Te 4s spectrum (Figure 4.1.6d). It is logical to assign this feature as the Te 4s-> 8t l u transition because it has a term value of 2.4 eV which matches very well with those of 8t l u bands in the other TeFg spectra and, in addition, because this dipole-allowed transition is expected to be strong. The dipole-forbidden transition to the 8a^g orbital may correspond to the very weak feature 1. No f-type continuum shape resonances (see above) can be identified above the Te 4s edge. However, no such continuum shape resonances are to be expected since s -» f type transitions are highly unfavorable in the atomic case. In addition, it should be noted that feature 2 in the Te 4s spectrum is much broader than features in the discrete region of the Te 4p and 4d spectra because of fast Coster-Kronig type autoionization of the Te 4s feature to the underlying Te 4p and 4d continua. Presumably such processes may also cause any shape resonance maximum in the Te 4s continuum to become too broad to be observable. Only two features can be identified in the Te 4p spectrum (Figure 4.1.6e). They are assigned as Te 4p^ / 2 -+ 8a l g (feature 1) and Te 4p^ / 2 - » 8 t l u (feature 2) transitions according to their term values (see Table 4.1.13) and relative intensity. No f-type continuum shape resonance is observed, as is expected from a consideration of dipole selection rules. The discrete region of the Te 4d spectrum (Figure 4.1.6f) is dominated by two broad bands (features 3 and 5) separated by 1.26 eV which is very close to the spin-orbit splitting of 1.25 eV reported in the Te 4d photoabsorption spectrum of TeF^ [133]. Therefore, these two bands are tentatively assigned as the spin-orbit partners of the dipole-allowed 8t u^ bands. In the high resolution (0.033 eV FWHM) Te 4d spectrum shown in Figure 4.1.7, two small shoulder (features 1 and 2) are observed on the lower TeFg Te4d 0 = 0 ° E o = 3 0 0 0 eV A E = 0.150 eV (9 = 0 ° E 0 =3000eV A E = 0.033 eV '//, T e 4 d 5 edge 3 4 5 6 7 8 Te4d 3 edge 48 49 50 51 52 53 ENERGY LOSS (eV) 54 55 Figure 4.1.7 Detailed Te 4d ISEELS spectrum of TeF6; (upper trace) low-resolution, (lower trace) high-resolution. oo K> 83 spin-orbit partners of the dipole-forbidden 8a-^  band. Extra small shoulders (features 4 and 6) can be observed on the 8t u^ bands. These shoulders, as well as the weak sharp features 7 and 8 are most likely due to Rydberg transitions. However, the term values of these features seem to be too low to fit into a predominant np Rydberg series. Furthermore, similar lowering in term values in the Te 4d spectrum relative to the other Te spectra has been observed on the 8a^ and 8t u^ bands (Table 4.1.14). Such a simultaneous lowering of term values is probably a result of underestimation of the Te 3d IPs. Therefore, the original estimates of the Te 4d IPs have been readjusted so that the term value for the 8t u^ band is equal to the average value of 3.0 eV estimated from the other TeFg spectra. With this adjustment, features 4, 6, 7 and 8 can be readily assigned as transitions to the 6p, 7p and 8p Rydberg levels as shown in Table 4.1.13 based on these readjusted term values. Three structures (features 9-11) are observed on the Te 4d continua. They are readily assigned as the dipole-allowed transitions terminated at the 9tj , l a 2 u and 2t_2u orbitals (f-type) respectively according to their relative positions from the Te 4d IPs. 4.1.2.5 Comparison of the ISEELS spectra of SF ,^ SeFg and TeFg It is now instructive to consider a direct comparison of the respective corresponding ISEELS spectra of the SFg, SeFg and TeF^ molecules. Accordingly ISEELS spectra of these hexafluorides with same relative energy scales are grouped into different figures under the following classifications : (Figure 4.1.8) F Is; (Figure 4.1.9) S 2s, Se 3s and Te 4s; (Figure 4.1.10) S 2p, Se 3p, Te 4p and Te 3p; and (Figure 4.1.11) Se 3d, Te 4d and Te 3d. The spectral characteristics of the discrete regions in any given group of these ISEELS spectra are very similar (note that spin-orbit splitting results in two series of lines in some of the spectra). This is a reflection of the situation that the lowest two (bound) virtual valence orbitals are of a'^  and t^u symmetry in all three hexafluorides. Recently, correlation of the term value of a bands with the bond strength of the relevant a bond has been discussed by Ishii et al. [60]. A similar correlation has been demonstrated in the monohaloethylenes (Chapter 5) and also the methyl halides [62]. 84 >-CO LU LU > _ l LU r r ( a ) 6 a l g 6 t l u 2 t 2 g £ = 0 ° 4eg E o =3000eV A E = 0.260 eV SF € Fls 1 23 1 1 1 * 6 5/7 8 9 I I 1 1 1 " 10""" 1 1 1 i ^ F Is edge 1 . 1 . 1 , 1 , 1 , 1 , 6 8 0 7 0 0 7 2 0 7 4 0 J I i u 7 6 0 7 8 0 0 = 0 ° E o =3700eV A E = 0.I35 eV SeF6 F l s 5e0 I 2 3 I I ... i W i s edge J — i — I — i I i l i l i I i l i i i- i i i i 6 8 0 8t l u 7 0 0 7 2 0 7 4 0 7 6 0 7 8 0 (C) 8a,g \ 9t|U V>\ l a 2 u 2t 2 u 0 = 0 ° E o =3000eV A E = 0.300 eV TeF6 Fls I 23 4 5 6 i 1 1 . . . 1 1 i_ v/. F Is edge J i i i i i j i i i L 680 700 720 740 760 E N E R G Y L O S S (eV) 780 Figure 4.1.8 Comparison of the F 1s ISEELS spectra of (a) SF6; (b) SeF6; (c) TeFg. 85 CO UJ UJ > LU cr (a) H S 2s edge I I I I I 1 2 3 4 5 SF 6 S2s 6a AE = 0.26 eV 260 v//- Se 3s edge 280 Sef^ Se3s 2 2 0 " ' " 7 a V V ft ••" AE= 0.13 eV 240 260 280 (c) I 2 Te 4s edge i i m. TeF6 Te4s e = o ° E0= 3000 eV AE = 0.13 eV 160 180 200 240 ENERGY LOSS (eV) Figure 4.1.9 Comparison of (a) S 2s spectrum of SF6; (b) Se 3s spectrum of SeF6; (c) Te 4s spectrum of TeF6-86 >-CO -z. LU LU > _J LU rr ( a ) S 2 p e d g e s % e = o° E D = 3 0 0 0 eV A E = 0 . 2 6 0 eV 3 4 7 9 1415 16 17 160 • I . I • l . l . 180 2 0 0 2 2 0 2 4 0 _ l I L . 2 6 0 i . p e 3 p 3 / 2 e d 9 e 7 a i g 7 t | U 3t2g % Se 3 p ^ edge (BACKGROUND SUBTRACTED) 0 = 0 ° E 0 = 3 7 0 0 eV A E = 0 . 1 3 5 eV (BACKGROUND SUBTRACTED) „ R t 9t, .; i ^ ^ r - ' 2 ' 3 4 5 6 7 ! I I I I I I • I . I . I i I i L 0 = 0 ° E D = 3 0 0 0 eV A E = 0 . 2 6 0 eV J L _ I I I I L _ J 1 — 1 _ 810 910 8 3 0 8 5 0 8 7 0 8 9 0 ENERGY LOSS (eV) Figure 4.1.10 Comparison of (a) S 2p spectrum of SFQ; (b) Se 3p spectrum of SeF6; (c) Te 4p and (d) Te 3p spectra of TeF6. 87 >-H CO UJ LU > _ l LU 7t„ V \ i Se 3d edges SeF^ Se3d e =0° E 0 3000 eV AE=O.I50eV 5e„ x 7 I 356 8 (BACKGROUND SUBTRACTED) J i I • I • I • I . I 120 140 1 6 0 TeF6 Te4d 9 =0° E 0 3000 eV (BACKGROUND SUBTRACTED) AE = 0.150 eV 13 9 10 II m i I I I I I I I I 4 0 6 0 (c)9 =0° E 0 3000 eV AE=O.I50eV 8 0 i 100 120 1 4 0 K Te 3da /gedqe 8o^ 8t|U 9t,u lOj,, 2t2(1  | Te3dii/2edqe ^..-••-'(BAC) (ORIGINAL DATA) J I I I L_l I I I I—L CKg*6L 570 590 610 630 650 670 TeF6 Te3d (BACKGROUND SUBTRACTED) 234 567891011 12 13 I II I I I II I I I I I ' • ' • L-J L_J_ 14 I 15 I 570 590 610 6 3 0 650 ENERGY LOSS (eV) 670 Figure 4.1.11 Comparison of (a) Se 3d spectrum of SeFg; (b) Te 4d and (c) Te 3d spectra of TeF6. 88 A search for such a correlation in the present work would be unrealistic because of inherent uncertainties of the estimated IPs for the SeFg and TeFg core levels (see sections 4.1.2.3 and 4.1.2.4). However, it can be seen that the term values of the and tj bands from the same class of excitation are fairly constant for the three hexafluorides (see Tables 4.1.18-20 below). Such observations are not inconsistent with the fairly close M - F bond energies of 79, 73 and 82 kcal/mole for SFg, SeFg and TeFg respectively [134] since both the bands in question have a (M-F) character. Furthermore, the energy separation between these two bands is very similar in the different F Is spectra (Note that intense transitions to both bands are only observed in the F Is spectra whereas one or other of these bands is heavily discriminated against in the other spectra due to dipole selection rules). The most pronounced effect on the discrete features of the spectra resulting from the change of the central atom is an increase of the t^  band intensity relative to that of the a, band in going from SF. to SeF, to TeF £ in the F Is lg o b 6 6 6 spectra. This indicates a relative increase in F 2p character in the t^  band as the central atom becomes larger in the MFg molecules. The very different spectral characteristics in the continuum regions of the spectra of TeFg compared with those of SFg and SeFg have been discussed above (section 4.1.2.4). Such differences cannot be rationalized by any simple model based on dimensional changes of the molecular potential as the size of the central atom is increased. As discussed above the involvement of 4f orbitals in the virtual valence orbitals of TeFg (or alternatively a shift to f-type outgoing channels in terms of shape resonances) is a better explanation of these observations in view of the availability of low-lying empty 4f orbitals in Te (rather than the empty 3d or 4d orbitals in SFg and SeFg respectively). It should be noted that the 4f orbitals are lower-lying than the 5d orbitals in Te [135]. Although the f orbitals are generally too diffuse to enter into bonding, so that there are few chemical effects from the presence of f electrons or unfilled f orbitals, there are, electronic effects of f electrons which show up in the absorption spectra and magnetic properties of the lanthanide series [135]. In addition, f orbital involvement can contribute 89 to bonding in complexes of high coordination number [135], which parallels the ability of TeFg to accept F~ ions to form TeF 7~ and TeFg 2 - adducts [131]. The ISEELS spectra of IF^ have recently been measured in order to investigate the involvement of iodine 4f orbitals in the virtual valence orbitals of this molecule [136]. In order to have a better understanding of electronic excitations in molecules with low-lying unoccupied f orbitals further experimental and theoretical studies of other molecules containing heavy atoms would be of considerable interest There is a dramatic intensity decrease in the above-edge shape resonances in the Se 3p spectrum of SeFg compared with those in the S 2p spectrum of SFg. Parallel situations are also observed in the present work for the respective F Is spectra and also the S 2s and Se 3s spectra of SFg and SeFg. These observations strongly support the earlier suggestion by Addison et al. [96] that the potential barrier is shallower and/or that poorer penetration of the d-type outgoing channels occurs in SeFg than in SFg. In addition, comparison of (dipole-allowed) continuum shape resonances from the (ligand) F Is orHtal (Figure 4.1.8) reveals that there is a decrease in the intensities of these features relative to those of the discrete (a^g and t-^ ) virtual valence features in going from S to Se to Te in the MFg molecules.. These observations similarly suggest that the magnitude of the potential barrier decreases in going from SFg to SeFg to TeFg. Such behavior is consistent with the increasing size of the central atom and thus of the molecule. 4.1.2.6 VSEELS spectra of SFg, SeFg and TeF g An overview of the long range (5-60 eV) valence-shell excitation spectra of SFg, SeFg and TeFg, measured at medium resolution (0.063-0.135 eV FWHM), is shown in Figure 4.1.12. In addition, higher resolution (0.028-0.030 eV FWHM) spectra (Figures 4.1.13-15) of these hexafluorides have been obtained in order to give a more detailed view of the lower energy regions of the spectra. The VSEELS spectra of SFg reported in the present work are in good agreement with previous studies [44,108-114]. The VSEELS 90 (a) "'-5t i „ , l t 2 u 3 e g U 2 g 4 t , u 5 a , g 2e a 3 t ) u 4a,g SR 6 VALENCE e = o° E 0 = 3 0 0 0 eV A E = 0 . 0 6 3 eV I 2 5 13 2225 29 32 I I I I I I I I I i | i I 34 I CO UJ L U > 5 LL) rr (b) 1 0 1»2u 2 0 3 0 4 0 5 0 6 0 W, B 6t l u4e g 2t 2 B 5 t l u 6 o l g W I / / / , . l 8 - 5 p I I I I " SeF 6 VALENCE ^ = o° E 0 = 3 7 0 0 eV A E = 0 . 1 3 5 eV x 3 1 2 3 7915 20 23 2 4 2 5 2 6 27 I I I I I I I I I I I I I I 28 I 1 0 ( C ) / I t , , — 6 p 2 0 3 0 lt|g 7t|U 5Cg StggBtlu 70|g mm € ^ 4 0 5 0 6 0 TeF6 VALENCE 0 = o° E Q = 3 0 0 0 eV A E = 0 . 0 7 0 eV I 23 7 12 19 24 25 26 I I I I I I I I I 10 2 0 3 0 4 0 5 0 6 0 ENERGY LOSS (eV) Figure 4.1.12 Long range VSEELS spectra of (a) SF6; (b) SeF6; (c) TeF6-91 Figure 4.1.13 Detailed VSEELS spectrum of SF6-t GO Z LU I-LU > UJ S e F 6 VALENCE / 6 - 0 ° E0= 1500 eV A E = 0.028 eV »2u 6 t j u 4 e g 2t , 2 g IO II I3I5I7 ISIS 8 IO 12 14 16 ENERGY LOSS (eV) I8 Figure 4.1.14 Detailed VSEELS spectrum of SeF6. vo >-t CO -z. L L I 10 L L I > 5 LU OL. 1% % 5 6 , 3 * 2 , 6 ^ 7 0 ^ m i s? V TeF6 VALENCE # = 0 ° E 0 = 3 0 0 0 eV A E = 0.030 eV 2345689 11 14 1518 2123 24 I I I I III Mill I Hill II11 I I I 10 15 20 25 30 35 4 0 ENERGY LOSS (eV) Figure 4.1.15 Detailed VSEELS spectrum of TeFg. 94 spectra of SeFg and TeFg are the first reported valence-shell excitation spectra of these molecules. The hatched lines shown in Figures 4.1.12-15 are the vertical IPs of SFg [137,138], SeFg [139] and TeFg [139] as obtained by photoelectron spectroscopy (PES). The ordering of the valence-shell orbitals is that given by theoretical calculations [128,129]. Although earlier partial photoionization cross-section measurements of SFg by Gustafsson [140] suggested a different ordering of the 3e and It, orbitals, Dehmer et al. [141] concluded that the same valence ordering of SFg used as in the present work is in fact the most consistent considering all available experimental and theoretical evidence. A further very recent experimental (PES) and theoretical (MS -Xa) study of SFg and SeFg [142] is also supportive of this assignment by Dehmer et al. [141]. In the ISEELS spectra of MFg (M = S, Se and Te) discussed above, the presence of broad and often intense inner-shell excitations to the virtual valence orbitals can be clearly seen whereas Rydberg transitions are generally either not present or at best very weak. In the light of these observations, the VSEELS spectra of these hexafluorides have been interpreted as being dominated by dipole-allowed transitions from the occupied valence orbitals to the empty virtual valence orbitals with occasional superimposition of sharper structures caused by transitions to Rydberg orbitals. In addition, since formally dipole-forbidden transitions are observed in the ISEELS spectra, some of the peaks observed in the VSEELS spectra may also have contribution from transitions which are formally dipole-forbidden. Such transitions may be vibronically or quadrupole allowed. The energies, term values and possible assignments of features in the VSEELS spectra of SFg, SeFg and TeFg are shown in Tables 4.1.15-17 respectively. In spite of an appreciable number of experimental and theoretical [44,55,56,108-114,119,143] studies of the valence-shell excitation of SFg, there is no general agreement on the spectral assignment Early works [55,108-111] tended to assign most excitations to Rydberg transitions. However, as pointed out by Robin [56] such assignments are not possible because of the magnitudes of the term values and even if Table 4.1.15. Energies, term values and possible assignments for features in the VSEELS spectrum of SF6 Feature^ Energy(b) Term Value(d> (eV) Possible Assignment Loss (eV) l t i g lt2u.5ti u 3eg l t 2 g 4tlu 5ai g 2eg 3 t i u ltig l t 2 u , 5 t i a 3eg l t 2 g 4 t i u 5ai g 2 e g 3tln 1 6.494 ? 2 9.596 6.1 - - - - - - - 6ai gt - - - - - - -3 10.982 4.7 7.0 - • - - - - - - -4 11.287 4.4 6.7 - - - - -:s U-516 4.2 6.5 - - - - > - • . 6 11.681 4.0 6.3 - - - - - - » 6 t j u »6aig - - - - - -7 11.821 3.9 6.2 - - - - - - -8 12.037 3.7 6.0 - - - - - - - - - -9 12.266 3.4 5.7 - - - - - _ - - -10 12.380 3.3 „ 6.2 _ _ > 4it ^Mgt _ _ _ _ 1 1 12.686 3.0 5.9 - _ 12 13.143 2.6 - - - - - - - - - - - - - -13 13.220 2.5 - - - - - - - • , 4p - - - - - - -14 13.309 2.4 - - - - - - - J 1 - • - - -15 13.970 - - 4.6 - - - - - - - - • 16 14.021 - - 4.6 - - - - - - - - - - - -17 14.059 - - 4.5 - - - - - - - . 6ti i i - - - - -18 14.122 - - 4.5 - - - - - > w 1 u - - -19 14.211 - - 4.4 - - - - - - - - - - - -20 14.415 1.3 - 4.2 - - - - - - - - - - -21 14.707 - 3.5 - - •- - - - 4s - - - -22 15.546 - - - 4.3 - - - - - - I - - -23 15.864 - - - 3.9 - - - - - - >-6tiu - - - -24 16.055 - - 2.5 3.7 - - - - - - 4p I - - - -25 17.021 - 1.0 - - 5.9 - - - - 6s - 6ai gt - - -26 17.314 - - 1.3 2.5 - - - - - - 5p 4p - - - -27 19.8«=> - - - - 3.1 - - - - - - - 4s - - -28 20.9<c> -5.2 - - - 2.0 6.1 - - 2t 2 gt - - - 5s 6ai gt - -29 23.2<c> - -5.2 - - - 3.8 - - - 2 t 2 g - - - 6tiu - -30 24.6<c> - - - - - 2.4 - - - - - - 4p -31 25.8<«1 - - - - - 1.2 - - - Sp -32 28.3<c> - - - - -5.4 - - - - - - - 2 t 2 g - -33 35.8<c> - -17.8 - - -12.9 - 3.5 5.4 - 4eg - - 4eg - 6tiu 6ai g 34 45.6<c> - - - - - - - -4.4 - - - - - - 2t2g rpw) 15.7 17.0 18.6 19.8 22.9 27.0 39.3 41.2 (a) See Figures 4.1.12 and 4.1.13. (b) Estimated uncertainty is ±0.020 eV. (d) Valence IPs from PES measurement [137,138]. t Forbidden transition. (c) Estimated uncertainty is ±0.5 eV. Table 4.1.16. Energies, term values and possible assignments for features in the VSEELS spectrum of SeF6 Feature**) Energy(b> Term Value'* (eV) Possible Assignment Loss (eV) l t l g lt2u.6tlu 4eg 2 t 2 g 5tiu 6ai g l t l g lt2u.6tlu 4eg 2t 2 g 5Mu 6a l g 1 7.285 8.5 7ai gt 2 9.067 - 7.9 8.2 - - - - 7ai g 7a l gt - -3 11.019 - - - 7.7 - - - - - "I - -4 11.139 - - - 7.6 - - - - _ . 7ai g - -5 11.379 - - - 7.3 - - - - - . - -6 12.933 2.9 - - - - - 7tlu - - - - -7 13.442 2.4 3.6 - - - - 5p 5s - - - -8 14.072 - 2.9 - - 7.4 - - 7tiut - - 7ai g -9 14.603 1.2 2.4 2.7 - - - 6p 5pt 7tlu - - -1 0 15.173 0.6 - 2.1 - - - - 5p - - -1 1 15.483 - 1.5 - - - - - 6s - - - -12 15.623 - • - - 3.1 - - - - - - - -13 15.763 - - - 2.9 - - - - - - -14 15.873 - - - 2.8 - - - - - • 7tiu - -15 15.952 - 1.0 - 2.7 - - - 7s - - -16 16.044 - - - 2.7 - - - - - - -17 16.174 - - 1.1 2.5 - - - - 6p. - -18 16.634 - - 0.7 2.1 - - - - \ 7p \ 5p - -19 16.845 - - 0.5 1.9 - - - - J r J V - -20 18.2<c> - - - - 3.2 7.4 - - - - 5s 7ai g 21 19.7(c) - - - - 1.8 - - - 6s 22 20.6<c> - - - - 0.9 - - 7s -23 23.5<c> - -6.5 - - - 2.2 - 3t 2 g - - - 7ti u/5p 24 28.0<c> - - - - -6.5 - - - - - 3t 2g -25 29.5<c> - - - - - - (4t) u -* 7a,g)(e) 26 32.4<c> - - - - - -6.7 - - - - - 3 t 2 g 27 37.3(c) - -20.3 - - - - 5 eg -28 44.8<c> - - - - -23.3 - - - - - 5eg -15.7 17.0 17.3 18.7 21.5 25.7 (a) See Figures 4.1.12 and 4.1.14. (c) Estimated uncertainty is ±0.5 eV. t Forbidden transition. (b) Estimated uncertainty is ±0.020 eV. (d) Valence IPs from PES measurement [139]. (e) Based on 4tj u TP of -38 eV [65]. Table 4.1.17. Energies, term values and possible assignments for features in the VSEELS spectrum of TeF6 Feature'3) EnergyC' Term Value<d> (eV) Possible Assignment Loss (eV) 1tlg 1>2u.7tl u 56g 3t2g 611 u 7 a i g 1tlg 1t2u.7ti u 5e g 3<2g 6 l l u 7 3 1 g 1 9.266 _ 7.7 7.9 _ _ _ 8 3 1 g 8 a 1 g T _ _ 2 12.073 3.9 - - - - - 8tlu - - - - -3 12.738 3.3 - - - 7.4 - 6st - - - 8 a 1 g -4 13.364 2.6 3.6 3.8 - - - 6p 6S - - -5 13.989 - - - 4.2 - - - - - - -6 14.772 1.2 2.2 - - - - 7P 6pt - - - -7 15.202 - - 2.0 - - - - " 1 * 6D - - -8 15.358 - 1.6 1.8 - - - - 7s J - - -9 16.141 - 0.9 1.1 2.1 3.9 - - 8s 7p 6p Btlut -10 16.415 - 0.6 0.8 - - - - 9s 8p - - -11 16.923 - - - 1.3 - - - - - - -12 17.275 - - - - 2.7 7.2 - - - - 6s 83! gt 13 17.471 - - - 0.7 - - - - - 8p - -14 18.683 - - - - 1.3 - - - - - 7s -15 20.834 -4.8 - - - - 3.7 - - -16 21.147 -5.1 - - - - 3.4 » 9tiu - - - >8tl u 17 21.382 -5.4 - - - - 3.1 - - -18 21.695 -5.7 - - - - 2.8 - - -19 22.047 - - -4.8 - - 2.5 - - "I -20 22.321 - - -5.1 - - 2.2 - - •9tlu - y 6p 21 22.555 - - -5.3 - - 2.0 - " J - - -22 23.064 - - - -4.9 - 1.4 - • - - 1 > 911 II - 7p 23 23.494 - - - -5.3 - 1.0 - - - J *" 1 u - 8p 24 26.4<c> -10.4 - -9.2 -8.2 - - 1a2ut - 1a2ut 1a2li - -25 31.7<c> - - - - - -7.2 - - - - - 1 a 2 u 26 36.6<c> -20.6 - -19.3 -18.3 - - 212 u - 212 u 2I2U - -IP«0 16.0 17.0 17.2 18.2 20.0 24.5 (a) See Figures 4.1.12 and 4.1.15. (b) Estimated uncertainty is ±0.020 eV. (c) Estimated uncertainty is ±0.5 eV. (d) Valence IPs from PES measurement [139]. t Forbidden transition. 98 this was possible, several dipole-forbidden excitations would then have extremely large intensities. Therefore, the most prevailing current view is to attribute most excitations to transitions terminating at the virtual valence orbitals, consistent with the interpretations of the core excitation spectra (see refs. [44,56,112-114,119] and the present work). However, Robin [56] considered it surprising to see the 3d orbitals placed in the valence manifold and therefore he rejected assignment of "d-type" 2u and 4e shape resonances to g continuum structures observed in the core excitation as well as the valence excitation spectra of SFg. Instead Robin [56] attributed the continuum structures to shake-up excitations. Possible complexity of these continuum features is further discussed in recent studies of the S Is core photoionization of SFg which conclude that the continuum feature analogous to that currently assigned as the 4e shape resonance is due to double excitation in S Is [144]. In addition, Ferret! et al. [144] consider that similar contributions from doubly excited states may occur in the S 2p spectrum [145] of SFg. However, the observed intensities of these continuum features in the inner-shell excitation spectra of SFg, in particular in the S 2p and F Is spectra (Figure 4.1.1), seem anomalously high for shake-up or double-excitation transitions. Furthermore, evidence from experimental and theoretical studies of partial photoionization cross sections for valence electrons in SFg [140,141,146] and also e-SFg scattering experiments [147] clearly supports the presence of 2u and 4e shape resonances. Therefore, a spectral analysis involving •<-g g 2u and 4e shape resonances in the core and also the valence ionization continua seems 2g g to be the most reasonable interpretation. As demonstrated in many previous studies from this laboratory, for example see references [64-66,148], term values for Rydberg transitions are essentially transferable between inner-shell and valence-shell excitation (see also Chapter 2 section 2.2). Therefore, term values of Rydberg states weakly observed in the S 2p spectrum of SFg (Table 4.1.2) have been used in the present work to predict the location of the Rydberg transitions in the VSEELS spectrum of SFg (Figure 4.1.12). Following this approach, 4s and/or 4p Rydberg excitations from the seven least tightly bound valence orbitals have been identified with term values in the range of 3.1-3.5 eV and 2.4-2.5 eV respectively 99 (see Table 4.1.15). Higher members of these ns and np Rydberg series are located by fitting to the well known Rydberg equation (Eq.2.1). In a higher resolution optical spectrum measured by Sasanuma et al. [112] vibrational progressions were observed corresponding to the 4t^ u -»ns, (n=6 and 7) states. Some evidence for these states can be seen on the lower energy side of the 5a^ •* 6t^ u band (feature 29) in the present VSEELS spectrum (Figure 4.1.13). In addition, the window resonances located in the range 24.5-26.5 eV (features 30-32), first noted by Codling [108], are also clearly visible in the present spectrum. In contrast to the situation for Rydberg excitations, term values for transitions to the. virtual valence orbitals are generally not transferable between ISEELS and VSEELS because relaxation as well as correlation effects are expected to be significantly different for core excitation and valence excitations. The present assignment of virtual valence bands in the VSEELS spectrum of SFg is made with the assumption that the term values of transitions terminating at the same virtual valence orbital are essentially constant irrespective of the particular initial valence orbital as discussed in Chapter 2 section 2.2. Such an approach has been demonstrated to provide a generally satisfactory assignment for the VSEELS spectra in previous studies (for example see refs. [54,66,148]). With the above assumption, features observed in the VSEELS spectrum of SFg can be tentatively assigned to transitions terminated at various virtual valence orbitals. With such an assignment of the experimental peaks, the term values (vertical) are almost constant i.e. 5.4 to 6.5 eV for the 6a l g bands, 3.5 to 4.4 eV for the 6 t l u bands, -5.2 to -5.4 eV for the 2L, bands and -12.9 to -17.8 eV for the 4e bands (see Table 4.1.18). Such ^g g assignments also appear to be reasonable since the intensities of the corresponding features reflect expectations based on considerations of dipole selection rules. The VSEELS spectra of SeFg and TeFg can be interpreted in a similar manner to that of SFg. The possible assignments of features observed in the VSEELS spectra of SeFg and TeFg are presented in Tables 4.1.16 and. 4.1.17 respectively. Vertical term values of excitations to various virtual valence as well as to first members of the ns and np Table 4.1.18 Summary of term values for ISEELS and VSEELS spectra of S F 6 100 Initial Orbital 6 a 1 g 6 t l u Final Orbital 2 t 2 g 4 e g 4s 4p F 1s 6.77 0.89 -4.2 -17.6 2.49 Core S 2s 7.8 4.2 -2.1 -14 3.0 S 2p 3 / 2 8.12 4.48 -3.041 r-15.2 2.96 2.20 S 2p 1 / 2 8.22 4.84 -2.94J 2.96 2.16 3 t i u 5.4 — -4.4 2e g -- 3.5 ~ -5 a i g 6.1 3.8 -- -- 2.4 Valence 4tm 5.9 -- -5.4 -12.9 3.1 1t2g -- 4.3 - - 2.5 3e g 6.2 4.4 - - 2.5 5tiu.1t2u 6.5 -- -5.2 -17.8 3.5 1 t i g 6.1 4.2 -5.2 - 3.3 2.5 Table 4.1.19 Summary of term values for ISEELS and VSEELS spectra of SeF 6 Initial Orbital Final Orbital 7 a i g 7 t i u 3 t 2 g 5 e g 5s 5p F 1s 8.4 2.4 -5.4 -22 Se 3s 11.0 5.4 --Core Se 3p3/2 10.4 4.7 -3.5"] , -16.6 Se 3p 1 / 2 10.4 2.2 -3.6 J Se 3d 5/2 9.1 4.8 " ) • -17 3.4 Se 3d3/2 9.1 4.9 - J — 3.4 6 a i g 7.4 2.2 -6.7 -- 2.2 5 t i u 7.4 - -6.5 -23.3 3.2 Valence 2t2g 7.7 2.7 - - 2.1 4 e g 8.2 2.7 -- -- 2.1 ' 6 t i u . 1 t 2 u 7.9 2.9 -6.5 -20.3 3.6 2.4 1 t 1 g 8.5 2.9 2.4 101 Table 4.1.20 Summary of term values for ISEELS and VSEELS spectra of TeF 6 Initial Orbital Final Orbital 8a-\ g 8ti u 9t-j u 1a2u 2 t 2 u 6s 6p C o r e Valence F 1s 8.3 2.9 -3.7 -10.7 -18.2 2.9 Te 3 p 3 / 2 8 3 -4 -12 -16 Te 3 d 5 / 2 8.2 3.0 -2.1 -10.2 -18.2 Te 3 d 3 / 2 8.3 3.7 -2.7 -10.2 -16.8 Te 4s 8.3 2.4 Te 4 p 3 / 2 6.7 3.6 Te 4 d 5 / 2 4.8 3.0 ' L 1 -22 2.7 Te 4 d 3 / 2 5.0 3.1 \ " 3 J 2.8 7 a i g 7.2 3.7 -7.2 2.5 6tm 7.4 3.9 2.7 3t2g 4.2 -4.9 -8.2 -18.3 2.1 5 e g 7.9 3.8 -4.8 -9.2 -19.3 2.0 7ti u.1t2u 7.7 3.6 2.2 1 t i g 3.9 -4.8 -10.4 -20.6 3.3 2.6 102 Rydberg orbitals for these two molecules are summarized in Tables 4.1.19 and 4.1.20 respectively. It should be noted that as far as bound excitations are considered (i.e. positive term values), such assignments give a similar excitation pattern in SFg, SeFg and TeFg if the corresponding relevant valence orbitals (Tables 4.1.15-17) are concerned, (for example, excitations to (6a l g, 6t l u, 4p, 5p), (7a g^, 7t^u> 5p, 6p, 7p) and (8a^ g, 8t u^> 6p, 7p, 8p) orbitals are observed from the 3e , 4e and 5e„ orbitals of SFC, SeF, and TeF, g g g . 6* 6 6 respectively). Exceptions to this pattern are generally dipole-forbidden transitions and in some cases these may be obscured by other much stronger dipole-allowed transitions. For example, the dipole-forbidden l t ^ -» 8a-^ transition is not apparent in TeFg whereas the corresponding lt^ -» 6a^ transition in SFg and lt^ g •* 7a l g transition in SeFg are clearly visible as small shoulders on the lower energy sides of the 0-^, 5t^u)-» 6a^ and (lt^, 6t u^)-* 7a^g bands respectively. The fact that the energy difference between the first two IPs decreases from a value of 1.3 eV in SFg and SeFg (Tables 4.1.15-16) to 1.0 eV in TeFg (Table 4.1.17) supports the view that the weak lt^ -•' 8a^ band may be located under the leading edge of the intense (1^. ^t^) •+ 8a^g bands in TeFg. Considering now transitions to the continuum shape resonances (i.e. virtual valence orbitals with negative term values), the excitation patterns from the various valence orbitals are again very similar in SFg and SeFg because both molecules are expected to exhibit d-type (t, and e ) shape resonances in their ionization continua by excitation from orbitals of "u" type symmetry. In contrast interpretation of the ISEELS spectra of TeFg suggested that the continuum shape resonances (9t u^, la2u and 2t2u) in TeFg are "f-type", and therefore, transitions to these shape resonances are dipole-allowed from valence orbitals of g symmetry only (i.e. the It, , 5e 3t, and 7a, orbitals). It should ig g g^ -ig be noted that features observed above 30 eV energy loss in the VSEELS spectra of SeFg and TeFg may also have contributions from excitations of the (essentially F 2s) inner-valence orbitals (i.e. 3e^, 4t^u and 5a^ in SeFg and 4eg, 5t u^ and 6a^ orbitals in TeFg) similar to excitations of the F 2s orbitals (2e^, 3t u^ and 4a-^ ) observed in SFg. Since no experimental values of these F 2s IPs of SeFg and TeFg have been reported, all features observed in these energy regions are tentatively assigned to shape 103 resonances associated with excitation from the outer valence orbitals of SeF, and TeF,. 6 6 However, since feature 25 in the VSEELS spectrum of SeFg (Figure 4.1.12b) cannot be fitted into any such scheme of shape resonances and in view of its relatively sharper profile, it may correspond to the 4t^u-» 7a^ transition because the 4t u^ IP is estimated to be at an energy of ~38 eV [128] i.e. ~8.5 eV above feature 25. 4.1.3 Conclusion In contrast to the ISEELS spectra, the VSEELS spectra (Figure 4.1.12) of SFg, SeFg and TeFg are generally quite similar in spectral distribution. In particular each VSEELS spectrum shows an intense broad band at low energy, which is followed by a number of sharper structures corresponding to Rydberg bands as well as additional somewhat broader valence bands. However it can be seen that Rydberg transitions are of somewhat greater relative intensity in the VSEELS spectra than in the ISEELS spectra. This can be understood because the delocalized valence orbitals are essentially on the periphery of the potential barrier whereas the highly localized core orbitals are enclosed by the barrier. The higher energy regions of the VSEELS spectrum are dominated by conunuum shape resonances beginning with an intense broad band in the range 22-23 eV. It is noteworthy that the intensities of these conunuum shape resonances drop appreciably from SFg to SeFg to TeFg in both the VSEELS and ISEELS spectra. In contrast the intensity of the lt- g^ np Rydberg transitions in the VSEELS spectra shown on Figure 4.1.12 increases from SFg to SeFg to TeFg. These observations as well as the findings in the present ISEELS study all lend support to the view that a weakening of the potential barrier effects occurs as the atomic number of the central atom in the MFg molecule increases through the series S, Se and Te. Such behavior may be understood in terms of Coulombic potential barrier model because when the size of the central atom increases, the "Fg ligand cage" size also increases and consequently the effective Coulombic potential barrier associated with the electronegative "Fg ligand cage" diminishes in magnitude. 104 It is of particular significance that the ISEELS spectra of TeFg clearly indicate 4f orbital participation in the unbound molecular orbital manifold. This together with the S 3d and Se 4d participation in the ISEELS continuum structures of SFg and SeFg respectively clearly supports the existence of molecular orbital type continuum resonances trapped by a potential barrier. In addition, the participation of low lying f-type virtual orbitals as demonstrated in TeFg will have important implications for the interpretation of X-ray absorption near edge structure (XANES) experiments on species containing high Z elements. 105 INVESTIGATION OF POTENTIAL BARRIER EFFECTS IN "CAGE" MOLECULES : II. INNER-SHELL AND VALENCE-SHELL ELECTRONIC EXCITATION OF ClFj 4.II.1 Introduction In the preceding section the potential (Coulombic) barrier effects as reflected in the inner-shell and valence-shell electron energy loss spectra of the SFg, SeFg and TeFg series of molecules have been investigated. In particular, it was found that as the size of the MFg cage molecule increases in going from S to Te, there is apparently a weakening of the potential (Coulombic) barrier as indicated by the reduction in intensity of the continuum structures (shape resonances) and an increase in intensity of Rydberg transitions through the series in both the ISEELS and VSEELS spectra. In a further attempt to investigate possible Coulombic barriers, this section focus attention on the core excitation spectra of a species, namely CIF^, in which the central atom (CI) has an electronegative ligand cage on only one side of the molecule. CIF^ is a T-shape molecule (Figure 4.II.1), as visualized either by using the valence-shell electron pairs repulsion (VSEPR) model [149] or by assuming sp d hybridization of the valence electrons surrounding the the chlorine central atom. In the resulting trigonal bipyramidal arrangement of electron pairs, two fluorine atoms occupy the axial positions, while the remaining fluorine atom and the two lone pairs of electrons occupy the three equatorial positions. Therefore, CIF^ has a partly-open and partly-caged structure with respect to the CI atom. This is in marked contrast to the situations in HCl (essentially no cage) and SFg (essentially a full octahedral cage). In the direction along the Cl -F interatomic axes the excited electron should encounter a potential barrier similar to that in SFg, whereas in the other directions the excited electron should experience essentially no potential barrier. Thus CIF^, and. related species with partial cages such as BrF^ [150] and IF^ [136] should provide unique tests for the existence of actual Coulombic barrier effects. In particular it is of great interest to study whether the inner-shell as well as the valence-shell electronic excitations of C1F, are dominated by Rydberg-type and direct ionization continuum A X Figure 4.11.1 The structure of the CIF3 molecule. o as 107 transitions as in the case of HCl [151], or whether the spectra are dominated by intense shape resonances as in the case of SFg [68]. Given the partly-caged nature of Cl in ClFj, the spectra might be expected to exhibit both types of behaviour if Coulombic barrier effects are indeed occurring. In the present study, the ISEELS spectra of C1F3 including the Cl 2p, 2s and F Is excitation regions as well as the VSEELS spectrum of CIF^ have been measured. The present work reported the first studies of the inner-shell and the valence-shell electronic excitation spectra of CIF^. There are, however, several theoretical studies on the electronic structure of CIF^ [152-156] as well as a report of the He(I) photoelectron spectrum of CIF^ [157]. 4.II.2 Results and Discussion 4.II.2.1 Electronic Structure of C1F 3 The T-shape CIF^ molecule is of C 2 y symmetry (see Figure 4.II.1). The ground-state electron configuration together with the normally unoccupied virtual valence orbitals may be written as follows [152-157]: Core orbitals ( l a p 2 ( 2 a / ( l b / ( 3 a / ( 4 a / ( 2 b / ( 5 a / ( l b / Cl Is F ls(eq) F ls(ax) Cl 2s Cl 2p Valence orbitals ( 6 a / ( 7 a / ( 3 b / ( 8 a / ( 9 a / ( 2 b / ( 4 b / ( 5 b / ( 3 b / ( l a / (10a/ ( 6 b / (11a/ ( 4 b / Unoccupied valence (virtual^ orbitals (12a/ ( 7 b / - minimum basis set or (12a/ ( 7 b / (5b2)° ( 2 a / ( 8 b / (13a/ (14a/ - extended basis set (including Cl 3d orbitals) 108 An extended basis set (with the inclusion of CI 3d) Gaussian 76 calculation [360] using the equilibrium geometry [158] gives the energies and characters of the valence (occupied) and virtual valence (unoccupied) orbitals of CIF^ as shown in Table 4.II.1. In particular, the calculations indicate that the 12a^ virtual valence orbital has a localized character of a (Cl-F ) because it is mainly constructed from the CI (3pz> 3s) and F (2p ) AO's. The 7b, orbital is constructed from Cl(3p ) and F (2p. 2s) AO's and, eq z i x ax ' x therefore, has largely a (F -Cl-F ) character. The characters of the remaining virtual 3.X clX valence orbitals are as follows: 5b.(d ), 2a~(d ), 8b, ( d ) , 13a, (d 2) and 14a, (d 2_ ,). yz, j\.y x Ax. J. L X. A MM y The orientations of the five d orbitals relative to the CIF^ molecular framework, are also indicated on Figure 4.II.1. It should be noted that lobes of the higher energy d 2_ 2 and x -y d z 2 orbitals (Table 4.II.1) point towards the F ligands whereas lobes of the lower energy d , d and d orbitals point between the ligands. AZ. * * j J ^ Within an independent particle, frozen orbital, picture the ISEELS and VSEELS spectra reported in the present work are conveniently discussed in term of transitions to a common manifold of virtual valence orbitals (shape resonances) as well as Rydberg orbitals. 4.II.2.2 Assignment of Inner-Shell Spectral Features 4.11.2.2.1 CI 2p Spectrum The low-resolution (0.14 eV FWHM) long-range CI 2p, 2s ISEELS spectrum of ClFj between 195 and 300 eV is shown in Figure 4.II.2a. An energy loss spectrum of CI 2p excitation between 195 and 220 eV, obtained with a higher resolution of 0.063 eV FWHM is also presented in Figure 4.II.2b. The CI 2p^/2 ionization potential (213.02 eV) has been determined by XPS measurement [118]. The CI 2 p ^ IP w a s estimated according to an expected CI spin-orbit splitting of 1.60 eV as given in Ref. [159]. The absolute energies, term values and possible assignments of features corresponding to the CI 2p excitation are showed in Table 4.II.2. Table 4.II.1 Calculated Energies and L o c a l Characters of the Valence molecular orbitals of C l F s ^ M O E n e r g y C h a r a c t e r M O E n e r g y C h a r a c t e r M O E n e r g y C h a r a c t e r (eV) (eV) ( e V ) 6 a j - 4 4 . 5 9 F e q 2s 5 b x - 1 5 . 9 5 n ( F e q 2 p x ) 1 2 a t 5 . 4 9 0 * ( C 1 - F e q ) 7 a , - 4 0 . 2 2 F a x 2s 3 b 2 - 1 5 . 0 0 7 b ! 9 . 1 3 t J * ( F a x - C l - F a x ) 3 b ! - 4 0 . 1 2 F a x 2s 1*2 - 1 3 . 6 9 n ( F a x 2 p y ) 5 b 2 1 9 . 2 7 CI 3 d y z - 2 8 . 4 5 CI 3s 6 b i - 1 3 . 1 2 n ( F a x 2 p z ) 2 a 2 1 9 . 9 8 CI 3 d x y 4 b i - 1 9 . 2 0 C ( F a x - C l - F a x ) 1 0 a ! - 1 2 . 5 1 n ( F a x 2 p z ) 8 b i 2 0 . 9 1 CI 3 d x z 2 b 2 - 1 8 . 8 9 7 I ( C 1 - F e q ) 4 b 2 - 1 0 . 7 3 7 t * ( C l - F e q ) 1 3 a t 2 3 . 4 7 Q 3 d z 2 9 a ! - 1 8 . 8 0 0 ( C 1 - F e q ) l l a i - 9 . 4 8 C ( F a x - C l - F a x ) 1 4 a i 3 3 . 2 6 CI 3 d x 2 - y 2 (a) G A U S S I A N 76 calculation [360] using minimum basis set + CI 3d. 110 10 CO LU % 0 LU > 5 LU c r 10 0 Res Res A CIF 3 Cl 2p, 2s c9 = o° E0= 3700 eV AE=O.I4eV 12 3 5 21 2223 II I I JU, II 24 _ l _ JJ L s2p 3/2.1/2 edges 200 220 0 = 0 ° E 0 =3000eV AE=0.063 eV 240 2s edge _J L 260 280 300 CIF3 Cl 2p 45 6 9 II 16 20 21 2P3/2.1/2 e d ^ e s 200 210 ENERGY LOSS (eV) 220 Figure 4.11.2 (a) Low-resolution Cl 2p, 2s; (b) high-resolution Cl 2p ISEELS spectra of CIF3. Table 4.II.2 Energies, term values and possible assignment of features i n the C l 2p spectrum of C1F 3 F e a t u r e ^ ) Energy Loss( b ) Term Value (eV) Poss ib le Ass ignment (eV) 2 P 3 / 2 2 P l / 2 2 P 3 / 2 2 p i / 2 1 2 0 1 . 7 5 11 .27 1 2 a i 2 2 0 3 . 2 4 — 11 .38 — 1 2 a r 3 2 0 6 . 2 5 7 .3( e ) 7 b j 7 b ! • 4 2 0 8 . 9 2 4 . 1 0 — y 4 s — 5 2 0 9 . 1 9 3 . 8 3 — — 6 2 1 0 . 2 5 2 .77 — 4 p i — 7 2 1 0 . 5 4 — 4 . 0 8 — ~r*4s 8 2 1 0 . 7 3 — 3 . 8 9 — J 9 2 1 1 . 0 0 2 . 0 2 — 4p2 — 1 0 2 1 1 . 3 4 1.68 — 5 s / 3 d — 1 1 2 1 1 . 6 9 1.33 — 5 p i — 1 2 2 1 1 . 9 1 1.11 2 .71 5 p 2 4 p i 1 3 2 1 2 . 1 0 0 . 9 2 — 6 s — 1 4 2 1 2 . 3 5 0 . 6 7 — 6 p 2 — 1 5 2 1 2 . 5 4 — 2 . 0 8 — 4 p 2 1 6 2 1 2 . 8 4 — 1.78 — 5 s / 3 d C l 2 p 3 / 2 W (213.02)< c> 0 — 0 0 — 17 2 1 3 . 3 3 — 1.29 — ' 5 p i 1 8 2 1 3 . 5 2 — 1.10 — 5 p 2 / 6 s 1 9 2 1 3 . 8 2 — 0 . 7 9 — 6 p 2 2 0 214.4<c» - 1 . 4 — 1 3 a j ( d z 2 ) — C l 2 p w 2 IP (214.62)< c> — 0 — 0 0 2 1 215.9<d> — - 1 . 3 — 1 3 a i ( d z 2 ) 2 2 222.4(d > - 9 . 3 — 1 4 a i ( d x 2 _ v 2 ) — ' 2 3 2 2 4 . 3 ( d ) — - 9 . 6 — 1 4 a i ( d x 2 : 2 4 246.0(d) - 3 2 . 2 ( e ) E X E L F S (a) See Figure 4 .H.2. (b) Estimated uncertainty is ±0.08 e V . (c) C l 2p3/2 IP from X P S measurements [118]. C l 2 p i / 2 IP is estimated by assuming C l 2 p 3 / 2 - 2 p i / 2 spin-orbit splitting to be 1.60 eV [159]. (d) Estimated uncertainty is ±0.4 e V . (e) Broad structure, term values with respected to average C l 2p IP (213.82 eV) . 112 Features 1 and 2 are attributed to the spin-orbit components of the excitation of the CI 2p electron into the lowest unoccupied virtual valence orbital (i.e. CI 2p^/21/2 transitions) since they have an energy separation of ~1.5 eV which is close to the average spin-orbit splitting of 1.60 eV [159]. In addition, the intensity ratio of these two features is also close to the statistical ratio of 2:1 expected for spin-orbit split partners. The large term value of feature 3 (7.3 eV with respect to the CI average 2p IP) and its broad nature indicates that it is also due to a transition to a virtual valence type orbital. It is, therefore, logically assigned as the CI 2p-» 7b^ transition. Owing to the broad nature of feature 3 the individual spin-orbit components of the CI 2p-> 7b^ transition are not resolved. It should be noted that the observed decrease in intensity of the CI 2p-» 7b^ transition relative to that of the CI 2p -*\2a.^ transition is not unexpected given that the 12a^ orbital has both CI 3p and CI 3s characters whereas the 7b^ orbital has mainly CI 3p character (see above section 4.II.2.1). It has been demonstrated in previous studies [16,65,66] that the intensities of transitions from the highly localized inner-shell (atomic like) orbitals to virtual valence orbitals are generally in keeping with expectations based on atomic dipole selection-rules Thus the p-» p type nature of the CI 2p-» 7b^ transition renders it unfavourable. Conversely, the CI 2p-» 12a^ transition is favorable (i.e. p -vs). The sharp features 4-19 observed between 208-214 eV are tentatively assigned as Rydberg series converging to the CI ^V^/i and 2p^2 IPs. In particular, an ns series (5 =2.13, n=4,_ 5, 6), two np series (6 =1.77, n = 4, 5; s P i 5 =1.42, n = 4, 5, 6) and an nd (5 . = 0.20, n = 3) series are observed (see Table 4.II.2). p 2 a Two intense resonances, (features (20+21) and (22+23)) are observed above the CI 2p IPs. The spacing of features 20 and 21 and features 22 and 23 (see Table 4.II.2) suggest that they are spin-orbit partners. The dashed line below these structures is the estimated contribution of the CI 2p ionization continuum (this is essentially equal to the continuum observed for the CI 2p spectrum of HCl (see Figure 4.II.7)). It should be noted that resonances corresponding to features 20-23 are also observed at similar term values on the CI 2s and F Is continua (see Tables 4.II.3-4 and Figure 4.II.6). Therefore, these features are ascribed as resonance enhanced transitions to the inner-well valence 113 orbitals (shape resonances) trapped by the the potential barrier of the F ligands. The dominating outgoing channels for these shape resonances are believed to be d-type (i.e. A = 2) since they are most intense in the Cl 2p spectrum but much weaker in the Cl 2s spectrum (see Figure 4.II.6). Similar d-type resonances have been observed in the 2p spectra of other molecules containing third-row elements, for example see references [65,160,161]. Therefore, features 20 and 23 are attributed to the spin-orbit components of the 13a^(dz2). shape resonances according to their term values of -1.4 and -1.3 eV (Table 4.II.1) with respect to the appropriate IPs. Similarly, features 22 and 23 can be assigned as the spin-orbit components of the 14a^(dx2_y2) shape resonances. Under C 2 y symmetry, there are five possible d-type exit channels (5b2, 2a2> 8b^ , 13a^  and 14aj) as shown in the extended MO scheme (see section 4.II.2.1). All of these channels are dipole-allowed for Cl 2p excitation. However, apparently only two d shape resonances are observed on the Cl 2p ionization continuum and also in the Cl 2s and F Is continua (see Figure 4.II.6 and discussion in section 4.II.2.4 below). A consideration of the relative energies of all the virtual orbitals according to the Gaussian 76 calculationt strongly supports the assignment of features (1 + 2), 3, (20 + 21) and (22 + 23) to the 12alf 7bp 13a1(dz2) and 14a^(dx2_y2) final orbitals since the experimental energy spacings are comparable. Likewise, it can be seen that the 8b, (d ), 2a9(d ) and 5b~(d ) orbitals are predicted to be in a closely spaced group in the range 2 to 4 eV below the 13a^(dz2) orbital. The above assignments and the measured term values of features 21 and "23 (Table 4,11.2) therefore imply that the d , d and d group of orbitals are in fact located 1-3 eV below the r J xz xy yz ionization edge (i.e. they are bound). This is the region where Rydberg transitions are expected (Table 4.II.2) and contributions from the three lower energy d-type resonances are evidently small. The large intensity of the above edge resonances and the low intensity of the below edge resonances may reflect the fact that lobes of the d 2_ 2 and x —y d j orbitals point towards the F ligands whereas those of the d d and d orbitals z x z xy yz, point between the ligands (see Figure 4.II.1). On this basis, it might be expected that the t As is well known calculations such as Gaussian 76 give a reasonable guide to relative energies (and therefore spacings), even though the absolute magnitudes are much less reliable. 114 largest Coulombic barrier effects (and thus the most intense resonances) would be experienced by the 14a^(d x 2_ y 2) and 13a^(dz2) orbitals. An additional very broad and weak structure (feature 24) is observed at ~32 eV above the CI 2p direct ionization continua. It is tentatively assigned as an extended energy loss fine structure (EXELFS) caused by scattering of the outgoing ionized electron by neighbouring atoms, in analogy to the well known EXAFS type structures observed in X-ray absorption spectra. Such EXELFS structures have also been observed in the CI 2p continua of the chloromethanes [125] and simple freons [162], 4.II.2.2.2 CI 2s Spectrum A CI 2s spectrum (0.14 eV FWHM) between 264-344 eV is shown in the upper section of Figure 4.II.3. Improved visibility of the details of the spectrum is given by the background subtracted spectrum which is shown in the lower section of Figure 4.II.3. The CI 2s IP is taken from XPS measurements [118]. Energies, term values and possible assignments of all features observed in the CI 2s spectrum are summarized in Table 4.II.3. The structures in the CI 2s spectrum are lifetime broadened compared to those in the CI 2p spectrum (Figure 4.II.2) due to fast Coster-Kronig type autoionizing transitions from the 2s to the underlying 2p and/or valence-shell continua. Features 1 and 2 are readily assigned as transitions to the virtual valence 12a^ and 7b ^  orbitals according to their large term values, 10.9 and 8.7 eV respectively as shown in Table 4.II.3. It is interesting to note that both transitions to the 12a^ and 7b^ orbitals are intense and of similar intensity. These observations are consistent with the fact that both the 12a^ and 7b-^  orbitals carry significant CI 2p character (see section 4.II.2.2.1) and such s-> p type transitions are highly favorable based on pseudo-atomic type selection rules. Feature 3 is assigned as the CI 2s •+ 4p Rydberg transition since it has a term value of 2.1 eV. Similarly, features 4 and 5 are attributed to continuum shape resonances 13a^(dz2) and 14a^(dx2_y2) respectively because their term values are reasonably close to those of corresponding features observed in the CI 2p spectrum. As t co z. LU LU > LU cr 3 4 C I F 3 CI 2s e = o ° E0=3700eV AE = O.I4eV 2 8 0 3 0 0 3 2 0 3 4 0 ENERGY LOSS (eV) Figure 4.11.3 Detailed CI 2s I S E E L S spectrum of CIF3-, (upper trace) original, (low trace) background subtracted. Table 4.II.3 Energies, term values and possible assignment of features i n the C l 2s spectrum of CIF3 F e a t u r e d Energy L o s s W Term Value Poss ib le Ass ignment (eV) ( e V ) (final orbital) 1 2 7 3 . 2 1 0 . 9 1 2 a ! 2 2 7 5 . 3 8.7 7 b i 3 . 2 8 2 . 0 2.1 4 p C l 2s IP (284.04)< c> 0 °° 4 285<d> - 1 13a i (d z 2) , 5 295<d> - 1 1 1 4 a i ( d x 2 2) (a) See Figure 4.II.3. (b) Estimated uncertainty is ±0.2 e V . (c) C l 2s IP from X P S measurement [118]. (d) Estimated uncertainty is ±0.1 e V 117 mentioned in section 4.II.2.2.1, the decrease in intensity of the continuum shape resonances in the Cl 2s spectrum relative to those in Cl 2p spectrum is consistent with the d character in these shape resonances and the fact that s+ d type transitions are not favorable according to pseudo-atomic selection rule arguments. 4.II.2.2.3 F Is Spectrum Figure 4.II.4 shows the F Is ISEELS spectrum of CIF^ measured at a resolution of 0.28 eV FWHM. In CIF^ the fluorine atom located at the equatorial position is chemically different from the two equivalent fluorine atoms located at the axial positions (see Figure 4.II.1). As a result, there are two different F Is IPs, designated as the F ls(ax) and F ls(eq) IPs respectively. The positions of these two IPs as obtained by XPS measurements [118] are shown in Figure 4.II.4. Energies, term values and possible assignments of features in the F Is spectrum are shown in Table 4.II.4. The F Is spectrum of CIF^ is dominated by an intense broad peak (~3 eV FWHM) at 684.8 eV (feature 1). Based on symmetry considerations, all transitions from the F ls(ax) and F ls(eq) core orbitals to the 12a^ and 7b^ virtual valence orbitals are dipole-allowed. However, consideration of the local characters (Table 4.II.1) of these virtual valence orbitals (i.e. a (Cl-F ) for 12a, and a (F -Cl-F ) for 7b,), indicated ©C| X 3.X clX 1 that the F ls(ax) ->7b^  and F ls(eq)-> 12a ^  transitions would be expected to be the dominant transitions. Feature 1 has term values of 7.4 and 10.0 eV with respect to the F ls(ax) and F ls(eq) IPs respectively. These values are reasonably close to the term values for the transitions to the 7b^ and 12a^ orbitals observed in the Cl 2p and Cl 2s spectra (see Tables 4.II.2-3). As a result feature 1 is assigned as a superimposition of the F ls(ax)->7b^ and F ls(eq)-+ 12a ^  transitions. Furthermore, transfer of term values places the dipole-allowed but understandably (see above) weaker F ls(ax) •* 1 2 a a n d F ls(eq) ->7b^  transitions at the low and high energy edges of feature 1 respectively. Consideration of the broad band shape of feature 1 suggests that these two transitions may indeed be making some contribution to this intense feature. Feature 2 is assigned as >-(f) t= |0 JJ > LU t F Is edge Ax Eq 2 3 4 CIF Fls 3 E 0 =3000eV AE=0.28eV i i 680 690 700 710 720 730 ENERGY LOSS (eV) 740 750 Figure 4.11.4 F 1s ISEELS spectrum of CIF3. 119 Table 4.II.4 Energies, term values and possible assignment of features i n F Is spectrum of CIF3 F e a t u r e d Energy L o s s W Term Value (eV) Poss ib le A s s i g n m e n t ( e V ) F ls(ax) F ls(eq) F ls(ax) F ls(eq) 1 6 8 4 . 8 7 . 4 1 0 . 0 7 b i 1 2 a ! 2 6 8 9 . 7 2 .5 — 4 p — F ls(ax) IP (692.22)< c> 0 — ~ — 3 6 9 4 . 0 -0.9<e> 1 3 a i ( d z2) F ls(eq) IP (694.76)< c> — 0 — °° 4 706<d> - 1 3 ( e ) 1 4 a , ( d T2 v 2 ) (a) See Figure 4.U.4. (b) Estimated uncertainty is ±0.2 eV. (c) F ls(ax) and F ls(eq) IPs from X P S measurements [118]. (d) Estim?;ed uncertainty is ±1 eV. (e) Broad structure, term values with respected to average F Is IP (693.49 eV) . 120 the F ls(ax) •* 4p Rydberg transition according to the term value of 2.5 eV with respect to the F ls(ax) IP. Two broad, discrete structures (features 3 and 4) are observed on the F Is continua at similar term values to corresponding features on the Cl 2p and Cl 2s continua. Therefore, they are likewise attributed to the 13a^(dz2) and 14a^(dx2_y2) shape resonances. The number of possible transitions together with the lifetime broadening expected for an excited state with an F Is hole precludes any identification of the separate series of transitions leading to the respective ionization edges for axial and equatorial F atoms. 4,11.2.3 Valence-Shell Spectrum of C1F 3 A long range medium resolution (0.055 eV FWHM) VSEELS spectrum of C1F 3 is shown in Figure 4.II.5b. A high resolution spectrum at 0.0030 eV FWHM, showing more detail of the 4-16 eV energy loss region, is shown in Figure 4.II.5a. The hatched lines in Figure 4.II.5 indicate the locations of the valence IPs as determined by He(I) photoelectron spectroscopy [157], assigned with the aid of a DVXoM calculation [155]. The energies, term values ' and proposed assignments of the spectral features are summarized in Table 4.II.5. Since CIF^ is a low symmetry (C 2 v) molecule and the valence IPs are. closely spaced, there are a large number of possible dipole-allowed valence transitions. Therefore, the VSEELS spectrum of CIF^ reported in the present work have been tentatively assigned using two assumptions as presented in Chapter 2 section 2.2. Firstly, it has been assumed that the term values for Rydberg transitions are transferable between the ISEELS and VSEELS spectra. Secondly, in order to locate the valence -»virtual valence transitions and valence •+ Rydberg transitions, it is assumed that the term values corresponding to a particular virtual valence or Rydberg orbital are largely independent of the originating valence (normally occupied) orbital. The expected positions of the various transitions with respect to the valence-shell IPs are indicated by the manifolds on Figure 4.II.5b. Table 4.II.6 shows the ISEELS term values and also those found for the VSEELS spectra on l ( a ) 12a, 1__ 4s 4 ^ 10 I 4s 7 ^L I " I I | 4 b 2 5s 6s 7s K l 5s 11 a, 5Pi 12a, ; 4s 7b, 6b, # = 0° E0=3000 eV AE = 0.030eV 4s 5s 6s l a 2 10a, I III II 12 14 16 I2a, l _ 7b, _1_ I I I 5 6? I ;4b, CIF 3 VALENCE 0 # = 0< E0=3000 eV AE = 0.055 eV I2a, 7b, —r 4s I TT S 67 3b, 10 15 20 25 ENERGY LOSS (eV) 30 Figure 4.11.5 (a) Detailed; (b) long range VSEELS spectra of CIF3. Table 4.II.5 Energies, term values and possible assignments of features i n the V S E E L S spectrum of CIF3 F e a t u r e ^ Energy Loss( b ) Term Value ( e V ) Possible Assignment (eV) 4 b 2 l l a i 6 b ! l a 2 l O a i 3 b 2 5 b i 4b 1 2 b 2 4 b 2 l l a j 6 b t l a 2 l O a j 3 b 2 5 b i 4 b i 2 b 2 1 5 .881 6.77 — — — — — — — 1 2 a i — — — — — — — 2 9 . 0 1 4 3 . 6 4 — — 6.35 — — — — "1 — — 1 2 a t — — — — 3 9 . 1 2 2 3 . 5 4 — — — 6 . 9 5 — — — M s — — — 1 2 a i — — — 4 9 . 2 3 0 3 . 4 2 J 5 9 . 6 8 6 2 . 9 6 4 . 3 8 4 p i 7 b i 6 1 0 . 2 8 5 — 3.79 4 . 5 5 : — — 6 . 9 9 — — — 4 s 7 b i — • — 1 2 a x — — 7 11 .017 1.63 — 3.81 4 . 3 4 — — — — 5 s — 4 s 7 b ! — — — — 8 11 .317 1.33 2.75 5 p i 4 p i 9 11 .548 1.10 — — 3.81 — — — — 5P2 — — ' 4 s — — — — 1 0 11 .747 0 . 9 0 — — — 4 . 3 2 — — — 6 s — — — 7 b j t — — — 1 1 12 .144 0.51 1.93 2 . 6 9 — — — 6 . 8 6 — 7 s 4P2 4 p i — — — 1 2 a i — 1 2 1 2 . 3 6 0 — 1.71 — — 3.71 4 .91 ' — 7.14 — 5 s — — 4 s 7 b i — 1 2 a i 1 3 1 3 . 7 3 9 — — 1.09 1.62 — 3 . 5 3 — — — — 5 p i 5 s — 4 s — — 14 14.5<c> — — — 0 . 9 1.6 2.8 4 . 5 — — — — 6 s 5 s 4 p i 7 b ! — 1 5 15.3(c> — — — — 0.8 2 . 0 3.7 4 . 2 — — — — 6 s 4P2 4 s 7 b ! 1 6 17 1 8 17.1<c> 19.6<c> 24.2< c) - 4 . 5 - 5 . 5 - 5 . 2 - 4 . 7 I • • J excitation from 1 3 a i ( d Z 2 ) more or deeply bound orbitals ip(d) 12 .65 14.07 1 4 . 8 3 15 .36 16.07 17.27 1 9 . 0 19.5 (a) See Figure 4.II.5. (b) Estimated uncertainty is ±0.020 e V . (c) Estimated uncertainty is ±0.1 eV. (d) IPs are from P E S measurements [157] and assigned by D V X a M calculation [155]. (e) Excitation from more deeply bound orbitals is also a possible alternative assignment (see text), t F o r b i d d e n transit ion 123 Table 4.II.6 Summary of term values for assignments of I S E E L S and V S E E L S spectra of CIF3 I n i t i a l Term Value (eV) O r b i a l 1 2 a i 7 b ! 13ax(dz2) H a ^ d ^ 2) 4 s 4 p i 4 p 2 3 d C l 2 p 3 / 2 11 .27 . 1 -1.4 - 9 . 3 3 . 8 3 2 . 7 7 2 . 0 2 1.68 r»7.30 C l 2 p i / 2 11 .38 -1 -1.3 - 9 . 6 3 . 8 9 2 .71 2 . 0 8 1.78 C l 2s 1 0 . 9 8.7 - 1 - 1 1 — 2.1 — F ls(ax) 7 . 4 1 1 — 2.5 — P - -0.9 r -13 F ls(eq) 1 0 . 0 —— J 4 b 2 6 . 7 7 — - 4 . 5 — 3 . 6 4 2 . 9 6 — — l l a i — 4 . 3 8 - 5 . 5 — 3 . 7 9 2 . 7 5 i . 9 3 — 6 b ! — 4 . 5 5 — — 3 . 8 1 2 . 6 9 — — l a 2 , 1 0 a i 6 . 3 5 4 . 3 4 — — 3 . 8 1 — — — 3 b 2 6 . 9 5 4 . 3 2 — — 3.71 — — 5 b i 6 . 9 9 4 . 9 1 — — 3 . 5 3 2.8 2 . 0 — 4 b i 6 . 8 6 4 . 5 - 5 . 2 — 3.7 — — — 2 b 2 7.14 4 . 2 - 4 . 7 — — — — — 124 the basis of the above assumptions. . The first structure observed in the VSEELS spectrum of CIF^ (Figure 4.II.5) is a weak and broad band at ~5.9 eV (feature 1). This feature is followed by a much more intense broad band with a vertical transition energy of 9.68 eV (feature 5). It should be noted that the least tightly bound (4bj) and the next, least tightly bound (lla^) occupied valence orbitals have localized characters of 7r(Cl-F ) and a(F -Cl-F ) [156] eq ax axy respectively. Similarly the normally unoccupied virtual valence orbitals 12a^ and 7b^ have * • localized character of a (Cl-F ) and a (F - C l - F ) respectively (see section 4.II.2.1). CQ 3.X 3.X Therefore, based on spatial overlap considerations, feature 1 is logically assigned as the 4bj •* 12a^ transition and feature 5 is assigned as the l l a ^ - * 7b^ transition. According to these assignments the term values for transitions to the 12a^ and 7b^ orbitals are 6.77 and 4.38 eV respectively (Table 4.II.5) which are ~3-4 eV lower than those for the corresponding inner-shell transitions to the 12aj and 7b^ final orbitals (Table 4.II.6). Since the loss of shielding caused by the removal of a (delocalized) valence electron should be less than that cause by the removal of a (localized) core electron from the centre of the molecule, the term values for corresponding transitions to the virtual valence orbitals would be expected to be lower in the case of VSEELS spectra. Therefore, the present assignments are consistent with this expectation. In addition, the present VSEELS assignments give an energy separation of ~2.4 eV, between the 12a^ and 7b^ orbitals which is also reasonably close to corresponding separations observed in the ISEELS spectra (see Table 4.II.6). Based on the assumption that the term values for virtual valence transitions are transferable between various originating valence orbitals additional transitions to the 12a-^  virtual valence orbital from the (^JOa^), 3b2> 5b^, 4b^ and 2b2 orbitals and transitions to the 7b^ virtual valence orbital from the 6bp (^.lOa-jX Vo^, 5bp 4b^ and 2b2 orbitals have been located (Table 4.II.5). Additional transitions may also be expected but are evidently either very weak or obscured by other more prominent bands. 125 Using the averaged term values of 3.86 eV (5g=2.13) for 4s Rydberg transitions observed in the Cl 2p spectra (Table 4.II.1), features (2-4), 6, 7, 9, 12, 13 and 15 are attributed to transitions to the 4s Rydberg orbital from the 4b 2 > l l a ^ 6^, (la 2, lOa^), 3b2, 5b1 and 4b^ orbitals respectively. Higher members of the ns Rydberg series (n>4) have also been located according to the term values calculated by the Rydberg formula (T=13.6/(n-6s)). Similarly, the np^ and np 2 Rydberg series from various valence levels have been assigned based on the average term values of 2.74 eV (6 =1.77) and 2.05 Pi eV (5p 2 = 1.42) for the 4p^ and 4p 2 Rydberg orbitals as observed in the Cl 2p spectrum (Table 4.II.1).' In general, all relatively sharper structures, in particular features 7-13, observed in the VSEELS spectrum, have been attributed to Rydberg transitions. It should be noted that most of these sharper Rydberg features are superimposed on an underlying background due to the presence of virtual valence transitions as indicated in Table 4.II.5. There is generally a very reasonable correspondence between the predicted manifolds and observed spectral features in Figure 4.II.5. As discussed in section 4.II.2.2, continuum shape-resonances have been observed in all the inner-shell excitation spectra of CIF^ (Figures 4.II.1-4) and in particular these features are very intense in the Cl 2p spectrum. The expected positions of the d-type shape resonance in the VSEELS spectra can be estimated as follows. Using a term value difference of 9.3 eV between transitions with 13a^(dz2) and 7b^ as final orbitals and, likewise, a 17.9 eV difference between transitions with 14a^(dx2_y2) and 7b^ as final orbitals as observed in the inner-shell excitations (Table 4.II.6) would predict the term values for the 13a^(dzJ) and 14a 1(d x 2_ y 2) shape resonances in valence-shell excitation to be ~-5 and ~-14 eV respectively, assuming the same differences. With these considerations, the weak features 16-18 observed above 17 eV may be 13a^(dz2) shape resonances originated from the 4b 2 > l l a ^ and (4b^, 2b2) orbitals respectively, while no structures have term values corresponding to the 14a, (d 2_ 2) shape resonance. l x —y Alternatively, features 16-18 may originate from excitations of inner-valence electrons which are more deeply bound than the 2b 2 orbital. Since there are no published measurements of the IPs for these more deeply bound inner-valence orbitals, no definite 126 assignments can be given. Overall, the present interpretation of the VSEELS spectrum of CIF^ suggests that for valence-shell excitation the continuum shape resonances, if present, are weak compared with the situation in ISEELS. Such significant weakening of the contribution from continuum shape resonances in the VSEELS spectra as compared to those in the ISEELS spectrum can be partly understood by the fact that the normally occupied valence orbitals are relatively more diffuse than the core orbitals and they are probably concentrated on the periphery of the potential barrier. As a result excitations from the valence orbitals will be expected to be less subject to potential barrier effects than are the core excitations. It should be noted that even though the continuum shape resonances in the VSEELS spectra of SFg, SeFg and TeFg have been found to be weaker than those in the corresponding ISEELS spectra (Chapter 4.1), these resonant features still have significant intensity in the VSEELS spectra. It is also interesting to note that SO2, which has a one dimensional barrier along the S - 0 axis, demonstrates resonance effects in the S Is and S 2p excitation spectra (see Ref. [163] and Chapter 7.II) but photoemission experiments on the valence ionization of S 0 2 found no evidence for potential barrier effects [164] Therefore, the observation of significantly diminished potential barrier effects in the VSEELS spectrum of CIF^ may also be due to the partly-open and partly-caged structure of the molecule. 4.II.2.4 The Role of Potential (Coulombic) Barrier Effects in CIF^ and Other Molecules The present work indicates that potential barrier effects are particularly pronounced in the ISEELS spectra of ClFj. The Cl 2p, 2s and F Is ISEELS spectra of C1F3 are shown on the same relative energy scale in Figure 4.II.6, aligned with respect to the Cl 2p 3 / 2 > Cl 2s and the average F Is IPs respectively. It is clear that these spectra exhibit transitions to discrete and continuum resonances as well as transitions to the respective Rydberg orbitals and ionization continua. In particular, all types of transitions, i.e. those to continuum shape resonances and Rydberg orbitals as well as direct ionization, 127 F l s A . V 2 3 AxEq 6 8 0 7 0 0 7 2 0 740 7 6 0 ENERGY LOSS (eV) Figure 4.11.6 Comparison of the (a) CI 2p; (b) CI 2s; (c) F 1s ISEELS spectra of CIF3. 128 are very intense in the Cl 2p spectra of, CIF^. In order to further explore the validity of the potential (Coulombic) barrier model, it is instructive to compare the corresponding central atom core shell spectra of SFg (S 2p), C1F 3 (Cl 2p) and HCl (Cl 2p) as shown in Figure 4.II.7. Details of the geometries and the spectral features are also shown on the top of Figure 4.II.7. The Cl 2p spectrum of CIF^ (this work) is evidently very different from the S 2p spectrum of the "fully-caged" molecule SFg (see Chapter 4.1 and Ref. [68]) which is dominated by the below and above edge shape resonances with only very weak contributions from Rydberg transitions and the direct ionization continuum. On the other hand the observation of intense shape resonances in addition to Rydberg structures and ionization continuum in the Cl 2p spectrum of CIF^ contrasts sharply with the Cl 2p spectrum of HCl. For HCl no Coulombic ligand barrier around the Cl is possible and it is of interest that the Cl 2p spectrum exhibits intense Rydberg transitions and a very prominent ionization continuum but shows no evidence of continuum shape resonances. Therefore it can be concluded that the co-existence of intense shape resonances and Rydberg transitions with a prominent direct ionization continuum in the Cl„2p spectrum of CIF^ is consistent with the partly-open and partly-caged molecular shape of CIF^ (see Figure 4.II.1 and discussion in section 4.II.1). Apparently, excitation channels "facing" the potential cage provided by the three F ligands will experience an effective potential (Coulombic) barrier which will result in intense shape resonances with suppression of Rydbergs and direct ionization. On the other hand, excitation channels "facing" the open-side of the molecule will not experience the potential (Coulombic) barrier and exhibit normal Rydberg transitions and a prominent direct ionization continuum. The three spectra shown in Figure 4.II.7 clearly demonstrate the changing situation from (a) prominent inner-well resonances with suppressed Rydberg and direct ionization structures to (c) prominent Rydberg and direct ionization structures and no resonance structures with (b) representing the in-between case exhibiting both types of behavior. These observations lend strong support to the potential (Coulombic) barrier model for the inner-shell excitation spectra of molecules with strongly electronegative ligands. 129 O C T A H E D R A L C A G E F U L L BARRIER Very weak Rydbergs Very weak continuum Strong MO resonances 11 II II _ l L 2pedge%a 200 220 _l_ 240 - J I' ' TTT; 'iro T - S H A P E C A G E PARTIAL BARRIER Prominent Rydbergs Prominent continuum Prominent MO resonances 210 ENERGY LOSS (eV) NO C A G E NO BARRIER H—Cl: Strong Rydbergs Strong continuum Weak MO resonances (Ryd) 200 220 240 260 280 300 RydtMrg i i "\'r'i i n^ -210 Figure 4.11.7 Comparison of potential barrier effects in central atom core excitation spectra of SF6 (S 2p), CIF3 (Cl 2p) and HCl (Cl 2p). 130 Further understanding of the nature of Coulombic potential barrier comes from considering the C 2 v symmetry of the ClFj molecule which implies there are five possible d-type outgoing channels of 5b 2(d y z), 2a 2(d x y), Sb^d^), Da^d^) and M a ^ d ^ ^ ) , characters respectively. Considering the relative orientation of the d orbitals and the CIF^ molecular framework (Figure 4.II.1) it can be seen that only lobes of the 13a^(dz2) and 14a,(d 2_ 2) channels will experience direct "head-on" repulsion from the potential barrier 1 x — y near the F and F atoms whereas the lobes of the other three d orbitals are either eq ax all pointed away from or point between the F ligands. Therefore the observation of only two and not five intense shape resonances (see also discussion in section 4.II.2.2.1) on each of the ISEELS spectra of CIF^ (Figure 4.II.5) is consistent with the above discussion suggesting that the potential barrier is most effective near the F ligands. It is noteworthy that the calculations (see Table 4.II.1) place the 13a^(dz2), channel at lower energy than the 14a, (d 2_ 2) channel. This again can be understood with a Coulombic potential barrier i x y picture because the 14a, (d 2 2) channel experiences repulsion from iw£ axial F atoms i x —y whereas the 13a^(dz2) channel experience repulsion from the single equatorial F atom only and, therefore, the 14a^(dx2_y2) channel is raised to higher energy than the 13a^(dz2) channel. These combined observations strongly suggest that the ideas of the Coulombic potential barrier model [34,35] may well have considerable physical significance and be of importance in molecules with electronegative ligands. Coulombic potential barrier effects may well be of importance in addition to the angular momentum (centrifugal) barrier effects described in the shape resonance (scattering) model [36,77,78,165]. Such centrifugal potential barrier effects have been observed in other types of molecules such as N 2 and CO [36,77,78] which do not possess electronegative ligands. The present work also suggests that a detailed theoretical reinvestigation of the role and function of Coulombic barriers in molecules such as CIF^ and SFg would be informative. In the meantime ISEELS studies of other "partial caged" molecules such as BrF^ [150] and IF^ [136] are of interest 131 CHAPTER 5 AN INVESTIGATION OF SUBSTITUTIONAL EFFECTS IN MONOHALOETHYLENES (C 2H 3X, X = F, CI, Br and I) : I. INNER-SHELL EXCITATION BY ELECTRON ENERGY LOSS SPECTROSCOPY 5.1.1 Introduction There have been numerous studies of the spectral properties of the halogenated ethylenes [54-56,166-183]. However, most of these studies are confined to excitation of valence electrons at energies below 20 eV because of the lack of continuum light sources in the far-UV and X-ray regions prior to the comparatively recent common availability of synchrotron radiation. There are only three reported studies of inner-shell excitation of monohaloethylenes: the CI Is photoabsorption spectrum of monochloroethylene using (weak) bremsstrahlung radiation [183], C Is excitation in monofluoroethylene by parent ion photoionization yield [176], .and the C Is and F Is inner-shell electron energy loss spectroscopy (ISEELS) of monofluoroethylene at low resolution recently reported by McLaren et al. [182]. This work reports a comprehensive ISEELS study of inner-shell excitation processes for the monohaloethylenes at energy losses (i.e., equivalent photon energies) up to ~740 eV. The following specific excitations have been studied: C Is, F Is in C 2H 3F; C Is, CI 2p, 2s in C 2H 3C1; C Is, Br 3d, 3p in C ^ B r ; and C Is, I 4d in C 2H 3I. Additional related work on valence-shell excitation with electron energy loss spectroscopy and photoionization using He(I), He(II) photoelectron spectroscopy, will be featured in part II of the present work. Monohaloethylenes, which are commonly called vinyl halides, provide an interesting chemical situation because of the interaction between the ir system of the vinyl group and the different halogen substituents. From a theoretical point of view, monohaloethylenes have served as models of substitutional effects on unsaturated systems. In simple models, 132 commonly used to explain perturbations caused by the substituent it is usual to invoke an inductive-electron-withdrawing (or donating) tendency of the substituent, as well as the resonance interaction resulting from the mixing of the substituent out-of-plane, nonbonding, p orbital with the 7r (or ir ) orbitals of the vinyl group. The inductive and resonance effects between the vinyl group and the halogen atom have been applied successfully in the interpretation of the photoelectron spectra of a variety of singly and multiply substituted haloethylenes [181-190], in which the energies for photoionization of electrons from the occupied valence orbitals are used as a probe of the substitutional effects. Complementary studies investigating the effects of halogen substituents on the empty virtual-valence orbitals of fluoroethylenes and chloroethylenes have been made using the techniques of electron-transmission spectroscopy [191,192] and electron, attachment [193] through the formation of temporary and stable negative ions, respectively. The transition of a core electron to one of the empty virtual-valence orbitals may show similar related effects. Therefore, in addition to studying the fundamental core-excitation spectroscopy, a further purpose of the present work is to investigate the sensitivity of the core-to-valence excitations to the substitutional effects of the different halogens in the vinyl halides. Robin [56,194] has found that for alkyl halides the term values for transitions terminating at a (CX) orbitals can be correlated with the bond strengths of the a bond appropriate to the a (CX), no matter whether the originating orbital is a core or valence orbital. It is of interest to investigate whether such a correlation also exists in the core to a (CX) bands, and also the valence to a (CX) bands of the monohaloethylenes. 5.1.2 Results and Discussion 5.1.2.1 General Considerations There are four types of virtual-valence orbitals for the monohaloethylenes (H 2C=CHX), which are /(CC), a*(CX), a*(CH), and a*(CC), in order of ascending energy above the occupied valence orbitals. The various ISEELS core spectra of monohaloethylenes are expected to be dominated by transitions from the core orbital of 133 interest to the low-lying ir (CC) and a (CX) virtual-valence orbitals, and by transitions to the Rydberg orbitals. In contrast, transitions terminating at the a (CH) and a (CC) virtual-valence orbitals have been located in the ionization continuum as weak and broad resonance features in the inner-shell excitation spectra of alkanes [195], and cyclic hydrocarbons [196]. In addition, a (CH) character has also been attributed to some transitions below the C Is IP in these molecules [195,196]. It should be noted that the • • • present notations of a (CC), a (CH) and a (CC) only provide general indications of the localized character of the virtual-valence orbitals. The real nature of a particular virtual orbital could be more complicated as a result of mixing of adjacent a orbitals and Rydberg orbitals. In CH^F, the o (CF) orbital was found to be mixed with the 3s Rydberg orbital [56,62,197]. ISEELS and NEXAFS studies of a variety of organic species indicate that features assigned as C Is •+ 3p Rydberg states also possess C Is •* a (CH) character [195,196,198]. The spectral features of the ISEELS spectra of the monohaloethylenes will be tentatively assigned, in a unified manner, in terms of excitations to a common manifold of virtual-valence and Rydberg orbitals. Most of the assignments are based on the term values (T) of these features relative to the ionization potential (IP) of the appropriate core level (i.e., T = IP - E, where E is the excited-state energy). 5.1.2.2 Halogen Inner-Shell Spectra Although all transitions are dipole allowed under the low C symmetry of the monohaloethylene molecules, the relative intensities of transitions to different levels below the various core edges may vary because of differences in spatial overlap between the respective initial and final orbitals. It is for this reason that the transitions terminating in the a (CX) orbital might be expected to be more prominent than those for the IT (CC) orbital in the halogen atom core spectra. Therefore, in order to identify the a (CX) bands, the halogen inner-shell spectra of the monohaloethylenes are considered first, and then the C Is excitation spectra. This is followed by a discussion of the relation between 134 the various transition terra values and the halogen substituent 5.1.2.2.1 F Is Spectrum of C ^ F Figure 5.1.1 shows the spectrum of monofluoroethylene (FWHM = 0.26 eV) in the region of F Is excitation. This is consistent with the lower-resolution (0.6 eV FWHM) ISEELS spectrum of C 2 H 3 F recently reported by McLaren et al. [182], except that the better resolved spectrum in the present work shows extra structure (feature 2). The energies, term values, and suggested assignments are given in Table 5.1.1. The F Is IP listed is the value reported from XPS measurements [118]. The first intense peak in the spectrum has a term value of 4.36 eV and is tentatively assigned as the F ls-» o (CF) transition. A small shoulder (feature 2) is located ~0.4 eV above feature 1 and may be due to vibrational excitation of the a (CF) transition, since the observed spacing is comparable with C-H stretching frequencies (~0.35 eV) observed in infra-red spectra of alkenes [199]. However, it should be noted that feature 2 has a term value of 4.01 eV, which is close to that for the 3s Rydberg feature observed in the C Is excitation spectrum of C 2 H 3 F (see Table 5.1.6). Therefore, the F Is •* 3s transition is an alternative assignment for feature 2. Feature 3 is attributed to the F Is -»3p transition according to a term value of 2.61 eV. This assignment is also supported by the fact that almost identical term values are observed for the two C Is ->3p transitions (see Table 5.1.6). Feature 3 is as intense as feature 1, and similar relatively intense bands have also been observed and assigned to 3p Rydbergs in the F Is spectra of CHjF [62], HF [124], and F 2 [124]. Similar bands observed in the F- Is spectra of other more highly substituted fluoroethylenes have, however, been assigned differently by McLaren et al. [182] (see below). In the purely atomic case, an s -»p type transition is highly favorable according to dipole selection rules, and therefore, as has often been noted in other similar situations IOH t CO LU LU > 6 5-LU (T 12 3 * T Is edge 0 = 0 ° Eo = 3 0 0 0 eV A E = 0 2 6 0 eV 5 ~ • 6 8 0 6 9 0 7 0 0 7 1 0 7 2 0 7 3 0 7 4 0 ENERGY LOSS (eV) Figure 5.1.1 F 1s ISEELS spectrum of C2H3F. Table 5.1.1 Energies, term values and possible assignments for features in the F 1s spectrum of C2H3F (a) Feature Energy L o s s ^ (eV) Terra Value (eV) Possible Assignment ( f i n a l orbital) 1 688.90 A.36 (vert) o*(CF)(v=0) 2 689.25 4.01 o*(CF)(v=l)/3s 3 690.65 2.61 3p F Is IP (693.26) ( c ) 0 A 695 ( d ) -2 * 0 ^ shape resonance 5 7 1 9 ( d ) -26 EXELFS (a) See Figure 5.1.1. (b) Estimated uncertainty is ±0.10 eV. below the F 1s edge. (c) F 1s IP from XPS measurement [118]. (d) Estimated uncertainty is ±1 eV. 137 in ISEELS spectra, the F ls-» 3p transition is expected to be relatively intense since the F Is orbital is highly localized and atomic-like. McLaren et al. [182] have assigned features in the F Is spectrum of CjH^F corresponding to peaks 1 and 3 of the present work as due to F Is •• ir (CC) and F Is -» a (CF), respectively. According to this assignment [182], the term values of both transitions would be approximately 1.5-2 eV lower than the term values of the corresponding C ls+ ir (CC) and C ls-> a (CF) excitations (see Table 5.1.6). Such a shift would be opposite to the normal expectation that the term values of transitions originating from a F Is orbital should be larger than those for the corresponding transitions originating from a C Is orbital because of the antishielding effect [56,194]. In addition, the present assignments are supported by the observation of F Is •* a (CF) and F Is.-*3p bands in CH^F [62], with similarly spaced term values as features 1 and 3. Features 4 and 5 are ~2 eV and ~26 eV, respectively, above the F Is IP. Similar features are also observed in the F Is spectra of each of the other, more highly substituted, fluoroethylenes [182]. Feature 4 is attributed to a a -type shape resonance since a corresponding structure also exists on the C Is continuum of C 2 H 3 F (see feature 23, Figure 5.1.6 and Table 5.1.6). The broad nature and high (negative) term value of feature 5 lead to the suggestion that this structure is due to an "EXAFS-type" feature associated with back scattering of the ionized F Is electron by the adjacent carbon atom, similar to the extended energy loss fine structure (EXELFS) observed earlier in ISEELS studies of the chloromethanes [125]. -5.1.2.2.2 CI 2p and 2s Spectra of C 2H 3C1 The broad range (0.13 eV FWHM) ISEELS spectrum of CI 2p excitation in C^H^Cl is shown in Figure 5.1.2b. A high-resolution (0.065 eV FWHM) spectrum below the CI 2p 3 / 2 j y 2 edges is shown in Figure 5.1.2a. The CI 2s excitation shown in the insert spectrum of Figure 5.1.2b is obtained at a resolution of 0.090 eV FWHM. The energies, term values, and possible assignments of these spectra are presented in Table 138 15-10->-b CO LU t o (a) Eo A E G 2000eV 0.065 eV 0° H H ;c=c H Cl Cl 2 p I 2 3 4 5 6 I I I I I I 8 9 10 II I I I I 12 13 Cl 2 p y 2 e d g e 2p ( / 2 edge 200 202 204 206 208 UJ > _J LU _ r r 5 (b) Eo = 3000eV A E = O.I30eV a = 0° 1 0 -5-H Cl Cl 2 p 15 Cl 2p 3 / 2 . 2p 1 / 2 edges V-v:**''" \ Cl 2s e • 0° E. « 3000eV : > * v , . AE « 0.090 eV 16 I 1 7 ^ ^ 1 Cl 2s edge ^ 0 268 278 190 210 230 ENERGY LOSS (eV) 250 Figure £.1.2 (a) High-resolut ion Cl 2p; (b) low-resolut ion Cl 2p, 2s (insert) ISEELS spectra of C2H3CI. 139 Table 5.1.2 Energies, term values and possible assignments for features in the CI 2p, 2s spectrum of C 2 H 3 C I Feature (a) Energy loss (b) Term Value (eV) Possible Assignment (eV) 2p 3/2 2p 1/2 2 p 3/2 2 p l / 2 1 200.50 •6.13 — — 2 200.78 5.85 — — 3 201.10 " 5.53 — r O*(CCD — 4 201.42 5.21 (vert) — — 5 201.74 4.89 — > — 6 202.14 — 6.21 " 1 7 203.18 3.45 5.17 (vert) 4s J • o*(CCl) 8 204.06 2.57 -- 4p — 9 204.90 1.73 3.45 3d 4s 10 205.22" 1.41 ~ 5s — 11 205.70 0.93 2.65 4d 4p 12 206.62 — 1.73 — 3d CI 2 p 3 / 2 IP 13 (206.63).C c ) 207.54 0 0.81 aa 4d CI 2 p 1 / 2 IP 14 (20&;35)C d ) Ce) 210.8. ' — -3.3 (f) 0 as * ed/or O - shape resonance 15 2 2 0 . 0 ( e ) -12.5 (f) CI 3d/ 0 (CC) shape resonance 2s 2s 16 17 CI 2s IP 272.3 257.3 (277.66) (c) 5.4 2.4 0 o (CCD 4p (a) See Figure 5.1.2. ~-(b) Estimated uncertainty is ±0.04 eV. (c) CI 2p 3 / 2 and CI 2s IPs from XPS measurements [118]. (d) Estimated CI 2p 1 / 2 IP- See text, section 5.1.2.2.2. (e) Estimated uncertainty is ±0.5 eV. (f) Term values with respect to the mean CI 2p 3 / 2 ii/2 IP (207.49 eV). 140 5.1.2. The IPs for the CI 2 p ^ 2 and CI 2s edges are known from XPS measurements [118]. The CI 2p^2 edge is estimated from the separation of the two sharp features 7 and 9 in the CI 2p spectrum (Figure 5.1.2a) by assuming that they are the spin-orbit partners for transitions terminating at the same final orbital. On this basis the estimated energy difference between the CI 2p^ / 2 and 2p 1 / 2 edges is 1.72 eV, which is close to the CI 2p spin-orbit splitting observed in methyl chloride [62] and the values (1.62, 1.63 eV) deduced from high-resolution CI 2p ISEELS spectra of C l 2 [200] and HCl [201]. Features 1-5 of the first band in the CI 2p spectrum (Figure 5.1.2a) are assigned as vibrational levels of the CI 2p^ / 2 •* a (CC1) transition. The vertical term value of this transition is 5.21 eV, which is similar to the value of 5.32 eV reported for the corresponding transition in methyl chloride [62]. The relative spacing of features 1-5 is ~0.3 eV, which suggests excitation of the C-H stretching vibrational mode. It is worth noting that features 1 and 2 of the F Is -* a (CF) transition in ^ H ^ F have similar spacing, end are likewise assigned to the C-H stretching mode (see above discussion and Table 5.1.1). The CI 2p 1 / 2+ a (CC1) transition is expected to be located under feature 7 at about 203 eV by using the same vertical term value (5.21 eV) as for its spin-orbit partner (feature 4, Table 5.1.2). This is not inconsistent with the spectral shape in this region, where feature 7 is apparently lying on top of a broad structure (Figure 5.1.2a) with intensity about half of that of the CI 2p^/2-» a (CC1) structure. Features 7 and 9 are assigned as spin-orbit partners of transitions terminating at the 4s Rydberg orbital. This assignment is consistent with their sharp band shapes and the term values of 3.45 eV, which are typical for a 4s Rydberg transition. Other features (8, 10-13) are assigned similarly as Rydberg states according to their term values relative to the appropriate CI 2p edges. Note that the transitions from the atomic-like CI 2p core to 4s Rydberg orbitals are much more intense than transitions to the 4p orbitals, as may be expected from a consideration of the corresponding atomic case where transitions from p to p are dipole forbidden, while transitions from p to s are dipole allowed. This 141 atomic-like selection rule has been found to be a good general guide in the interpretation of previous ISEELS spectra. There are two broad structures located above the Cl 2p continua. The first of these structures (feature 14) may be attributed to the delayed maxima of the Cl 2p continua, which, in analogy to the atomic case, may be expected to occur at an energy somewhat above threshold owing to the centrifugal barrier to the dominant 2p -*ed ionization channel [44,132]. Similar structures have been observed in the Cl 2p continua of C l 2 [200], HCl [201], and the chloromethanes [125]. Additionally, a a* shape resonance may be contributing to this feature since a continuum feature at a similar relative energy has been observed in the C Is spectrum of C 2H.jCl (see feature 20, Table 5.1.7, and Figure 5.1.6). The second structure (feature 15), which is about 12.5 eV above the threshold, is interpreted as a Cl 2p ->C1 3d shape resonance.t Similar intense 3d-type shape resonances have been observed in the 2p spectra of other molecules containing third-row elements .[66,160,161]. It can be noted that a C Is -> a (CC) resonance is found at about the same relative energy (see feature 21, Table 5.1.7), which may also contribute to the observed continuum structure. The Cl 2s spectrum shows much broader structures than that for Cl 2p because of band broadening induced by fast Coster-Kronig type autoionization to the underlying Cl 2p continua. The Cl 2s ISEELS spectrum of CjH^Cl is dominated by a broad band (feature 16) with a term value of 5.4 eV, which suggests that the terminating orbital is a (CC1) since a similar term value of 5.2 eV is found for a (CQ) in the Cl 2p spectrum. The second weak broad band (feature 17) is attributed to the Cl 2s •* 4p Rydberg transition in keeping with the term value of 2.4 eV, which is also similar to that observed for Cl 2p •+ 4p transitions. t The Cl 3d shape resonance can be described in term of an extend MO scheme with the inclusion of Cl 3d orbitals into the minimum basis set 142 5.1.2.2.3 Br 3d and 3p Spectra of C 2 H 3 Br Figure 5.1.3b shows the low-resolution energy loss spectrum (0.13 eV FWHM) for excitation of the Br 3d electrons of C2H.jBr in the 60-170. eV region. It is obvious that the Br 3d excitation structures are lying on top of the strong "tail" of the valence-shell continua. A high-resolution spectrum (0.080 eV FWHM) below the Br W^^Z/l eciges *s presented in Figure 5.1.3a. The hatched lines shown in the figure are estimated values of the Br 3d^ / 2 and -^3/2 ^ ° measurement has been reported for Br 3d in C 2 H 3 Br. Therefore the Br 3d^ / 2 IP of CjH^Br is taken to be equal to the Br 3d^ / 2 IP of CH^Br (76.4 eV) [118], since a consideration of the Br 3d^ / 2 IPs for a wide range of similar bromine-containing molecules [118] shows that the variation is only ±0.3 eV. The value for the Br 3d^ / 2 IP is determined from the spin-orbit splitting evaluated from the separation of sharp features 8 and 10, in the same manner as described in section 5.1.2.2.2 for the CI 2p^ / 2 ^ / 2 IPs of C ^ ^ C l . The spin-orbit splitting of 0.9 eV derived in this manner, for Br 3 d ^ 2 ^ 2 is similar to that (~1 eV) estimated earlier for CH^Br [62]. The energies of features observed in the Br 3d spectra of CjH^Br, along with term values and suggested assignments, are summarized in Table 5.1.3. The general appearance of the Br 3d spectrum of C 2 H 3 Br resembles that of CH^Br [62]. In addition, the term values of the first two broad bands (5.3 and 5.4 eV relative to the respective Br 3 d 5 / 2 3 / 2 edges) are also very close to those (5.5 eV) for similar bands in the Br.3d spectrum of CH^Br [62], where -they -are assigned as Br 3d^ / 2 -*a (CBr) and Br 3 d 3 / 2 - » a (CBr) transitions. Because of these similarities, the first two bands (features 1-2. and features 3-6) in the Br 3d spectrum of CjH^Br are • assigned as transitions terminating at the a (CBr) orbital. Additional small shoulders, just visible in these two bands, are most likely due to vibrational excitation. The observed energy differences between these vibrational features (see Table 5.1.3) suggest that the C - H stretching mode, as well as the C=C stretching mode, may be excited in these * transitions. The relative intensity of the Br 3d^ / 2 3 / 2 -» a (CBr) transition. is expected to be 3:2 according to the statistical weights, but experimentally the two bands are 10 t oo 0 LU LU > ^ 15 _ J LU cr 10 0 ( a ) 68 ( b ) valence continuum 9 Eo A E 0° 3000 eV 0.080 eV \ H H C=C H Br Br 3d 1 2 3 456 7 l _ l L_LU L 8 9 tO 1112 J U U_ B r 3 d 5 / 2 , 3 / 2 e d < 3 e s 70 72 74 76 78 9 Eo A E 0° 3000eV 0.130 eV Br 3d 5 / 2 , 3 / 2 ^ Q 6 5 60 80 100 120 140 ENERGY LOSS (eV) 160 ure 5.1.3 (a) High-resolution; (b) low-resolution Br 3d ISEELS spectra of C2H3Br. 144 Table 5.1.3 Energies, term values and possible assignments for features in the Br 3d spectrum of C2H3Br Feature ( a ) Energy l o s s ( b ) Term Value (eV) Possible Assignment ( 6 V ) 3 d 5/2 3 d 3/2 3 d 5/2 3 d 3/2 1 70.97 5.4 (vert) — 2 71.42 5.0 — 3 71.99 — 5.3 4 72.30 — 5.0 5 72.45 — 4.9 6 72.62 — 4.7 7 73.15 3.3 — 8 74.09 2.3 3.2 9 74.77 1.6 ' — 10 75.03 1.4 2.3 11 75.75 — 1.6 12 75.91 — 1.4 3 d 5/2 IP (76 .4 )( c ) 0 — 3 d 3/2 13 IP (77 .3) ( C ) 128 ( d ) _A 1(e) 0 } o * ( C B r ) o * ( C B r ) 5s 5p 5s Ad 6p 5p Ad 6p ef (a) See Figure 5.1.3 for Br 3d spectral features. (b) Estimated uncertainty is ±0.04 eV. (c) Br 3d IPs are estimated values. See text, section 5.1.2.2.3. (d) Estimated uncertainty is ±1 eV. (e) Term values with respect to the mean Br 3d5/2i3/2 IP (76.9 eV). 145 essentially of equal intensity. It is therefore likely that the spin-orbit intensity ratio for the Br 3 D5/2 3/2^ 0 ( C B r) transitions is modified by the exchange interactions of the core orbital with the penetrating final orbitals, as discussed by Schwarz [121]. A similar redistribution of spin-orbit intensity has been observed for CH^Br [62]. The other weaker features below the Br 3d^ / 2 3/2 limits may be assigned as transitions terminating at the 5s, 5p, 4d, and 6p Rydberg orbitals (see Table 5.1.3) according to their term values. With regard to the relative intensity of the different Rydberg excitations, it is obvious that the pseudo-atomic selection rule is again operative so that the transitions to the 5p are much more intense than those to the 5s or 4d Rydberg orbitals. In analogy with CH^Br [62], a weak maximum (feature 13) is observed on the Br 3d continua. This feature is ascribed to a delayed onset of the Br 3d continua as a result of the centrifugal barrier effect on the dominating ef outgoing channel for the ionized electron. The Bi 3p ISEELS spectrum of C ^ B r (0.13 eV FWHM) is shown in Figure 5.1.4. The top section of the figure shows the spectrum as recorded, while the lower part shows the spectrum after subtraction of an estimated sloping background given by the straight line. Such a subtraction procedure enhances the visibility of the peaks in the Br 3p spectrum. Unfortunately, there is no reported XPS measurement of the Br 3p^ / 2 Y/2 ^ s °^ e i t r i e r ^H^Br o r ^H^Br. Therefore, an average energy difference of 113.3 ±0.2 eV between the Br 3 p j / 2 and Br 3d^ / 2 IPs has been obtained by considering all compounds for which the Br 3p,j / 2-3dj / 2 energy difference is available from XPS measurements [118]. By this means, the Br 3p^ / 2 IP of CjH^Br is estimated to be 189.8 eV with an uncertainty of 0.5 eV. Similar procedures for estimating otherwise unknown edge positions were satisfactorily used in the interpretations of ISEELS studies of dimethyl sulfoxide and sulfur dioxide reported in Chapter 7. The spin-orbit splitting of Br 3p^ / 2 and Br 3 p y 2 edges is estimated to be 6.5 eV from the energy difference of features 1 and 3, and this value compares well with the corresponding average splitting (7 eV) reported for bromine [87]. The energies, term values, and possible assignments for features observed in Figure 5.1.4 are given in Table 5.1.4. 146 ENERGY LOSS (eV) Figure 5.1.4 Br 3p ISEELS spectrum of C2H3Br; (a) original, (b) background subtracted data. 147 T a b l e 5.1.4 Energ ies , term va lues a n d poss ible ass ignments for features in the Br 3p spectrum of C 2 H 3 B r Feature ( a ) Energy l o s s ^ Term Value (eV) Possible Assignment (eV) 3 p 3/2 3 p l / 2 3 p 3/2 3 p l / 2 1 184.67 5.1 — o*(CBr) — 2 186.84 3.0 — 5s Br 3 p 3 / 2 . I P (189.8) ( C ) 0 — CO — 3 191.16 — 5.1 — o*(CBr) 4 193.30 — 3.0 — 5s Br 3 p 1 / 2 IP (196.3) ( C ) — 0 — GO (a) (b) (c.) S e e Figure 5.1.4 for B r 3 p spectral features. Est imated uncertainty is ±0.06 e V . Br 3p IPs are est imated va lues . S e e text, sect ion 5.1.2.2.3. 148 The strongest band (feature 1) in the Br 3p spectrum (Figure 5.1.4) is assigned as the Br 3p^/2"» a (CBr) transition since its term value of 5.1 eV is close to the value of 5.4 eV for a (CBr) in the Br 3d spectrum (Table 5.1.3). Consequently, feature 3 can be attributed to the Br ?>Vy2 "* ° (CBr) transition according to its term value of 5.1 eV relative to the Br ^V^,^ IP- Similarly, features 2 and 4 are assigned as spin-orbit mates terminating • at the 5s Rydberg orbitals, according to their term values of 3.0 eV. In addition, the 5s Rydberg transitions are expected to be relatively strong because of the highly favorable p •+ s transition resulting from Br 3p excitation. The bands (features 3 and 4) belonging to the Br 3p ^ 2 series are much broader than their spin-orbit partners (features 1 and 2) associated with the Br 3p 3 / 2 u i r u t- Ah features observed in the Br 3p spectrum are expected to be somewhat broadened because of the possibility of fast autoionization to underlying valence and Br 3 d ^ 3 /2 continua, which shortens the lifetime of Br 3p excited states. In this sense, the Br 3p^ / 2 transition would be expected to be even broader due to the possibility of extremely rapid Super-Coster-Kronig type processes involving the additional autoionization channel provided by the Br 3p 3 / 2 continuum in the same subshell. Realizing that features 3 and 4 are on top of the Br 3p 3 / 2 continuum also accounts for the fact that the area ratios (feature 1/feature 3) and (feature 2/feature 4) appear to be somewhat different from the expected value of 2, neglecting exchange interaction [121]. 5.1.2.2.4 I 4d Spectra of C 2H 3I Figure 5.1.5b shows the low-resolution (0.29 eV FWHM) I 4d electron energy loss spectrum of C2H.JI in the 48-100 eV region. A high-resolution spectrum (Figure 5.1.5a, 0.070 eV FWHM) shows details of the structures below the I 4 d ^ edges. The energies, term values, and possible assignments for features observed in Figures 5.1.5a and 5.1.5b are given in Table 5.1.5. The I 4d ^ 2 I P *s estimated in a similar manner as that for the Br 3d^- (see above) by evaluating I 3d^ /2~I 4 d ^ splittings of all iodine compounds for which the necessary XPS data exist [118]. The estimated I 4 d ^ IP is 149 (a) 10 e = o ° Eo = 3000 eV A E = 0.070 eV H. H :c=c H I I4d to LU L U > _ J LU cr 5 h 123 0 101 50 ( b ) /I V i 4 5 _L_L 6789 I I II 10II 12 I I I 52 \ I 4 d 5 / 2 4d3/2edges L L_ 54 56 58 6 0 62 •e = o ° Eo = 3 7 0 0 eV A E = 0 .290 eV o i 2 4 69 10 I I M I 13 J _ I4d 5/2,3/2 J edges 14 J _ 50 6 0 7 0 8 0 ENERGY LOSS (eV) 9 0 Figure 5.1.5 (a) High-resolution; (b) low-resolution I 4d ISEELS spectra of C 2 H 3 I . 150 Table 5.1.5 Energies, term values and possible assignments for features in the I 4d spectrum of C2H3 I "1 Feature (a) Energy l o s s ^ Terra Value (eV) Possible Assignment^ (eV) 4 d5/2 A d3/2 A d5/2 A d3/2 1 50.85 6.2 1 2 50.97 6.0 (vert) > c*(CT) 3 51.18 5.8 J 1 4 52.6A — 6.0 (vert) 5 53.00 — 5.6 > 0 (CI) 6 7 5A.A9 5A.6A 2.5 2.A 6p 8 55.03 — 3.6 6s 9 55.19 1.8 — 5d 10 56.1A — 2.5 11 56.32 — 2.3 6Pj_ 12 56.81 ~ 1.8 5d 1 A d 5/2 IP (57.0)( c ) 0 — 00 1 A d 3/2 IP (58.6)( c ) -- 0 OO 13 68 ( d ) 10 ( e ) double excitation 1A 83 ( d ) 26 ( e ) €f (a) See Figure 5.1.5. (b) Estimated uncertainty is ±0.03 eV. (c) I 4d IPs are estimated values. See text, section 5.1.2.2.4. (d) Estimation uncertainty is ±1 eV. (e) Term values with respect to mean I 4ds/2,3/2 IP (57.8 eV). (f) 6p / y and 6p± refer to in-plane and out-of-plane 6p Rydberg orbitals. 151 57.0± 0.3 eV. The I 4d /^2 3 /2 spin-orbit splitting is estimated to be 1.6 eV from the energy difference between sharp features 2 and 4, as well as between features 6 and 10. This value is the same as the I 4d^ /23/2 splitting reported for CH^I [62]. The I 4d spectrum of CjH^I is also very similar to that of CH^I [62], and therefore the assignment can proceed, as in the case of the other monohaloethylenes, through a comparison with the corresponding methyl halides. The first two bands (features 2 and 4) are attributed to spin-orbit mates terminating at the a (CI) orbital in keeping with a term; value of 6.0 eV, which is the same as the term value of corresponding a (CI) transitions in CH^I. In addition, the broad nature of these bands suggests that the terminating orbitals are of valence-type. Evidence for vibrational structures can also be observed, as in the case of the corresponding a (CX) transitions for the other monohaloethylenes. Spin-orbit pairs, features (6,10) and (7,11), are attributed to I 4 d ^ ^ •* 6p//r in-plane, and I 4 d ^ 3 /2 "*6p_j_ out-of-plane Rydberg transitions according to their term values of 2.5 and 2.3 eV, respectively. In a photoabsorption study of the "perfluoro" effect in the valence to 6p Rydberg states of C 2 H 3 I and [168], the in-plane p^ Rydberg state is found to have a higher term value (2.45 eV) than the out of plane pj_ Rydberg (2.11 eV) because of more effective inductive stabilization for the in-plane orbital. This will be discussed further in part II of this chapter considering the VSEELS spectra of C J H J X . Feature 8 has a term value of 3.6 eV relative to the I 4d^ / 2 edge, which' implies that it may be the transition to the 6s Rydberg orbital. The I 3d^2"* 6s spin-orbit partner is evidently obscured by the tail of the second broad band (features 4 and 5). Features 9 and 12 are assigned similarly to transitions to 5d Rydberg orbitals, based on their term values. of 1.6 eV with respect to the appropriate I 4 d ^ 3 / 2 I p s- ^ should be noted that the lower intensities of 6s and 5d Rydberg features relative to 6p Rydberg features are again in keeping with pseudo-atomic selection rules. An exceedingly broad structure (feature 14), more than 20-eV wide, is found at about 26 eV above the I 4^/2 2/2 * P s - A consideration of the position and width of 152 this feature suggests that it is the delayed onset of ionization in the high angular momentum (ef) ionization channel [132,202,203]. Closely similar delayed-onset peaks have been observed in the I 4d spectrum of CH^I [62], and I 2 [204] and the Xe 4d spectra of XeF 2 and XeF 4 [205]. There is a second weak broad maximum (feature 13) located at 68 eV, approximately 10 eV above the I 4 d ^ / 2 j / 2 edges. By comparison with the assignment of similar features observed at energies of 66 and 72 eV in CH^I [197], and 61 and 71 eV in I 2 [204], feature 13 is assigned to double-electron excitation. A noteworthy characteristic of the above interpretations of the halogen inner-shell excitation spectra of the monohaloethylenes is the absence of transitions to the ir (CC) orbital. The assignment of the lowest transitions to a (CX) final orbitals seems the most reasonable conclusion, given the evidence discussed above. This assignment is contrary to that proposed for CjH^F by McLaren et al. [182]. The apparent absence of ir (CC) transitions in the present assignments seems surprising on first sight, since transitions to it (CC) and a (CX) are both dipole-allowed for the low-symmetry (Cg) monohaloethylene molecules. However, attempts to locate the n (CC) transition or reassignment of the spectra on the basis of the TT (CC)-O (CX) energy separations (see Tables 5.1.6-9 below) found in the following consideration of the C Is excitation of the monohaloethylenes, do not produce any satisfactory or consistent results. It is clear that the transitions to the * it (CC) orbital are either absent, or at best are weak features hidden under the low-energy side of the broad and much more intense a (CX) bands^ (see Figures 5.1.1-5). Additionally, the high degree of similarity in band shape and band positions between the halogen inner-shell excitation spectra of the monohaloethylenes and the corresponding methyl halides, where of course there is no C=C bond, strongly implies that the 7r (CC) transition is at best very weak or not present in the halogen core spectra of the monohaloethylenes. Therefore the present proposed assignments with the absence of 7r (CC) transitions in the halogen core-excitation spectra are considered as the most reasonable interpretation. These observations indicate that the spatial overlap of the 153 ir (CC) final orbital and the originating highly localized (i.e. atomic-like) halogen core orbital is extremely small. Under C symmetry only the out-of-plane pi orbitals on the halogen substituent have the proper symmetry for mixing with the ir (CC) orbital of the vinyl group. Therefore, transitions from CI 2p and Br 3p will be severely discriminated against in transitions to the ir (CC) orbital on the basis of atomic-like (i.e. p-+ p) selection rules. However, the observation that transitions from F Is, CI 2s, Br 3d, and I 4d to the ir (CC) orbital are apparently not present suggests that there is little halogen p character in the ir (CC) orbital, which in turn implies that the ir (CC) orbital is essentially localized about the C=C bond. This contrasts with the better-known situation that the normally occupied TT(CC) orbital is considered to interact significantly with the occupied halogen valence (out-of-plane) p_j_ orbital, as shown by PES studies [184-190]. However, it should be noted that in this situation the halogen valence p orbital is energetically close to the normally occupied 7r(CC) orbital, but is much lower in energy than the ir (CC) orbital. Therefore, it is not unreasonable to expect that there is little mixing between the ir (CC) and the occupied halogen valence p orbitals. Furthermore, the experimental observations also suggest that there is little or no mixing of the normally unoccupied halogen out-of-plane pj_ orbital character into the ir (CC) orbital. This is "not unexpected since the ir and np term values differ by ~3 eV. 5.1.2.3 C Is Spectra of Monohaloethylenes The two carbon atoms in CjH^X are not chemically equivalent since one is bonded to two hydrogen atoms, while the other is bonded to one hydrogen atom and the halogen atom. For ease of discussion, in the present work they are designated as C(CH) and C(CX), respectively. Figure 5.1.6 shows the broad-range C Is ISEELS spectra of the four monohaloethylenes in the 280- 340 eV region. These spectra were measured at resolutions in the range 0.130-0.155 eV FWHM, as indicated in the figure. A C Is ISEELS spectrum of C-H. measured by Hitchcock and Brion [206] is also shown at the bottom 154 (a) e = o° Eo = 3 0 0 0 eV AE = O.I55eV I vs I0III9 23 24 i 11 j -l 1 I 1 C,,C2 Is edges 1 r Ul LU LU > _ l LU cr (b) i 1 r i 1 r 9 = 0 ° Eo = 3 0 0 0 eV AE = 0.130 eV 16 9 17 19 20 21 C h C 2 Is edges "1 1 I <~. I 1 I 1 I 1 T (C) e = o° Eo = 3 0 0 0 e V AE = 0.130 eV I6I3OB20 2I II If (_J I » C„C2 Is edges - 1 — T (d) - i — r " T — r i i i t9 = 0 ° Eo = 3 7 0 0 eV AE = 0.143 eV /••JI234578IOII u f ff rr_j i •C|,C2 Is edges "I ' 1 ' 1 r T | I | I | I | 9 - 2 ° Eo = 2 5 0 0 eV AE = 0.6 eV (e) I ijTIOII 12 (3 14 'i—LJJL- ' 1 C Is edge i 1 1 1 1 1 1 1 1 1 1 1 1 — 280 300 320 340 ENERGY LOSS (eV) 360 Figure 5.1.6 Low-resolution C 1s ISEELS spectra of (a) C 2 H 3 F ; (b) C2H3CI ; (c) C2H3Br; (d) C 2 H 3 I (this work); and (e) C 2 H 4 ( t a k e n from Ref [206]). 155 (a ) C i s EXCITATION \ 9 = 0° Eo « 3000 eV A E • 0.080 eV I 2 3 4 V J S 6 7 B 9 O II 12 O B 17192022 " " ' _ III ' ' ' "I 1 r -i 1—r ' C2 Is edges -l 1 1 1 1 1 1 1 r e - o° Eo = 2000eV A E - 0.065 eV (b) co LU h-LU > LU cr 1234 36 7 8 910-13 14 B16 17 B I III III I I III! II I I I • C, B C2 IS edges — i 1 1 1 r 9 • 0° Eo « 2000eV A E • 0.075 eV (0 M ' \ 234 3-11 1213 M S 1617 B 19 ' m m i n i i i i I I i I 1 H *Br C| C2 Is edges i — i — r 1 r (d) i 1 r 9 • 0° Eo = 3700 eV A E » 0.143 eV C, Cj Is edges T — I — I — i — r i—i—I—i—r 9 =0° Eo = 3000 eV A E * 0.065 eV (e) C Is edge i 1 1 1 1 1—f 282 286 i 1 1 1 r 290 294' 298 ENERGY LOSS (eV) 1—1—1—' Figure 5.1.7 High-resolution C 1s ISEELS spectra of (a) C2H3F; (b) C2H3CI; (c) C2H3Br; (d) C2H3I; and (e) C2H4. 156 T a b l e 5.1.6 Energies, term values a n d possible assignments for features in the C 1s spectrum of C 2 H 3 F Feature Energy loss Term Value (eV) Possible Assignment (eV) C(CH) Is C(CF) Is C(CH) Is C(CF) Is 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 C(CH)ls IP 17 18 19 20 21 22 C(CF)ls IP 23 24 28A.84 285.05 285.26 285.44 287.03 287.09 287.24 287.54 287.72 288.56 289.52 289.64 290.21 290.42 290.84 290.96 (291.10) 291.59 291.77 291.95 292.49 292.79 293.03 (c) (293.48) (c) 294.4 299.3 (d) (d) 6.26 6.05 5.84 5.66 4.01 (vert) 2.54 1.58 1.46 0.89 0.68 TT*(CC) 6.45 3 6.39 (vert) o (CF)/3 6.24 5.94 5.76 — J 3.96 3.84 2.64 2.52 1.89 1.71 1.53 0.99 0.69 0.45 0 -2.1 -7.0 (e) (e) Tt (CC) 3p 3d/4s 4p 4d/5p 5d o (CF)/3s } 3p "7 y. 3d/4s 4p 4d/5p 5d 6d/6p a . shape resonance * * o (CC)/o j shape resonance (a) See Figures 5.1.6a and 5.1.7a. (b) Estimated uncertainty is ±0.04 eV. (c) C Is IPs from XPS measurements [118]. (d) Estimated uncertainty is ±0.15 eV. (e) Term values with respect to mean C Is IP (292.29 eV). 157 Table 5.1.7 Energies, term values and possible assignments for features in the C 1s spectrum of C 2 H 3 C I Feature Energy loss Terra Value (eV) Possible Assignment (eV) C(CH) Is C(CC1) Is C(CH) Is C(CC1) Is 1 284.68 6.31 (vert) — 2 284.96 6.03 — > TT*(CC) 3 285.12 5.87 4 285.32 5.67 5 285.92 — 6.43 6 286.02 4.97 6.33 (vert) o*(CCD V- n*(CC) 7 286.28 — 6.07 8 287.36 3.63 4.99 4s o*(CCl) 9 288.04 2.95 10 288.32 2.67 — >- 4p 11 288.48 2.51 12 288.62 — 3.73 — ") > 4s 13 288.74 — 3.61 — J 14 289.44 — 2.91 15 289.60 1.39 2.75 5p f 4p 16 289.94 — 2.41 17 290.72 — 1.63 — 3d C(CH) Is IP (290.99) ( c ) 0 — « 18 291.38 — 0.97 — 4d C(CC1) Is IP (292.35) ( C ) — 0 — 19 293 .1 ( d ) -1.5 20 295 .6 ( d ) -4.0 21 3 0 4 . 0 ^ -12.3 o (CC) shape resonance o ^ shape resonance * o 2 shape resonance (a) See Figures 5.1.6b and 5.1.7b. (b) Estimated uncertainty is ±0.04 eV. (c) C Is IPs from XPS measurements [118]. (d) Estimated uncertainty is ±0.10 eV. (e) Term values with respect to mean C Is IP (291.67 eV). 158 Table 5.1.8 Energies, term values and possible assignments for features in the C 1s spectrum of C2H3Br _ . (a) Feature Energy loss Term Value (eV) Possible Assignment (eV) C(CH) Is C(CBr) Is C(CH) Is C(CBr) Is 1 28A.62 6.33(vert)— ~ \ — 2 28A.95 6.00 • n*(CC) — 3 285.07 5.88 4 285.18 5.77 J 5 285.63 — 6.A3 6 285.73 5.22 6.33(vert) o*(CBr) 7 8 285.87 286.00 6.13 6.06 ( . V u (CO 9 286.12 — 5.9A 10 286.23 5.83 / 11 286.36 — 5.70 J 12 286.83 — 5.23 — o"(CBr) 13 287.2A 3.71 — 5s ~ IA 287.9A — A.12 ~ 5s 15 288.A3 2.52 — 5p — 16 289.2A 1.71 — Ad . — 17 289.53 — 2.53 — 5p 18 290.3A — 1.72 — Ad C(CH) Is IP (290.95) ( c ) 0 — CD — 19 290.98 — 1.08 5d C(CBr) Is IP (292.06) ( c ) — . 0 00 20 292.9 ( d ) -1 .A * shape resonance 21 295.5 ( d ) -3 .9 * ° 2 shape resonance (a) See Figures 5.1.6c and 5.1.7c. (b) Estimated uncertainty is ±0.04 eV. (c) C 1s IPs are estimated values. See text, section 5.1.2.3. (d) Estimated uncertainty is ±0.10 eV. (e) Term values with respect to mean C 1s IP (291.51 eV). 159 Table 5.1.9 Energies, term values and possible assignments for features in the C 1s spectrum of C2H3 I _ . (a) Feature Energy l o s s ^ Term Value (eV) Possible Assignment (eV) C(CH) Is C(CI) Is C(CH) Is C(CI).Is 1 284. ,66 6.43 n*(CC) 2 285. ,37 5.72 6.43 o*(CI) n*(CC) 3 286.13 — 5.67 — o*(CI) 4 287. ,36 3.73 — 6s — 5 288. ,01 — 3 .79 — 6s 6 288, ,24 2.85 — 6p — 7 289. ,12 1.97 2.68 5d 6p 8 289. ,88 1.21 1.92 5d 9 290. ,53 — 1.27 — 7p C(CH) Is IP (291. , 0 9 ) ( C ) 0 -- CO — C(CI) Is IP (291, , 8 0 ) ( d ) — 0 ~ CO 10 292, , 8 ( e ) -1. , 4 ( f ) * 0 1 * 0 2 shape resonance 11 294, 2(e) -2, , 8 ( f ) shape resonance (a) See Figures 5.1.6d and 5.1.7d. (b) Estimated uncertainty is ±0.24 eV. (c) C(CH) Is IP from XPS measurement [118]. (d) C(CI) Is IP is estimated value. See text, section 5.1.2.3. (e) Estimated uncertainty is ±0.6 eV. (f) Term values with respect to mean C Is IP (291.45 eV). 160 of Figure 5.1.6 for comparison. The discrete structures observed below the respective C Is IPs are shown in greater detail in the high-resolution spectra of C 2 H 2 F , C 2 H 3 C I , C 2 H j B r , and CjH^ presented in Figure 5.1.7. A higher resolution C Is spectrum of C2H.JI was not obtained due to limited sample availability. To aid the comparison in the pre-edge regions, a relevant section of the low-resolution spectrum of C 2 H 3 I is shown in Figure 5.1.7d. A low-resolution C Is ISEELS spectrum (0.6 eV FWHM) of C J H J F , reported earlier by McLaren et al. [182], is consistent with the present work, given the large difference in energy resolution. The energies, term values, and possible assignments for all features observed in the C Is excitation spectra of C^H^F, € 2 ^ 0 , CjH^Br, and C 2 H 3 I are given in Tables 5.1.6-9 respectively. The C Is IPs are taken from XPS data [118] where measurements are available. Since the C(CH) Is and C(CBr) Is IPs of C2H.jBr, and also the C(CI) Is IP of C 2 H 3 I , have not been reported, they have been estimated by interpolation from plots of the available C Is IPs of the other monohaloethylenes versus the electronegativities of the respective halogen atoms. The predicted IPs are reasonably close to the corresponding C Is IPs of related molecules listed in XPS data tables [118]. The C Is excitation spectra of the monohaloethylenes can be interpreted as two overlapping series of transitions associated with excitation from the two chemically distinguishable C Is orbitals to a common manifold of unoccupied orbitals. Transitions terminating at the same final orbital can be identified by nearly equal term values relative to the respective C Is IPs. A discussion of the C Is spectra is given below. 5.1.2.3.1 C Is-* TT*(CC) Transitions The low-energy region of the high-resolution C Is spectra of €2^1^, C^H^Cl, and C^H^Br (Figure 5.1.7) consists of two intense broad bands, each of which exhibits several partially resolved features. Similar intense bands are also apparent in the low-energy region of the somewhat lower resolution C2H.JI spectrum (see Figure 5.1.7d). The first band is due to the transition from the C(CH) Is orbital to the ff (CC) lowest 161 unoccupied molecular orbital. The absolute energies of these transitions in each of the monohaloethylenes is within 0.2 eV of the energy for the it (CC) band in the C Is spectrum of C 2 H 4 [206,207]. Likewise, the vertical term values (Tables 5.1.6-9) for the (CC) peaks in C 2H 3X are all very similar (6 .26-6.43 eV) to the value of 6.21 eV observed for C 2 H 4 [206,207]. The second band, which is in each case the most intense structure in the C Is spectrum, can be identified as the respective C(CX) Is •* n (CC) transition in accord with the vertical term values, which are in the range 6.33-6.43 eV with respect to the C(CX) Is IPs. The term values of it (CC) features for the monohaloethylenes are summarized in Tables 5.1.6-9. Discussion on the trends of these transitions is given in section 5.1.2.3.5. It should be noted that the absolute energy of the second band decreases significantly from 287.09 eV for C J H J F . to 285.37 eV for C 2H 3I so that it is closer in energy to the first band in going from C-jH^F to C-jH^I (see Figure 5.1.7). This shift of absolute energy can be rationalized as mainly due to inductive stabilization of the C(CX) Is orbital by the halogen substituent (i.e., greatest for C2H.jF). This parallels the decrease in the C(CX) Is IP in going from F to I. Since inductive effects diminish with distance, the effect on the C(CH) Is orbital is expected to be much less than that on the C(CX) Is orbital. As a result the absolute energies of the first bands are quite constant and similar to that for C 2 H 4 [206,207]. The high resolution C Is ISEELS spectrum of C 2 H 4 [206,207] shows that the C-H stretching mode is excited in the C Is -*it (CC) transition [208]. Since both of the C Is-* 7r (CC) bands in each of the monohaloethylenes are quite broad and structured in the high-resolution spectra, it is likely that vibrational excitation is involved. A study of the high-resolution C Is excitation spectra of the isotopically substituted molecules [62], of a theoretical study similar to that reported by Barth et al. [208] for C 2H 4 > would be informative. It should be noted that the vibrational envelope of the first band is quite different from that of the second band, which is less well resolved. This suggests that * different vibrational modes may be excited in the C(CH) Is -*it (CC) and 162 C(CX) ls-»7r*(CC) transitions. The relative intensities of the two C Is ir (CC) transitions can be considered as probes of the relative spatial distributions of the ir (CC) orbital over the two different carbon-atom sites. It can be seen from Figure 5.1.7 that the first band (C(CX) Is -*ir (CC) transition) is less intense than the second band (C(CH) ls->7r (CC) transition) in the C Is spectrum of each monohaloethylene. Although the C(CH) Is-* a (CX) transition is predicted to occur at the same energy position as the C(CX) Is •+ ir (CC) transitions, its contribution to the intensity of the second band will be small because of poor spatial overlap between C(CH) Is and a (CX) orbitals (see section 5.1.2.3.2). These considerations suggest that the ir (CC) orbital is polarized toward the C(CX) carbon atom. Similar changes in intensity have been observed in the ISEELS spectra of more highly substituted fluoroethylenes [182] and also in fluorobenzenes [209], where the increase is attributed to a stronger potential barrier in more highly fluorinated species. It is noteworthy that the intensity of the second band relative to the first band is largest for monofluoroethylene (see Figure 5.1.7). This observation is consistent with the fact that fluorine is the most electronegative halogen substituent 5.1.2.3.2. C Is ->a*(CX) Transitions Discussion of the halogen inner-shell excitation spectra in preceding sections of the * present work suggests that low-lying a transitions may also exist in the C Is spectra. Identification of these will be difficult in the C Is spectra because they will likely be located in heavily overlapping regions of the valence and Rydberg transitions. However, their intensities are expected to be comparable to or even stronger than the Rydberg transitions in considering the a (CX) valence character. In addition, the term values of o (CX) transitions in the halogen spectra, though generally not transferable, can provide guidelines on the possible relative positions of the a (CX) transition in the C Is spectra. With this in mind the broad intense features observed in the spectra can be tentatively assigned as a (CX) transitions through the comparison of term values with their 163 counterparts in the halogen-excitation spectra, as well as in the C Is spectra of the methyl halides [62]. Feature 11 in the spectra of CjH^F is the most probable candidate for the C(CF) Is •* a (CF) transition, considering the arguments presented above. Feature 11 is relatively intense and broader (0.5 eV width) than the Rydberg states. The term value of 3.96 eV for feature 11 is quite close to that of 4.36 eV observed for the a (CF) transition in the F Is spectrum. The C(CF) ls+ a (CF) assignment for feature 11 therefore seems reasonable considering the larger antishielding effect in the F Is excitation. A similar decrease of a (CF) term value from 5.3 eV in the F Is spectrum to 4.65 eV in the C Is spectrum has been reported in the study of CH^F [62]. The assignment of feature 11 as a transition to the a (CF) orbital is also consistent with the findings of a recent ISEELS study of CjH^F and more highly substituted fluoroethylenes in which an increase of tie C ls-> o (CF) term value from 3.8 to 5.0 eV with degree of fluorination has been observed [182]. The presently observed C(CF) ls-» a (CF) term value in CjH^F agrees well with the value and assignment of reference [182], and the observed trend with increased fluorination lends confidence in the assignment of feature 11 to the o*(CF) final orbital. In CH^F the C Is -» a (CF) transition was found to overlap the C Is -» 3s transition [62]. In addition, since the a (CF) and 3s Rydberg orbitals are valence-shell conjugates (i.e., same symmetry and same nodal pattern), they are believed to interact strongly and mix extensively [62,197]. Similar Rydberg valence mixing could possibly occur for the a (CF) and 3s Rydberg states in C2H.JF. The valence/Rydberg mixing may be the reason why the a (CF) band (feature 11) is sharper and more noticeable than the a (CX) bands observed in the C Is excitation spectra of CjH^Cl, C2H 3Br, and C 2 H 2 I . Assignment of feature 11 as the C(CF) Is •+ 0 (CF)/3s transition implies that the peak due to C(CH) Is -*a (CF)/3s is located in the second band of the spectrum and • * overlaps with the C ls(CF)-> n (CC) transition. The C(CH) Is-* a (CF)/3s band, assigned to peak 6, will be discussed further below. 164 The broad band in the energy-loss range 287.0-288.5 eV in the C Is spectrum of CjHjCl (Figure 5.1.7b) has a shoulder on the low-energy side (feature 8) that may be attributed to the C(CC1) ls-> a (CC1) transition. The term value of 4.99 eV is close to the a (CC1) term values (5.21, 5.4 eV) observed in the Cl 2p, 2s spectra of C 2H 3 C1, and is comparable with those reported (5.00 eV in C Is, 5.32 eV in Cl 2p, and 5.6 eV in Q Is) for CH^Cl [62]. Similarly, feature 12 in the C Is excitation spectrum of CjH^Br is assigned as the C(CBr) Is -*o (CBr) transition because of a term value of 5.23 eV, which is consistent with the observed values (5.1, 5.4 eV) in the Br 3p, 3d spectra of CjHjBr and those (5.3 eV in C Is, 5.5 eV in Br 3d) reported for CH^Br [62]. In the C Is spectra (Figure 5.1.7d) of CjH^I the C(CI) Is •* a *(CI) transition (feature 3) is located well below the region of Rydberg transitions (features 4-9) since, as can be seen from the halogen-core spectra of the monohaloethylenes, the largest a (CX) term value occurs for C2H3I. Feature 3 has a term value of 5.67 eV (Table 5.1.9) relative to the C(CI) Is IP, which is similar to that reported (5.65 eV) for the C ls-> a (CI) transition in CH^l [62]. Similarly, the I 4d-» a (CI) term value in C ^ I (see Table 5.1.5) and CH^I are very close (6.0 and 6.1 eV respectively). Thus far, only the C(CX) Is •* a (CX) transitions have been considered in detail. Although the C(CH) Is •* a (CX) transitions are dipole-allowed under C g symmetry, they are expected to be less intense because of the localized nature of the C(CH) Is orbital, which results in poorer spatial overlap with the a (CX) orbital than in the case of the C(CX) Is orbital. Somewhat surprisingly, the C(CH) Is •* a (CX) transition is predicted to be located under the C(CX) ls-» 77 (CC) band in all four species by using the same respective term values as observed for the C(CX) Is -*a (CX) transitions in each of the C2H2X molecules. This result can be rationalized if it is realized that the energies of * both C(CH) Is and ir (CC) orbitals are not significantly affected by the halogen atom, » while the energies of both a (CX) and C(CX) Is orbitals are significantly affected, but in opposite directions. The a (CX) orbital is stabilized, but the C(CX) Is orbital is 165 destabilized in going from F to I. As a result, the absolute spectral energies, which are a measure of the energy difference between the originating and the terminating orbitals, are both decreased for the C(CH) ls-» a (CX) and C(CX) Is ->ir (CC) transitions for the C2H.JX molecules in going down the series from F to I. Such an accidental degeneracy of the two states may be removed by perturbing the system with another substituent In fact in the C Is spectrum of 1,1-difluoroethylene, reported by McLaren et al. [182], the second ir (CC) transition band is shifted to a higher energy of 289.6 eV. By a deconvolution analysis [182] of the spectrum, a feature (#2 in Ref. [182]) with a term value of 3.8 eV (note that the C(CX) ls-> a*(CF) term value in C ^ F is 3.96 eV) is believed to contribute to a new peak that appears [182] on the low- energy side of the * (shifted) ir (CC) band in C F ^ H j . Therefore, a study of the C Is spectra of the other, more highly substituted haloethylenes and related compounds such as haloethanes and haloacetylenes, would possibly clarify the existence of degenerate ir and a transitions in the monohaloethylenes. 5.1.2.3.3 ' Is Rydberg Transitions All features below the C Is IPs in Figure 5.1.6, except those assigned to ir and a transitions in previous sections, are attributed to Rydberg transitions originating from C(CH) Is or C(CX) Is orbitals. The assignments of these features for the monohaloethylenes are listed in Tables 5.1.6-9. The Rydberg transitions are heavily overlapped in most regions of the spectra and are expected to exhibit splitting due to the C g symmetry. The assignments given here have been tentatively made on the basis of the observed term values. An unusual feature of the C Is spectra is the prominent structure, due to transitions to d Rydberg orbitals converging to the C(CX) Is ionization limits, which have been identified for each monohaloethylene. No such nd Rydberg features can be clearly identified leading to the C(CH) Is limit, which is consistent with the situation in C^^A [206,207]. In particular in monofluoroethylene, the C(CF) ls-> 3d transition' includes three 166 sharp structures (features 17-19), each 0.18 eV apart In contrast the corresponding first members of d Rydberg series in other monohaloethylenes appear as single sharp peaks only. It should be noted that a vibrational spacing of 0.18 eV has also, been seen in the C ls-> 7r (CC) transitions of €2^2^- Therefore, the cause of the splitting observed in the 3d Rydberg feature may be vibrational excitation rather than the splitting of the degeneracy of the d Rydberg orbitals due to the nonspherical symmetry of the molecule. The peaks assigned as the first members of d Rydberg series in C2H3C1 (peak 17) and CjH^Br (peak 18) are particularly prominent In general the C(CH) ls-> Rydberg transitions for the monohaloethylenes resemble those of C 0 H 4 [206,207], with the p Rydberg feature being the strongest relative to the s and d Rydberg features. This, as noted above, is in sharp contrast to the situation for the C(CX) Is -> Rydberg transitions where the p Rydberg features are not prominent Stohr et al. [198] have given convincing evidence, from NEXAFS studies of a variety of organic species, that features assigned as C Is-* lowest np Rydberg also have significant C Is-* 0 (CH) character. Presumably this np Rydberg and 0 (CH) mixing is responsible for the apparent decrease in intensity of the C(CX) Is-* np transition relative to that of the C(CH) Is •* np transition because of the presence of only one C-H bond on the C-X carbon. In addition, it is interesting to observe that the intensity of the C(CH) ls-» np Rydberg transition relative to that of the C(CX) ls-> nd or ns Rydberg transitions increases in going from C2H.JF to C-^^^- These observations indicate that the addition of a halogen substituent has substantial effect on the nature of the molecular Rydberg orbitals. A theoretical calculation that can provide transition energies supplemented with oscillator strengths of the transitions, similar to the multichannel quantum defect calculations reported for core excitation of SC^ (see Ref. [210] and Chapter 7), would likely provide a better understanding of these observations. 5.1.2.3.4 C ls-> Continuum Features At least two features are observed in the C Is continuum of each 167 monohaloethylene. These features may be due to any of the following processes: (i) one electron promotions to the a (CH) or a (CC) orbitals; (ii) double excitations; (iii) shake-up continua; and (iv) shake-off continua. In a theoretical study of C 2 H 4 the -1 • lowest shake-up state (C Is , ir-* ir ) is predicted in the 6-7 eV region above the C Is edges [208]. The position of the first continuum feature in the C Is spectra of the monohaloethylenes relative to the average C Is IP drops from -2.1 eV in C 2 H 3 F to -1.4 eV in CjH^I, while there is a significant change from -7.0 eV to -2.8 eV for the second feature. Therefore, these two features seem to be located at too low an energy for shake-up continua (with the possible exception of C2H.JF) as well as for shake-off continua, which will begin at even higher energy than shake-up. A recent study of the C Is spectra of the various fluoroethylenes suggests [182] that these continuum features • • • could be one-electron transitions to orbitals with 0 characters (o (CF), a (CH) and a (CC) mixing) or (C Is •* ir , ir •* ir ) double excitation. However, no conclusive assignment could be made [182], even from a consideration of the intensity variation with degree of fluorination. In the following discussion (section 5.1.2.3.5), term values of these two features have been demonstrated to be correlated with the electronegativities of the halogen substituent The fact that the correlation slopes for these two features, and also that for the a (CX) transitions, are larger than those for the ir (CC) and Rydberg transitions suggests that there is significant a character in these two continuum features. In fact, in the present work a single broad maximum at similar term values has been observed in both the F Is spectrum of CjH^F (-2 eV, feature 4, Table 5.1.1) and the Cl 2p spectrum of C ^ C l (-3.3 eV, feature 14, Table 5.1.2). Therefore, the first two features observed in the C Is continua of the monohaloethylenes are tentatively assigned * * * as a resonances, designated as and in Tables 5.1.6-9. A third weak structure (feature 21) is observed at 12.3 eV above the C Is IPs of C 2 H 2 C I (note that a resonance feature occurs at a very similar term value in the Cl 2p spectrum of C 2 H 2 C I ; see Figure 5.1.2 and Table 5.1.2). This feature is tentatively assigned to the a (CC) shape resonance because its term value matches the expected 168 * • position of the a (CC) shape resonance calculated from the correlation equation of a (CC) term values with C=C bond length postulated by Sette et al. [58]. Similarly, the broad resonance (feature 24, Figure 5.1.6a) in C2H.jF may also be considered to have the same a (CC) origin. Shape resonances attributed to a (CC) final orbitals have also been identified in the ISEELS spectra of C 2 H 4 (feature 14, Figure 5.1.6e), fluoroethylenes [182] and fluorobenzenes [209] at similar term values. The interpretation of the ~300 eV * feature in CjH^ as a a (CC) shape resonance has been discussed in recent NEXAFS studies [211]. Such an interpretation is strongly supported by the observation that the band maximum of this feature shifts in accordance with the changes in C=C bond length when C 2 H 4 is adsorbed on different metal surfaces [198,212]. 5.1.2.3.5 Correlation of Term Values with Substitutional Effects The term value for the final state in a given transition can be thought of as the ionization potential of the excited state in a frozen orbital model. Therefore, correlation of term values in groups of related molecules is an extensively used method for spectral analysis [62]. In Figure 5.1.8 the term values of various series of corresponding transitions in the C Is ISEELS spectra of the monohaloethylenes are plotted as a function of the electronegativity of the halogen substituent For a given halogen substituent and a given final orbital, the term value used is the average of the very similar values observed for transitions from the C(CX) Is and C(CH) Is initial orbitals. A good correlation is observed in each case, as can be seen in Figure 5.1.8. The rapid decline of the a (CX) term value from C^H^I to C^H^F contrasts sharply with the constancy of the IT (CC) term value. This is a clear indication of the close relationship of the a (CX) orbital with the halogen substituent The indication from the correlation diagram (Figure 5.1.8) that C Is -»a (CF) and C Is •+ 3s transitions would be predicted to occur at approximately the same energy in C^H^F based on the trend of the term value above for C 2HjI, C 2HjBr and C ^ ^ C l strongly supports the present suggested assignment (see above) of overlapping a (CF) and 3s Rydberg transitions (i.e., 169 2 3 4 ELECTRONEGATIVITY Figure 5.1.8 Correlat ion between term v a l u e s for C 1s (average) excitations and the electronegativity of the halogen substituent in the monohaloethylenes . 170 Rydberg/valence mixing). It can be seen that the a (CX) stabilization increases on going down group VII (i.e., as the electronegativity decreases). The dominating property of the halogen in this case may be the electron capacity (i.e., polarizability), which increases with increasing atomic number [213]. The term values of Rydberg transitions vary more slowly (Figure 5.1.8). This can be understood in terms of the diffuse nature of Rydberg orbitals, which renders them • less sensitive to the halogen substituent On the other hand, the valence character of * and a 2 shape resonances results in larger substitutional effects on their term values, as indicated by a greater correlation slope (Figure 5.1.8). 5.1.2.3.6 Correlation of a*(CX) Term Values with C-X Bond Strength Robin [56,194] has shown that the term value corresponding to the final states in C Is-* a (CX) transitions can be correlated with the corresponding C-X bond strength in methyl halides (Figure 5.1.9, dashed line). Clearly, a similar correlation exists in the monohaloethylenes, as shown in Figure 5.1.9 (solid line). Since there are no published C-X bond strengths (D(R-X)) for the monohaloethylenes, the C-X bond strengths listed in Table 5.1.10 are estimated values based on a thermodynamical analysis of the following reaction: H\ / \ f n — ? H X H =C + -X 5.1 For this process D(C-X) = AHj(R.) + AH°(X-) -where R represent the vinyl group and the AHp(RX)) are obtained from thermochemical AH°(RX) f 5.2 heats of formation (AH°f(R-), AH°f(X-), and data [214,215]. C ( C X ) I S ^ C T * ( C X ) T E R M VALUE (eV) Figure 5.1.9 Correlation between C - X bond strength and C ( C X ) 1s-»o*(CX) term values in the monohaloethylenes (solid line) and methyl halides (dashed line, data taken from Ref. [194]). Table 5.1.10 C - X bond strengths and C ( C X ) 1s-»o*(CX) term values for the monohaloethylenes Molecule C-X Bond Strength (kcal/mole) o*(CX) Term V a l u e ( a ) (eV) C 2H 3F 118(2) 3.96 C 2H 3C1 86(2) 4.99 C 2 H 3 Br 74(2) 5.23 C 2 H 3 I 61(5) 5.67 (a ) T ( o * ( C X ) ) = 7 .48 - 0 .0297 D ( C - X ) w i th cor re la t ion coef f ic ien t = 0 .998 , s e e F igure 5.1.9. 173 The strong correlation observed is not surprising in the sense that as the C-X bond becomes stronger, the energy of the a(CX) bonding orbital is lowered (i.e., becomes more stable), while the a (CX) and-bonding orbital energy is raised. Therefore, the term value will strongly reflect the antibonding nature of the a molecular orbital, and thus also the strength of the relevant o bond [60,194]. The connection between bond strength and a term value has been discussed by Ishii et al. [60] and has been applied to locate low-lying inner-shell excitations to a virtual valence orbitals in molecules involving O-O and O-F bonds. Robin [56,194] observed a similar correlation of a term values with bond strength in the valence and core excitation of the methyl halides. However, in a study of the valence-excitation spectra of 02^11*, CjH^I and a variety of more highly substituted haloethylenes, Robin [56,194] claimed that the a band term values of the haloethylenes are quite constant irrespective of the type of halogen atom or the extent of halogenation. * This sharply contrasts with the present observations, in which the a term value for C Is excitation in the monohaloethylenes is strongly dependent on the identity of the halogen atom This difference in behavior could be related to different correlation and relaxation effects in the valence and core excitation. This problem will be investigated further in part II of the present work on the study of the valence-shell electron energy loss spectra of the monohaloethylenes. 174 5.1.3 Conclusion High resolution electron energy loss spectra of various inner shells below 740 eV of the four, monohaloethylenes have been measured. The spectral features are tentatively assigned in terms of excitations to a common manifold of virtual valence and Rydberg orbitalst. The halogen inner-shell excitation spectra are dominated by a (CX) transitions, • . .. while the ir (CC) transitions are apparently absent, or unresolved and of very low intensity. Various transitions observed in the C Is excitation spectra correlate well with the electronegativity of the halogen substituent. Based on the correlation analysis, substitutional effects are found to be most prominent for a -type bands, while the ir and Rydberg levels are less influenced. A linear correlation of the C-X bond strengths * and the term values of C Is •+ a (CX) transitions is also demonstrated for the monohaloethylenes. t. After completion of the present studies, work comparing low-resolution C Is ISEELS spectra and electron transmission spectra (ETS) has been brought to my attention [216]. While the general features of these C Is spectra [216] are consistent with the present, much higher resolution results, the assignments to ir* and a* states for C 2 H 3 I are reversed, and in addition for C 2 H 3 F the (unresolved) a* levels are predicted to be at too high an energy. These differences in interpretation probably arise as a result of assumptions made by Benitez et al. [216] in comparing the ETS and ISEELS spectra. It should be noted that the ETS spectra involve final states that correspond to the negative , ion rather than the neutral molecule. 175 AN INVESTIGATION OF SUBSTCTUTIONAL EFFECTS IN MONOHALOETHYLENES (C 2 H 3 X, X = F, CI, Br, I): II. PHOTOIONIZATION BY He(I) AND He(II) PHOTOELECTRON SPECTROSCOPY AND VALENCE-SHELL EXCITATION BY ELECTRON ENERGY LOSS SPECTROSCOPY 5.II.1 Introduction In part I of this, chapter, the core electronic excitation spectra of the monohaloethylenes (CjH^X, X=F, CI, Br and I) measured by the technique of inner-shell electron energy loss spectroscopy (ISEELS) are reported. In that work the interaction of the halogen substituent with the empty virtual valence orbitals was investigated. In particular it was found that the normally unoccupied ir orbital is highly localized on the C = C bond and not influenced to any significant extent by the halogenation. In part II of this study, it is mainly concerned with the effect of the halogen atom on the ionization or excitation of the occupied valence molecular orbitals (MO), as reflected in (i) the He(I) and He(II) photoelectron spectra (PES) and (ii) the low momentum transfer, valence-shell electron energy loss spectra (VSEELS) of these molecules. The interactions between the halogen atom and the normally occupied ir MO in the monohaloethylenes are of particular interest and can be summarized as follows [186,188,217,218]:. (a) A resonance (conjugation) interaction between the ir(CC) and the out of, plane halogen npj_ nonbonding orbitals. These two orbitals have the same symmetry(a"). As a result the 7r(CC) orbital will be destabilized or stabilized, depending on whether the binding energy of the halogen npj_ orbital is greater than or less than that for the ir orbital in ethylene (IP=10.51 eV [217]). Such resonance interaction would tend to decrease the nonbonding character in the halogen npj_ orbital and 176 result in a broad peak in the PES spectairn rather than a sharp peak characteristic of a purely nonbonding halogen npj_ orbital. Similarly the resonance interaction would be expected to cause a shifting of the 7r(CC) -» 7r (CC) transition energy in the VSEELS spectrum as the halogen atom is varied, (b) An inductive effect, which depends on the electronegativity of the halogen atom, and a resulting stabilization of the 7r(CC) MO. The electronic structure of monohaloethylenes has been the subject of numerous experimental studies by PES and also theoretical calculations [181,185,186,188-190,217-230]. These studies have all suggested that the the outermost valence molecular orbital in all four monohaloethylenes is of TT(CC) character and that the halogen np^ and np_j_ orbitals are located at higher binding energies. However an examination of the band shapes in the PES spectra and the fact that in the case of CjH^I the ionization energy of the iodine 5p nonbonding orbital (unlike the situation for CjH^F, CjH^Cl and C 2 H j B r ) is lower than and very close to that of the 7r(CC) orbital in ethylene suggests that the above generalized orbital assignment may in fact be reversed in the case of 0 2 ^ 1 . In particular in C 2 H 3 I a strong overlap (mixing) between the 7r(CC) and I 5p_|_ orbitals is likely to occur and the character of the resulting MO (nonbonding np or bonding TT) will be less well defined than in the other monohaloethylenes. Thus, a further motivation of the present work is to investigate the nature of the highest occupied molecular orbital (HOMO) in monoiodoethylene. This has been done in the present work by new measurements of the He(I) and He(II) PES spectra. The use of He(II) radiation is useful because comparison of the He(I) and He(II) intensity in a given molecule may permit differentiation between nonbonding and bonding orbitals of the halogen atoms on the basis of changes in the relative cross-section as a function of photon energy [231]. No previous He(II) PES measurements have been reported for C 2 H J I . Existing studies of the valence-shell electronic excitation of monohaloethylenes include UV and far-UV photoabsorption (at photon energies <10.5 eV) for all four monohaloethylenes [166-168,171-175], a semi-empirical MO-CI calculation for C~H,F 1-7T 1 / ' [179] and an SCF MO calculation for C ^ C l [180]. While electron energy loss spectra have been reported in the range below 16 eV for C ^ F [169,170,178] and CjHjCl [177]. These EELS measurements are at low impact energy and large scattering angle (i.e. large momentum transfer) and therefore include significant contributions from dipole forbidden transitions. Robin has given a comprehensive summary of existing work on the valence-shell electronic excitation of both the singly and multiply substituted haloethylenes [54-56]. VSEELS spectra at small momentum transfer (i.e. high impact energy and low scattering angle) are equivalent to photoabsorption measurements [6-9]. In the present work such VSEELS measurements up to ~28 eV energy loss (i.e. photon energy) have been used to investigated further details of valence-shell excitation processes in all four monohaloethylenes. If indeed a 'reversal' of the orbital ordering occurs in s u c h effects should also be manifested in the VSEELS spectra. A linear correlation between the C-X bond strength and the term values * corresponding to the C Is -*o (CX) transition has been demonstrated in part I of this chapter, the ISEELS work for the monohaloethylenes. However, in sharp contrast a study of the valence excitation spectra of C2H.jBr and C 2 H 2 I as well as a variety of more highly substituted haloethylenes the term values for the ir-* 0 (CX) and np •* a (CX) transitions were found to be fairly constant irrespective of the identity of the halogen or the extent of halogenation [54-56,194]. It is however noteworthy that a good correlation of bond strength with a (CX) term values in both the valence and the C Is excitation of the methyl halides has been observed [54-56,194]. Therefore, it is of interest in the present work to investigate whether such a correlation exists between the C-X bond strength and the term values for excitation to the a (CX) bands in the VSEELS spectra of the monohaloethylenes. 5.II.2 Experimental Methods The He(I) and He(II) PES spectra of monohaloethylenes have been measured by Dr. A. Katrib as part of the present collaborative study. These spectra were recorded on 178 a modified Perkin-Elmer Model PE-16 photoelectron spectrometer. The He(ll) radiation was obtained by using a quartz discharge tube, operating at very high current and low He pressure. The VSEELS spectra were obtained with the same high resolution electron energy loss spectrometer used to measure the ISEELS spectra reported earlier in part I of this chapter. The design and operation of this instrument have already been described in detail in Ref. [91] and Chapter 3. 5.II.3 Results and Discussion 5.II.3.1 Photoelectron Spectra He(I) and He(II) photoelectron spectra of the monohaloethylenes provided by Dr. A. Katrib are shown in Figure 5.11.1. While the He(II) spectrum of monoiodoethylene is the first such measurement, all other photoelectron spectra have been reported earlier at both He(I) and He(II) energies [185,186,188,217-227]. The present spectra are in. excellent agreement with the earlier work. Figure 5.II.2 shows more detailed spectra obtained in the present work for the first band of each of the photoelectron spectra of C 2 H j F , C2H.JCI, and CjH^I. The spectrum of the first band for C 2 H j B r in Figure 5.II.2 has been reproduced from the earlier work of Chadwick et al. [186]. The measured ionization potentials and vibrational frequencies are shown in Tables 5.II.1-4. As discussed in the introduction of the present paper all earlier PES studies have assigned the first band as effectively 7r(CC) in character for all four monohaloethylenes with the orbitals of principally halogen np^ and npj_ character being located at higher binding energies. The assignments for monofluoroethylene, monochloroethylene and monobromoethylene shown in Tables 5.II.1-3 are identical to those reported earlier [181,185,186,188-190,217-230]. However the present proposed assignments for monoiodoethylene shown in Table 5.II.4 are different from those reported in previous studies [217,220,221,226,227]. The band assignments in the present work to 7r(CC), np/7 and np_j_ are indicated on the various spectra shown in Figure 5.II.1. 179 i i i J 1 L 20 22 24 — 10 12 14 16 18 20 22 24 8 10 12 14 16 18 20 22 Hel?p" Tec J I I L J I I 10 12 14 16 18 20 22 24 Hen 7£C nPi npPI •°ca, J I I I L 8 10 12 14 Hell 16 18 20 22 J I i i i J L 10 12 14 16 18 20 22 24 8 10 12 14 16 18 20 22 IONIZATION ENERGY (eV) Figure 5.11.1 He(l) and He(ll) photoelectron spectra of C2H3F, C2H3CI, C2H3Br and C2H3I . 180 First band He I cnrr' m 850 520 I s F n l r f l l R T ! rTll r T T l First band He I H H x H CI PES cm 1300 118 h i h i h i h i h i 10 11 Rrst band He I H^ H X H X I PES cm-' 1260 p-rn r-300 1 1 1 1 i i i 10 IONIZATION ENERGY (eV) Figure 5.11.2 Detai led first band He(l) photoelectron spectra of C2H3F, C2H3CI, C 2 H 3 B r (taken from Ref. [186]) and C2H3I. 181 Table 5.II.1 Vertical ionization potentials and associated vibrational structures in the first band of the photoelectron spectrum of C 2 H 3 F Orbital Dorminant Ionization Energy Associated Vibrational Vibrational Character (eV) Structure (cm-1) Assignment 2a" n(CC) 10.57 7a' n(F 2p//) 13.80 6a' a(CF) 14.60 5a' 16.00 1a" n(F 2px) 16.70 4a' 17.90 3a' 20.10 1530 C=C stretch 1270 C - F stretch 850 C H 2 deformation 500 C-Fbend 182 Table 5.II.2 Vertical ionization potentials and associated vibrational structures in the first band of the photoelectron spectrum of C 2 H 3 C I Orbital Dorminant Ionization Energy Associated Vibrational Vibrational Character (eV) Structure (cm"1) Assignment 2a" Jt(CC) 10.16 1300 C=C stretch 930 C-CI stretch 550 C-CI bend 7a' n(CI 3p//) 11.64 1200 C H 2 deformation 560 C-CI stretch 1a" n(CI 3pJ 13.10 6a1 a(CCI) 13.60 5a' 15.40 4a' 16.40 3 a' 18.80 2a' -23 183 Table 5.II.3 Vertical ionization potentials and associated vibrational structures in the first band of the photoelectron spectrum of C2H3Br Orbital Dorminant Ionization Energy Associated Vibrational Vibrational Character (eV) Structure (cm"1)<a) Assignment*0) 2a" 7C(CC) 9.80 1340 C=C stretch 740 C-Br stretch 290 C-Br bend ' 7a' n(Br 4p//) 10.90 1200 CH2 deformation 460 C-Br stretch 1 a" n(CI 3pjJ 12.28 6a' a(CBr) 12.94 5a' 15.02 4a' 16.21 3a' 19.20 (a) From Ref. [186]. 184 Table 5.II.4 Vertical ionization potentials and associated vibrational structures in the first band of the photoelectron spectrum of C 2 H 3 I Orbital Dorminant Ionization Energy Associated Vibrational Character (eV) Structure (cm 2a" n(I 5 P l ) 9.26 1260 300 7a' n(I 5p//) 10.04 1160 1a" 7l(CC) 11.50 6a' a(CI) 12.30 5a' 14.60 4a' 15.60 3a' 18.50 2a' 19.70 Vibrational Assignment C=C stretch C-I stretch C H 2 deformation 185 The present reassignment of the monoiodoethylene PES spectrum is based on the following general observations as well as the considerations discussed in the introduction to the present work (see above). Firstly, it is clear from Figures 5.II.1 and 5.II.2 that the first band of the He(I) PES spectrum of monoiodoethylene is very different from that in the other three molecules. In particular the , first band in monoiodoethylene is much narrower (note the He(I) PES spectra are at higher resolution than the He(II) PES spectra) and this suggests a possible dominant non-bonding character rather than 7r(CC) character. In addition the first band of monoiodoethylene exhibits a much less complex vibrational structure than the first bands of C J H J X (X = F, Cl and Br). It can also be seen from Figure 5.II.1 that the intensity of the band assigned as np^ in each case decreases relative to the bands assigned as 7r(CC) and np_|_, in going from He(I) to He(II) photon energies. On the basis of these relative intensities the assignments as shown (Figure 5.II.1, Tables .5.11.1-4) • are. consistent with the fact that the expected interaction (mixing) between 7r(CC) and npj_ would result in these two bands having similar behaviour and being different from the np^ band. The similar broad shape of the 7r(CC) and np_L bands as well as narrower shape of the np^ bands, which are particularly noticeable in CjHjCl, C2H^Br and C2H3I, a r e a * s o consistent with the present assignments. The relative narrowness of the np^ bands is also in keeping with the non-bonding (atomic-like) character of this orbital. Robin [54-56] has discounted the possibility that the first two sharp bands in the monoiodoethylene PES spectra (Figure 5.II.1) are due to spin-orbit splitting because of the low symmetry of this molecule (Cg) which is expected to quench the spin-orbit coupling. As discussed below the present reordering of the PES assignments in monoiodoethylene is also consistent with the assignments and interpretation Of the VSEELS spectrum of monoiodoethylene (see section 5.11.3.2). The orbital ordering resulting from the present PES measurements, including the new reassignment of the monoiodoethylene spectrum as discussed above, is summarized in the correlation diagram shown in Figure 5.11.3. As has been pointed out earlier [168,188,190] the near equivalence of the 7t(CC) IP in C-EL and C , H , F indicates the the resonance (destabilizing) and inductive 186 > 9 >-CD c c 10 LLl LTJ II o 1 4 < 15 o h-LL I > 17 16 H2C=CH2 H2C=CHF H2C=CHCI H2C=CHBr H2C=CHI 7T(C=C) CT(C-H) X(^)a ' -ex(C-X)a'. , ' 7 X ( P ± ) a — ' I \ I I Figure 5.11.3 Energy levels correlation diagram for C 2 H 4 , C2H3F, C 2 H 3 C I , C 2 H 3 B r and C2H3I . 187 (stabilizing) effects (see introduction) are of the some order of magnitude and opposite in sign for C ^ F . The decrease in the 7r(CC) IP in going to C 2HjCl and C^H^Br can be attributed to the predominating resonance interaction as the electronegativity (and thus the stabilizing inductive effect) of the halogen atom decreases. In the case of CjH^I, it is important, to note that the IP of the 5p electron of the free iodine atom is 10.45 eV [217] which, unlike the situation for Br, CI and F, is less than the 7r(CC) IP of C 2H 4. As a result the reverse situation may be expected to occur i.e. a net stabilization of the 7r(CC) orbital in ^H^I and a destabilization of the interacting 5pj_ molecular orbital due to the resonance interaction. In addition the 7r(CC) ; will be further stabilized by the iodine inductive effect Thus in the case of C^H^I both the resonance and inductive effects are of the same sign and result in a significant stabilization of the 7r(CC) molecular orbital. 5.II.3.2 Valence-Shell Electron Energy Loss Spectra 5.II.3.2.1 General Considerations The VSEELS spectra for C ^ X (X = F, CI, Br and I) are shown in Figures 5.II.4—7 and summarized, along with tentative assignments, in Tables 5.II.5-8 respectively. Photoabsorption spectra of these molecules have only been reported up to an energy of ~10 eV [166-168,171-175]. The features observed in the VSEELS spectra obtained in the present work are consistent with the photoabsorption measurements in the common energy range. The VSEELS spectra can be interpreted as transitions involving excitations from the 7r(CC) orbital, halogen lone-pair and (predorninandy) C-X, C-H and C-C a orbitals to the unoccupied virtual valence and Rydberg orbitals. Low lying transitions to the 7r (CC) and o (CX) virtual valence orbitals have been observed in the inner-shell excitation spectra of the monohaloethylenes in part I of the present work and hence the VSEELS spectra of these molecules are expected to show similar manifolds of final states. Valence-shell excitation spectra are generally quite difficult1 to interpret because of the possibility of numerous overlapping transitions from various closely spaced outer-valence occupied orbitals. In contrast, the interpretation of inner-shell excitation spectra is generally 188 much more straightforward due to the wide energy separation of the various initial core orbitals. Accordingly, term values of Rydberg transitions observed in the inner-shell excitation spectra of the monohaloethylenes have been used as a guide to locate Rydberg transitions in the respective valence-shell excitation spectra. This is possible because the term values for Rydberg levels have been demonstrated to be transferable between ISEELS and VSEELS spectra [64-66]. In contrast, term values for virtual valence final states are in general not transferable between VSEELS and ISEELS spectra. However, further insight can be gained from a consideration of the term values of corresponding transitions for related compounds, such as ethylene and alkyl halides, which provide * * predictions of the expected positions of the transitions to the o (CX) and ir (CC) orbitals. With the foregoing considerations in mind tentative assignments of the VSEELS spectrum of each monohaloethylene was achieved as follows.' Once transitions from the highest occupied orbital (2a") to various virtual valence and Rydberg orbitals are identified, their term values are used to predict the positions of other expected valence to virtual valence and valence to Rydberg transitions from the remaining more tightly bound occupied orbitals. 5.II.3.2.2 VSEELS Spectrum of C ^ F Details of the VSEELS spectra of ...C^H^F are shown in Figure 5.II.4. Energies, term- values and possible assignments of all features in the VSEELS spectrum of ^iiyF are shown in Table 5.II.5. Since the first and second vertical IPs are well separated by 3.23 eV, features observed below the first IP (10.57 eV) can be considered as being largely due to excitation originating from the highest occupied 2a" orbital which is of mainly ir character (Table 5.II.1). The prominent broad band between 6.5 to 8.5 eV (maximum at 7.51 eV) in the VSEELS spectrum of monofluoroethylene (Figure 5.II.4) is , thus assigned as the 7r(2a")-> ir (CC) band typically observed in olefinic compounds * [54-56]. It is noteworthy that the ir-* ir (CC) band maximum for ethylene and multiply substituted fluoroethylenes is seen to be located in all cases in the range of 7.5-7.6 eV [54-56]. Sharp vibronic structures corresponding to the 7r(2a") -» 3s (features 1-6) and 189 00 LU h-£lO-_ J UJ R -oH a' IT' ! 7?? 1 0 , 2 , 4 1 6 , 8 , 9 2 | 2 2 24 26 I I I I I I M l I II II I | || | M I I I I | e = o ° Eo = 3000 eV AE = 0 0 3 5 eV 7 a' I—L. 8 3s T T 56 9 =8 2a" 10 IT* »•' I —L . 3 P y 4^a? 3d L_ i 5 3s T r 56 i 4 a ' 3s 3p t—I 1 — T T 56 I V y 3s T T 56 | 6 a ' r-L. 3s TT 56 = f 3 a ' ** ** ^ p y «Li9? 10 21 26 27 6 8 12 10 14 16 18 ENERGY LOSS (eV) 20 22 Figure 5.11.4 Long range and detailed (insert) V S E E L S spectra of C2H3F. Table 5.II.5 Energies, term values and possible assignments for features in the VSEELS spectrum of C 2 H 3 F Feature(">Energy(b> Term Valued (eV) Possible Assignment Loss(eV) 2a" 7a' 6a' 5a1 la" 4a' 3a1 2a" 7a 1 6a" 5a' la" 4a' 3a 1 1 6.998 3.57 2 7.181 3.39 3 7.353 3.22 — — — — — — 3s/a*(CF) — — — — — — 4 7.509 3.06 — — — — — — Jt*(CC) — — — — — — 5 7.649 2.92 6 7.795 2.77 7 8.090 2.48 8 8.277 2.29 — — — — — — > 3 P — — — — — 9 8.449 2.12 — — — — — — — — — — — 10 8.667 1.90 11 8.776 1.79 12 8.854 1.72 13 9.010 1.56 — — — — — — > 4s/3d — — — — — — 14 9.041 1.53 — — — — — — — — — — — — 15 9.182 1.39 16 9.353 1.22 — — — — — — > > 4p — — — — — — 17 9.447 1.12 — — — — — — — — — — 18 9.509 1.06 19 9.712 0.86 — — — — — — 5s — — — — — — 20 9.822 0.75 5p 21 9.993 0.58 3.81 — — — — — 6s/5d^ 22 10.180 — 3.62 — — — — — — — — — — — 23 10.352 — 3.45 — — — — — '3s/o*(CF) — — — — — 24 10.539 — 3.26 — — — — — — — — — — — 25 10.726 — 3.07 — — — — — — — — — — 26 10.9<c> — 2.9 — — — — — — n*(CC) — — — — — 27 11.4(c> — 2.4 3.2 — — — — — 3p 3s/o*(CF) — — — — 28 12.5<c) — 1.3 — 3.5 — — — — 4s — 3s/o*(CF) — — — 29 12.9<c> — — 1.7 — 3.9 — — — — 3d — 3s/o*(CF) — — 30 14.9(c) 3.0 JI*(OQ — 31 16.3W 1.6 3.8 3d 3s/a*(CF) 32 18.5W 1.6 3d (a) See Figure 5.II.4. (b) Estimated uncertainty is ±0.016 eV. (c) Estimated uncertainty is ±0.1 eV. (d) See Table 5.U.1 for IPs. 191 7r(2a")-» 3p (features 7-10) Rydberg transitions are superimposed on the lower energy and higher energy sides of the 7r(2a") -*ir band respectively. According to the term value range of 1.56-1.81 eV for 3d Rydberg transitions observed in the ISEELS spectra of C 2H.jF (Chapter 5.1), the 7r(2a") -»3d transition is believed to be overlapping the 7r(2a")-+ 4s transitions (features 10-15). Features 16-21 are readily assigned as transitions to higher members of the s, p and d Rydberg series originating from the 2a"(it) orbital according to the quantum defects of 1.05(6 g), 0.66(6 ), and 0.24(6^) derived from the transitions to the 3s, 3p and 3d Rydberg levels. In cis-C 2H 2F 2, trans- C 2 H 2 F 2 and C^HF^ it ->O (CF) bands having term values in the range of 3.4-3.6 eV have been identified as weak shoulders on the leading lower energy tail of the it -* it bands [54-56]. This leads to the expectation that the 7r(2a")-* a (CF) band will be located in the 6.8-7.0 eV energy loss region in the VSEELS spectra of monofluoroethylene. This prediction is also in keeping with the MO-CI calculation performed by Salahub [179] who found that the first transition in the * fluoroethylene series was of it-* a type located at energy range of 5.4-7.0 eV. However, superimposition of the weak 7r(2a") -+a (CF) band and the 7r(2a") -> 3s and * * 7r(2a")-> it (CC) bands in C 2H^F may obscure the' transition to the a (CF) orbital. It is also noteworthy that, in the ISEELS spectra of CjH^F, the o (CF) orbital is found to be mixed with the 3s Rydberg orbital (part I section 5.1.2.2). Presumably, such 3s/a (CF) Rydberg/valence mixing may also be occurring in the case of valence-shell excitation. Features located above the first IP (features 26-32) in the VSEELS spectrum of CjH^F are generally broader and correspond to excitations from the more tighdy bound orbitals. Line broadening is usually expected in these higher excited states because they can interact with the underlying continua and undergo rapid decay through autoionization and predissociation. Additionally, overlapping of structures is expected in this higher energy region because the more tightly bound valence orbitals in monofluoroethylene are more closely spaced. Manifolds of discrete transitions associated with the higher IPs are shown on Figure 5.II.4 using the relative energy positions of transitions originating from the 192 7r(2a") orbital as discussed above. Most of the observed broader features are attributed to transitions terminated at the 3s/a (CF) and 7t (CC) orbitals. In addition, some of these features are considered to also have contributions from transitions terminated at the Rydberg levels. 5.II.3.2.3 VSEELS Spectrum of C ^ C l The 7r(2a") •+ ir (CC) band in C ^ C l is located between 6-8 eV and has maximum intensity at 6.684 eV (Figure 5.II.5). This band is broad and structureless but has sharp vibronic structures associated with the 7r(2a") ->4s (features 2-4) and the 7r(2a")-> 4p (features 5-7) Rydberg transitions superimposed upon it, as observed in the first band of ^tl^F (Figure 5.II.4). The 7r(2a")-» 3d transition is believed to contribute to features 8 and 9 based on their term values of 1.71 and 1.61 eV. Energies, term values and possible assignments of all the features observed in the VSEELS spectrum of C2H.JCI (Figure 5.II.5) are summarized in Table 5.II.6. * The 7r(2a") -*ir band also has a small change of slope (feature 1) on the lower energy side of the peak profile (Figure 5.II.5). This shoulder can be seen more clearly and located at about 6.3 eV in the slightly higher resolution photoabsorption spectrum of C 2 HjCl reported by Berry [167] and discussed by Robin [54-56]. Similar shoulders have been observed in the spectra of other higher substituted chloroethylenes and they have been assigned as the ir(2a") -*o (CC1) bands. This parallels the observation in C2H^F that the lowest transition is the ir -» a transition and the same prediction is. obtained by SCF calculation for the chloroethylenes [180]. Additionally the term value of 3.9 eV obtained by assigning the shoulder (feature 1) at 6.3 eV as a 0 (CC1) band is close to the term value of 3.99 eV found for the o*(CCl) band observed in CHjCl [194]. Above the 7r(2a")-> IT (CC) band, a number of sharp peaks are observed up to 11 eV and they are assigned as s, p and d Rydberg series converging to the four lowest IPs (see Figure 5.II.5 and Table 5.II.6). In addition, valence -» virtual valence transitions are believed to be lying under these sharp features. In particular, transfer of term values io-t CO LU r -- o-LU _ l UJ c r o air'Ap 3d 545 Ll L. 4 4s n r 5 67 (a) 0 - V 4 p 3d545 fc , 4s 5 67 CT*7r'4p 3d545 ., 'I ' T^U-i4o' 5 67 H H H XI VALENCE crV'4p 3d545 4s 5 67 9 = 0 ° A Eo = 3000 eV \ A E = 0.041 eV • .„ „* X 0-V*4p 3d545 . . CT7r*4p 3d 545 t-l L T r 5 67 4s a'7T*4p 3d545 Ll L. 4s "5 67 II III II I I I III I I I I 2 5 8 101316 22 25 26 27 -1—1—1—1—1—1—1—1—r 5 10 15 (b) r r 4 p u 25 6a' • f ja" I 234 5 6 7 89 10 12 14 1516 17 1921 22 23 24 25 13 5 7 9 ENERGY LOSS (eV) Figure 5.11.5 (a) Long range and (b) detailed VSEELS spectra of C2H3CI. Table 5.II.6 Energies, term values and possible assignments for features in the VSEELS spectrum of C2H3CI FeatureWEnergy(b> Term Valued (eV) Possible Assignment Loss(eV) 2a" 7a 1 la" 6a' 5a' 4a" 3a' 2a" 7a' la" 6a' 5a' 4a 1 3a' 1 6.3<c> 3.9 _ a* (CO) 2 6.684 3.48 3 6.783 3.38 — — — — — — }>4s/7t*(CC) — — — — — — 4 6.981 3.18 5 7.576 2.58 "] 6 7.774 2.39 3.87 Y 4p a* (pa) 7 7.933 2.23 J 8 9 8.449 8.549 1.71 1.61 3.19 3.09 — — — — — >"} »4s/7l*(CC) 10 9.084 1.08 2.56 — — — — — 5 P -I 11 9.203 0.96 2.44 — — — — — 4d • 4p 12 9.322 0.84 2.32 3.78 — — — — 6s J 0*(CO) — — — — 13 9.547 — — 3.54 — — — — — 1> 7t*(CC) — — — • — 14 9.679 — — 3.42 3.92 — — — — o*(CCD — — — 15 9.997 — 1.64 3.10 — — — — — 3d 4s — — — — 16 17 10.235 10.513 — 1.41 1.13 2.59 3.37 3.09 — — — — 5s 5p -I 4s/Jt*(CC)_ _ — 18 19 10.700 10.831 — 0.94 0.81 2.40 2.27 — — — — — 4d 6s > 4p — — — — 20 10.950 — 0.69 2.15 — — — — — 6p , — — — — 21 11.046 — 0.59 — 2.55 — — — — 7s/5d — 4p — — — 22 11.5C) — — — — 3.9 — — — — — a* (CO) — — 23 24 12.0<d> 12.3<d> — — — 1.6 3.4 3.1 — — — — — 3d 1 *4s/re*(CC) — 25 12.9<d> — — — — 2.6 3.6 — — — — — 4p 4s/7t* (CC) — 26 14.9(d) 1.5 3.9 3d o*(cci) 27 17.1C) — — — — — — 1.7 — — — — — — 3d (a) See Figure 5.H.5. (b) Estimated uncertainty is ±0.022 eV. (c) Feature observed in VUV photoabsorption spectra [166,167]. (d) Estimated uncertainty is ±0.1 eV. (e) See Table 5.II.2 for IPs. 195 for the ir(2a")-» a*(CO) and 7r(2a")-> 7r*(CC) transitions places the la" -+o*(CC\) (feature 12), la'"-*ir*(CC) (features 13,14), 6a' -*o*(CC\) (feature 14) and 6a' -» TT*(CC) (features 16,17) transitions in the 9-11 eV region (Table 5.II.6) as indicated by the various, manifold shown on Figure 5.II.5. The presence of underlying valence •+ virtual valence transitions is also consistent with the apparent rising background and overall intensity distribution in this region. The higher energy region of the spectrum consists of mainly broad bands attributed to excitations of the deeper 5a', 4a' and 3a' levels, and a largely structureless continuum above 18 eV. No transition from the 2a' level can be identified. 5.II.3.2.4 VSEELS Spectrum of CjHjBr The VSEELS spectrum of C2HjBr is shown in Figure 5.II.6. Energies, term values and possible assignments of all the features observed in the VSEELS spectrum of 0 2 ^ ^ are summarized in Table 5.II.7. The intense broad band between 6 to 7 eV in the VSEELS spectrum of C ^ B r (Figure 5.II.6) is attributed to the 7r(2a") -• 7r*(CC) transition. A small shoulder (feature 1) is clearly seen at the low energy side of this 7r(2a")-> 7r (CC) band and it is tentatively assigned as the 7r(2a") -*a (CBr) transition. It is unfortunate that the photoabsorption spectrum of this molecule reported by Schander and Russell [172] did not extend to low enough energy to include this band. Based on this photoabsorption spectrum [172], Robin has assigned the band at 6.45 eV (assigned as 7r(2a")-+ f (CC) in the present work) as the 7r(2a") -*a (CBr) band [54-56]. However, as observed in the spectra of singly and multiply substituted fluoroethylenes and chloroethylenes [54-56], the ir -* ir (CC) band is always more intense than the ir -*o (CX) band because of better spatial overlap in the former case. Therefore, it is much more reasonable to assign the newly observed weak band (feature 1) at 5.7 eV as the ir(2a") -*o (CBr) transition and the much stronger band at 6.5 eV as the 7r(2a")-* it (CC) transition. It is ; noteworthy that in the photoabsorption spectra of 1,1-dibromoethylene, cis-dibromoethylene and trans-dibromoethylene [172] a band, assigned as the ir-* o (CBr) transition by Robin [54-56], is clearly observable in each case on 196 io->-H CO LU h-^ 0 I 7a' e = o° Eo = 3000 eV A E = 0.042eV H H X H Br VALENCE a- TT SPi.' -in® 5, _ l4a' i i II | 6 7 8 9 1 0 Vi-cr TT' 5Px' 6, 7,8, -ISl-SP,,! 7T* 5 p J " 5 s ? -id, 7'8l T TTTT-^ 6 7 8 9 1 0 KZ 1 I I II K 6 7 8 9 I 0 K * 3a' |5a 6 7 8 9 I 0 P * 1 16a TT" 8 10 4d| 5, 1 ^ ITTT v . 5 s 6 7 8 9 1 0 n — r 6 22 29 35 l a ' l r 38 39 40 ~TT 41 43 4 4 LU > < I 0 LU cr 5-0 8 r 10 12 14 16 18 (b) 5 P „ I 6. 7| 8| l 7T l U 5p[ 4d 6' I 5 ' s ^ | 1 6 1 I I I 7 8 9 10 1 7 8 9 ENERGY LOSS (eV) Figure 5.11.6 (a) Long range and (b) detailed V S E E L S spectra of C 2 H 3 B r . 197 Table 5.II.7 Energies, term values and possible assignments for features in the VSEELS spectrum of CaH^Br Feature^Energy**5) Term Value(d) (eV) Possible Assignment Loss(eV) 2a" 7a' la" 6a' 5a' 4a' 3a' 2a" 7a" la" 6a' 5a' 4a' 3a' 1 5.7«=> 4.1 — — — 0*(CBr) — — — — — — 2 6.449 3.35 -1 3 6.478 3.32 — — — - — — — r Jt*(co — — — — — — 4 6.566 3.23 J 5 6.778 3.02 -1 6 6.932 2.87 3.97 Y 5s a*(CBr) 7 7.167 2.63 J 8 7.439 2.36 3.46 — — — — — 5p//1 9 7.548 2.25 3.35 10 7.782 — 3.12 — 1 1 1 7.920 — 2.98 >• Ss 1 2 8.061 — 2.84 — — — — — — , — — — — — 13 8.127 1.67 4.15 4d — 1 . <J*(CB0 3.85 1 4 8.427 1.37 — — — — — 6s — J — — — — 15 8.668 1.13 2.23 5P// 16 8.735 1.07 — — — — — — 6p x — — — — — — 17 8.815 — 2.08 — — — — — — 5Pl — — — — — 18 8.926 — — 3.35 4.01 — — — — — \ • **<OQ C*(CBr) — — — 1 9 9.028 0.77 — 3.25 — — — — 7s — J — — — — 20 9.110 0.69 — — — — — — 7p — — — — — — 21 9.277 0.52 — 3.00 — — — — • 8s — 1 — — — . — 22 9.313 0.49 1.59 8p 4d } • 5s — — — — 23 9.431 0.37 — 2.85 — — — — 9s - J — — — — 24 9.556 0.24 1.34 — 3.38 — — — 10s -, — 1 — — — 25 9.673 — 1.23 — 3.27 — — — * 6s _ J — — — 26 9.761 — 1.14 J 27 9.834 — 1.07 2.45 — — — — — 6P// 5P// — — — — 28 9.878 — 1.02 — — — — — — 6p ± — — — — — 29 9.987 — 0.91 — 2.95 — — — — 5d — — — 30 10.025 — — 2.26 — — — — — — *Pl *• 6s — — — 3 1 10.128 — 0.77 — 2.81 — — — — 7s — — — — 32 10.245 — 0.65 — 2.69 — — — — 7p — V — — — 33 10.318 — 0.58 — — — — — 8s — — — — — 34 10.391 — 0.51 — — — — — — 6d — — — — — 35 10.523 — 0.38 — 2.42 — — — — 9s — 5P// — — — 36 10.618 — 0.28 — — — — — — 10s — — — — — 37 10.677 — 0.22 — 2.26 — — — — 1 Is — 5p ± — — — 38 11.5<c> — — — 1.5 3.6 — — — — — 6s Jt*(CQ — — 39 12.0<c> — — — — 2.9 4.2 — — — — — 5s a*(CBr) — 40 12.5<c) — — — — 2.5 3.7 — — — — — 5p Jt*(OC) — 4 1 13.8<c> — — — — 1.3 — — — — — — 6s — — 42 13.9<c> — — — — — 2.3 — — — — — — 5p — 43 14.5<°) 1.7 4.7 4d a*(CBr) 44 17.6<c> — — — — — — 1.6 — — — — — — 4d (a) See Figure 5.II.6. (b) Estimated uncertainty is ±0.015 eV. (c) Estimated uncertainty is ±0.1 eV. (d) See Table 5.II.3 for IPs. 198 the low energy side of the it-* it (CC) band. Superimposed on the higher energy side of the 7r(2a") it band (features 2-4) are three weak features (5-7) separated by 0.14 eV, corresponding to the frequency of the C=C symmetric stretch, decreased from its value of 0.20 eV in the ground state [232]. These features are assigned as 7r(2a")-»5s Rydberg transitions, corresponding to a vertical term value of 3.02 eV. The higher energy region of the VSEELS spectrum of C 2H 3Br can be interpreted in a similar manner to that discussed for C^H^Cl in the preceding section. Below the second IP, the" large number of sharp peaks in the 7-11 eV region of the spectrum, are assigned as s, p and d Rydberg series converging to the lowest four IPs (see Figure 5.II.6 and Table 5.11.7). Assuming transferability of term values for the 7r(2a") -* it (CC) (features 2-4) and 7r(2a")-» a (CBr) (feature 1) transitions places the 7a' -*o (CBr) (feature 6), 7a' -» TT'(CC) (features 8,9), 6a' -* o*(CBr) (feature 18) and 6a' it*(CC) (features 24,25) transitions in the 8—11 eV region. It is apparent that the Rydberg features (see above) are superimposed on these broader valence -> virtual valence transitions. In contrast, features observed above the second IP are generally broad and most of them are attributed to a (CBr) and it (CC) final states associated with transitions originating from the deeper 5a' , 4a' and 3a' orbitals. The "perfluoro effect" has been investigated by Schander and Russell [168], who compared VUV spectra of monobromoethylene and • monoiodoethylene with the perfluoro analogs, trifiuorobromoethylene and trifluoroiodoethylene. Fluorine substitution has been found to have little effect on it ionization potentials but it has increased the term values of 3s Rydberg transitions (i.e. stabilize the final state) with a symmetry [168] in keeping with the general expectations of the perfluoro effect [54-56]. However, in these compounds, there are two np Rydberg series : namely the in-plane np//(a') Rydberg series of o-symmetry and the out-of-plane np^a") Rydberg series of 7r-symmetry. Applying arguments based on the perfluoro effect Schander and Russell [168] concluded that the np Rydberg series of higher term value is the np//(a') Rydberg series (a symmetry) since the term value of this series showed a marked increase in magnitude 199 with fluorination. On the other hand little shift in term value with fluorination was observed on the second np Rydberg series of lower term value and it was therefore assigned as the npj_(a") Rydberg series (7r symmetry). Two np Rydberg series originating from each of 2a", 7a', la" and 6a' orbitals have been located (see Table 5.II.7) and accordingly the higher term value Rydberg series is labelled as the np^ series and the one with lower term value is labelled as the npj_(a") Rydberg series. 5.11.3.2.5 VSEELS Spectrum of C ^ I The VSEELS spectrum of C-2^£ *s s n o w n m F i g u r e 5.II.7. Energies, term values and possible assignments of all the features observed in the VSEELS spectrum of CjH^I are summarized in Table 5.II.8. The first observable structure in the VSEELS spectrum of &2^^ ' s a w e a ^ broad and structureless band (feature 1) with the band maximum at * 4.919 eV. Similar weak features, assigned as iodine 5p-» a (CI) have been observed in the far-UV spectra of iodo-alkanes and allyl iodide in the same energy region [171,233]. The observation of this same type of structure in the spectra of these iodine-containing molecules [171,233] both with and' without a ir system also strongly suggests that the highest occupied 2a" orbital in CjH^I has mainly iodine 5p character instead of 7r(CC) character, which is consistent with the assignment of the least tightly bound orbital in the PES spectra of Cj^^ a s i o a m e ^Pj. (see* section 5.11.3.1 on PES). Therefore, feature 1 is * » assigned as the iodine 5p_L(2a")-» o (CI) transition, which again places the a (CX) orbital lower than the ir (CC) orbital as is also observed in the other monohaloethylenes (see preceding discussion). The second structured broad band centered at 5.70 eV is attributed to the iodine 5pj_(2a") •+ 7r (CC) transition. The progressive shifting of the band maxima of the 2a" -> 7r*(CC) transition from 7.509 eV in C ^ F to 5.70 eV in C ^ I down the monohaloethylene series is consistent with the fact that the IP of the 2a" level increases across the series (see Tables 5.11.1-4) whereas the ir (CC) orbital energy is relatively constant as shown in the earlier ISEELS studies of these molecules (part I section 5.1.2.2). 200 io-o-( a ) 6"y i_ 6Pj| 5d 7 ' 6 I I I I ~l Ml g 6s 7 8 910 ti? 6p. i IL p l a ' I 6 p T ~ 5 d 7 T 6 1 U L 7 8 910 ^ I 7a ' 6p i 7 t II 1—I T 6p/ 5d 1 6 I , I I 6s 1 \ 7 6 910 12a 6p I 7, ,• T» 6p/ 5d 7' 6 I I , I , L. 6s II m - i 8 910 [g 5 a ' .C=C H VALENCE 6p | Jj \£A 6p^ 5d 7 ' 6 g 4 a ' 910 {g 6P.. L I L — I -6s 6f2 5jd 7 ' 6 ~ I III g 7 8 910 fg 6P„ I 7 i ;3a' 2a ' 6Pj| 5d 7 h—S^n-^ 8 910 Jg 6 a » ' 6pJ 5a 7 1 6 -i—j 1 i 'in 6s 7 8 910 g I 29 30 31 12 14 16 18 20 IO" a* 7T* 6p„ , 7, I 7 0 5d 7i i 6s 1 I I I 7 8 9 10 cr i _ 5 -x l O e = o° Eo = 3000 eV A E = 0.024eV l a " 27 28 4 5 6 7 8 9 ENERGY LOSS (eV) Figure 5.11.7 (a) Long range and (b) detailed V S E E L S spectra of C2H3I. 12 Table 5.II.8 Energies, term values and possible assignments for features in the VSEELS spectrum of C2H3I Feature(*)EnergyO>> Term Valued (eV) Possible Assignment Loss(eV) 2a" 7a' la" 6a' 5a' 4a' 3a' 2a" 7a' la" 6a' 5a' 4a' 3a' 1 4.919 4.34 cr*(0) 2 5.575 3.68 3 5.700 3.56 — — — — — — > 7l*(CC) — — — — — — 4 5.810 3.45 J 5 6.060 — 3.98 — — — — — — o*(CD — — — — — 6 6.232 3.03 7 6.279 2.98 8 6.342 2.92 9 6.467 — 3.57 — — — — — — 7t*(CC) — — — — — 10 6.920 2.34 — — — — — — 6 P / / — — — — — — 11 7.077 — 2.96 — — — — 6s — — — — — 12 7.202 2.06 — 4.30 — — — — 6p x — — — — — 13 7.616 1.64 2.42 — — — — — 5d 6 P / / 14 7.858 1.40 — 3.64 — — — — 7s — — — — 15 16 7.984 8.093 1.17 2.06 3.52 3.41 4.32 — — — 7p// 6pj. , 7l*(CQ 0*(CT) — — — 17 8.171 1.09 — 3.33 — — — — 7PX . — - — — — — 18 8.390 — 1.65 — — — — — — 5d — — — — 19 8.546 0.87 — 2.95 — — — — 6d — 6s — — — — 20 8.750 0.71 1.29 — 3.55 — — — 8s 7s — Jt*(CC) — — — 21 9.016 — 1.02 — — — — — — 7P1 — — — — — 22 9.156 — — 2.34 — — — — — — 6 P / / — — — — . 23 9.313 — 0.73 2.19 2.99 — — — — 8s 6p ± 6s — — — 24 9.500 — 0.54 — — — — — — 9s — — — — — 25 9.672 — 0.37 — — — — — — 10s . — — — — — 26 9.829 — 0.21 1.67 2.47 — — — — l i s 5d 6 P / / — — — 27 10.6 — — — 1.7 4.7 — — — — — 5d o*(d) — — 28 11.1 — — — — 3.5 4.5 — — — — — Jl*(CC) o*(CD — 29 12.1 3.5 rt*(cc) — 30 13.2 2.4 6 P — 31 14.2 4.3 o*(C3) (a) See Figure 5.II.7. (b) Estimated uncertainty is ±0.016 eV. (c) Estimated uncertainty is ±0.1 eV. (d) See Table 5.II.4 for IPs. 202 It is also noteworthy that the. relative intensity of the 2a" -» IT ( C C ) band decreases from C J H J F to CjH^I. Such a decrease in intensity may be a result of increasing halogen npj_ character in the 2a" orbital so that the spatial overlap with the final state is getting poorer compared with the pure IT-* IT situation. Moreover, as the IT character is decreasing in the 2a" orbital in going from ^ri^F to ^ r l ^ , there must be an increase of 7r character in the la" orbital. Taking the same term value of 3.56 eV as for the iodine 5pj_(2a") -* IT ( C C ) transition places the 7r(la")->7r ( C C ) transition under the intense sharp features 14-17 (Table 5.II.8). The region of the spectrum below the second IP (7a" at ~10 eV) exhibits transitions attributable to excitation of the four highest occupied levels. Four sets Of sharp intense structures (features 6-8, 11, 19 and 23) can be assigned as transitions to the 6s Rydberg level according to term values which are all within the range of 2.95-3.14 eV with respect to the appropriate IP. Similarly, p as well as d Rydberg series converging to these four IPs have also been identified using term value considerations (see Figure 5.II.7 and Table 5.II.8). Additional a ( C X ) and IT ( C C ) bands from the other occupied levels can be tentatively identified in the same manner as in the cases of the other monohaloethylenes by transferring the term values of corresponding bands originating from the 2a" orbital. The predicted bands match closely with the broad structures observed in the spectrum (see Figure 5.II.7, Table 5.II.8) A window resonance has been reported to be located at 9.35 eV by Robin [54-56] (i.e. between features 23 and 24). If this was the situation, the window resonance observed would be the result of interference of the 2a" continuum with a discrete transition originating from a more deeply bound orbital and it is required that the symmetry of this discrete transition is the same as that of the 2a" continuum states (i.e. a" symmetry) [54-56]. However, all possible discrete transitions that fall in this energy region (i.e. 7a' -» 8s(a'), 7a' -* 9s(a'), la" ->6pj_(a") and 6a' -*6s(a')) as shown in Table 5.II.8 have a final state symmetry of a'. Therefore, it is considered that the assignment of the dip between features 23 and 24 as a window resonance [54-56] is 203 questionable. Furthermore, it should be noted that the present spectrum (Figure 5,11.7) in this region can equally well be interpreted as having peaks (features 23-26) which fit the 7a' •* ns series (n = 8-ll) as shown in Table 5.II.8. In summary, it can be seen that the present interpretation of the VSEELS spectrum of monoiodoethylene is consistent with the PES results (section 5.II.3.1) which indicate that the 2a" HOMO orbital is of predominantly iodine 5pj_ character rather than of 7r(CC) character. 5.II.3.3 Comparison of the ISEELS and VSEELS Spectra of the Monohaloethylenes The term values for transitions to the 7r (CC) and a (CX) virtual valence orbitals as well as those for transitions to the first members for the ns, np and nd Rydberg series observed in the ISEELS and VSEELS spectra of the monohaloethylenes are summarized in Table 5.II.9. In general, the term values in the VSEELS are lower than those for the corresponding ISEELS transition and the decrease is most significant on transitions to the virtual valence orbitals. The term values are decreased by ~3 eV for 7r (CC) transitions, ~1 eV for a (CX) transition (except C 2H 3F) and <0.5 eV for most Rydberg transitions. Such a lowering in term values can be explained, in particular in the case of transitions to virtual valence orbitals, since the loss of shielding in a core excitation is larger than that occurs in a valence excitation. A further factor contributing to the lowering in term values for the valence -» virtual valence transitions is due to the much larger electron interaction (correlation) between the excited electron and the valence hole in valence -> virtual valence transitions as compared to that between the excited electron and the lower-lying localized core hole in core •* virtual valence transitions. In this regard, the larger lowering in term values observed for the n (CC) transitions than that for the corresponding o (CX) transitions is consistent with the expected larger electron interaction (correlation) experienced by the excited electron in the more * delocalized 7r (CC) orbital. Table 5.II.9 Summary of term values for assignments of ISEELS and VSEELS spectra of the monohaloethylenes Molecule Initial Term Value (eV) orbital 7C*(OQ o*(CX) ns<b> np<b> md ( b ) C ls<a> 6.33 3.99 3.99 2.59 1.56 C 2H 3F F Is ' — 4.36 — 2.61 — Valence^ 3.0 3.6 3.6 2.4 1.6 C ls<a> 6.32 4.98 3.62 2.93 1.63 C 2H 3C1 Cl 2p • — 5.19 3.45 2.61 1.73 Cl 2s — 5.4 — 2.4 — Valence^ 3.4 3.9 3.3 2.6 1.6 C ls<a> 6.33 5.23 3.92 2.53 1.72 C 2H 3Br Br 3d ' — 5.4 3.3 2.3 1.6 Br 3p — 5-1 3.0 — — Valence ( a ) 3.5 4.1 3.0 2.3 1.6 C ls<a> 6.43 5.70 3.76 2.77 1.95 C 2H 3I I 4d — 6.0 3.6 2.4 1.8 Valence ( a ) 3.5 4.3 3.0 2.3 1.7 (a) Averaged term value where applicable. (b) n=3,4,5,6 for the first members of the ns and np and m=3,4,5 for the first members of the md Rydberg series in C2H3F, C2H3CI, C^^Br and C2H3I respectively. 205 In parallel with the observations of the ISEELS study in part I of the present work, the variations in term values for various types of transitions observed in the VSEELS spectra as the halogen substituent is changed (i.e. the substituent effects) are most prominent in the case of a (CX) transitions, while the ir (CC) and Rydberg transitions are much less influenced. However, it should be noted that the variations in term values observed in the VSEELS spectra are larger than those observed in the ISEELS spectra for _ the 7r*(CC) (0.5 eV in VSEELS, 0.11 eV in C Is) and ns Rydberg (0.6 eV in VSEELS, 0.37 eV in C Is) transitions. This again may be considered to be a result of the larger electron interaction between the excited electron in the n (CC) orbital or in the ns Rydberg orbital (the most penetrating Rydberg orbital) and the valence hole as compared to the situation in the C Is core excitation. 5.II.3.4 Correlation of a (CX) Term Value with C-X Bond Strength In part I of this chapter, a linear correlation between the C-X bond strength and the term values corresponding to the C Is -*a (CX) transitions in the ISEELS spectra has been demonstrated. In Figure 5.II.8 (solid line), the respective term values of the observed 2a" -» a (CX) bands in C2H.JCI, Q,^^1 and ^2^3* a r e P l o t t e d a S a m s t m e calculated C-X bond strengths of these molecules as reported in part I section 5.1.2.3.6. All the 2a" •* a (CX) term values and C-X bond strengths used, as well as the resulting linear correlation equation and correlation coefficient, are summarized in Table 5.II.10. It is clear that a similar linear correlation with the C-X bond strengths exists for the valence excited a (CX) bands as for the corresponding core excited bands (Chapter 5.1 section 5.II.2.3.6). A similar observation of the correlation of a term values with bond strength in both the valence (dashed line, Figure 5.II.8) and core excitation (dashed line, Figure 5.1.9) of the methyl halides has been reported [54-56,194]. It is noteworthy that, according to the correlation equation for the valence-shell of the monohaloethylenes, the term value of the 7t(2a") ->0*(CF) band would be 3.32 eV (see Table 5.II.10) which is * reasonably close to the term values of 3.4-3.6 eV for TT •+ a (CF) transitions observed in studies of more highly substituted fluoroethylenes [54-56]. Additionally, this predicted term _0) E 120 o u f 100 o IZ UJ " N \ C 2 H 3 F C H , F ^ 80 CO Q O m x i 6 0 - C2H3X — CH3X N . C H 3 C I C 2 H 3 C I \ \ C 2H 3Br% CH3Br 3.0 4.0 5.0 :(CX)TERM VALUE (eV) Figure 5.11.8 Correlation between C - X bond strength and o*(CX) term values in the va lence excitation spectra of monohaloethylenes (solid line) and methyl halides (dashed line , data taken from Ref. [194]). Table 5.II.10 C - X bond strengths and 2a"-»a*(CX) term values for the monohaloethylenes Molecule C-X Bond Strength*") o*(CX) Term Value ( b ) (kcal/mole) (eV) C 2 H 3 F 118(2) (3.32)(£) C 2 H 3 C I 86(2) 3.9 C 2 H 3 B r 74(2) 4.07 C 2 H 3 I 61(5) 4.34 (a) See Chapter 5.1 section 5.1.2.5. (b) T(o*(CX)) = 5.40 - 0.0176 D(C-X) with correlation coefficient = 0.994, see Figure 5.II.8. (c) Estimated value by equation in (b). 208 value (3.32 eV) places the o (CF) bands in the same energy region as the 3s Rydberg bands as was suggested above in the VSEELS spectrum (Figure 5.II.4, Table 5.11.5) and also in the study of ISEELS spectrum of ^ ii^F (part I section 5.1.2.3.5). In an earlier study of the valence-shell photoabsorption spectra of C2H2Br, ^ H^I and a variety of multiply substituted haloethylenes, the rr ->o (CX) band term values were found to be reasonably constant irrespective of the type of halogen or the extent of halogenation [54-56,194]. However, this apparent lack of correlation is not surprising in view of the range of molecules included in the comparison reported by Robin [54-56,194] and the small correlation slope of -0.0176 eV kcal -* mole found in the present study for a given series (i.e. monohaloethylene) (see Table 5.II.10). 5.II.4 Conclusions The He(I) and He (II) PES spectra and the VSEELS spectra of the monohaloethylenes have been obtained in order to clarify the nature of the outermost orbital in C ^ I . In contrast to previous studies [181,185,186,188-190,217-230], the HOMO orbital in ^ H^I is reassigned as predominantly of iodine 5p_j_ character rather than the 7r(CC) character observed for the HOMO orbitals of C J H J F , C 2H 3C1 and C ^ B r . Such a reassignment is supported by the observations that the first band in the He(I) PES spectra of C^H^I is sharper than and has a different vibrational profile from those of the other three monohaloethylenes. In addition, a comparison of the relative intensity changes between He(I) and He(II) PES spectra as well as observations in the VSEELS spectra also support the above reassignment This phenomenon can be understood, in the case of CjH^I, by considering that both the inductive and resonance effects are stablizing the 7r(CC) orbital whereas these effects are acting opposite to each other in the cases of C^^F, CjH^Cl and CjH^Br. Therefore, it can be seen that the halogen substituent has considerable influence on the normally occupied 7r(CC) orbital. This is in sharp contrast to the negligible substitutional effects experienced by the normally unoccupied IT (CC) virtual valence orbital as observed in the ISEELS work in part I. This difference in 209 behavior is probably a direct reflection of the very different respective energy separations * of the TT(CC) and ir ( C C ) orbitals from the halogen valence np orbital. The VSEELS spectrum of each monohaloethylene, like the corresponding ISEELS spectrum, is tentative assigned to a common manifold of ir ( C C ) and a ( C X ) virtual valence and Rydberg levels in a unified manner. The term values for various types of transitions observed in the VSEELS spectra are, in general, lower than those for the corresponding transitions in the ISEELS spectra as a result of a smaller loss of shielding and a larger' electron interaction in the valence-shell excitation. Finally, a linear correlation between the term values of the 2a" ->o ( C X ) transitions and bond strength of the relevant C - X bond is demonstrated. l 210 CHAPTER 6 A CASE STUDY IN CO-ORDINATION CHEMISTRY : INNER-SHELL AND VALENCE-SHELL ELECTRONIC EXCITATION OF NICKEL TETRACARBONYL BY HIGH RESOLUTION ELECTRON ENERGY LOSS SPECTROSCOPY 6.1 Introduction Transition-metal carbonyl complexes have been the subject of a large amount of spectroscopic measurement [234-248] and theoretical investigation [235,240,246,249-259] due to their high photochemical and catalytic activity [260-265]. They have been used as prototype models for the bonding of CO to transition-metal surfaces [243,266-268], and serve as model systems for many organometallic complexes. An understanding of the electronic structures of transition metal carbonyl complexes is thus of great importance to diverse areas of practical importance. Although Ni(CO)4 is the simplest example of the tetrahedral metal carbonyl species, its valence-shell photoabsorption spectrum has received little attention, and to date no core excitation spectra have been reported. Previous photoabsorption studies of Ni(CO)^ [239,240] have been limited to excitation energies below the quartz cut-off frequency (~7 eV), and even within this limited region the spectrum is not well understood. Valence-shell [234,235] and core-level [236,237] photoelectron spectra of Ni(CO)4, and the corresponding ionization energies have, however, been reported. Unlike the valence-shell electronic excitation spectrum, the C Is and O Is inner-shell spectra of carbon monoxide exhibit very strong resonantly enhanced features, * * namely the dominant ls-» TT resonances below the respective Is edges, and the Is -*o resonances which lie in the continuum [75,76,207,269]. In contrast the Is -•Rydberg transitions make only a relatively small contribution to the spectrum. The resonant 211 enhancement is caused by interaction of the outgoing core electron with the anisotropic molecular field in selected inner-shell excitation channels. This interaction results in strong local concentrations of oscillator strength. Quantitative theoretical interpretations of these core spectra include the MO treatment of Rescigno et al. [270] and the MS - X o calculations of Dehmer and Dill [271]. In the MO picture core electrons are excited to the empty IT , o and Rydberg levels, while in the MS - X a approach the resonant enhancements appear as phase shifts in the Jt = 2 (IT ) and A = 3 (a ) outgoing partial waves, and are often referred to as d-wave and f-wave resonances respectively. Such strong core excitation resonances are a commonly observed feature of inner-shell excitation and photoabsorption spectra in a wide variety of "molecules [5]. In addition to these interesting features, inner-shell spectra are potentially useful because they are easier to interpret than their valence-shell counterparts due to the unambiguous assignment of the initial (core) orbital. This is in marked contrast to the typical valence-shell situation where unequivocal assignment of the transitions is often complicated by the many closely spaced occupied valence orbitals. With the above considerations in mind it is of interest to investigate the applicability of ISEELS methods to a study of transition metal complexes. In particular the metal-ligand (dir •* pir) back-donation envisaged in models of the bonding in • transition metal carbonyl complexes involves receptor orbitals that correspond to the ir orbitals of free CO. If localized ir and a resonances of the type observed in CO also occur in metal carbonyls then a study of the energies and intensities of these spectral features should provide a sensitive probe of the metal-ligand bonding. Also the high energy resolution uniquely available in the ISEELS technique in the soft X-ray equivalent photon energy region should provide further information through analysis of the vibrational structure in the C Is -* IT spectra of the metal carbonyls. With the above aims of further elucidating the electronic structure of Ni(CO)^ and its highly excited electronic states, electron energy loss spectroscopy have been utilized to obtain the first reported core level (C Is, O Is and Ni 3p) electronic spectra of gaseous 212 nickel tetracarbonyl at vibrational resolution under experimental conditions .where electric dipole transitions dominate the spectra [6-9]. The high resolution valence-shell excitation spectrum is also obtained over an extended energy range. 6.2. Results and Discussion 6.2.1 Electronic Structure of Ni(CO) 4 The electronic structure of Ni(CO) 4 has been the subject of many theoretical studies (for example [235,240,249-252,255,259]). There is still not full agreement on the exact ordering of the valence energy levels, but the qualitative details of metal-ligand bonding are well understood. A qualitative MO diagram indicating the occupied and unoccupied orbitals of Ni(CO)^ is given in Figure 6.1. The orbital ordering of the occupied valence levels is based on the calculation of Ref. [249]. In Ni(CO) 4 the Ni 3d orbitals are split by the tetrahedral ligand field into a higher energy triply degenerate (9t^) and a lower energy doubly degenerate (2e) set of orbitals, while the CO 5a and Iff orbitals are split into a manifold of levels with an energy spread ~4 eV [234,235] as illustrated in Figure 6.1. The CO 4a orbitals are largely unaffected by complex formation and produce the very closely spaced 6^ and 7a^ levels. The metal-ligand bonding is considered to be generated through synergic CO Ni a electron donation and Ni 3d •* CO rr back donation. The relative importance of the ligand -» metal and metal -» ligand interaction in bond formation has been the subject of considerable controversy [235,240,249-252,255,259]. For ease of reference in the following discussion, the group of lowest energy unoccupied MOs (the 2tj, lOt, and 3e) will be referred to as the Ni(CO)^ rr orbitals, since they are largely CO ir in character. Similarly the manifold of levels derived from the CO 5a and lrr orbitals (the le, ll^ Sa^, ltj and 8L,) will be referred to as CO (5a + lrr). 213 Ni(C0) 4 MO DIAGRAM Q -UJ CL ZD O O O -zl ZD cr' ( C-0) P O S S I B L E O - ' ( N i - C ) 6CT (a* ) / / / / / / / / / C O N T I N U U M / / / / / / / / / C O N T I N U U M / / / / / / / / / R Y D B E R G S R Y D B E R G S R Y D B E R G S 4p 4s l i t? 9a, 3e IQtg Tt7 27T (TT* ) UJ O UJ _1 < > -UJ cr o o 3d _2s_ 4 \ 8a, \ \ 7 t 2 x * le 6 t 2 . 7a, . 3 t 2 4a, 2 t 2 3a, 9ta 2e 8t2 . . -» •> ' 5<T ITT 4CT 2cr (C Is) to- (0 Is) Ni Ni(CO), CO Figure 6.1 Qualitative molecular orbital energy level diagram for Ni(CO)4. 214 6.2.2 C Is and O Is Excitation Spectra Figures 6.2 and 6.3 show C and O Is excitation spectra of Ni(CO) 4 respectively, obtained with 3 keV incident electrons at a modest resolution of 0.305 eV FWHM. In each case the corresponding spectrum of free CO is also given on the same energy scale in order to facilitate comparisons between the spectra of the metal complex and the free ligand. The energies, term values and proposed assignments for the numbered features in these spectra are given in Tables 6.1 and 6.2. It is immediately obvious that the C and O Is spectra of nickel tetracarbonyl show some obvious similarities with the respective corresponding spectra of free CO, and several of the features may be assigned by analogy. Clearly the counterparts of the below edge ir and above edge a resonances of CO [75,76,207,269] are present at quite similar corresponding energies and intensities in both the C Is and O Is Ni(CO)^ ISEELS spectra (Figures 6.2 and 6.3). The overall degree of similarity is particularly striking for the O Is spectra. There are also however some noteworthy differences in the region between the ir and a resonances particularly in the C Is spectra. In contrast the O Is spectra of Ni(CO) 4 and CO are much more similar in overall shape although as expected the Rydberg structure is less well resolved in the larger Ni(CO) 4 molecule. It is also noticeable that the relative contribution of Rydberg states is significantly less in the O Is spectra of both Ni(CO) 4 and CO. The most intense peaks in the Ni(CO) 4 spectra occur at approximately 6.2 eV below the respective C Is and O Is ionization thresholds [236,237], and may be confidently identified with Is -*ir (LUMO) transitions in Ni(CO) 4. It should be noted (Tables 6.1 and 6.2) that the C Is and O Is ionization energies for the metal complex are both ~2.3 eV lower than for free CO. This has been shown to be due to greater orbital relaxation in Ni(CO) 4 accompanying ionization, rather than any significant differences in the Is orbital energies [268,272]. Figure 6.4b shows a high resolution (0.068 eV) scan of the C Is •* ir peak in 215 (a) i.o A .5H t w i.o _ J UJ .51 .9 = 0 ° E 0 =3000eV AE = 0.305 eV 77"* CT* s h a k e - u p C 1s edge (b) n i — r 23 4 5 6 TT* cr* <C\s edge CO Cis x 8 Ni(C0)4 Cis x 4 2 3 45 6 7 8 o1 280 290 3 0 0 310 320 ENERGY LOSS (eV) 330 Figure 6.2 C 1s ISEELS spectra of (a) CO; (b) Ni(CO)4. 216 530 5 4 0 5 5 0 5 6 0 5 7 0 5 8 0 ENERGY LOSS (eV) Figure 6.3 O 1s ISEELS spectra of (a) CO; (b) Ni(CO)4. Table 6.1 Energies, term values and possible assignments for features in the C 1s spectrum of Ni(CO)4 and CO(a) NI(CO) 4 Feature Energy Loss Term Value Possible Assignment (eV) (eV) (Final Orbital) 1 287.61(3) 287.84(3) 288.08(3) 6.17 5.95 5.72 Jt* (v=0) 71* (v=1) Jt* (v=2) 2 289.34(3) 4.44 Ni 4p, 4s 3 290.21(3) 3.57 CO 3so; delocalized n* 4 291.71(12) 2.07 Ni 5p, 5s Rydberg 5 292.19(12) 1.59 Rydberg 6 293.75(20) 0.03 Rydberg Is edge 293.78<b> 7 296.2(4) -2.4 Double excitation; -?=3 o*(Ni-C) shape resonance 8 300.2(4) -6.5 Double excitation -f=3 a*(Ni-C) shape resonance 9 304.4(4) -10.7 =^3 o*(C-0) shape resonance C O Feature Energy Loss Term Value Possible Assignment (eV) (eV) (Final Orbital) 1 287.40(2) 8.70 2n* (v=0) 287.65(1) 8.45 2JC* (v=1) 287.89(3) 8.21 2jt* (v=2) 2 292.54(5) 3.5 3so 3 293.52(5) 2.6 3pjt 4 294.9(1) 1.2 4s, 3d C 1s edge 296.1<c> 5 300.8(1) -4.7 Double excitation 6 304.0(4) -7.9 =^3 o*(C-0) shape resonance (a) From Ref. [75], calibrated using Ref. [93]. (b) From Ref. [236]. (c) From Ref. [87]. Table 6.2 Energies, term values and possible assignments for features in the O 1s spectrum of Ni(CO)4 and CO( a) N i ( C O ) 4 C O Feature Energy Loss (eV) Term Value (eV) Possible Assignment (Final Orbital) Feature Energy Loss (eV) Term Value (eV) Possible Assignment ' (Final Orbital) 1 533.93(12) 6.18 it* 1 534.21(9) 8.2 2TX* 2 536.51(12) 3.6 CO 3sa; delocalized it* 2 3 538.90(8) 539.88(8) 3.5 2.5 3sa 3prc 3 538.61(12) 1.5 Rydberg 4 541.0(1) 1.4 4s, 3d O 1 s edge 540.1 l( b) O 1 s edge 542 .4 ( ( ° ) 4 542.9(4) -2.8 -*=3 a*(Ni-C) shape resonance 5 546.0(40) -5.9 -*=3 a*(IMi-C) shape resonance 6 0 550.07(40) -10.0 -^ =3 a*(C-0) shape resonance 5 551.0(1) -8.6 /=3 a'(C-O) shape resonance (a) From Ref. (75], calibrated using Ref. [93). (b) From Ref. [236]. (c) From Ref. [87]. 219 Ni(CO) 4. A spectrum of the corresponding transition in free C O at the same resolution is shown direcdy above in Figure 6.4a for comparison. The latter spectrum exhibits vibrational structure consistent with earlier high resolution ISEELS work [75,207,269]. The intensity distribution in the partially resolved structure in the high resolution spectrum of Ni(CO) 4 (Figure 6.4b) is not consistent with that expected on degeneracy grounds (~3:3:2) for the three possible it final electronic states (2t^, lOt^ and 3e - see Figure 6.1). It is therefore more likely that these features are vibronic components of a single electronic transition. Also illustrated in Figure 6.4 are least squares fits of 3 equal width Voigt profiles [273-275] (Gaussian/Lorentzian mix) to the experimental data. The parameters of these fits for CO and Ni(CO) 4 are given in Tables 6.3 and 6.4 respectively. The best fits are obtained when the peaks . have regular spacings of 0.245 ±0.007 eV (1980±55cm _ 1) for CO and 0.235+0.007 eV (1900+55cm"1) for N i ( C O ) 4 . The vibrational spacing for CO is in good agreement with previous work (2040cm-1 [75]). * The fact that apparently only one (and not three) Is •+ IT band is observed for Ni(CO) 4 can be understood in terms of the lowering in symmetry from tetrahedral in the ground state to C j y in the Is excited state, due to the localized nature of the C Is (or O Is) hole. SCF-LCAO-MO calculations for C Is ionization in Ni(CO) 4 [250,272] show that this leads to a doubly degenerate set of tr orbitals (e symmetry in C^v) becoming localized on the CO group containing the core hole, while the remaining six IT orbitals are delocalized over the remaining ligands and have almost zero contribution from the * C Is ionized CO. The presence of a core hole also results in the localized IT orbitals having a substantially lower energy (~2.2 eV [272]) than the others. Similar considerations are expected to apply in the case of C Is excitation, and also O Is ionization and excitation. Thus, from considerations of spatial overlap, the probability of transitions to the localized IT orbitals will be much greater than for excitation into the other TT M O S delocalized over the 3 neutral ligands, and transitions to the delocalized IT orbitals would be expected to be located ~2 eV above the localized ls-» IT band for both C Is and O Is excitation. By analogy with the situation in free CO it might therefore also be expected that resonant enhancement [75,76,207,269-271] of the C Is-* IT transitions would 220 287 288 289 ENERGY LOSS (eV) Figure 6.4 High-resolution ISEELS spectra of the C1S-»JI* transitions in (a) CO; (b) Ni(CO)4. The dashed lines show fits of 3 Voigt profile curves to the experimental data (see Tables 6.3 and 6.4 for details). Table 6.3 Parameters of the least squares fit of the C 1s-»n* transition in the C 1s ISEELS spectrum of CO Peak No. Position FWHM Relative Gaussian/Lorentzian (eV) (eV) Area Area Ratio 1 287.40(a) 0.175 1.0 0.691 2 287.65 0.175 0.153 0.691 3 287.89 0.175 0.015 0.691 (a) From Ref. [93]. Table 6.4 Parameters of the least squares fit of the C 1s-»n* transition in the C 1s ISEELS spectrum of Ni(CO)4 Peak No. Position FWHM Relative Gaussian/Lorentzian (eV) (eV) Area Area Ratio 1 287.61 0.232 1.0 0.889 2 287.84 0.232 0.370 0.889 3 288.08 0.232 0.083 0.889 222 also occur in the case of the localized it orbitals in Ni(CO) 4. This resonant enhancement in both Ni(CO) 4 and free CO is clearly present in the C Is and O Is spectra (Figures 6.2 and 6.3). This, together with the fact that the respective Is -» it and Is-* a transition energies are almost identical for CO and Ni(CO)^ in both the C Is and O Is spectra provides convincing evidence of the localized character of both the core hole and the occupied 7r and a orbitals in C Is and O Is excited Ni(CO)4- These observations are entirely consistent with the earlier theoretical predictions of the C Is hole localization in core ionized Ni(CO) 4 [250,272,276]. The term values of the localized Is-* it bands in both the C Is and the O Is spectra are ~2.2 eV lower for nickel tetracarbonyl than for free CO gas (Tables 6.1 and 6.2). Assuming that the C Is orbital energies are almost equal in both compounds [268,272] would therefore imply a higher energy for the it final orbitals of Ni(CO) 4 > Although of course the it orbitals involved in these transitions do not correspond directly with those in the neutral transition metal complex, and we have the complicating factor of severe orbital relaxation, the decrease in term value nevertheless provides direct experimental evidence of how the energies of the CO lit (it ) orbitals are perturbed by their interaction with the Ni atom during complex formation. Further evidence is provided by the increase in C Is-* it transition energy of ~0.2 eV, although the O Is -*it energy is reduced by ~0.2 eV, presumably due to relaxation effects. Comparisons of term values and transition energies between a range of transition metal carbonyl complexes * should allow experimental determinations of the relative extent of dff -» pw back bonding in these types of compounds. ISEELS studies of other carbonyl complexes are currently being carried out in this laboratory [277]. The most striking difference between the vibrationally resolved C Is -»it band of Ni(CO) 4 and that of CO (Figure 6.4) is the relatively more intense excitation of the higher vibrational components v=l and v = 2 (v = vibrational quantum number) in the Ni(CO) 4 spectrum. This can be rationalized by realising that greater orbital relaxation accompanying C Is excitation of the transition metal complex (by analogy with the 223 ionization process) will result in greater changes in molecular electron density distribution, and hence larger changes in the C-0 internuclear distance. The value of 0.235 eV (1900cm"1) obtained for the vibrational peak separation in Ni(CO) 4 is typical of a C = 0 stretching frequency, and the progression therefore can be confidently assigned to a C-0 stretching mode. Since the core electron is being promoted into a it orbital localized on just one of the CO groups, vibrational excitation will be limited largely to this same ligand. The derived vibrational frequency (1900cm"''-) can therefore be compared directly with the frequency of the C-0 progression in the C Is •+ it transition in free CO (1980 cm"1), and with the vibrational modes of the ground state of Ni(CO) 4. Because of the presence of several C-0 stretching modes in the neutral nickel tetracarbonyl complex with different associated frequencies, it is inappropriate to compare • any one of these frequencies to the more localized C-0 stretching in the C Is -*it transition. However, since the C-0 stretching frequencies of different symmetries do not differ widely in Ni(CO) 4, a mean value of V Q Q , obtained by averaging the frequencies, weighted according to the degeneracy of each normal mode, yields a characteristic frequency, VQQ> which is approximately independent of the vibrational interaction force constants. This average value is 2075cm 1 [240], which will be used in the following discussion. There is a substantial reduction of ~ 170cm 1 in vibrational frequency between ground state Ni(CO)^ and the C Is-* it excited state, similar to the reduction of ~ 160cm"1 observed for the corresponding processes in CO [75,207,269]. A reduction in frequency is expected due to the promotion of an electron into the C-0 it anti-bonding orbitals. Compared directly with CO gas, the stretching frequency of the excited vibration • -1 in the C Is -*it transition for the metal complex is slightly lower (1900 vs 1980cm ). This result is entirely expected since Ni 3d -*tr back bonding should reduce the C-0 bond order. The fact that this reduction in frequency (~80cm *) is similar to the difference between the C-0 stretching frequency in ground state CO and the averaged 224 C-0 stretching mode of ground state Ni(CO) 4 (~70cm_1) suggests that a study of the vibrational^ resolved core excitation spectra should also provide a reasonable assessment of the metal-ligand bonding situation in metal carbonyl complexes. This observation also supports the view that for the Is -*ir excitation in Ni(CO) 4 the final orbital is effectively localized on the CO group containing the core hole. The vibrational frequency observed in this work can also be compared with the averaged vibrational frequencies ("QQ) °f m e isoelectronic transition metal carbonyl - 2-complexes [Co(CO)4] and [Fe(CO)4] (see Table 6.5). It can be seen that the • _ i C Is -»7r excited Ni(CO) 4 containing a C Is hole has a C-0 frequency (1900cm ) approximately the same as the [Co(CO) 4]~ ion (1910cm-1). The lower value for [Co(CO) 4]~ compared to ground state Ni(CO) 4 (2075cm-1) arises due to more diffuse Co 3d orbitals, hence greater back bonding and higher ir electron density. The extra ir electron density in C Is excited Ni(CO) 4 compared with ground state Ni(CO) 4 is therefore similar to the additional n electron density due to back bonding in the [Co(CO) 4]~ ion, but less than in the dianion [Fe(CO) 4] 2~. The final feature of the C Is high resolution spectra worthy of note is that the vibrational components of the Ni(CO) 4 spectrum are substantially broader than those of the CO spectrum. Since the instrumental resolution was equivalent for collection of these two spectra, the bands of the metal complex must be intrinsically wider. One explanation is that the excited states have shorter lifetimes, and hence their energies are less well defined. However, this effect should give rise to purely Lorentzian broadening of the peaks, whereas the curve fits reveal that their shape is more Gaussian-like in Ni(CO) 4 (see Tables 6.3 and 6.4). A more likely explanation is therefore that there is simultaneous excitation of additional vibrational modes in Ni(CO) 4, possibly the Ni-C modes, which have very low vibrational frequencies (~ 400cm ^). Interpretation of the remainder of the features in the inner-shell spectra is less straightforward. The Is delocalized ir transitions are expected (see above discussion) at Table 6.5 Averaged vibrational frequencies of selected isoelectronic transition metal carbonyl species Compound Averaged Vibrational Frequency (cm"1) Ni(CO)4 2075(a) [Co(CO)4]" 1910<b) [Fe(CO)4]2" 1790<b) (a) From Ref. [240]. (b) Calculated from the vibrational frequencies given in Refs. [358,359]. 226 * ~2.2 eV higher excitation energy than the Is •» localized 7r bands, so that either or both of the peaks (numbers 2 and 3, Table 6.1, Figure 6.2) in the C Is spectrum with term values of 4.44 and 3.57 eV (or weak underlying features) could be assigned to these transitions. However, excitations to the 9a j (Ni 4s) and l i t , (Ni 4p) orbitals should also occur in this energy range, plus orbitals derived from the CO 3so Rydberg orbitals (by analogy with the term values for the corresponding transitions in CO [75,269]). Indeed, that the O Is spectrum does not contain a peak with a term value of 4.44 eV argues for its assignment in the C Is spectrum to orbitals containing none, or very little O 2s or 2p character. The preferred assignment is therefore to l i t , (Ni 4p) and/or 9a^ (Ni 4s) at a term value of 4.44 in the C Is spectrum, and delocalized rr plus orbitals of mainly CO 3sa character at a term value of ~3.6 eV (C Is and O Is spectra). Likewise the lack of a band at a term value ~2 eV in the O spectrum suggests final orbitals with little O character for the peak (number 4) at T=2.07 eV in the C Is spectrum. Therefore the most reasonable assignments for this band would be to Ni 5s and/or 5p Rydberg orbitals. The remaining pre-edge features in both spectra must be assigned in a general fashion to transitions to Rydberg orbitals. The extremely large number of orbitals involved and the lack of any sharp structure in the spectra at smaller term values prevents any further analysis. The C Is and O Is spectra of Ni(CO) 4 show similar features to the corresponding spectra of free CO above their respective ionization edges. By analogy the very broad features at term values of -10.7 eV (C Is) and -10.0 eV (O Is) may be assigned to localized o molecular shape resonances (see above discussion) in the J =3 (f-like) ionization channels. Such continuum shape resonances have been observed previously in a number of small molecules [5,58,124,278-281], including free CO [75,76]. Recently it has been proposed that there is some correlation between the energy of a shape resonance maximum above the ionization threshold, and bond length [58,279-281]. However, the relative magnitudes of the a term values in Ni(CO) 4 and CO (Tables 6.1 and 6.2) are not consistent with the longer C-0 bond in Ni(CO) 4 [282]. There are at least two possible reasons for this discrepancy: 227 (a) The much greater orbital relaxation accompanying ionization of the transition metal complex results in lower ionization energies, hence larger (more negative) term • * values for the Is •* o resonances, even though the absolute a virtual orbital energies may be lower due to increased C-0 bond lengths. (b) The different symmetries of CO and Ni(CO) 4 result in different symmetries for the a orbitals in the two molecules, and thereby quite possibly different energies. Note * also that the Is -*a peak in Ni(CO) 4 may be split by symmetry into two bands, although no splitting is evident in the spectra. The C and O Is spectra of Ni(CO) 4 also exhibit additional features in the continuum at term values ~-2.5 and ~-6 eV. The X-ray photoelectron spectra of a number of 1st row transition metal carbonyls [238] show shake-up bands at ~2 and ~5-6 eV above the main core level photoelectron bands. These have been attributed to metal 3d -* localized CO ir excitations [272] (triplet coupled at ~2 eV and singlet coupled at ~5-6 eV) or [250] a combination of metal 3d-> CO ir (~2 eV) and ligand localized CO ir-* ir transitions (5-6 eV). These small features may therefore be associated with these shake-up processes. Alternatively, these peaks are sufficiently wide that they may be additional a molecular shape resonances which have no counterpart in the CO Is spectra. Possibly they are associated with the Ni-C bonds, and may be described as Ni-C a shape resonances. The fact that these transitions are weaker in the O Is spectrum than the C Is spectrum is consistent with this assignment Recent NEXAFS results on CO on metal surfaces also suggest the possibility of metal-molecule levels in the near continuum [283]. Note in Figure 6.2 the great differences between the C Is spectra of Ni(CO) 4 and CO in this spectral region - the intense peak in the CO spectrum at T=-4.7 eV which is assigned to double excitation processes [63,75] is absent in the Ni(CO) 4 spectrum, and is replaced with the broader features present in both the C Is and O Is spectra as discussed above. However it cannot be ruled out that part of the intensity in this region of the C Is spectrum is due to double excitation processes, * * possibly C Is-* IT accompanied by 9*^ and/or 2e (Ni 3d) -* ir . The intense double excitation peak at T=-4.7 in the C Is CO spectrum is also lost when CO is 228 chemisorbed on metal surfaces, as shown by recent NEXAFS results [281,284]. 6.2.3 Nickel 3p Spectrum The energy loss spectrum of Ni(CO) 4 near the Ni 3p ionization edge is shown in Figure 6.5. The upper portion of the figure shows the original experimental data, below this is the same data after subtraction of the estimated contribution of the underlying valence-shell continuum. The valence contribution was estimated by fitting a curve of the —B form AE (A and B are constants, E=energy loss) to the experimental data below ~69 eV, and extrapolating it to higher energy. The energies, term values and proposed assignments of the numbered features in the Ni 3p spectrum are given in Table 6.6. An estimate of the Ni 3p 3^ 2 ionization energy of nickel tetracarbonyl was made by considering the constant energy difference between the 2p.y2 a n d 3 p ^ c o r e ionization energies for the free Ni atom [127] and also for Ni(CH 3COCHCOCH 3) 2 [285] (see Table 6.6). Using the same Ni 2p 3 / 2~3p 3 / 2 difference for Ni(CO)^ together with the available 2 p 3 / 2 ionization energy data for Ni(CO) 4 provides an estimate of 74.2 eV for the Ni 3p 3 / 2 ionization energy of Ni(CO) 4. The 3p 3 / 2 ^ 2 spin-orbit splitting of ~2.0 eV observed for the free Ni atom [127] is also applied to Ni(CO)4-Although the Ni 3p transitions are of quite low intensity, and are" superimposed on the intense underlying valence-shell continuum, at least 3 features can be distinguished (see Figure 6.5). The term values of the two peaks below the (estimated) Ni 3p ionization edge are rather low for their assignment to 3p -*it transitions. Although term values of transitions to the same final virtual valence orbitals are generally expected to decrease in going from core excitation to valence excitation [56], the term values for peaks 1 and 2 in the Ni 3p spectrum are lower than those proposed for the valence * * level -»it transitions (see below). Since 3p -* it excitations are allowed by electric dipole selection rules, it may be that the it orbitals are localized too heavily on the ligand 101 >-b co UJ UJ 5 > UJ rr 0 (a) Ni(C0) 4 Ni 3p 6 = o° E0= 3000 eV AE = 0.305eV \ \ \ Ni 3p edges *"-. I 2 (b) ; ~ . , w . ^ Original Data Background Subtracted 60 70 80 90 100 ENERGY LOSS (eV) 110 Figure 6.5 (a) Original; (b) background subtracted Ni 3p ISEELS spectra of Ni(CO)4. 230 Table 6.6 Energies, term values and possible assignments for features in the Ni 3p spectrum of Ni(CO)4 Feature Energy loss (eV) Term Value (eV) 3 p 3 / 2 3P1/2 Possible Assignment 3p3/2 3pi / 2 1 2 Ni 3 p 3 / 2 edge Ni 3 p 1 / 2 edge 3 70.94(12) 72.85(12) ~74.2(a) ~76.2(a) 78.40(24) 3.3 -4.2 3.4 0 -2.2 Ni 4s (9ai) Ni 4s (9a-|) 3p-»ed maximum; Double excitation; or a*(Ni-C) shape resonance (a) Estimated from the following data: Species 2P3/2 3P3/2 3 P 1 / 2 Reference (eV) (eV) (eV) Ni(CO)4 861.1 [236,237] N i ( C H 3 C O C H C O C H 3 ) 2 860.5 73.5 [285] Ni(g) 860.0 73.0 75.0 [127] 231 atoms for these transitions to be intense enough to be detected. The two observed peaks are therefore assigned to 3 p 3 / 2 " * N i 4 s (term value 3.34 eV) and 3p^ / 2 4s ( t e r m value 3.4 eV). The splitting of these two peaks (~2 eV) is comparable to the spin-orbit splitting of the 3p hole state in the free Ni atom [127,286]. That the relative intensities of the lower energy and higher energy peaks do not reflect the 3p^ / 2 '• -*P1/2 s p i 1 1 - 0 1 ^ degeneracies may be caused by exchange interactions in the excited state being of comparable magnitude to the spin-orbit interaction [287]. The broad feature ~4.5 eV above the estimated 3p ionization threshold may be the consequence of a cross- section maximum in the p +ed ionization channels. Such maxima are well known from atomic cross-section studies by photoabsorption and photoelectron spectroscopy [288]. Alternatively this feature may be a a (Ni-C) shape resonance or may result from double excitation processes. 6.2.4 Valence-Shell Spectrum Figure 6.6 shows the high resolution valence-shell electron energy loss spectrum of Ni(CO) 4 > together with a spectrum of CO obtained at the same resolution (0.068 eV FWHM). The insert shows a least squares fit of 4 Gaussian curves to the first band system in the Ni(CO) 4 data. Spectra obtained at higher resolution (0.036 eV FWHM) revealed no additional structure. It is apparent from Figure 6.6 that the percentage of CO gas in the sample of the metal complex is very low, and that all the identified features are due solely to Ni(CO)4- The energies, term values and proposed assignments of the numbered features in the Ni(CO) 4 spectrum are presented in Table 6.7, while the fit parameters are given in Table 6.8. The electronic photoabsorption spectrum of gaseous nickel tetracarbonyl has been previously reported over the energy range 4 to 6.5 eV [240]. The present data shows almost quantitative agreement with Ref. [240], except that a careful deconvolution procedure yields a lower estimate of the energy (by ~0.7 eV) for the first band (see Figure 6.6b). Gray [239] has reported a spectrum of Ni(CO). in a solid CO matrix. 232 (a) .0 >-b .5 co LJJ > o -<5 1.0 LU cn (b) o -69 = 0° E 0 =3000eV AE=0.068eV II ii o-9ta edge ^ ^ 2e edge (5cr + l7r) edge Ni (C0) 4 VALENCE CO VALENCE ; 4 a edge I 2 T T 3 4 J 1 1 I I 6 7 8 9 10 II 12 8 12 16 20 ENERGY LOSS (eV) 24 28 Figure 6.6 VSEELS spectra of (a) CO; (b) Ni(C0)4. The insert shows a fit of 4 Gaussian curves (see Table 6.8 for details) to the low energy band system from 4 to 8 eV in the Ni(CO)4 spectrum. Table 6.7 Energies, term values and possible assignments for features in the VSEELS spectrum of Ni(CO)4 Feature Energy Loss (eV) 9t2 Term 2e Value (eV) 5a+1rc 4a Possible Assignment 9t2 2e 5a+l7t 4a 1 4.49(12) 4.23 Ni 4s, 7i*(10t2) 2 5.36(12) 3.30 4.25 7t*(1 0t2) Ni 4s, it* 3 5.97(12) 2.75 3.70 ' - - JI*(2tl) T t * 4 6.71(12) ... 2.93 * - - — 5 10.1(2) 4.0 6 12.5(2) 1.6 1 • Ni 4s, r t * 7 13.7(20) 0.4 4.6 . . J Ni 4s, j t 8 14.7(2) 3.6 - - 7C 9 15.6(2) 2.7 71 10 17.0(5) 1 1 18.9(5) 12 22.0(5) IP(a) 8.72 9.67 14.10 18.25 (a) From Ref. [234]. 234 Table 6.8 Parameters of the Least Squares Fit of the Valence-shell Electron Energy Loss Spectrum of Ni(CO)4 Peak No. Position Width Relative Area (eV) (eV) 1 4.49 0.667 0.261 2 5.36 0.909 1.0 3 5.97 0.582 0.774 4 6.71 0.881 0.766 235 Above the first ionization energy, ~9 eV, detailed assignments are complicated by overlap of the continuum oscillator strength with discrete transitions, and the lack of sharp structure in this spectral region. In addition, no suitable calculations are available so that comparison with theory is also not feasible. However, the following tentative assignments are proposed on the basis of the photoelectron spectrum of Ni(CO) 4 [234,235] and calculated molecular orbital energy levels [235,240,249-252,255]: (a) The broad structure (peak 5) at ~10 eV energy loss is due to transitions from the highest energy CO (5o + l7r) levels into the IT orbitals, plus a rise in the Ni 3d (9t2 and 2e) ionization continuum just above threshold. (b) The 'plateau' from ~12.5 to ~13.7 eV energy loss is caused by transitions from the main bulk of CO localized 5o and lir levels to the IT orbitals. Note that these arguments ignore the possibility of many-electron excitations and are based purely on an independent-particle picture. Now turn to the low energy band system from ~4 to 8 eV energy loss. A number of assignments for the region have been suggested previously. In Ref. [239] the lowest energy band was assigned to the transition 9t 2 (Ni 3d) ••lOtj (IT ), while the most intense peak was assigned in a general fashion to the remainder of the Ni 3d * (9t2 + 2e)-* 7r transitions. More detailed assignments have been attempted on the basis of extended Huckel molecular orbital calculations [240], and more recently using configuration interaction on CNDO-type molecular wavefunctions [255]. Four components of the 4-8 eV band are resolved in the present work, and the insert to Figure 6.6 shows that the spectrum fits well to 4 Gaussian curves, the parameters of which are given in Table 6.8. It is obvious from the figure that peaks 1 and 3, and 2 and 4 have similar widths, suggesting that these pairs may have a common initial state. Since transitions from the triply degenerate 9t^ set of Ni 3d orbitals are expected at lower energies than peaks due to promotion of electrons- from the doubly degenerate 2e set, this would lead to the conclusion that peaks 1 and 3 originate from the 9u MOs, and peaks 2 and 4 from the 2e MOs. However, even though the 236 transition matrix elements will certainly be quite different for the 9t> and 2e orbitals, it is unlikely that they are so different as to result in the experimental (l + 3)/(2 + 4) intensity ratio of 0.586, considering the 3:2 (9t2:2e) degeneracy ratio, and the fact that more of the low energy transitions involving the 2e orbitals are symmetry forbidden. More likely the observed band widths are the result of heavily overlapping transitions and/or associated vibrational progressions. The assignment given in Ref. [240] involved transitions from the CO localized (5a + Iff) levels giving rise to a number of components of the low energy band system (i.e. the bands below ~8 eV). However, relative to the (5a + Iff) ionization onset, the highest energy loss (lowest term value) component of this band system (peak 4) at ~6.7 eV energy loss would then have a term value of ~7.4 eV, which is higher than the term value for any band in the inner-shell spectra (see Tables 6.1-3). Since term values for a particular final virtual valence orbital tend to decrease in going from core excitation to valence excitation [56], such an assignment can be eliminated on these grounds. The configuration interaction calculations [255] came to the conclusion that not only is the manifold of ff orbitals accessible from ~4 to 6 eV, but so also are the Ni 4s and 4p orbitals, although many of the possible transitions were calculated to be weak. Their assignment of the most intense peak at ~6 eV was to metal •* ligand charge transfer, 9t> (Ni 3d)-> 2t^ (TT ), the first lower energy shoulder at ~5.5 eV to 9t 2 (Ni 3d)-> lOt, (ff ), and the lowest energy band to either 9t^ (Ni 3d) -» Ni 4s or lOt^ (ff ) triplet coupled. If the lowest energy transition is indeed the latter, then it must gain intensity through vibronic coupling, since exchange transitions are forbidden at the high, impact energies and zero degree scattering angle used in the present VSEELS work. These findings of the CI calculations [255] are reflected in the assignments shown in Table 6.7. Unfortunately these calculations [255] did not consider transitions involving the 2e (Ni 3d) electrons. Assuming transferability of term values observed for 9t 2 (Ni 3d) excitations, the excitations from the lower lying 2e (Ni 3d) orbitals can then be assigned either wholly to the peak centered at ~6.7 eV energy loss (feature 4), or partly to this 237 peak plus some intensity contributing to features 2 and 3. Similarly, tentative assignments for transitions originating from the (5o + l7r) and 4a orbitals have been proposed and summarized in Table 6.7. It is hoped that the results reported here will stimulate further theoretical investigations of the valence-shell photoabsorption spectrum of Ni(CO) 4 in order to clarify this situation. 6.3 Conclusions In the present work, the first ISEELS spectra of a gaseous transition metal complex, Ni(CO) 4 have been obtain. The VSEELS spectrum was also obtained over an extended energy range. The inner-shell spectra could be interpreted largely by analogy with those of free CO. In particular the C Is and O Is spectra both show intense Is 7r and Is -• a transitions for Ni(CO)^ and free CO. Vibrational structure associated with a C - 0 stretching mode was resolved in the Is -+ 7r bands of the C Is spectra. The Ni(CO) 4 inner-shell spectra can most readily be explained by invoking localization of the C Is or O Is hole on one of the CO groups. The shifts in ls-> i transition energies and term values from free CO were interpreted in terms of severe orbital relaxation upon creation of a Is hole, and also the influence of Ni-» CO n back * bonding. The changes in vibrational structure between Ni(CO) 4 and CO in the C Is -*ir band, in particular a lower C - 0 vibrational frequency in Ni(CO) 4 > also show direct evidence for metal-ligand back bonding. Due to the complicating factor of greater orbital relaxation accompanying creation of a Is hole in Ni(CO) 4 > quantitative estimates of the extent of back bonding are not possible. However, the ISEELS technique should be very valuable in the investigation of metal-ligand bonding when results for different metal carbonyls can be compared, since then the extent of orbital relaxation should be almost constant for a given series of compounds. A study of the group VIA metal hexacarbonyls are presently undertaking in this laboratory in order to investigate further the applicability of ISEELS spectroscopy to studies of transition metal-ligand bonding. CHAPTER 7 238 ELECTRONIC EXCITATION OF SULFUR COMPOUNDS : I. INNER-SHELL AND VALENCE-SHELL ELECTRONIC EXCITATION OF DIMETHYL SULFOXIDE 7.1.1 Introduction Dimethyl sulfoxide (DMSO) is one of the safest solvents in common laboratory and industrial use. A notable feature is that anion reaction rates are exponentially increased in DMSO by the selective solvation of cations, and as a result reactions can run up to one million times faster than in other solvents. Furthermore, a large variety of reactions such as alkylation, cyclization, condensation and etherification, etc. can be more selective and give higher yields in DMSO. DMSO is also resistant to hydrolysis and thermal decomposition. It is used in media covering a wide range of acidity and with reaction temperatures up to its boiling point at 189 °C. Therefore, all types of spectroscopic data for DMSO are of great interest, not only for increasing fundamental knowledge, but also to extend present understanding of why DMSO is such an effective, versatile and safe solvent In particular there is almost no published spectroscopic information on DMSO at vacuum UV and X-ray energies. In the present work the spectra for electronic excitation of the valence-shell and all inner-shell electrons of DMSO are reported. The results include the S Is X-ray photoabsorption spectrum obtained elsewhere using synchrotron radiation (see below), the inner-shell electron energy loss spectra (ISEELS) in the regions of S 2p, S 2s, C Is and O Is excitation and the valence-shell electron energy loss spectrum (VSEELS). The use of synchrotron radiation and crystal monochromators is particularly advantageous for the study of inner-shell excitation (i.e. photoabsorption) of third-row and higher row K shells such as sulfur Is excitation [163], which occurs at X-ray energies above ~2 keV. 239 However, optical monochromators are less efficient in the far vacuum ultraviolet and soft X-ray regions, particularly at photon energies below ~1000 eV. Such studies are also further complicated by surface contamination of grating and mirror materials in spectral regions corresponding to carbon Is and oxygen Is excitation. However, as is now well known, ISEELS [6-9] provides a viable and often superior alternative to photoabsorption techniques at energy losses (i.e. equivalent photon energies) in the 50-1000 eV region. At the high impact energy (1.0-3.7 keV) and zero scattering angle used in the present work, electric dipole transitions dominate and the ISEELS and VSEELS spectra are essentially identical to their optical counterparts. The virtual photon field of the fast electron provides a continuum excitation source and as such is an alternative to the use of monochromatized synchrotron radiation or soft X-ray tubes [6-9]. ISEELS is particularly advantageous for the study of high-resolution Is excitation spectra of second-row, (i.e. C, N, O, etc.) atom containing molecules [8,9]. However ISEELS at energy losses above 1000 eV is very difficult because of low cross sections. Thus synchrotron radiation and ISEELS have complementary strengths and limitations and their combined use is optimal for the study of inner-shell electronic excitation in a molecule such as DMSO, which contains both second-row and third-row atoms. Valence-shell excitation at reasonable resolution is easily observed by VSEELS by simply reducing the energy loss from that used in the ISEELS region using the same spectrometer. To date the electronic spectral information published for DMSO is limited to a valence-shell photoabsorption spectrum in the 5-10 eV region [289]. There is no information for inner-shell excitation, although the S Is X-ray emission spectrum has been reported [290]. The main purpose of the present work is to obtain the various inner-shell and valence-shell spectra of DMSO. In addition it is of interest to compare directly the relative energies and intensities of the various transitions and resonances measured below and above the different core edges (i.e. S 2p, S 2s, C Is, O Is and S Is). This comparison is very useful for experimental assignments of the symmetry and spatial extent of the empty (virtual) ground state orbitals because each of them are populated differently from different core levels in accord with dipole selection rules and overlap 240 considerations. Previously, a systematic analysis of K shell excitation spectra of simple molecules containing B, C, N, O and F has indicated a linear empirical relationship between the location of inner-shell a shape resonance features and the corresponding bond lengths [58,280,291]. Such a correlation has been extended to larger molecules including aromatics [291], cyclic hydrocarbons [196], heterocyclics [292], non-cyclic alkanes [195] and even to sulfur (i.e. third-row atom) containing molecules such as thiophene and thiolane [161]. In the latter work inner-shell excitations to a (S-C) were found to occur 4-6 eV below the ionization threshold and to have relatively small variation with bond length. By contrast, the inner-shell spectra of SF 4 and SFg [160] suggested a variation in the position of a (S-F) resonances with bond length which is of similar magnitude to that observed for resonances associated with bonds between lighter elements (e.g. C-C or C-O). Thus, a further motivation for the present work is to investigate the systematics of bond length correlations applied to inner-shell excitations to a orbitals associated with bonds involving third-row atoms. DMSO is an interesting molecule in this regard since there are two distinct bond lengths corresponding to the S-0 and S-C bonds. The present spectral data for DMSO have been combined with the inner-shell spectra of S 0 2 (Chapter 7.II and Ref. [210]) and with the S Is synchrotron radiation photoabsorption spectra of S0C1 2 > S0 2C1 2 [293], S02FC1, S 0 2 F 2 [294] and a variety of organo-sulfur compounds (CH 3SH [163], CHjSCHj, CH 3S 2CH 3 > C ^ S C ^ , C^H^SH, C 6H 5SCH 3 and (CH 3) 2S0 2 [295]) to probe the location of the virtual a*(S-0) and o (S-C) orbitals and the systematics of their positions with S-C and S-0 bond lengths. 7.1.2 Experimental Methods 7.1.2.1 ISEELS and VSEELS The high performance election energy loss spectrometer used to measure the ISEELS and VSEELS spectra presented in this work has been described in detail in Ref. [91] and Chapter 3. 241 7.1.2.2 Synchrotron Radiation Photoabsorption The S Is inner-shell excitation spectrum shown in the present work has been provided by Hitchcock, Tronc and Bodeur [296]. This spectrum was obtained using a double crystal monochromator and synchrotron radiation from the LURE ACO storage ring at the Universite de Paris- Sud. The details of the apparatus and operating procedures using conventional photoabsorption techniques have been described previously [163,297]. For o this study Ge(lll) crystals (2d = 6.54 A) were used in the monochromator. The photon energy scale was determined from the crystal Bragg angles, and calibrated by the S ls->t^ u transition in SFg at 2486.0 eV [107]. The conventional photoabsorption technique was replaced with a more efficient ionization yield technique [294,298]. The sample was contained in an ionization chamber 10 cm long, with the X-rays admitted through a 12 um polypropylene window. In order to obtain sufficient vapor pressure the inlet line, gas cell and sample ampoule were heated continuously to 40 °C. Valves connecting the gas cell to the sample and to a mechanical pump were adjusted to obtain a constant pressure of 0.6 torr as measured by a capacitance manometer directly connected to the gas cell. Higher pressures are required for photoabsorption since the cross sections are lower than in EELS. The photoabsorption was monitored by measuring the current induced by the X-ray ionization of the gas between a filament and a cylinder concentric with the X-ray beam. The potential between the filament and the cylinder was maintained at 250 eV by a high stability power supply. Recent studies [294] have shown that the ionization signal measured in this fashion is essentially indistinguishable (both above and below the S Is IP) from that obtained with traditional measurements of the X-ray flux transmitted with and without the gas. The ionization current technique has been used previously for studies of gases [298] in the soft X-ray region (0.1-1 keV) and the reasons for the observed equivalence with photoabsorption have been discussed [294,299]. Its use for higher energy studies of gases (3-5 keV) has been demonstrated recently [300]. 242 7.1.3 Results and Discussion 7.1.3.1 Electronic Structure of DMSO ln DMSO it is assumed from the short S-0 bond length (148.2 pm) [301] that a double bond exists between the S and 0 atoms. In addition the pyramidal geometry is consistent with there being a lone electron pair on the sulfur atom. As a result the molecule has a C g symmetry with the mirror plane passing through the S = 0 bond. Using the geometry [301] shown in Figure 7.1.1, a minimum basis set Gaussian 76 calculation [360], was performed, with the addition of S 3d orbitals, to give the electron configuration of the *A' ground state of DMSO, which is as follows: (a) Core orbitals -> - -> -.2 ,,2 ,, . *2 . ,2 ,-„.2 , , . ,2 (la' f (2a')" (la'T (3a' f (4a't (5a' f (2a")" (6a') V w ' » ^ ~ S Is O Is C Is S 2s S 2p (b) Valence orbitals (7a') 2 (8a') 2 (3a")2 (9a') 2 (10a')2 (11a*)2 (4a")2 (5a")2 (6a")2 (12a')2 (13a' ) 2 (7a")2 (14a',)2 (c) Unoccupied valence (virtual) orbitals (8a")0 (15a')° (16a' )° (17a')° (9a")° (18a')° (19a' )° (10a")° (lla")° (20a' )° (21a')° (12a")° (13a")° (22a')° Within this molecular orbital scheme, and considering the predicted orbital energies (see Table 7.1.1), the features observed below the core ionization threshold can be interpreted in terms of transitions from the core levels to the three lowest virtual valence levels (8a", 15a' and 16a' ). The coefficients of valence AO contributions to these three orbitals are summarized in Table 7.1.2. In the ground state, these three virtual valence orbitals are somewhat delocalized over the molecule. The calculated LCAO coefficients suggest that the 8a" orbital has largely o (C-S) character with some contribution from IT (S-O). The * * 15a' orbital is indicated to have mainly a (S-O) character while 16a' has both o (S-0) and o (C-H) characters. It is possible that these local characters will be emphasized in 243 Figure 7.1.1 The structure of the DMSO molecule. 244 Table 7.1.1 Calculated energies of the molecular orbitals of DMSO(a) M.O. Energy M.O. Energy M.O. Energy M.O. Energy M.O. Energy (eV) (eV) (eV) (eV) (eV) l a ' -2475.421 6a' -175.874 4a" -15.282 8a" 10.448 10a" 19.296 2a' - 548.101 7a' - 33.886 5a" -14.668 15a' 11.259 11a" 19.573 la" - 300.854 8a' - 27.090 6a" -13.314 16a* 13.503 20a' 23.393 3a' - 300.851 Sa- - 25.017 12a' -12.699 17a' 16.199 21a' 25.007 4a' - 240.367 ga' '.- 20.200 13a' -10.937 9a" 17.297 12a" 25.179 5a' - 175.982 10a' - 16.056 7a" - 7.933 18a' 18.180 13a" 29.870 2a" - 175.957 11a' - 15.706 14a' - 6.762 19a' 19.242 22a' 37.067 (a) Gaussian 76 [360] calculations using minimiun basis set + S 3d. Table 7.1.2 Valence atomic orbital coefficients and local characters of 8a", 15a' and 16a' virtual valence molecular orbitals of DMSO M.O. S 0 C " l H 2 H 3 L o c a l C h a r a c t e r 3 8 3 "* 3 py 3 p * V ) *l* V d 2 + d 2 - 2s 2p 2p 2p K x ry  v i 2s 2p 2p 2p I s I s I s 8 a " 1 5 a f 1 6 a ' 0 . 0 0 0 . 0 0 0 . 8 8 - 0 . 0 0 - 0 . 0 0 0 . 0 0 - 0 . 0 0 - 0 . 0 0 - 0 . 3 5 - 0 . 1 6 - 0 . 5 0 0 . 0 0 0 . 6 4 - 0 . 4 5 0 . 0 2 0 . 0 0 - 0 . 1 2 - 0 . 0 0 0 . 7 1 0 . 5 3 - 0 . 0 0 0 . 2 4 - 0 . 2 9 0 . 0 8 - 0 . 0 0 0 . 2 4 - 0 . 0 0 - 0 . 0 0 - 0 . 0 0 - 0 . 3 4 0.00 -0.07 0 . 1 5 - 0 . 0 0 0 . 3 1 - 0 . 1 7 - 0 . 1 9 0 . 0 0 0 . 4 3 • 0.28 ±0.20 $ 0 . 3 0 + 0 . 1 2 JO.19 r 0 . 2 0 * 0 . 2 0 1 0 . 1 7 r0 .24 JO.43 ± 0 . 2 9 - 0 . 0 2 TO.06 ± 0 . 0 4 ± 0 . 1 3 = 0 . 1 6 J0.18 "0.06 JO.12 JO.12 : 0 . 2 9 a * ( S - C ) , u * ( S - 0 ) o * ( S - 0 ) * * a ( S - 0 ) , o ( C - H ) (a). Gaussian 76 [360] calculations using minimum basis set + S 3d. 246 the core excited states. Transitions to virtual valence orbitals above 16a! are not considered to give rise to distinguishable spectral features because of their higher energy and thus preference for direct ionization. 7.1.3.2 Assignment of Inner-shell Spectral Features Figure 7.1.2 shows an overview of the S 2p, S 2s, C Is, O Is ISEELS and the S Is photoabsorption spectra on the same relative energy scale. The spectra are aligned at their respective ionization edges. For this purpose the S 2p^ / 2 edge is chosen for the S 2p region. Taking into account the spin-orbit splitting in the S 2p spectrum, the various pre-edge structures show considerable similarity in relative energy positions. The discrete structures corresponding to transitions from the various core levels to the virtual valence and Rydberg orbitals are discussed in detail below. Although all structures of the S 2s, O Is and S Is spectra are not highly resolvable, in part because of the inherent larger natural line width of these shorter-lived excitations, all the core spectra show structures at very similar term values (energies relative to the respective edges). Such similarity strongly suggests that features at similar term values are due to excitation from the various core levels to the same final orbital. This simplified view has been used to interpret the core excitation spectra in the present work. However, in contrast to the pre-edge region, there seems to be little correspondence between the structures in the various ionization continua. Features in the continuum are most prominent in the S 2p spectra. Corresponding features are also observed in the S 2p continua of other sulfur-containing compounds [69,160,161,210], especially those carrying several very electronegative ligands. These features have been attributed to shape resonances caused by centrifugal and/or electrostatic potential barriers near the electronegative atoms in the molecule [35,302]. Alternatively, they can be described in terms of transitions to additional virtual orbitals as given by an extension to the molecular orbital scheme by the inclusion of S 3d orbitals in the basis set [35,303]. CO z: LU I -LU > LU 247 TERM VALUE (eV) 10 o -10 _1_ -20 J -30 -40 _ l l_ AE (eV) 0.15 I 3710 12 14 S2p (ISEELS) 15 _l_ S 2P3 / 2 - i / 2 edges 160 180 *S2s edge 200 S2s (ISEELS) 220 0.15 240 260 J I 3 5 U_L Cis (ISEELS) 280 E C1s edge L 300 A 0 . I4 320 01s . . . • /13 7 8 I I l l l l I (ISEELS) 520 _l_ 01s edge _ l L 540 560 580 S1s (photoabsorption) 2470 2490 25I0 ENERGY LOSS (eV) Figure 7.1.2 Comparison of S 2p, 2s, C 1s and O 1s ISEELS spectra with S 1s photoabsorption spectrum of DMSO. 248 7.1.3.2.1 Sulfur 2p and 2s Spectra The low-resolution (0.13 eV FWHM) S 2p, 2s ISEELS spectta (160- 260 eV) of DMSO are shown in Figure 7.1.3. An energy loss spectrum of S 2p between 165 and 173 eV, obtained with higher resolution (0.095 eV FWHM), is presented as an insert in Figure 7.1.3. The absolute energies, term values and possible assignments are shown in Table 7.1.3. The S 2p^ / 2 ionization edge (171.91 eV) has been determined by XPS [118] while the S 2p^ / 2 and S 2s IPs of DMSO have not been reported. A spin-orbit splitting of 1.3 eV (similar to that for S 0 2 [87]) is assumed for S 2p^ / 2 ^ / 2 and gives an ionization energy of 173.21 eV for production of the S 2p^ / 2 hole. The energy of the S 2s edge is estimated by assuming an energy difference of 64.3 eV between the S 2p and 2s ionization edges. This value is derived from the S 2p and 2s IPs of a variety of sulfur-containing compounds [118], where it is found that the S 2p-2s splitting is largely independent of molecular type. The features 1-12 below the S 2p IP are tentatively assigned to transitions to the virtual valence and Rydberg levels from the spin-orbit split S 2p^ / 2 and S 2p^ / 2 levels. Features 1 and 2 are assigned to the transitions from the S 2p^ / 2 level to the first and second lowest virtual valence orbitals (8a" and 15a') because of the large term values of 4.73 and 4.27 eV respectively. Their spin-orbit partners from the S 2p^ / 2 level are assigned to features 3 and 6, which have similar term values. Features 4 and 5 are assigned to the S 2p3/2-> 16a' transition. Feature 3 is believed to have an additional contribution from the S 2p^ / 2 -• 4s transition. If this is the case then the sum of the oscillator strengths of S 2p^/2-+ 4s and S 2p /^2-+ 8a" transitions would possibly be larger than the oscillator strength of the S 2p^2-> 8a" transition (feature 1) as observed. Such an assignment is consistent with the expectation that the S 2p^ / 2 -* 8a" transition should have twice the oscillator strength of its spin-orbit partner (S 2p^/2"> 8a") which contributes to feature 3. Since the 4s Rydberg state has the same symmetry and a very similar energy to the 15a' and 16a' states, the possibility of valence/Rydberg mixing cannot be eliminated. As a result the 4s features may be broadened and do not appear as sharp peaks, which are typically observed in the S 2p spectrum of other 249 >-CO z: LU LU > _l LU cr S 2 p A (9 = 0 ° E 0 =3000eV A E = 0.095 eV 'V. 1 2 1 1 3456 78 1 1 II II 9 10 II 12 I I I 1 2P3/ 2 2p l / z edges 165 167 ) 169 1 1 171 173 Nil y J 1213 13710 14 15 _ i _ " S 2 p 3 / . edges 0 = 0 ° E 0 = 3700 eV AE = O.I28eV f S 2 s edge 160 180 200 220 240 ENERGY LOSS (eV) 260 Figure 7.1.3 High-resolution S 2p (insert) and low-resolution S 2p, 2s ISEELS spectra of DMSO. 250 Table 7.1.3 Energies, term values and possible assignments for features in the S 2p, 2s spectrum of DMSO Sulfur 2p (a) Feature v ' Energy L o s s ^ Term Value (eV) Possible Assignment (eV) 2 p 3/2 2 p l / 2 2 P 3 / 2 + 1 167.18 4.73 — 8a" -2 167.64 4.27 — 15a' 3 168.44 3.47 4.77 4s 8a" 4 168.66 3.25 — V 16a' 5 168.82 . 3.09 — J 6 168.98 — 4.23 - 15a* 7 169.62 2.29 3.59 4p ' 4s 8 169.78 — 3.43 - 16a' 9 170.26 1.65 — 5s,4d -10 170.82 — 2. 39 - 4p 11 S 2 p 3 / 2 edge 171.56 (171.91) ( c ) 0 1.65 CO 5s,4d 12 S 2 p 1 / 2 edge 172.63 (173.21) ( c ) — 0.58 0 -7s,6d 00 13 14 15 173.7<d ) 178 .3 ( d ) 189 .4 ( d ) -1. -5. -16. 2(e) 7(e) 8(e) 2p ->• S 2p > S 2p + S 3d Resonance 3d Resonance 3d Resonance Sulfur 2s -16 S 2s edge 231.33 (236.2) ( f* 4. 0 87 2s > CO 8a",15a' (a) Sea Figure 7.1.3. (b) Estimated uncertainty is ±0.04 eV. (c) S 2 p 3 / 2 IP from XPS [118]. S 2 p 1 / 2 IP estimated by assuming (S 2 p 3 / 2 - S 2 p 1 / 2 ) = 1.3 eV. see Rets. [87,210]. (d) Estimated uncertainty is ±0.5 eV. (e) Term values with respect to the mean S 2p edge (172.56 eV). ( f ) S 2s IP estimated by assuming (S 2p-S 2s) = 64.3 eV, see Refs. [118,210]. 251 sulfur-containing compounds [210,304]. A comparison with a solid state spectrum, in which the transitions to the (large orbit) Rydberg orbitals would collapse because of the external perturbation from the surrounding molecules, would be valuable for spectral assignment The third band (features 7, 8) in the S 2p spectrum (Figure 7.1.3, insert) is assigned to a combination of the S 2p3/2-> 4p, S 2p 1 / 2 + 4s (feature 7) and S 2p 1 / 2 •+ 16a' (feature 8) transitions. Other weak features (9-12) below the S 2p ionization edges are assigned to S 2p 3 / 2-» 5s, 4d, S 2p1/2-> 4p, S 2 p 1 / 2 -» 5s, 4d and S 2p1/2-> 7s, 6d Rydberg transitions respectively, based on a consideration of the respective term values. The quantum defects for the s, p and d series, calculated from the Rydberg equations, are 2.03, 1.57 and 1.12, respectively, and they are close to the values expected for sulfur-containing molecules [54]. A comparison with the 2p spectra of other molecules containing third-row elements [65,69,160,161,210] suggests that the three broad continuum structures (features 13-15) in the S 2p spectrum of DMSO are related to the S 2p-» S 3d resonances. The continuum features could also arise from shake-up processes. Shake-up features can be identified by the correspondence between the energy of their onsets and the positions of XPS shake-up satellites. However, no XPS satellite spectrum of DMSO has been published. Therefore all the S 2p continuum features are tentatively assigned to S 3d shape resonances. The S 2s excitation region of DMSO exhibits a broad band (feature 16) below the S 2s IP (Figure 7.1.3) and is structureless above the ionization threshold. Generally, features corresponding to S 2s excitation are expected to be broadened due to very fast autoionization (Coster- Kronig) processes to the S 2p subshell, which effectively shortens the lifetime of the excited state. Feature 16 has a term value of 4.87 eV and it is assigned as the S 2s -» 8a" transition in comparison with the first' S 2p excitation feature . Since this 2s structure has a width of about 1 eV and the S 2p spectrum indicates that the 8a" and 15a' levels are separated by only 0.5 eV, the S 2s •+ 15a' transition may also be contributing to the broad structure. The S 2s 1 6 a ' transition cannot be 252 seen. Differences in transition probability could arise from the different symmetries and spatial extents of the S 2s and S 2p orbitals which produce selective differences in the overlap with the unoccupied orbitals. The absence of continuum features in the S 2s spectrum is not unexpected since S 2s -*S 3d excitations are electric dipole forbidden in the atomic case. 7.1.3.2.2 C Is Spectrum The broad range (0.15 eV FWHM) ISEELS spectrum of C Is excitation in DMSO is shown in Figure 7.1.4. A high-resolution spectrum (0.095 eV FWHM) over the range 285-293 eV is shown as an insert to Figure 7.1.4. The energies, term values and possible assignments are presented in Table 7.1.4. The C Is binding energy is the value reported from XPS measurements [118]. The first band in the C Is region has two distinct features (1 and 2). They are assigned as C Is excitations to 8a" and 15a' respectively. Compared with the S 2p •* 8a", 15a' transitions, the term values for the C ls-» 8a", 15a' features have decreased by 0.7-0.9 eV. Similar shifts in the term values for inner-shell excitations to virtual valence orbitals have been observed by Hitchcock et al. between the C Is and S Is, 2s and 2p spectra of thiophene and thiolane [161]. Alternatively, features 1 and 2 could be assigned as a vibrationally resolved C Is-* 15a' transition, but this is considered unlikely because under C g symmetry all transitions are dipole-allowed. In addition, according to the ground state calculation in section 7.1.3.1 the 8a" orbital has mainly a a (C-S) character. Thus it can be expected that C Is-* 8a" will be stronger than the transition to the 15a' orbital, which is composed mainly of a (S-O) character. Feature 2 has a term value of 3.59 eV which is close to that assigned for the 4s Rydberg in the S 2p spectrum (see Table 7.1.3). Thus the C ls-» 4s transition may also contribute to the intensity observed in feature 2. The assignment of C ls-> a (C-S) transitions as contributing in the discrete portion of the spectrum of DMSO is supported by the observation of a triplet signal in the corresponding regions of the C Is signal of COS and CSj [305]. A measurable singlet-triplet splitting is expected from the states of valence character, but not for those of Rydberg character. >-t UJ LU > !5 LU cr 1 0 5 -0 280 til 1 69 = 0 ° A E o = 3 0 0 0 eV N \ 0 . 0 9 5 eV i v*_. - y 1 2 3 4 5 M i l I 1 1 1 , ^CIs edge , 285 Cis edge 287 289 291 293 6 =0° E 0 3000eV AE= 0.150 eV 290 300 310 320 ENERGY LOSS (eV) 330 340 Figure 7.1.4 High-resolution (insert) and low-resolution C 1s ISEELS spectra of DMSO. to Table 7.1.4 Energies, term values and possible assignments for features in the C 1s spectrum of DMSO F e a t u r e ^ Energy L o s s ^ Term Value (eV) Possible Assignment (eV) ( f inal orbital) 1 287.43 3.81 8a" 2 287.65 3.59 15a',4s 3 288.47 2.77 16a' 4 289.01 2.23 4p 5 289.87 1.37 5p Is edge (291.24) ( c ) 0 09 6 297.7 -6.5 S 3d Resonance or two electron exci-tation or shake up (a) See Figure 7.1.4. (b) Estimated uncertainty is ±0.04 eV. (c) C 1s edge energy from XPS [118]. 255 Feature 3 is assigned to the C Is 16a' transition with a similar- shift in term value of about 0.8 eV from the S 2p -»16a' transition. The intensity relative to the continuum and the term value of feature 3 (2.77 eV) are both very similar to those of strong features consistently observed in the C Is spectra of hydrocarbons [161,195,196,292]. These features have been attributed to C Is excitations to a mixed Rydberg/valence orbital, labelled 3p/7t (CH) for reasons discussed in Ref. [195], which has a strong a (C-H) character. This assignment is supported in numerous cases by MS - X a calculations and the observation in the C Is spectrum of the solid state of a corresponding feature, which has a polarization dependence consistent with the molecular orientation and the a (C-H) character [161,196,306]. This feature confirms the view that feature 3 in the C Is spectrum of DMSO arises from transitions to the 16a' orbital * since this has significant a (C-H) character (see Table 7.1.2). The other two features (4, 5) below the ionization edge are attributed to the C Is excitation to 4p and 5p Rydberg orbitals. The 4p term value (2.23 eV) is quite similar to that assigned for the 4p in the S 2p spectrum (2.39 eV, see Table 7.1.3). Similar Rydberg term values for excitation from different shells are expected due to the large size of Rydberg as compared with virtual valence orbitals. There is only one weak and broad structure in the C Is continuum. It is assigned, provisionally, as a transition from the C Is to a S 3d resonance, which is expected to be weaker than in the case of the S 2p continuum where a S 2p -»S 3d resonance would be highly favored (dipole allowed). C ls-> S 3d resonances would be further discriminated against because of the poor overlap of the C Is and S 3d orbitals and also because the C Is hole is remote from the center of the molecule so that it would experience a smaller potential-barrier effect Alternatively, this continuum maximum could be a double excitation or shake-up satellite. Finally, it should be noted that the monotonic decline in the C Is continuum, and particularly the absence of any maximum at the ionization edge, is very similar to the situation in the C Is spectrum of methane, but contrasts with the spectra of all hydrocarbons containing C-C bonds, which contain C ls-> a (C-C) resonances near the C Is IP. 256 7.1.3.2.3 O Is Spectrum The long-range, low-resolution (0.135 eV FWHM) 0 Is ISEELS spectrum of DMSO is shown in Figure 7.1.5. An expanded view of the region between 530 and 540 eV energy loss is shown as an insert to Figure 7.1.5 in order to demonstrate more clearly all the pre-edge structures. The energies, term values and possible assignments for all the O Is features are summarized in Table 7.1.5. The O Is IP is taken from the XPS value reported by Jolly et al. [118]. The prominent broad band in the region of 533 eV in the O Is excitation spectrum can be attributed to overlapping transitions from the O Is level to the three lowest unoccupied valence levels. This peak is much broader than expected from natural line widths of O Is excited states (~0.2 eV) and thus suggests overlapping electronic transitions, probably accompanied by extensive vibrational excitation, as is observed for the O Is -»it transition in CO [75,307]. Feature 1 is assigned to unresolved transitions to the closely space 8a" and 15a' levels. The small shoulder on the high energy side of the band (feature 2) is assigned as a transition to the 4s level based on the term value of 3.56 eV. Feature 3 is assigned as the O Is ->16a' transition. The weakness of this feature contrasts sharply with the strength of its C Is counterpart consistent with the proposed a (C-H) character. The respective term values of the transitions from O Is to the 8a", 15a' and 16a' states are quite different from those in the S 2p, 2s spectra, but are quite close to those for the corresponding transitions in the C Is spectrum (Table 7.1.4). This is to be expected, as both O and C are peripheral "ligands" and thus in a different situation to the central S atom in DMSO. Based on their term values, the remaining pre-edge features (4-6) are assigned to transitions to the ns/nd Rydberg series. As in the C Is spectrum, the O Is continuum features (8 and 9) have very low intensity, suggesting that O Is S 3d resonances are either very weak or non-existent This likely occurs for the same reasons of symmetry, poor overlap and remoteness from the center of the potential barrier field, as were discussed for the C Is and S 2s spectra above. 257 10 £ 5 GO LU t 10 UJ > LU cr o 12 3 4 56 7 II I 1 11 I 01s edge 5 3 0 535 5 4 0 I 34 7 II I III 8 J_ 9 I 01s edge 69 = 0° E 0 = 3500eV AE = 0.135 eV 530 540 550 560 570 ENERGY LOSS (eV) 580 Figure 7.1.5 O 1s ISEELS spectra of DMSO. Insert spectrum on the top left-hand corner is an expanded view of the 528-542 eV region. 258 Table 7.1.5 Energies, term values and possible assignments for features in the O 1 s spectrum of DMSO Feature^ 3^ Energy Loss^^ Term Value (eV) Possible Assignment (eV) (final orbital) 1 2 3 4 5 6 0 Is edge 7 8 532.66 533.11 533.91 535.06 535.71 536.01 (d) (536.67) (c) 536.8 540.7 556.1 (c) 4.01 3.56 2.76 1.61 0.96 0.66 0 -0.1 ' -4.0 -19.5 8a", 15a' 4s 16a' 5s, 4d 6s,5d < 7s,6d S 3d Resonance or two elec-tron excita-tion or shake up (a) See Figure 7.1.5. (b) Estimated uncertainty is ±0.08 eV. (c) Uncertainty is ±0.5 eV. (d) O 1s IP from XPS measurement [118]. 259 7.1.3.2.4 S Is Spectrum The S Is photoabsorption spectrum of DMSO obtained using synchrotron radiation and ionization current detection is shown in Figure 7.1.6. The estimated bandpass of the photon beam is 0.35 eV (FWHM) at hv - 2500 eV [163]. The energies, term values, possible assignments and S Is IP [308,309] are given in Table 7.1.6. The small shoulder (feature a) appearing on the lower energy side of the first band is not believed to be due to DMSO because the intensity of this feature relative to the 2476.0 eV main peak varies with temperature. In addition, the energy of this shoulder (2472.8 eV) is close to that of the main peak in S 0 2 (values of 2473.8 [163], 2473.3 [294] and 2473.2 eV [310] have been reported). The spectrum for S Is excitation looks somewhat- similar to the O Is spectrum. However, feature 1 has an observed bandwidth of about 1.8 eV so that the transitions to the three unoccupied virtual valence levels and the 4p Rydberg state are not resolved. Features 2 and 3 are assigned to excitation to the 5p and 6p levels, while features 4 and 5 are assigned as S 3d continuum resonance features. It should be noted that the mean term value for the compound peak 1 of S Is is ~4.6 eV, which is similar in magnitude to the values observed for the core-* virtual valence transitions for S 2p (4.3-4.8) and S 2s (4.9). These similar term values support the assignments and also indicate that the core •+virtual valence term values are reasonably transferable for those core holes located on the same atom, i.e. S in this case. In contrast, very different core ->virtual valence term values are found for C Is (2.8-3.8) and O Is (2.8-4.0) (see Table 7.1.10). Furthermore, the core -»Rydberg term values are, as expected, reasonably transferable between the different core levels, regardless of the atomic location. This confirms earlier observations for tetramethyl silane [65] and phosphorous compounds [66]. Feature 6 at 2536 eV (Table 7.1.6) is a broad continuum maximum observed in a longer range scan, which is not shown in Figure 7.1.6. This type of long-range modulation of the continuum intensity, usually called extended X-ray absorption fine structure (EXAFS), is caused by interference arising from backscattering of the ejected photoelectron from the 2460 2470 2480 2490 2500 2510 PHOTON ENERGY (eV) Figure 7.1.6 S 1s photoabsorption (ionization current) spectrum of DMSO obtained using synchrotron radiation. 261 Table 7.1.6 Energies, term values and possible assignments for features in the S 1s spectrum of DMSO (a) Feature Photon E n e r g y ^ (eV) Term Value (eV) Possible Assignment (f inal orbital) a 2472.8 - S0 2 i m p u r i t y ( c ) 1 2475.8 4.6 8a", 15a' (16a , ,4p) ( d ) 2 2479.0 1.4 5p 3 2480.0 0.4 6p S Is edge (2480.4) ( e ) 0 CD 4 2484.8 -4.4 S 3d Resonance 5 2488.7 -8.3 S 3d Resonance 6 2536< f ) -56 exaf s (a) See Figure 7.1.6. (b) Estimated uncertainty is ±0.3 eV. (c) The intensity of this feature relative to the 2475.8 eV peak varied with temperature and thus is believed not to arise from DMSO. The energy of this shoulder Is similar to that of the main peak in S 0 2 (2473.3 eV, see Refs. [163] and [294]). (d) These two states are likely represented in the shoulder on the high energy side of peak 1 (Figure 7.1.6). (e) From XPS, see Ref. [308,309], including later correction of this IP. (f) Broad continuum maximum observed in a longer-range scan (not shown in Figure 7.1.6). 262 neighboring atoms in the molecule. 7.1.3.3 Correlation of a (S-C) and a (S-O) Energies with Bond Length Previously a reasonable linear correlation between the position of a resonances and corresponding bond lengths has been demonstrated for small molecules containing C, N, O, or F [58,280]. It has also been found that such a correlation can be applied to the a (S-F) resonance of SF^ and SFg [160]. It is of considerable interest to see whether such a correlation exists for transition to a orbitals associated with S-C and S-0 bonds. In DMSO the 8a" virtual valence orbital is taken to be the a (S-C) orbital. Term values from the S 2p levels and C-S bond length of C 4H 4S [161], C^HgS [161], CS 2 [304,311] COS [304,311] and CH 3SH [312] are listed together with the presently reported values for DMSO in Table 7.1.7. Term values for S Is -*o (S-C) and * C ls-> a (S-C) excitations in a wide range of organo-sulfur compounds are also summarized in Table 7.1.7. Based on the average S 2p^ 2Y /2 t e r m v alues for the o (S-C) transitions of the above molecules, the correlation between the position (5=E-IP) of the S 2p-» o (S-C) transition, relative to the ionization edge, and S-C bond length (R) in pm is found to be given by the relation 6 = 3.72-0.047 R, with a correlation coefficient of r = 0.99. The linear correlation (Figure 7.1.7) between the S-C bond and the average position of * the S 2p-» a (C-S) feature has a slope of -0.047 eV/pm. This is much smaller than the slopes of the correlation lines for resonances in the K-shell spectra of second-row elements, which range between -0.30 and -0.55 eV/pm [58]. The reduction in sensitivity of o resonances to S-C bond length is consistent with the trends with Z (sum of core-ionized and back-scatterer atomic numbers) for 13<Z<18, outlined by Sette et al. [58]. In a previous study of thiophene and thiolane [161], the a (S-C) position was claimed to be relatively insensitive to the S-C bond length. 263 Figure 7.1.7 Correlation between o*(C-S) resonance position relative to the edge (5) and bond length for S 2p spectra of organo-sulfur compounds. Table 7.1.7 R(S-C) bond lengths and o*(S-C) energies (8) for various organo-sulfur compounds Molecule R(S-C) pm (a) cs2 155.3 COS 156.1 w C.HCSH O J C,Hc SCK„ O D J C 2^  5^ *2^  5 171.4 180(2) e 6 t 180(2) e s t 1 8 0 ( l ) e 8 t CH 3SCH 3 180.2 CH jS 2^ 3^ (CH 3) 2SO 180.6 180.7 ( b ) CH3SH 181.4 W 183.5 6 = m + nR Cor r e l a t i o n C o e f f i c i e n t S is E 0 (eV) 2473 2473 2473 2472 2473 2472 2472 2473 2475 2472 2472 7 ( c ) 4<e> 3 ( d ) (d) ,.(d) (d) j ( f ) ,(e) 1 P ( 8 ) (eV) (h) (h) 2478.0 2478.9 2477.6 2477.8(5) 2477.8(5) € 2477.5 2477.2 2477.5 2480.4 2477.7 2477.8 est -4.3 -5.5 -4.2 -5.5 -4.7 -4.8 -4.8 -4.0 -4.6 -4.9 -5.5 -4.33 -0.0027 0.05 S 2p E g*(eV) 166.7 167.6 166.3 167.8 165.0 164.9 IP*(eV) 170.4 171.2 170.6 171.56 169.9 169.9 -3.7 -3.6 -4.3 (1) ( i ) (e) -4.75 -4.9 -5.0 (j) (e) 3.72 -0.047 0.993 C Is E o (eV) 290.6 291.0 287.1 (k) (k) (e) 287.43 •Si) 287.6 X 287.8 (e) I P ( A ) (eV) 293.1 295.2 290.3 (k) (k) 291.24 291.4 290.5(5) -3.5 -4.2 -3.2 -3.81 -3.8 -2.7 -3.03 -0.019 0.037 t 8 = E 0 - IP = -T. $ S 2p values represent an average of the 2p 3 / 2 and 2p 1 / 2 values. (a) Bond length from [158] except those marked. (est) represent estimated values. (b) (CH 3) 2SO bond length ftom [301]. (c) COS and CS 2 S ls->o*(C-S) energies from [309,313]. (d) S ls->o*(C-S) energies from [295]. (e) S Is, S 2p and C Is data from [161]. (f) S Is data from [163]. (g) S Is data from [270] with corrected H2S value. (h) COS and CS 2 S Is IPs from [314] (quoted in [270]). (i) COS, CS 2 S 2p data from [304]. (j) CH 3SH S 2p and C Is data from unpublished McMaster spectra [312]. (k) COS, CS 2 Cis data from [311]. Note that the peak assignments have been revised from those reported in the original work [311]. (1) C Is IPs from [118]. 265 A systematic variation of the a (S-C) term value with R(S-C) is found only for S 2p-* a (S-C) excitation. The relationships between R(S-C) and the S ls-> a (S-C) and C ls-» a (S-C) term values (Table 7.1.7) are not well defined statistically, although in each case the least-squares slope of an attempted linear correlation is small (Table 7.1.7). In most previous treatments of the correlation between bond lengths and core-excitations, * * the term values of X ls-> a (X-Y) and Y Is -»a (X-Y) transitions were averaged as a means of increasing statistical precision. As the difference in atomic numbers of core-ionized and neighbor atoms increases, the term values for transitions to a (X-Y) become less similar in the X and Y core excitation spectra. This is illustrated in the present work where the a (S-C) term value is 1 eV smaller in the C Is than in the S 2p or S Is spectra (see Table 7.1.10 below). Also, in cases where the spectra are obtained with low-resolution (this is the situation for most of the data in Table 7 except that for S 2p) or where the spectral features are inherently broad such as in S 2s and S Is excitations, several transitions of different origin could overlap and the overall position of the feature will be dependent on the relative energies and intensities of these transitions. As a result the energies of such kinds of broad features may not be truly representative of the transition to the a orbitals concerned. Any uncertainty in the a position or assignment can easily obscure correlations with the bond lengths that have shallow slopes, as is evidently the case for a (S-C) and o (S-O) levels (see below). In addition to the C-S bond, DMSO also contains the S = 0 bond. The a (S-O) orbital of DMSO is taken to be the 15a' orbital. Data relating to correlations between R(S-O) and excitations to a (S-O) are summarized in Table 7.1.8. Unfortunately, out of those compounds containing S-O bonds and with inner-shell excitation spectra available, there is insufficient variation in the bond length (143.1-148.2 pm) to reveal any clear correlation. However, these spectra do show that the structures corresponding to the o (S-O) transitions occur at similar positions relative to the ionization edge in accord with the similar bond lengths. 266 Table 7.1.8 R(S-O) bond lengths and a*(S-O) energies (8) for various sulfur-oxygen compounds Molecule R ( S - 0 ) ( a ) 6 (eV) (c) (pm) S Is S 2 p 3 / 2 c i 2 so 2 140.4 -4.7<d> _ so 2 143 .1 ( b ) -5.4<e> - 4 . 4 9 ( 8 ) F 2 S 0 2 144.3 -3.6<f> -FC1S02 1 4 4 . 3 ( e s t ) .-4.0< f> 0 -ci 2so 144.3 - 5 . 9 ( f ) -(CH 3) 2SO 148.2 -4.6 -4.27 (a) Bond lengths from [158] except for FCIS0 2 and S0 2 . (b) Bond length from [315]. (c) 5 = E 0 - I P = - T . (d) C I 2 S 0 2 S 1s data from [243] (average of two transitions). (e) S 0 2 S 1s data from [163], recalibrated in [294] (average of two transitions). (f) S 1s data from [294,310]. (g) S 0 2 S 2p 3 / 2 data from [210]. 267 * • The present observation of relatively small variations in a (S-C) and o (S-O) positions with changing bond length contrasts markedly with the considerably greater slope of the proposed bond length correlation for o (S-F) energies [160]. The slopes of the correlation lines for 12<Z<18 were found to vary smoothly with Z [58,60]. Thus, it * seems unlikely that a large change in the slope would occur between the o (S-O) (Z=24) and a (S-F) (Z=25) correlations. This suggests that the slope of the proposed a (S-F) correlation may be exaggerated, possibly as an artifact of a restricted range of R(S-F) and the consideration of a very limited data set (only data for SF^ and SFg were considered in [160]). Previously Sodhi et al. [65] have shown that Si 2p -» Si 3d continuum resonance features can be correlated with bond length in an analogous fashion to the a resonances in compounds containing second-row atoms [58] (although note that both R and Z varied in the indicated Si 2p correlation). However, in the S 2p spectra of the molecules listed in Table 7.1.7 no clear trends between bond lengths and positions of the S 2p -> S 3d continuum resonances can be found. This may be related to the smaller range of ligand electronegativities in the S 2p data set than in the Si 2p data set [65]. 7.1.3.4 Valence-Shell Spectrum of DMSO The high-resolution VSEELS spectrum (0.032 eV FWHM) of DMSO in the range 5- 20 eV is shown in the lower section of Figure 7.1.8. This spectrum is consistent with the earlier-reported photoabsorption spectrum recorded in the 5-10 eV region by Clark and Simpson [289]. The hatched lines on the figure indicate the locations of the nine valence ionization thresholds below 21.22 eV as measured by He(I) photoelectron spectroscopy [225]. No estimate is available of the remaining four valence-shell IPs since neither He(II) nor XPS studies have been reported for DMSO. A more detailed VSEELS spectrum in the range 5-8 eV is shown on the upper section of Figure. 7.1.8. The energies of the spectral features and proposed assignments for the valence-shell spectrum are summarized in Table 7.1.9. The spectrum shows at least eight distinct bands with 268 11! M! j 11 4 s | 8o"l5a' r6a' 4 d 5 d 6 d - /In j 5s 6s7s U H U Ilia' 4p 5^€p7/>** 8o" l5o ' r6o ' 4 d 5 , 3 6 ( 1 - , , 4s ! • ' 5s 6s7s / . 11 I I .1.'. 5a 4/> T i l VA 5/9 €p7/0~ 8 a - l 5 a ' ! 6 o ' 4 d 5 d 6 d - ' 1 1 I 'V i l | 2 a ' 4/> S/jtv7/5- ' ^ - U 4s ! 8o"l5a' 16a' 4 d 5 d 6 d -5s 6s7s L 10a1 Ap 5p6p7p~ 0 II s CH CH, 3 ^' '3 VALENCE 9 = 0° E 0 = 1000 eV AE = 0.032 eV • 1 2 3 1 I I 6 7 8 9 Oil 12 13 14 I I I I I I I I I 1516 I I 17 _ L _ 18 8a'l5o' 16a' 4 d 4 s ' ! ' 5s 5 d 6 d -6s7s [/, , " 113a 4p I I I cz bp Gp-fp— 8o*l5a' 16a' 4 d  4 * ! ! ! 5 s 5 d 6 d -6»7s u 17a 4^ 5p €^7J»"* B a ' l S o ' r 6 o ' 4 d 5<J6d~ 4 s ! ! ! 5s 6s7s '/, L 14a 19 _ 1 _ 20 I LOW ENERGY REGION T " 5 3 4 5 6 —I I I I - r — 7 — r 8 10 14 18 ENERGY LOSS (eV) Figure 7.1.8 Detailed (insert) and long range VSEELS spectra of DMSO. The Rydberg manifolds show the transition energies predicted from quantum defects (see text for details). Table 7.1.9 Energies, term values and possible assignments for features in the VSEELS spectrum of DMSO F e a t u r e * 3 * E n e r g y L o s s ( b ) Terra V a l u e (eV) P o s s i b l e A s s i g n m e n t ( e V ) u a . 7 a " 1 3 a ' I 2 a ' , 6 a " 5 a " 4 a " , I l a ' 1 0 a ' ' V "V 1 5 .526 3 . 5 7 - - - - - U a ' > 8 a " , 4 s 2 6 . 0 2 9 3 . 0 7 - - - - - - U a ' * 1 5 a ' 3 6 .594 - 3 .51 - - - - 7 a " -»• 8 a " , 4 s 4 6 . 8 9 6 2 . 2 0 - - - - - - 1 4 a ' * 4 p , 16a* 5 7 .123 - 2 . 9 8 - - - - - 7 a " + 1 5 a ' 6 7 . 4 3 2 1 .67 - - - - - - 1 4 a ' * 4 d , 5 s 7 7 .924 1 .18 2 .18 - - - - - 1 4 a ' + 5p; 7 a * • 4 p , 16a* 8 8 . 3 9 4 0 . 7 1 1.71 - - - - - 14a* -»• 6 p ; 7a* > 4 d , 5 s 9 8 . 6 6 6 - - 3 . 4 6 - - - - 1 3 a ' -> 8 a " , 4 s 14a* l i m i t ( 9 . 1 0 ) ( c ) 0 - - - - - - 1 4 a ' > "> 10 9 . 2 9 8 - 0 . 8 0 - - - - - 7 a * + 5 d , 6 s 11 9 . 5 3 6 - 0 . 5 6 - - - - - 7 a " -»• 6 d , 7 s 12 9 .912 - - 2 . 2 2 3 . 5 - - 1 3 a ' * 4 p , 1 6 a ' ; 1 2 a ' , 6 a " + 8 a " , 4 S 7 a " l i m i t ( 1 0 . 1 0 ) ( c ) - 0 - - - - - 7 a " + » 13 1 0 . 5 4 3 - - 1.59 - 3 . 4 - - 1 3 a ' + 4 d , 5 s ; 5 a " + 8 a ' * , 4 s 14 10 .913 - - 1 .22 3 . 0 - 1 3 a ' -*• 5 p ; 5 a " * 1 5 a ' 15 11 .882 1.5 - 3 . 1 1 2 a ' , 6 a " * 4 d , 5 s ; 1 0 a ' + 1 5 a ' 1 3 a ' l i m i t ( 1 2 . 1 3 ) ( c ) - - 0 - - - - 1 3 a ' 16 12.175 - - - 1.2 1.7 - - I 2 a ' , 6 a " • 5p; 5 a * * 4 d , 5 s 17 1 2 . 9 6 5 - - 0 . 9 1.6 - 5 a " > 5 d , 6 s ; 4 a * . l l a , > 4 d , 5 s ; N 2 I m p u r i t y 1 2 a ' , 6 a " l l m i t s ( 1 3 . 4 ) ( c ) - 0 - - - 1 2 a ' , 6 a " > » 18 1 3 . 8 4 6 - - - - . 1 . 2 1 0 a ' -» 5p 5 a ' l i m i t ( 1 3 . 9 ) ( c ) - - 0 - 5 a " * » 4 a " , l l a , l l m i t s ( 1 4 . 6 ) ( c ) - - - - 0 - 4 a " , 1 1 a ' > » 10a l i m i t ( 1 5 . 0 ) ( c ) - - - - 0 1 0 a ' + » 19 15.197 - - - -20 16 .961 _ _ _ _ _ (a) See Figure 7.1.8. (b) Estimated uncertainty is ±0.021 eV. (c) Ionization limits from UPS measurement [225]. 270 partially resolved fine structure evident in each band. These features arise from transitions to unoccupied virtual valence levels and/or Rydberg levels. The valence spectrum is considerably more complex than the inner-shell spectra due to the large number of closely spaced initial state orbitals possible for valence excitation. Features 1 and 2 have term values equal to 3.57 and 3.07 eV relative to the ionization energy (9.10 eV [225]) of the HOMO orbital (14a* ). Feature 1 can be assigned as a transition to either the first virtual valence orbital 8a" and/or 4s Rydberg orbital from the 14a' orbital. At first sight, the sharp peak of feature 2 would most likely indicate a Rydberg-type transition. However, the term value of 3.07 eV does not correspond to any of the term values for Rydberg excitation in the core spectra, which would (unlike transitions to virtual valence orbitals) be expected to be transferable. Hence feature 2 is sharp primarily because of the non-bonding character of the initial orbital 14a' which is composed of mainly S(3s, 3px, 3pz) and 0(2px) character according to the Gaussian 76 calculation. Accordingly feature 2 is assigned as 14a' -»15a' so that the energy difference of the 14a' •* 8a" and 14a' -+ 15a' transitions is consistent with the energy separation of 0.5 eV as indicated by the core spectra, although the term values themselves are, as expected, much larger in the core spectra due to the localized nature of the hole. In valence excitation the valence-shell hole is delocalized and thus the term value is smaller. Based on the average energy separation of 0.8 eV between the 15a' and 16a' states in the core spectra, feature 4 is assigned as the 14a' -»16a' transition. In order to predict the positions of possible transitions from the other normally occupied valence orbitals to the 8a", 15a' and 16a' virtual valence orbitals, the term values of 3.57, 3.07, and 2.30 eV found for features 1, 2 and 4 (see above) have been subtracted from the ionization potentials of the various other occupied valence orbitals. A similar procedure has been found to be helpful in rationalizing the VSEELS spectra of other molecules [64-66]. As shown in Figure 7.1.8, the agreement obtained between the predicted energies of the additional valence-valence transitions (based on the above procedures) and major features in the spectrum is reasonable. 271 Based on the expected transferability of the term values for the Rvdberg levels between VSEELS and ISEELS spectra [64-66], the contribution from Rydberg transitions to the VSEELS spectra can be tentatively identified. From the S 2p, C Is and O Is ISEELS spectra of DMSO the mean term values for 4p and 4d Rydberg levels are predicted to be 2.30 and 1.64 eV respectively. By means of the Rydberg formula (Eq. 2.1) the quantum defects of the s, p and d series are 2.05, 1.57 and 1.12 respectively. The term values of the 5s, 6s, 7s, 5p, 6p, 5d and 6d Rydberg levels are calculated and listed in Table 7.1.10. Based on these term values the positions of the Rydberg transitions relative to the various ionization edges can be estimated in the same manner as described above for the virtual valence transitions. The predicted positions of the various manifolds are shown in Figure 7.1.8. The measured positions of all peaks assigned as valence -* virtual valence and valence -»Rydberg transitions are listed in Table 7.1.9. Feature 17 is believed to be at least partially due to an N 2 impurity because the intensity of this feature, relative to all the other features in the valence spectrum, varies with sample pressure in the collision chamber and the position of this feature (12.965 eV) is close to that of X * I + -• b 'TI transition of Since the IPs are known up to the level of g u 2 r 10a' only, features 19 and 20, which have energies higher than the IP of 10a' , are not covered in the above predictions. These features presumably arise from transitions of electrons from the deeper-lying valence orbitals that have IPs higher than the energy of the He(I) line. The preceding assignments, particularly in the case of the valence-shell, are considered to be tentative pending detailed and accurate theoretical calculations of transition energies and intensities. 7.1.4 Conclusions The spectra for all inner-shell and valence-shell electronic excitations in DMSO have been obtained for the first time. The spectral features are tentatively assigned in a unified manner, in terms of excitations to a common manifold of virtual valence and Rydberg orbitals. Although the details of theseN proposed assignments need to be examined by- accurate calculations and by further comparison to the spectra of related molecules, the 272 Table 7.1.10 Summary of term values for assignments of the ISEELS and VSEELS spectra of DMSO Term Value (eV) Final S 2p 3^ 2 S 2 p ^ 2 S 2s S Is C Is 0 Is Valence State 8a" 4.73 4.77 V ") 3.81 } 3.57 V 4.87 V4.6 f 4.01 15a' 4.27 4.23 1 J 1 J 3.07 > 3.59 4s . 3.47 3.59 — — J 3.56 3.57 16a* 3.25 3.43 — — 2.77 2.76 2.20 4p 2.29 2.39 — 2.23 - - 2 . 3 0 ( a ) 4d,5s 1.65 1.65 — — - - 1.61 1 .64 ( a ) , (1.56) 5p - - — — 1.4 1.37 — ( 1 . 1 6 ) ( c ) 5d,6s — - - - - — — 0.96 ( 0 . 9 0 ) ( c ) , (0.87) 6p — — — 0.4 — — ( 0 . 6 9 ) ( c ) 6d,7s - - 0.58 — — — 0.66 ( 0 . 5 7 ) ( c ) , (0.55) (a) Estimated as averaged value of inner shell term values. (b) The ns term values are calculated by the Rydberg formula using the quantum defect of 4s (2.05). (c) Calculated values by the Rydberg formula using the quantum defects of 4p (1.57) and 4d (1.12) determined from average term values in (a) above. 273 systematics of the present assignments are in accord with previous analyses of inner-shell spectra. In particular, when comparing term values for excitations to a common Rydberg and virtual valence orbital from all different initial orbitals (summarized for convenience in Table 7.1.10), it is found that there are essentially constant Rydberg term values and reasonably constant virtual valence term values for excitation from the S 2p, S 2s and S Is orbitals. There are, however, significant differences among the term values for excitations from the sulfur core, ligand (C Is, O Is) and valence-shell orbitals to the 8a", 15a' and 16a' virtual valence orbitals. The energy positions of excitations to a (S-C) and a (S-O) states with respect to the IPs are found to be consistent with the bond-length correlation concept when compared with the inner-shell spectra of other sulfur compounds. A good correlation between bond lengths and S 2p-» o (S-C) term values is found when all available S 2p spectra are examined. 274 ELECTRONIC EXCITATION OF SULFUR COMPOUNDS : II. INNER SHELL EXCITATION OF S 0 2 BY HIGH ENERGY ELECTRON IMPACT A COMPARISON WITH MUTICHANNEL QUANTUM DEFECT THEORY 7.II.1 Introduction Sulfur dioxide is one of the major gaseous pollutants that threatens both our metropolitan areas and natural environments. It is released into the atmosphere as a result of volcanic activity and the combustion of fuels used for heat, power and transportation. In recent years the absorption spectrum of S 0 2 in the UV and VUV regions of the electromagnetic spectrum has received considerable attention. Much of this research has been stimulated by the desire to understand mechanisms for the removal of SO2 from the Earth's atmosphere [316]. Furthermore, the finding of S 0 2 in the atmosphere of Venus [317-319] and the satellite Io of Jupiter [320,321] suggests that S 0 2 is a major source of S and O atoms in these planetary atmospheres. In addition to fundamental interest in the electronic structure and excited states and electronic transition probabilities of S 0 2 > such spectroscopic data may be of assistance in current terrestrial atmospheric pollution research as well as in the analysis of observational data from planetary atmospheres and interstellar space. To date no ISEELS spectra have been reported for either the S 2p, 2s or the O Is spectral regions of S 0 2 - There has, however, been discussion of these regions in earlier reported soft X-ray photoabsorption studies [69,97,122,123,303,322-326] obtained at lower energy resolution than that used in the present ISEELS work. Some apparent inconsistencies in these earlier works are investigated in the present more detailed, higher resolution ISEELS measurements. Existing valence shell studies of electronic excitation in S 0 2 include UV-photoabsorption [327-331], an EUV study by synchrotron radiation [332], valence-shell electron energy loss spectroscopy (VSEELS) at low impact energies and variable angle [333-335] and also several theoretical investigations [336,337]. 275 The ionization energies of the valence and core electrons in SC^ are known from measurements using photoelectron spectroscopy. He(I) photoelectron spectra of SC^, which give the outer valence IPs, have been reported by several groups [217,338,339]. The ionization energies of S 2p^ / 2 and O Is as obtained by X-ray photoelectron spectroscopy have been reported in a tabulation by Jolly et al. [118]. Most earlier ISEELS and many VSEELS spectra have thus far been interpreted on an empirical basis using the concepts of quantum defect and term values. The usual simple quantum defect treatment has assumed fixed values of quantum defect within certain specific ranges for each of s, p and d Rydberg excitations, respectively, regardless of principle quantum number. The related term values have alternatively been used and generally considered to be transferable for transitions involving different initial and the same final states. This transferability of term values has been shown by Sodhi et al. [64] to be a reasonably useful general concept for valence-shell to Rydberg and inner-shell to Rydberg transitions, but it is certainly not of high accuracy. However, transferability is not a generally useful concept for transitions to normally unoccupied virtual valence MOs from valence-shell or from core shell initial orbitals [64]. The assignments based on approximately constant quantum defects and term values are thus rather problematical, often speculative and at best tentative. Such assignments are probably reasonably adequate for widely spaced Rydberg levels with only simple symmetry manifolds, such as occur for instance in atoms and also in the very simplest molecules. However, the situation is much less satisfactory for larger molecules for final states of both virtual valence and Rydberg character. Much better assignments can be made if the measurements are interpreted on the basis of accurate concomitant theoretical calculations of the transition energies. Even more definitive spectral assignments and interpretation can be made if the calculated energies are supplemented by the calculation of the respective oscillator strengths. An excellent example of such a quantitative comparison of theory and experiment is the work by Hitchcock et al. [51] for the VSEELS electronic excitation spectrum of hydrogen fluoride, 276 in which time-dependent Hartree-Fock calculations of the energies and oscillator strengths permitted a very comprehensive and detailed assignment of the observed spectrum [51]. A powerful alternative theoretical approach for investigating molecular spectra is multichannel quantum defect (MCQD) theory, first suggested by Seaton [53]. This theory, which calculates a unique energy-dependent quantum defect for each particular (identified) initial and final state combination as well as the corresponding absolute oscillator strength, permits a very accurate quantitative prediction of the electronic spectrum. As such, MCQD theory is expected to provide a powerful interpretive tool in assigning the transitions from core orbitals as observed in ISEELS spectra. In the present study, detailed ISEELS studies of SC^ in the S 2p and 2s, as well as in the O Is regions are reported. In the present spectrometer [91] the S Is excitation (~2.48 keV) is not accessible under dipole-allowed conditions since the maximum value of the impact energy is at present 4 keV. However, Bodeur et al. [163] have recently published the S Is photoabsorption spectrum of SC^ obtained using synchrotron radiation, and these results are compared with the new ISEELS measurements for the other core levels in the present work. Multichannel quantum defect theory calculations of transition energies and oscillator strengths for transitions from all inner-shells of SC^ have also been carried out by Tong and Li [210] as part of the present collaborative study. The results of these calculations are used to interpret and assign the various experimental spectra. 7.II.2 Mutichannel Quantum Defect Theory Excited atoms or molecules usually consist of a residual ion and an excited electron (either bound or unbound). The excited states form infinite Rydberg series of states (which may be autoionizing states in appropriate cases) and adjoint continua. In the case of the molecules, the excited states may subsequently dissociate into various fragments (i.e., predissociated states). All these diverse and rich spectral phenomena can be treated in a unified manner by Quantum Defect Theory [53,340-344]. More specifically, the 277 structures in all . spectral regions are described in terms of a compact set of slightly energy-dependent physical parameters, namely the eigen quantum defects M q and the orthogonal transformation matrix U j C , which are related to the short-range scattering matrix [343] for collision (i.e. the interaction) between the low-energy electron and the residual ion. The eigen channel quantum defects u^, for eigen state o , measure the extra phase-shifts in the excited electron wavefunctions of all ionization channels i due to the non-Coulombic many-electron interactions between the excited electron and the residual ion within the reaction zone forming the excited atomic or molecular complex [3431. The orthogonal transformation matrix I_L o describes the couplings between the eigen channels a and the ionization channels i when the excited electron departs from the reaction zone through various ionization channels into the separation zone. As the excited molecules may be dissociated into a variety of fragments through various dissociation channels d, there exists a set of physical parameters [344] that describes the couplings between the eigenchannels o and the dissociation channels d. Theoretical calculations of the electronic part of the physical parameters, u_ and U j a > due to all core excitations in the SC»2 molecule has been performed by Tong and Li [210]. The main points of the computational procedure are outlined below. Based on the multiple-scattering method [271,345-347], the wavefunction of the molecular excited electron can be calculated in a self-consistent molecular potential. This self-consistent molecular potential, which is calculated by the multiple-scattering, self-consistent field method [346-348], has the following form: (1) a central potential within each atomic sphere; (2) a central potential with the appropriate Coulombic tail outside a sphere that inscribes the atomic spheres; (3) a constant potential in the interstitial region. The calculated wavefunctions of the molecular excited states are expressed in the following form [346,347]: (1) Within each inequivalent atomic sphere n: 7.1 278 where the r . is the radial distance with respect to the atomic center. The £ is the orbital energy in atomic units. Here (r R) is an A -partial-wave radial wavefunction that satisfies the Schrodinger equation of the central potential and is normalized within the atomic sphere. The expansion coefficients o are determined by matching boundary conditions in the multiple-scattering method. The symbol r represents an irreducible representation of the molecular point group and the a is the index for the degeneracy in the r-irreducible representation. The P n is the set of the equivalent atomic spheres. In each atomic sphere i belonging to P , the wavefunction is expanded into partial wave JL with {L : 4 < A 3. The Y „ „ is the real spherical harmonic. The symmetric n n A m adapted coefficient m (i) is the normalized projection coefficient of the spherical harmonic A m into the symmetric state ro. The index 6 represents the repeat time of projection. (2) In the interstitial region: ^ T O ( f , e) = 2 I I B 8 . W W N Y v v m S , ; n n 7.2 E I I B 0 n.(kr) Z Z X » i) Y ( r.) Z B 8 j . (kr) Z X™-6 Y (r) where the r is the radial distance with respect to the center of the outer sphere inscribing all the atomic spheres. The j ^ and n^ are the spherical Bessel function and spherical Neumann function, respectively, where the k is the wave number with respect to the interstitial constant potential. The expansion coefficients, ^ Q and B ^ Q are determined by matching boundary conditions. (3) Outside the outer sphere given by equation (3): ¥ T O ( r ) = Z Z C 5 F " ( r ) I X ] ° ' 5 Y / m ( r ) The wavefunction is expanded into various partial waves(^ 6). The expansion coefficients c are also determined by matching the boundary conditions. The (r) is the ^-wave radial wavefunction, which has the appropriate asymptotic forms, as follows: 279 A: For continuum states £ >0, F"( r ) f ( r ) cos JCU - g ( r ) sin 7m A ct ct ' 7 4 where u. Q is the eigen quantum defect The f (r) and g (r) are the regular and irregular Coulombic wavefunctions, respectively. Thus, U. = Q 8 7 5 which is the electronic part of the transformation matrix with the ionization channel i consisting of the partial wave(-t*8). The eigenchannel a is characterized by the r irreducible representation. B: For discrete bound states e = — — < 0, 2 V 2 F a ( r ) i ^ — [ u i ( r ) s i n j c ( v + n a ) - v l ( r ) _ c o s j c ( v + . | x o ) ] , 7.6 n where the u^ (r) and v (r) are exponentially increasing and decreasing Coulombic functions, respectively [349]. Thus, v = n - [ia , 7 j 2 where n is an integer. The N is normalization factor. The N can be regarded as the n n c "density of states" for the bound states, namely N 2 = V 3 + (— a) n • , 7.8 n de n 3 which is the famous v rule for Rydberg series [350]. In a specific eigenchannel c, the uo and I L Q vary smoothly with the excited orbital energy. Figure 7.II.1 shows a typical example of the quantum defect variation in a particular symmetry manifold and exit channel. The eigen quantum defects are shown for the a eigenchannels in the A-^  representation of the C 2 v molecule SO^ Based on the transformation matrix U.a, the dynamics of the eigenchannels can be further characterized: for example, the eigenchannel sa-^  belonging to the A^ representation has a dominant s 280 2.56 2.24 H 1.92 | _ I.60H O W 128-1 LU Q 0.96 Z> 0.64-i < 0.32-| O o.O -0.32 -0.64 I P 7sai 5sa, 6sai 4 s a i •pa, 7pai 5 pa, 6pai ' 4 f 0a, 3 d 2 0 | 4 d 2a, NEUTRAL EXCITED STATES i i i sa, pa. f 2a. d 2a, CONTINUUM i -4 -3 -2 -I I TERM VALUE (eV) Figure 7.11.1 Calculated quantum defects for eigen channels in the Ai representation of SC*2. 281 partial wave outside the outer sphere. The eigenchannels with dominant p partial waves outside have three distinct dynamical characters belonging to the A^, and B2 irreducible representations, respectively. There are six eigenchannels with the A^ character up to f-wave outside the outer sphere, as shown in Figure 7.11,1, namely sa^, pa-^ , d 2ap dga^, f ^ and f ^ . The detailed dynamical characters of the eigenchannels in the Ap Bj, B J and A 2 representations have been discussed elsewhere [347]. In each eigenchannel it can be seen that the calculated eigen quantum defect indeed varies smoothly. Thus, all the bound states in a specific eigen channel form a Rydberg series converging to the threshold. With the calculated wavefunction, the oscillator strength f can be calculated and then the oscillator strength density is defined in terms of the density of states N n as * = f -NJ dE n n 7.9 which is a continuous function of the orbital energy in each eigenchannel. As the momentum transfer in the present experiments is negligibly small (i.e., forward scattering at high incident electron energies), the electron-impact inelastic collision processes are dominated by dipole selection rules and the resulting spectra are equivalent to photoabsorption processes (see Chapter 1). Below the sulfur 2p edge of SC>2, photoexcitations from the sulfur 2p core state give rise to six eigenchannels, namely sa^, pa^ dga^, pb2, db^ and da 2 > with appreciable oscillator strength densities due to electron-dipole allowed excitations. Because of the spin-orbit interaction, there are two ionization edges O-V-^/i and 2p^ / 2) separated by about 1.3 eV [87]. Thus, each above-mentioned eigenchannel is separated into two Rydberg series (denoted without a bar and with a bar) converging to the 2 p j / 2 and 2p^ / 2 limits with oscillator strength densities determined by the statistical branching ratios 2/3 and 1/3, respectively. In the present calculation, the couplings between the channels converging to the 2p^ / 2 and 2p^ / 2 limits are neglected. Similar MCQD calculations of transition energies and oscillator strengths for the S Is, S 2s and O Is excitations in the S0 2 molecule have also been carried out Comparison of the MCQD calculations with the experimental spectra is made 282 in section 7.II.3 below. 7.II.3 Result and Discussion The SC>2 molecule is of C 2 v symmetry. The ground state independent particle electron configuration together with the unoccupied virtual valence orbitals may be written as [351-354], (a) Core ( l a p 2 ( l b 2 ) 2 ( 2 a / ( 3 a / ( 2 b / ( l b / ( 4 a / » y • <— V ' S Is 0 Is S 2s S 2p (b) Valence (5 a i) 2(3b 2) 2(6 a i) 2(7 a i) 2(4b 2) 2 (2b 1) 2(5b 2) 2 ( l a 2 ) 2 ( 8 a i ) 2 (c) Unoccupied valence (virtual) MQ (3b1)°(9a1)°(6b2)° - Minimum Basis Set or (3b1)0(9a1)0(6b2)0(4b1)0(10a1)0(7b2)0(lla1)°(12a1)0 - Extended Basis Set (including S 3d orbitals). The antibonding virtual MOs in the Minimum Basis Set have been designated as bj, a* and b 2 in the present work. Where the possibility of the extended basis set (i.e., inclusion of S 3d orbitals) is considered the additional virtual MOs have been labelled 4b^, lOaj, 7b 2 > l l a ^ and 12a^ [303]. The various spectra are conveniently discussed with respect to these configurations and to the dipole-allowed transitions shown in Table 7.II.1. 7.II.3.1 Inner-Shell Spectra 7.II.3.1.1 S 2p Spectrum Figure 7.II.2(upper section) shows the S 2p excitation between 163 - 176 eV at 6 = 0° and E o = 2000 eV. The energy resolution is 0.055 eV FWHM. The 2p 3 / 2 Table 7.11.1 Transitions in SO2 from the 1 A i ground state for C2v symmetry (a ) Final Configuration ' Final State Dipole Allowed From Ground State Hole State Occupied Virtual M.O. a. b l B l Yes a l a l ' A l Yes a. b2 . B2 Yes b l b l A l Yes b l a l B l Yes b l b2 A 2 No b2 b l A 2 No V a l B2 Yes b2 b2 A l Yes a2 b l B2 Yes a 2 a l A 2 No a2 b2 B l Yes (a) (2s) ' 1 hole is of z-^ symmetry. (2p)- 1 hole is of a-i, or b 2 symmetry. 284 CO LU h -10-LU > _l LU 5 i 0 I -"O _] __J 2 O CO 0 . o u EQ = 2000 eV 6 = 0° AE = 0.055 eV S0Z S 2p i i 9 0 | I 2 I 3 14 I I " " 22 30 SS 37 3941 17 2 l l 23 24 27 29 131 321 36 / 36' I 142 _ l LU I 1 L_J U U I I I I I 6 12345178 I I I I I I I I e fe 20 23 2 » l / 2 ' e d « e _ l U I I L U _ I 1 U l_LLL_ 2p 3/2 edge (a) EXPERIMENT i n mm 163 165 167 169 (b)MCQD CALCULATION —r-171 (4sa,)f T f 163 173 i k 175 t t f /dEtlO-'eV 1 ) P3 2 I m i_L bit. (4s5,)? df/dE(IO"W) .0 1 iii, h 1 j k I m p r s v K 2P 165 167 169 171 173 ENERGY LOSS (eV) 175 Figure 7.11.2 (a) S 2p excitation spectrum of SO2 (163-176 eV); (b) multi-channel quantum defect theory calculation. 285 Table 7.11.2 Experimental and calculated data for S 2p excitation in SO2 Experimental MCQD Calculation P e a k ( a ) Term Value(eV) Energy Loss^^ Energy Osci l lator Assignment:3^ Desig*- a ) Number 2 p 3 /2 2 P l / 2 (eV) (ev) Strength nation 1 10.41 - 164.39^ 2 10.28 - 164.52 3 10.16 - 164.64 4 10.03 - 164.77 164.57 8.50X10" 1* 3b : * b l 5 9.90 - 164.90 6 9.78 - 165.02 7 9.67 - 165.13 8 9.56 - 165.24^ 9 - 10.36 165.74 ^ 10 - 10.22 165.88 165.87 4.30X10- 1 1 3 b l _ * b l 11 - 10.09 166.01 12 - 9.98 166.12 13 - 9.87 166.23 14 - 9.75 166.35^ 15 5.58 - 169.22 169.45 1.24xl0 - 2 9 a l * a l 16 4.49 - 170.31 170.40 1.84xl0- 2 6b 2 * b 2 17 - 5.55 170.55 170.75 6.20xl0 - 3 9 a x _ * a l 18 4.14 5.44 170.66 19 4.01 5.33 170.79 286 Table 7.II.2 continued Experimental MCQD Calculation P e a k ( a ) Term Vc ilue(eV) Energy Loss^^ Energy Osci l lator Assignment:3^ Desig^ a ) Number 2 p 3 /2 2 P l / 2 (eV) (ev) Strength nation 20 3.49 - 171.31 171.05 1.98x10-2 4sa 1 f 21 - 4.41 i7i.69 y ("171.70 9.20X10"3 6b 2 _ * b 2 22 3.02 4.32 171.78 f (172.03 1.76xl0 - 3 4pa 1 g 23 2.89 4.19 171.91J 24 - 3.62 172.48 172.35 9.90xl0- 3 4sa 1 f 25 2.14 - 172.66*)- 172.78 2.53xlO- 3 4pb 2 h 26 2.09 - 172.71J 172.90 3.33x10-3 3dbx i 27 1.76 3.06 173.04*) 172.95 2.10x10-3 5sa 1 j 28 1.49 2.79 173.31J 173.18 2.20x10-3 3dax k 173.27 2.30xl0- 4 3da 2 Jl 173.33 8.80x10" 4pa x i "173.52 4.20xl0 - 1* 5pa x m 173.74 8.00xl0 _ l + 5pb2 n 29 1.24 2.54 173.58) 173.75 1.73xlO- 3 4dbx 0 30 1.03 2.33 173.71VI 4 173.81 7.87xl0 _ , t 6sa^ P 31 0.91 2.21 173.897 173.88 9.30xl0 _ l t 4daL q 173.92 2.70xl0 _ l t 4da 2 r 174.06 1.70X10"14 6pax s J74.08 1.27x10-3 4pb 2 h 287 Table 7.II.2 continued Experimental MCQD Calculation P e a k ( a ) Number Term ilue(eV) Energy L o s s ^ (eV) Energy (ev) Osci l lator Strength Assignment^ Desig^ a ) nation 2 p 3 / 2 2 p l / 2 r174.15 8 . 0 0 X 1 0 - 1 1 6pb 2 t 174.15 4 .00xl0 - u 5dbL u 32 0.52 1.82 174.28 11 S174.19 4.30xl0- 4 7sa ^ V 174.20 1.67xlO- 3 3db1 i 174.22 5 .10xl0 - 4 5da : w 174.24 2.10xl0 - 1 4 5da 2 X 174.25 v. 1.03x10-3 5 5 8 ^ j '174.31 9.00x10-5 7pax y 33 0.35 1.64 174.45 ' 174.48 l . l O x l O - 3 3da1 k ,174.57 l . lSxlO- 1 * 3da 2 I 34 0.25 174.55 higher 35 0.19 174.61 Rydbergs 2 P 3 / 2 0 (174.80)( C ) 174.82 2 . 1 0 X 1 0 - ' 4 5pa 1 m 175.04 4.00*10 - 1 + 5 P b 2 n 36 1.20 174.90") 175.05 8 . 7 0 X 1 0 - 1 * 4db : 0 f i l l 37 1.09 175.OlJ 175.11 4 . 0 0 X 1 0 - 4 6sa 1 P 175.18 4.70 x10- 4 Ud&l q ^175.22 1.30*10-1* 4da 2 r 288 Table 7.II.2 continued Experimental MCQD Calculation (a) Peak. Number Term Value(eV) 2p 3/2 2p 1/2 Energy Loss (eV) (b) Energy (ev) Osci l lator Strength Assignmen l a ) D e s i g £ a ) nation 38 39 AO Al A2 2p A3 AA A5 1/2 0.75 0.65 0.53 0.A1 0.32 0 175.35) 175.A5j IV 175.57 175.69 175.78 (176.10) 175.5A 175.61 9.00xl0 - s 2.00xl0-1< A.00-10 - 4 2.10xl0 - l t 2.50*10-1+ 1.10*10-1+ A.50xl0 - 5 6paj 6pb2 5db 7sa 5da 1 s t u 5da 2 7pa 1 higher Rydbergs (c) -3.0 - 1 1 . 3 ° -19.5 d 178.A 186.7 19A.9 l l a 1 (4-lOaj + 7b 2) Resonances Suggested assignment ? Resonance only (no MCQD calculation) 12ay Resonance (a) For peak numbers, assignments and designations see Figure 7.II.2 for peak numbers 1-42 and Figure 7.II.4 for peak numbers 43-47. A bar placed over a symbol indicates a state associated with the 2 p 1 / 2 limit. (b) Estimated uncertainty is ±0.04 eV for features 1-42 and ±0.25 eV for peak numbers 43-45. (c) 2 p 3 / 2 , , / 2 ionization limits from X-ray PES [87]. (d) Term value with respect to the mean of the 2p3 / 2,i/2 edges (175.4 eV). 289 ionization edge (174.80 eV) has been assigned using the XPS value for SOj reported by Siegbahn et al. [87]. Very similar values have been recorded by Jolly et al. [118] and also by Jen and Thomas [355]. A value of 176.1 eV has been assigned to the 2p^ / 2 ionization edge according to a yi s P m 0TD^ splitting of 1.3 eV [87]. The fine structure in the first two weak bands between 164 and 167 eV will be discussed separately in a following section (see also Table 7.II.3 and Figure 7.II.3). The energies of all the clearly identifiable observed peaks (numbered 1-42) and both of the S 2p edges are listed in Tables 7.II.2 and 7.II.3. The results of the MCQD calculation below the S 2p edges (see section 2 above) are shown in the lower portion of Figure 7.II.2. It can be seen that there is a very good quantitative agreement between the calculation and experiment for both the energies and the intensities (oscillator strengths). The calculated energies, oscillator strengths and assignments are shown in Table 7.II.2. The results of the MCQD calculation indicate that as the excitation energy increases, it first leads to the antibonding states b-^  and b^ with respect to the 2p^2 afld ^V\/2 ec*2es, respectively, and then to two pairs of antibonding states (a^, a-^ ) and (bj, b 2). The states a^ and b 2 are too close to be resolved experimentally, as shown in Figure 7.II.2. Thus, the sum of the oscillator strengths of the a^ state and the b 2 state is larger than the oscillator strength of the state, while that for the a^ state alone is only one half of the oscillator strength for the a^ state. It is interesting to note from the work of Bodeur et al. [163] that below the sulfur Is edge (where there is no spin-orbit interaction) there are clearly three antibonding states (b^, a-^ , bj) and that the assigned b 2 state is higher in energy than the a^ state by about 0. 5 eV (see discussion of Figure 7.II.6 below). The work of Bodeur et al. [163] is consistent with the present assignments based on the MCQD calculations. Above the antibonding (valence) states, the two series of Rydberg states (f-y, f-y, see Table 7.II.2, Figure 7.II.2) are found. For the low excitation energies (energy losses), the experimental 2 line-profile width A is smaller than the inverse of the density of states N . Therefore in this region, in order to make a reasonable comparison with the observed (total electronic; 1. e., summed over all final rovibronic states) line intensities at lower energies, the 290 • « » theoretical intensities for the bp a^, b 2 and 4sa^ states and their spin-orbit counterparts have been plotted in histogram form with the f /A as height and the A as width. As the excitation energies increase, the experiment linewidth A becomes comparable to the inverse of the density of states. Therefore in order to provide a simple visual comparison with experiment in the upper energy region the oscillator strength densities (df/dE) for the higher Rydberg states have been plotted. The observed peak I consists principally of the Rydberg states, 5pa^, 5pb2> 4db^, 6sa^, 4da^ and 4da2. Similarly, the observed peak II includes the Rydberg states 6pa^, 6pb2> 5db^, 7sa^, 5da^ and 5da2- The peaks I and II are mainly composed of the highest Rydberg states converging on the 2 p 3 / 2 ionization edge. The pair of peaks, III and IV, include the corresponding transitions, leading to the 2p-jy2 edge. As the excitation energies approach the edges, the A becomes larger than the inverse of the density of states and the spectra become quasi-continuous. There is overall excellent correspondence between the calculated spectra and the measured electron energy-loss spectra for almost all features below the S 2p 3 / 2 ^ / 2 edges of S0 2. It can be seen that essentially all details of the experimental spectrum, including unresolved and partially resolved bands are reproduced by the calculation. On this basis the assignments, as shown in Table 7.II.2, can be made with considerable confidence. A more detailed, higher resolution (0.046 eV FWHM) spectrum of the 164-167 eV region is shown in the lower section of Figure 7.II.3. The soft X-ray photoabsorption measurements for this region, as reported earlier by Krasnoperova et al. o [122] at a somewhat lower resolution of 0.05 A (i.e., 0.12 eV) FWHM, are shown in the upper part of Figure 7.II.3. Obviously, considerably more detailed information is available from the high resolution ISEELS spectrum, which shows the presence of five vibrational progressions. In addition to the broad features (a-f) visible in the soft X-ray spectrum, the spectrum is significantly further resolved in the ISEELS spectrum so that at least 14 vibrational features (numbered 1-14) are now observed. The energy values of the corresponding but broader features in the soft X-ray spectrum are consistent with the more detailed ISEELS results (see Table 7.II.3). However, the energy values shown in the photoabsorption spectrum and those given in the corresponding table of Ref. [123] are 291 CO _ z LU LU > __ UJ rr 10 7 SO S2p (!64-l67eV) Krasnoperova 1976 (PHOTOABSORPTION) AE~O.I2eV e f This work (ISEELS) A E = 0 . 0 4 6 eV E o = 2 0 0 0 e V I 2 3 4 5 6 7 8 I I I I I I I I 9 10 II 12 13 14 I I 1 I I I 164 165 166 ENERGY LOSS (eV) 167 Figure 7.11.3 Comparison of the (a) photoabsorption, taken from Ref. [122] and (b) high-resolution electron energy loss S 2p excitation spectra of SO2 in the energy range 164-167 eV. 292 Table 7.11.3 S 2p excitation of SO2 at high resolution (164-167 eV) Possible Feature (a) Energy Loss (eV) Term Value ( e v / b ^ 2 p 3 /2 2 p l / 2 Assignment fc) Photoabsorptionv ' 1 164.39 10.41 — B :(v=0) 164.3 2 164.52 10.28 B ^ v - l ) 3 164.64 10.16 Bx(v=0) 164.6 4 164.77 10.03 B 1(v=l) 5 164.90 9.90 — A^v-0) 164.8 6 165.02 9.78 — A 1(v=l) 7 165.13 9.67 A1(v=2) 8 165.24 9.56 — Ai(v=3) 9 165.74 — 10.36 Bl(;v=0) 165.7 10 165.88 10.22 B 1(v=l) 11 166.01 10.09 AL(v=0) 166.0 12 166.12 — 9.98 A 1(v=l) 13 166.23 — 9.87 Ax(v=2) 14 166.35 — 9.75 Ai(v=3) (a) Estimated uncertainty is ±0.04 eV. (b) S 2p3/2,1/2 edges are at 174.8 and 176.1 eV respectively, see Ref. [87]. (c) See Ref. [123]. 293 inconsistent In the present work the tabulated values have been assumed to be coned It can be seen that, even taking into account the resolution differences in the two experiments, the spectral shapes and relative intensities in the ISEELS and soft X-ray spectrum are not identical, especially for the first band. The energies, term values and possible assignments of the first two bands in the S 2p region are shown in Table 7.II.3. In accord with the MCQD calculations and also the large term values (~10 eV), these two bands are nominally assigned as the 2p-jy2 "* 3b^ and 2p^ 2 •* 3b^ transitions (denoted as b^ and b^ in the present work). The transfer of an electron from the S 2p orbital (2b2, lbj, 4ap to the 3b^ orbital should produce excited states A 2 > B^ and (see Table 7.II.1). In the dipole approximation the transition to the A 2 excited state is formally forbidden. Thus taking into account the spin-orbit splitting only four dipole-allowed transitions would be expected on such a simple model. However, the spectrum in Figure 7.11:3 is much more complex with at least three vibrational progressions in the lower energy band and two in the higher energy band. It would therefore appear that extensive final state mixing is taking place to give five final electron states, as has been predicted in calculations by Shklyaeva et al. [123] using the Hartree-Fock-Roothaan method with allowance for spin-orbit coupling. The intensity distribution observed in the vibrational^ resolved spectrum is in reasonable agreement with the theoretical predictions of the electronic state intensities [123]. The intensities in the energy loss spectrum appear to be in somewhat better agreement with the calculation [123] than those in the photoabsorption spectrum [122]. The nomenclature for the states as shown on Figure 7.11.3 and Table 7.II.3 is that given by Shklyaeva et al. [123]. These states are in fact considered to be mixed. It is of further interest to note that Shklyaeva et al. [123] predict approximately equal intensities for the two separate band systems rather than the 2:1 statistical ratio expected on the basis of a simple model. This prediction of equal intensities is in general agreement with experiment. The small but significant difference in band shapes between the electron energy loss and photoabsorption spectra (Figure 7.II.4) may be due to non-dipole contributions in the former owing to the finite momentum transfer, or it may more likely be due to a spurious contribution to 294 the photoabsorption spectrum. The vibrational quantum for the vibrationally resolved states is found to be 0.12 eV, which is very similar to that for ^ (ap in the ground state of C10 2 [356], which is the so-called (Z + 1) analogy (or core equivalent) species for the S 2p excitation of SOj. A comparison with the magnitude of v ^ (0.14 eV) in the ground state of SOj [331] shows that the removal of a S 2p electron has slightly weakened the S-0 bond. Significant broad structures (features 43-45) can be seen beyond the S 2p ionization edges shown in Figure 7.II.4. These features can be described as o -shape resonances (or inner-well states) trapped by a centrifugal potential barrier [36]. Alternatively they can be described in terms of an extension to the molecular orbital scheme [302]. In the case of S 0 2 such further MOs beyond the 6b 2 are only possible if the description is extended beyond the minimum basis set to include the sulfur 3d orbitals. In this regard Kondratenko et al. [303] have suggested in a calculation of the energies and oscillator strengths for excitation of the quasi-stationary states near the ionization edge that feature 43 is composed mainly of the l l a ^ final state with weaker contributions from the 10a^ and 7b 2 states. Considering further the calculations of Kondratenko et al. [303] it is possible that the broad resonance (feature 45) at ~ 194.9 eV (term value ~-20 eV) is due to the occupation of the highest extended basis set MO, namely the 12a^ , which has a predicted term value of 14.26 eV. The weak broad feature 44 at ~ 186.7 eV may be due to excitation to the 7b 2 extended basis set continuum resonance, which would be formally dipole forbidden. This interpretation is supported by the fact that a very strong resonance appears at the equivalent position in the S Is spectrum (Figure 7.II.6). 7.II.3.1.2 S 2s Spectrum In the region immediately preceding the S 2s edge two prominent peaks (features 46 and 47) can be seen with term values of 10.03 eV and 5.35 eV (Table 7.II.4). These peaks are assigned to transitions originated from the S 2s orbital to the 3b, (b^ and >-t GO UJ 10 > LxJ so2 S 2p,2s Eo = 3000 eV e = 0° AE = 0.090 eV S2p (a) EXPERIMENT 24 S 2p edges Sll 1516 20131 43 _ U I I l l I I I I 44 _ l _ 45 _ l _ (b)MCQD > CD CM i — r-2 I -CALCULATION 0 < 0 * rti4sai 240 220 230 S2s • V * , ^ * Q l * b 2 * W \ 4sai 46 47 S 2s edge 160 170 180 190 200 210 220 ENERGY LOSS (eV) 230 240 Figure 7.11.4 (a) S 2p, 2s excitation spectrum of SO2 (160-240 eV) ; (b) mult i -channel q u a n t u m defect theory ca lcu lat ion. M O 296 Table 7.11.4 Experimental and calculated data for S 2s excitation in SO2 Experimental MCQD Calculation Peak Term Value (a) Energy Loss Energy Osci l lator Assignment Number (eV) ' (eV) (eV) Strength 46 10.03 229.07 228.87 2.69*10 - 2 * b l 47 5.45 233.65 233.75 234.70 235.35 6.9Qxl0&q