w 'r ' v Aw :‘I .. FW'II A \A'AIrAIthA'IILI :A IzhAAAHI- AII~A J AAI'AA o .A,IAA. b} gnu ‘1“??? ‘I‘AI l". ‘A. than, ‘ ‘1 A" ‘ ‘2 f‘." 3.} ' J‘ ‘ AA .A IAN. A .. 35;] \Ib'il‘.’ 5“qu I «IAI‘ 1:]. It‘lh“ . .M Y 11f; ‘ ' SAN r.:i*}fi?§?: . ’Xf .: L?‘ ‘ - “:12“ 32.14; 2% I. ., 301:... . a" 31%;- , , . 'h ‘ ' .I . ‘I 3 'f'fif'H-r‘ 3“ 471W}. ARV"; {1&3} “'3 ?I.'-""LI 4* ’e L I fig. A3; A. a A A A I u" ' I ~_ - v A u " -1. A . . L 'p u A. AA A. A I A A A .v 13". “ H hfiisl'l ;' h. l" Ii“ HilA A ‘V' A ‘Iv H _' J“- !' JAIV'M‘. 1| 'A A. ‘I ‘ I " ‘ ,"“|‘fl.l JR ‘ -' .bllll "‘ ”l'lj‘I gill!" 11"!A"rt ‘ ' . .Arl'QUA' I“. 'II"I'LA for AI.AIA .A,A AA'I. 15A,t ‘ J ""A Ivl‘hf' 'IA Allfi-UA l ,h“ :‘IIAE. ‘AA"';IA,:;§:"AI II? .” ’ A A A , [5.4, $9 .A'l A AA‘ANAi |" “fly“ :}Im:rl' “18"?" | I'll.‘jA.".. #91151?“ A‘. "I". H A»; §, {A‘J'AI AIM" ",4 9mm! ‘3' ”\“I “It“. " " .‘ .7“._ ‘_ f . ii 'r r ‘ ~1 .. V 1 ‘{ 1:13 ”1 . l V . ‘I *1, Y A“ This is to certify that the thesis entitled PHOTOIONIZATION MASS SPECTROMETRY: IONIZATION AND FRAGMENTATION OF CH3CN AND CDBCN presented by David M. Rider has been accepted towards fulfillment of the requirements for Ph . D . degree in Chemistry éwufi blajor professor Date February 11, 1980 0-7639 ‘pl-‘ls ol.‘ " fli=\\\ ' ~ .1 (lg-1:5: *- v‘» ~- y find!!!" .‘ ’3» /:V V OVERM FINE§z 25¢ par day per ‘1th MRNLNG LIBRARY MATERIALS: Place in book return to remove charge from circulation records PHOTOIONIZATION MASS SPECTROMETRY: IONIZATION AND FRAGMENTATION OF CH3CN AND CD3CN BY David M. Rider A DISSERTATION submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Chemistry 1980 ‘4‘. ‘f‘ L? //’Ll:‘ ABSTRACT PHOTOIONIZATION MASS SPECTROMETRY: IONIZATION AND FRAGMENTATION OF CH3CN and CD3CN BY David M. Rider Photoionization mass spectrometry (PIMS) is a powerful technique for investigating ionization processes and fragmenta- tion of molecular ions. Accurate ionization and fragment ion appearance potentials can often be determined and ion frag- mentation pathways and mechanisms elucidated. In this disser- tation a PIMS investigation of acetonitrile (CH3CN) and acetonitrile-d3(CD3CN) is reported. Photoionization of acetonitrile with 5843 (21.2 eV) photons causes extensive fragmentation of the parent ion. The following ions are produced with sufficient intensity to be + 2 2 2 3 Photoionization efficiency (PIE) curves of these ions and studied: CH3CN+, CH CN+, HC N+, H CN+, CH , and CH2+. their deuterated counterparts from the onset of ionization (~1016 A, 12.2 eV) to 600 R (20.7 eV) have been measured. The parent ion PIE curves (CHBCN+ and CD3CN+)demonstrate that direct ionization and autoionization both contribute to the production of ions in the region studied. Autoionizing Rydberg series converging to the first and second excited electronic states of the parent ions are observed and assigned. David M. Rider Jahn-Teller interactions in the ground state of the parent ion are indicated by the observation of the threshold for one quantum of excitation of the doubly-degenerate CCN bending mode (v8). The measured ionization potentials are 12.194 1 0.005 eV for CH3CN and 12.235 i 0.005 eV for CD3CN. Appearance potentials for the fragment ions listed above are determined, and where a more reliable heat of formation is not available in the literature,one is calculated. At their thresholds CHZCN+ and CDZCN+ are found to be produced exclusively from parent ion states which are populated by autoionization. The experimental appearance potentials and relative intensities of the remaining fragment ions indicate that H-atom migrations in the parent ion are important in the fragmentation mechanisms. To Sasha ACKNOWLEDGMENTS Words are not sufficient to express my thanks and grati- tude for help from the many individuals who made the completion of this dissertation possible. First and foremost I thank my wife, Sasha, for her loving care, devotion, and understanding throughout our stay at Michigan State University. I thank Professor G. E. Leroi for his guidance, encourage- ment and patience as my advisor and for his perspective on the human aspects of research. Dr. Edward Darland's contributions were unmeasurable. He was always willing to help solve instrumental problems and was a wonderful teacher and friend. I also want to acknowledge his help in generating several of the computer-drawn figures in this thesis. Naomi Hack I thank for many stimulating discussions about nonscientific things, for her willingness to listen and her outlook which helped to keep problems in perspective. Marty Rabb was a tremendous help in solving electronic problems. He was always willing to listen and provided many useful suggestions. The support of Ron Haas, Scott Sanderson, Steve Veile and Dick Johnson of the electronic shop, and Russ Geyer, Len Eisele, 'Dick Menke and Deak Watters of the machine shop is gratefully acknowledged. Rose Manlove I thank for the thankless job of typing the thesis. iii Financial support from the Office of Naval Research, the National Science Foundation and the General Electric Co. are gratefully acknowledged. iv LIST OF LIST OF CHAPTER CHAPTER A. B. TABLE OF CONTENTS TABLES. . . . . . . . . . . FIGURES . . . . . . . . . I. INTRODUCTION. . . . . . II. AN OVERVIEW OF PIMS . . The PIMS Experiment . . . . Photoionization Processes 1. Direct Ionization . . . 2. Autoionization. . . . Fragmentation Processes . . 1. Small Molecules . . . . 2. Statistical Theory of Mass Spectra. . . . . . Complementary Experimental Techniques. . . . . . . . . 1. Photoelectron Spectroscopy. 2. Electron Impact Energy Loss Spectroscopy . . . CHAPTER III . EXPERIMENTAL APPARATUS A. B. AND PROCEDURES. . . . . Introduction. . . . . . . . The Instrument. . . . . . . 1. Light Source. . . . . . 2. Monochromator . . . . . 3. Ionization Region . . . 4. Mass Spectrometer and Ion Optics. . . . . . . V Page No. - vii - viii . . 1 . 4 . 6 . . 10 . . 10 . . 20 . . 28 . . 29 . . 31 . . 39 . . 40 . . 41 . . 44 . . 44 . . 44 . . 44 . . 56 . . 58 . . 58 Page No. 5. Ion and Photon Transducers. . . . . . . . . . 59 6. Vacuum System . . . . . . . . . . . . . . . . 61 7. Data Collection . . . . . . . . . . . . . . . 62 C. Experimental Details. . . . . . . . . . . . . . . 59 1. Samples . . . . . . . . . . . . . . . . . . . 63 CHAPTER IV. RESULTS AND DISCUSSION. . . . . . . . . . . . 66 A. Previous Studies. . . . . . . . . . . . . . . . . 66 B. The Photoelectron Spectrum. . . . . . . . . . . . 66 C. The Photoionization Mass Spectra of CH3CN and CD3CN. . . . . . . . . . . . 71 D. Parent Ion Curves . . . . . . . . . . . . . . . . 75 1. General Observations. . . . . . . . . . . . . 75 2. Autoionization Structure: 1040 — 940 R. . . . . . . . . . . . . . . . . 80 3. Autoionization: 900 - 800 A . . . . . . . . . 92 4. The Threshold Region: 1000 - 1040 A . . . . . . . . . . . . . . . . 97 E. Fragment Ion PIE Curves . . . . . . . . . . . . .113 1. CHZCN+, CD2CN+. . . . . . . . . . . . . . . .116 2. CHCN+, CDCN+. . . . . . . . . . . . . . . . .121 + + 3. CH2 , CD2 . . . . . . . . . . . . . . . .125 + + 4. CH3 , CD3 . . . . . . . . . . . . . . . . . .128 + + 5. HZCN 'DZCN . . . . . . . . . . . . . . . . .134 CHAPTER V. CONCLUSIONS AND SUGGESTIONS FOR FURTHER WORK. . . . . . . . . . . . . . .140 APPENDIX A. DIGITAL MASS SCANNER. . . . . . . . . . . . .145 LIST OF REFERENCES. . . . . . . . . . . . . . . . . . . .162 vi LIST OF TABLES Page No. 11-1. Photoionization Processes. . . . . . . . . . . . . ll IV-l. Vibrational Modes and Frequencies of CH3CN+ and CD3CN+ Excited in the Photoelectron Spectra Compared to the Frequencies of the Neutral. . . . . . . . . 69 IV-2. CH3CN and CD3CN Photoionization Mass Spectra . . . . . . . . . . . . . . . . . . . 73 IV-3. Rydberg Series Converging to the First Excited State of CH3CN+ and CD3CN+......................84 IV-4. Rydberg Series Converging to the Second Excited State of CH3CN+...........-..........96 IV-S. First Ionization Potentials of CH3CN and CD3CN. . . . . . . . . . . . . . . . . .107 IV-6. Heats of Formation of some Neutral Species and some Useful Conversion Factors. . . . . . . . . . . . . . . . . . . . . .117 IV-7. Calculation of AHfO of CHZCN+ and CD2CN+ . . . . . . . . . . . . . . . . . . . . . .122 IV-8. Summary of Appearance Potentials and Heats of Formation . . . . . . . . . . . . . .139 vii II-l. 11-2. 11-3. III-1. III-2. III-3. III-4. IV-1. IV-2. IV-3. IV-4. IV-5. IV-6. IV-7. IV-8. IV-9. IV-lO. IV-ll. LIST OF FIGURES Block Diagram of a Photoionization Mass Spectrometer . . . . . . . . Effect of the Franck-Condon Factor on Direct Ionization Thresholds of Diatomic Molecules. . . . . . . Krypton PIE Curve . . . . The MSU PIMS Instrument . . Rare Gas Lamp High Power Switching Circuit I O O O O O O O O O O O 0 Spectral Distribution of The Helium Continuum and Hydrogen Pseudo-continuum Lamps . . . . . . . . . . . . . Channeltron (CEM 4816) Wiring Diagrams for Positive and Negative Ions. . Photoelectron Spectrum of CH3CN 21.2 eV Photoionization Mass Spectra of CH3CN and CD3CN. . . . . . . CH CN+ and CD3CN+ PIE Curves: 603 - 1040 A. . . . . . . . . Total PIE of CH3CN. . . . . . . . CH3CN+ PIE Curve: 940 - 1040 A. CD3CN+ PIE Curve: 940 - 1040 8. Transition Energy Diagram for Excitation from the 5al Orbital . CH CN+ and CD3CN+ PIE Curves: 80 - 900 A . . . . . . . . . . CH CN+ and CD3CN+ PIE Curves: 1000 - 1040 A . . . . . . . . . . The Effects of a Triangular-Slit Function on a Step Function . . Integration of a Gaussian Transition Probability. . . . . . viii Page No. 15 27 . . 67 72 76 81 93 . olOl . .104 IV-12. IV-l3. IV-14. Fit of the CH CN+ and CD CN+ PIE ThreShOldgo C O . . § 0 O C O C O CHZCN+ and CD CN+ 2 PIE Curves. . . . . . . . . . . . . CHCN+ and CDCN+ PIE Curves. . . . . CH2+ and CD2+ PIE Curves. . . . . . + + CH3 and CD3 PIE Curves. . . . . . H CN+ and D CN+ PIE Curves. . . . 2 2 Block Diagram of Digital Mass Scanner . . . . . . . . . . . . . . Scanner Computer Interface. . . . . . Scanner Computer Interface. . . . . . Computer Interface Board Layout . . . Main Logic Board Opto-Isolator Line Receivers. . . . . . . . . . . . Main Logic Board Time Base Circuit. Main Logic Board Scan Time Multiplier Main Logic Board Twelve-Bit Counter and Multiplexer Circuit . . . . . . . Thumb Wheel Circuits. . . . . . . . . Manual Control Circuits and Counter Enable. . . . . . . . . . . . Main Logic Board Layout . . . . . . . Analog Circuit. . . . . . . . . . . . ix .105 .118 .123 126 .129 .135 .146 150 .151 .152 .153 .154 .155 .156 .157 158 .160 .161 CHAPTER I INTRODUCTION Although acetonitrile (CH CN) is a relatively small and 3 simple molecule, the ionization and consequent fragmentation of CH3CN has not been accurately and completely studied. The recent detection of acetonitrile in the comet Kohoutekl and in interstellar space in the region of the constellation Sagitarius2 has rekindled interest in its ion chemistry. Fur- thermore, acetonitrile has been found in discharge chambers simulating rare-earth conditions, suggesting that an investi- gation of its ion chemistry might be relevant to an understand- ing of abiogenic synthesis.3’4 A wide range of molecular species has been discovered in interstellar clouds of gas and dust, and the question of how these molecular species are synthesized remains unanswered.5 The temperatures of interstellar gas clouds are generally in the range 4-25° K, and occasionally as high as 1000 K.6 At these low temperatures most neutral-neutral reactions which could account for the observed molecules proceed at negligible rates due to kinetic reaction barriers. Reactions between ions and molecules often have very low activation barriers, and it has been suggested that many interstellar molecules are synthe- sized via ion-molecule reactions. Interstellar gas clouds are bathed in a flux of cosmic radiation which can account for the presence of ions in these environments; a quantitative model 1 99 . «Ha 2 based on ionization by cosmic radiation and subsequent ion- molecule reactions has been proposed to account for many of the observed molecules.7 Much of this model is based on the thermochemistry of the species involved in possible ion- molecule reactions, and for the theory to be further tested there is a need for thermochemical parameters such as heats of formation, bond dissociation energies, electron and proton affinities, etc., of the relevant neutrals and ions. Photoionization mass spectrometry (PIMS) has become a viable experimental method in recent years.8 The most success- ful and important application of the technique has been to determine thermochemical parameters of ions. Indeed, PIMS is generally recognized as the best and most accurate technique for such investigations.9 With the above comments in mind, acetonitrile was selected as a good candidate for a PIMS inves- tigation. PIMS is primarily a spectroscopic technique in which the yield of ions from a sample gas is measured as a function of ion mass and photon energy. Thus, relative photoionization and fragmentation cross sections are measured, and they pro- vide a wealth of information about the ions. Such quantities as ionization and appearance potentials, vibrational and elec- tronic energy level spacings, and relative rates of various ionization and fragmentation processes can often be directly determined from PIMS experiments. From these, thermochemical parameters, fragmentation mechanisms and a variety of other information can be deduced. PIMS, coupled with a few other techniques, can often fully characterize molecular ionization and fragmentation processes. This thesis is a report of the photoionization mass spec- trometry of acetonitrile. The objective was to learn as much as possible about the photoionization and fragmentation of this molecule, to derive thermochemical parameters where the data merit such a calculation, and to elucidate fragmentation pathways and mechanisms. The deuterated analog, CD CN, was 3 included to help in the analysis of the data and to determine the isotopic dependence of various processes. Chapter II of this thesis presents a general overview of PIMS. It includes a brief description of the PIMS experiment and contains a discussion of ionization and fragmentation processes which are important in PIMS. The experimental details of the acetonitrile experiments are presented in Chapter III. This chapter also includes a brief description of the MSU PIMS instrument along with a discussion of some of the instrumental problems encountered during the course of the investigation. The data are reported and discussed in Chapter IV, and Chapter V provides a summary with conclusions and suggestions for future experiments. CHAPTER II AN OVERVIEW OF PIMS The experimental objective of PIMS is to measure the yield of ions produced by the absorption of photons by a neutral molecule, as a function of photon energy and ion mass. Although the name PIMS emphasizes mass spectrometry, the more important aspect of the experiment is the variation of the photon energy. The experiments described in this thesis are really vacuum ultraviolet absorption experiments in which the mass spectrometer serves as a very special detector. The mass spectrometer serves as a means for observing photon absorptions that lead to the ejection of an electron from a molecule as well as those absorptions in which sufficient energy is im— parted to the ion to cause fragmentation. A PIMS experiment,like other spectroscopic techniques, is directed toward measuring transition energies, and the results often allow one to infer properties of the initial and final states. Due to the nature of photoionization processes, PIMS often provides the most information about the ground states of the ions; however, features in the spectra often yield informa- tion about higher energy states, also. It is the purpose of this chapter to present an overview of PIMS. Section A is a brief description of the general experimental setup and is followed by a discussion of photoionization processes in Section B. Section C is a discussion of fragmentation of 4 polyatomic ions,and Section D is a very brief description of photoelectron spectroscopy and electron impact energy loss spectroscopy, two techniques which aid the interpretation of PIMS data. First reports describing mass spectrometric analysis of ions produced by photoionization of gaseous samples were pub- lished in 192910 and 1932.11 However, early investigators were severely limited by the available vacuum and light source technology, and only fairly recently has PIMS become a viable experimental technique. In recent years it has been demon- strated to be useful for the study of unimolecular reaction kinetics of polyatomic ions?'12’13 the study of kinetics and 13,14,15 thermodynamics of ion-molecule reactions, and the determination of ionization potentials of atoms and molecules and of appearance potentials of fragment ions.8'9 Even though a wide variety of applications have been found for PIMS, key references are somewhat scattered and in- complete. No comprehensive reviews of PIMS have been published since the one by Reid8 in 1971, which is now somewhat dated. Nevertheless, that review, together with references 9,12,14, 16-19 , provide a useful introduction to the experimental methodology of PIMS, photoionization and fragmentation 9,12,18-22 9,12,14 processes and applications of the technique. Since the introduction to PIMS presented here is intention- ally brief and is intended to serve only as a framework for the discussion of the acetonitrile data, the reader is urged to consult these references for more detailed information. A. The PIMS Experiment A block diagram of a photoionization mass spectrometer is shown in Figure 11-1. The apparatus consists of a photon source, photon monochromator, interaction region, ion optics and mass spectrometer, and photon and ion detectors. The output of the photon source is dispersed by the monochromator and the output of the monochromator is passed through the sample gas in the interaction region. The intensity of the transmitted photons is converted to an electrical signal by the photon detector. The ions are extracted at a right angle to the photon beam with an electrostatic potential and are focused onto the entrance of the mass spectrometer with electrostatic ion lenses. The mass spectrometer allows ions of only a single mass-to-charge ratio to impinge on the ion detector, where the ion arrival rate is converted into an electrical signal. The experiment entails very simply the measurement of the intensity of an ion of selected mass as a function of the photon energy. Since the intensity of the available light sources is not constant as a function of energy, the photon intensity must also be measured so that the ion intensity can be normalized. It would be most desirable to measure the absolute photoionization cross section, but this quantity is difficult to determine because then the absolute photon intensity and the number density of neutral molecules in the photon beam must be known. Instead, a relative quantity, the I SAMPLE PHOTON SOURCE NONOCHRONATOR INTERACTION ooooooé) ”L .62"... ION OPTICS AND NASS SPECTROMETER PHOTON DETECTOR ooooooé) ION DETECTOR ION \. Figure TON NAL WW “I GO II-l. Block Diagram of a Photoionization Mass Spectrometer. /' SIGNAL photoionization efficiency (PIE), defined as: = ion intensity (E) (II-1) transmitted photon intensity(E) ' PIE(E) is calculated. Although properly the incident photon inten- sity should be used to calculate the PIE, under typical operating conditions (low sample pressure) the amount of light absorbed by the sample is sufficiently small that the PIE calculated from the above relationship is directly propor- tional to the partial photoionization cross section of the ion being observed.23 The plot of the PIE of an ion as a function of photon wavelength or energy is referred to as the PIE curve. The study of photoionization processes is experimentally difficult due to limitations imposed by experimental technology and the nature of the fundamental photoionization processes. The most severe difficulties arise because of the region of the electromagnetic spectrum in which the transitions of interest lie. The lowest ionization potential of most atoms and molecules is greater than 8 eV.,and therefore one is re- quired to work in the vacuum ultraviolet region (VUV) (2000-300 A). First of all, most substances, including oxygen and nitrogen, absorb strongly in this region and as a result the optical path must be maintained at pressures less than 10-5 torr. This in itself is not a severe restriction, but it is complicated by the absence of suitable optical window materials in the VUV. Below the short wavelength cutoff of lithium fluoride (~1050 R, or 11.8 eV) there are no known window materials which can be used to isolate the lamp or sample gas from the rest of the instrument. Since most VUV light sources are electrical discharges in a gas, the lack of a suitable window material requires the light sources to be "isolated" from the rest of the apparatus via high speed pumping. The fact that most substances absorb strongly in the VUV also means that reflectivities are low. The best available optical substrates have reflectivities on the order of 15-30% at 1000 A, so that every reflection employed to disperse or focus the light throws away a large fraction of the incident light. This problem is further complicated by the lack of intense light sources. The best available light sources in the VUV region have peak intensities of only about 108-1010 photon sec.l 3-1. Thus it is imperative that reflecting surfaces be kept to a minimum; normally a single diffraction grating is the sole optical element employed. Another serious problem stems from the necessity to use low sample pressures in order to avoid scattering of the ions and reactive collisions of the ions with the sample gas; these processes decrease the intensity of the ions and introduce artifacts into the data. Most experiments must be performed with sample pressures of less than 10-3 torr. Low sample pressures coupled with low intensity light sources and the low reflectivity of optical elements result in low ion count rates. In favorable circumstances, ion count rates may be several thousand per second, but they are often 10 as low as only a few ions per second. Fortunately, the utili- zation of ion counting techniques and digital computers allows sufficiently long integration times that useable data can often be obtained with such weak signals, but only at the cost of very long experiments. Data runs for a single ion may last as long as several days! B. Photoionization Processes When molecules are irradiated with high-energy photons, processes such as photoexcitation, photoionization or some combination of the two may occur simultaneously to produce a multitude of final states. In the following pages of this section,those processes which result in an ion and are there- fore important to PIMS are discussed. Section 3.1 deals with direct ionization and Section B.2 with autoionization. Sections 8.3 and B.4 provide a discussion of fragmentation of molecular ions resulting from photoexcitation of neutral molecules. These processes are summarized in Table 11-1. 1. Direct Ionization Irradiation of molecules with visible and ultraviolet light may induce transitions of valence electrons to the lowest excited states of the molecule. As the energy of the radiation is increased,transitions to higher excited states of the molecule become energetically possible and eventually an energy is reached at which the excited valence electron is no longer bound by the potential of the nuclei. The final state is then described as the product of a quantized state Table 11-1. 11 Photoionization Processes. AB AB AB AB AB AB+ + e' + - AB* + AB + e A++B+e- + — AB* + A + B + e .4. _ AB*+e + A+ + B + e Direct ionization Autoionization Direct dissociation Predissociation 12 of an ion core and a continuum state of a free electron. As the radiation energy is further increased the molecules con- tinue to absorb, leaving the ion in the same final state with the electron carrying away the excess energy as kinetic energy. The absorption is continuous. Further increasing the radi- ation energy may make transitions to excited states of the ion possible; the free electrons will be produced with a distri- bution of energies, where the distribution will reflect the difference in energy between that required to reach the final states of the ion and the radiation energy. Transitions to the various final states of an ion will occur simultaneously. This is direct ionization. Direct ionization is exactly analogous to an ordinary one- electron transition observed in the visible or ultraviolet region of the spectrum, except for the one important difference that the electron is excited into a continuum state instead of a bound state. The probability of direct photoionization is proportional to the square of the dipole transition moment integral:22 2 - 2 M =|1 (II-2) where fl is the dipole moment operator, 43 is the wave function of the initial state of the neutral molecule, and wfis the wave function of the ion and the photoelectron. From the Born-Oppenheimer approximation and the assumption that the electronic parts of the wave functions change negligibly in the time it takes for a transition to occur, one findszz'24 13 that 2 " 2 2 M = | l |< wfvlwiv>l ' (”‘3’ Here wfe and wie are the electronic parts of the final and initial state wave functions respectively, and wfv and wiv are the vibrational wave functions of the ion and the neutral molecule. The wave function wfe describes both the ion and the free electron and may be represented as a linear combina- tion of two Slater determinants -- one for the ion and the other for the free electron.25 Rotational motion is tacitly neglected since the experimental conditions required in PIMS are generally insufficient to resolve rotational structure in direct ionization, and the integrals of equation (II-3) are to a good approximation independent of rotation.22 Qualitatively it is found, both theoretically26 and experimentally,18 that the electronic part of (II-3), I< ll’fe Ifilwie> I2, rises sharply to a maximum at the ionization threshold and decreases to small values for exci- tation energies much greater than threshold. In the region accessible to PIMS ( 20 eV) the decrease is usually less than 50% of the maximum at threshold. Near the ionization threshold -- up to one or two eV above -- the photoionization cross section can usually be taken as a step-function which 27'28 However, as pointed out by is constant above threshold. Rosenstock,9 there is no guarantee that the range of validity of this assumption is this large. If one neglects the vibrational part of equation (II-3) 14 and considers a system in which several electronic states are available, the photoionization cross section or PIE curve would be a series of steps, each step corresponding to the threshold for an ionic state. All ionic states with a thresh- old energy less than a given photon energy will be populated at a rate proportional to the cross section at that photon energy. In the absence of competing indirect processes this step-like behavior is indeed observed for atoms (see reference 18 for examples). The vibrational part of equation (II-3) is the well-known Franck-Condon factor. It is simply the overlap between the vibrational wave function of the neutral molecule (wvi) and that of the ion (va). The effect of the Franck-Condon factor on the photoionization cross section is to attenuate the electronic factor and to introduce additional step-like struc- ture to the PIE curves. The threshold for an electronic state of a molecular ion becomes a series of steps, where each step represents a threshold for a different vibrational state. The height of each vibrational step above the previous step is proportional to the Franck-Condon factor for the corres- ponding transition. The overall shapes of the electronic thresholds can be quite varied, depending on the differences in the bond lengths between the neutral and the ion. This is shown pictorially for a diatomic molecule in figure II-2. If the electron is removed from a non-bonding orbital the interatomic distance of the neutral and the ion will be nearly the same, and 15 / I I I \ > \ I 0 l/ \ ’ I ’ I / I ‘i’ I \/ 2 M, = 012156 am y. = own an. r. = 0.14002 mu : 2 3 \K//// v. = 0.1207: m. 1 1 L L 1 1 1 1 1 1 01° OJ? 0.“ 0 I0 0.12 0.14 0.30 012 01‘ 0J6 0.1. c, an 42* C :8 2 :5: ‘Utb . I'll ,,...0lo ...... OI OlIJOS OIIICSOYOQlOHIII) cl 3 l )- ENERGY Figure 11-2. Effect of the Franck-Condon Factor on Direct Ionization Thresholds of Diatomic Molecules (from Reference 9.) 16 essentially only the ion vibrational ground state will be accessible from the neutral ground state. The threshold for the electronic state of the ion would be only a single step. If the electron is removed from an orbital which is slightly bonding or antibonding, in which case the internuclear distance in the ion is a little greater or smaller than that of the neutral, additional ion vibrational levels become accessible and a vibrational progression will be observed, with an intense (0 + 0) transition and progressively less intense transitions to the higher vibrational levels. The threshold will resemble a staircase with a large first step and progres- sively shorter steps to higher energy. If the electron is removed from an orbital which is strongly bonding or antibond- ing the internuclear distance in the ion will be significantly greater or smaller than that of the neutral and the (0 + 0) transition may no longer be accessible. The threshold region will exhibit a long progression of vibrational steps. The Franck-Condon factor has the same effect for polyatomic mole- cules except that many bond lengths may change and the thresh- olds may be overlapping progressions of steps if more than one vibrational mode of the ion is excited. The structure may be very complicated. In the absence of competing processes this step-like structure is often observed in the threshold region of parent ion PIE curves. The PIE curves for NO,29 HCCH,3O and NH331 are good examples. Resolved vibrational steps provide useful information about the vibrational structure of the ion, and about the 17 orbital of the neutral molecule from which the electron is removed. The energy spacings of vibrational steps are a direct measure of the vibrational spacing of the excited vibrational modes of the ion, and from them the excited modes can often be identified. From the assignment of the excited vibrational modes, the heights of the vibrational steps, and the overall shapes of the vibrational progressions, deductions can be made about the bonding characteristics of the orbital from which the electron is removed and the geometry of the ion. Although individual rotational transitions within a vibronic photoionization transition are not resolved, they do affect the direct ionization cross section curve shape by producing some tailing at the onset of a vibronic step and rounding of the top of the step. This is a result of the spread of threshold energies of the individual rotational transitions. The exact shape of the steps depends on the thermal distribution of populated rotational states of the neutral, the differences in the moment of inertia of the neutral and the ion, and rotational selection rules. The electronic selection rules for direct photoioniza- tion are very much simplified compared to those for absorp- tion and emission between two bound states. The most impor- tant and most common direct photoionization transitions are one-electron processes, and they are restricted by the usual dipole selection rules. However, the final state of the system is an ion plus a free electron,and the free electron can carry away whatever angular momentum is needed to satisfy 18 the dipole selection rules. As a result all one-electron direct photoionization dipole transitions are allowed.21’22 When spin-orbit interactions are small, the spin selection 21’22 The electron rule AS=0 can also always be fulfilled. spin of the final state of the system is the sum of the spin of the ion and the spin of the free electron. The free elec- tron can always leave the molecule with the spin it had in the neutral so that AS=0. Most neutral molecule ground states are singlets, and since the spin of a free electron is 1%, the most common ionic states are doublets. The most common direct photoionization transitions occur from singlet molecular ground states to doublet states of the positive ion.22 The vibrational part of equation (II-3), the Franck-Condon factor, puts some restrictions on the allowed vibrational transitions within a direct photoionization transition. In order for the Franck-Condon factor to be non-zero, the inte- grand (vawiv) must be totally symmetric in the point group 21'22'24 The vibrationless to which the molecule belongs. ground state of a molecule is always totally symmetric and therefore only totally symmetric vibrational modes of the ion can be reached by direct photoionization. The ground vibra- tionless states of the ion are likewise totally symmetric and the (0 + 0) transition is always allowed. The vibrational ' wave functions of symmetric vibrational modes are totally symmetric regardless of the vibrational quantum number, so excitation of such modes is allowed with any number of quanta. Antisymmetric modes are antisymmetric for odd and symmetric for even vibrational quantum numbers, and such modes can be 19 excited in direct photoionization only in units of two quanta. For polyatomic molecules belonging to point groups for which the normal modes are not just symmetric and antisymmet- ric, the vibrational selection rules must be determined by the direct product of the symmetry species of the initial and final vibrational levels. This product must contain the totally symmetric representation for the transition to be allowed. One still finds, when the neutral is in the ground vibrational level, that for all point groups the transition to the ground vibrational level of the ion is allowed. Also totally symmetric modes can be excited with any number of quanta and the single excitation of a single non-totally symmetric mode is 33233 allowed. The allowedness of multiple excitations and combinations of non-totally symmetric modes must be determined by using the direct product. Often one is most interested in measuring the energy of the transition from the ground vibrational state of the neutral to the ground vibrational state of the ion (the adiabatic ionization potential). Although this transition is always allowed by symmetry, there is no guarantee that it will be observed in direct ionization. For a transition to be observed not only must it be allowed by symmetry, but there must also be overlap between the vibrational wave functions of the initial and final states. Photoionization of N0232 is a good example of a case where, because of a large geometry change, the ground vibrational state of the lowest electronic state of the ion is not populated by direct ioniza- tion. 20 2. Autoionization Autoionization, in contrast to direct ionization, is a two-step process. Molecules have bound, neutral states corresponding to excitation of electrons of lower energy (more tightly bound) than those in the highest occupied orbital. Many of these excited states lie above the lowest ionization potential of a molecule, and if they are populated by absorp- tion of radiation they may interact with the continuum states lying above the first ionization potential, leading to ejec- tion of an electron. This is autoionization -- the absorption of a photon to produce an intermediate, highly-excited discrete state of the neutral which ejects an electron to form an ion. Since radiation is absorbed in a transition to a discrete state of the neutral, autoionization is a resonant process and ions will be produced only at the energy of the transi- tion. Autoionization will often appear as a sharp, peak-like structure superimposed on the normally featureless direct ionization continuum. However, autoionization exhibits a great variety of asymmetrical line shapes and may appear as a "window resonance" in which there is a decrease or dip in the 18,33,34 41 (e.g. see the COZPIE. ) The lifetimes of autoionizing levels can vary a great deal, 14 photoionization cross section, ranging from ~10- seconds all the way to ~10.6 seconds;9 for states with short lifetimes, the autoionization structure may be so broadened that it is indistinguishable from the con- tinuum for direct ionization. Autoionization structure on a PIE curve may also be broadened by emission and/or by 21 predissociation into neutral fragments, in which case it may also be indistinguishable from the continuum. Autoionization can occur from valence states lying above the lowest ionization potential of a molecule, but generally it originates from Rydberg states that are members of series converging to the second, third and higher ionization poten- tials of the molecule. The distinction between Rydberg states and valence states is not always clear, but in the molecular orbital framework, molecular Rydberg orbitals can be thought of as being formed from atomic orbitals with larger principal quantum numbers than those used to calculate the valence shell orbitals.33 Being composed of atomic orbitals of high principal quantum numbers, Rydberg orbitals have a large spatial extent and in a somewhat naive but useful approximation Rydberg states can be considered as a one-elec— tron system in which the electron in the Rydberg orbital is at so large a distance from the remaining ion core that the core appears as a point charge (H-like atom.)33 The above picture is reinforced by the fact that nearly all transitions to Rydberg states(visib1e in PIMS experiments as autoionization structure) can be fit to a series described by the Rydberg equation: R E = IP - ——z . (II-4) hv (n-d) Here Ehv is the energy difference of the transition and IP is the ionization potential of the molecule to which the series converges as n, referred to as the principal quantum number, 22 runs serially touso mHm :oummux .mIHH ouswflm 27 a: 00m 0mm 00m 0V0 ONQ Com _ _ _ _ _ mawxw .H. “3*, w: nu 1. I B m .. s so._.... ... m: .o m o.. ...«._m_._. mo 28 C. Fragmentation Processes Fragmentation processes can be very complex, and therefore they are often more difficult to analyze on a detailed level than direct ionization or autoionization. As an example, con- sider the four-atomic molecule, ABCD. When a parent ion is detected, the overall process: ABCD + hv + ABCD+ + e- has occurred. The energy of the photon (hv) is precisely deter- mined by the monochromator setting, and the internal energy of the parent ion is easily calculated by subtracting the elec- tron's kinetic energy from the photon energy. Thus, the energetics and composition of the system are well character- ized. If the fragment ion A+ is detected, then ABCD + hv + A+ + (BCD) + e-. The ion mass and photon energy may be accurately measured and even the kinetic energy of the electron can be measured, but this is only part of the system. The neutral fragment or fragments are not detected with PIMS and their composition may be uncertain (they could be BCD, B+CD, BC+D or B+C+D). Moreover, the ion and the neutral fragments may also carry away kinetic and internal energy. The total system is thus not easily characterized compositionally or energetically. The appearance potential of the fragment ion, the lowest photon energy at which the particular ion is detected, may well correspond to the produc- tion of neutral and ionic species having significant internal and kinetic energy, which are not simply related to the known energies of the incident photon and ejected electron. 29 Fortunately, thermochemical information is often helpful in reducing the complexity of the process; on this basis many fragmentation pathways can be eliminated, and kinetic energy releases can be estimated. However, accurate estimates require accurate thermochemical data, which may not be available for all species of the system. 1. Small molecules (diatomic or triatomic) For small molecules it is often useful to consider fragmentation as the result of processes which are directly analogous to direct ionization and autoionization, where a neutral molecule undergoes a transition (directly or indirectly) to a repulsive potential energy surface or a repulsive region of a bound potential energy surface of an ion. Direct dissociation results from a direct transition to a respulsive surface and follows the same principles as direct ionization. The probability of the transition is determined primarily by Franck-Condon overlap of the initial neutral state and the continuous states of the unbound ionic surface. Direct dissociation is very fast (on the order of a vibrational period) and most of the excess energy will be carried away as translational kinetic energy of the products.21 The products may, however, be formed in excited electronic states if the dissociative surface does not correlate with the ground state of the ion. Predissociation, sometimes referred to as preionization or autoionization when ion fragments are formed, implies an intermediate undissociated excited state. Two possibilities 30 arise: Case (1). The intermediate state is an excited state of the neutral that undergoes a further transition to a repulsive surface of the molecular ion. This is directly analogous to autoionization as discussed in the previous section. Case (2) The intermediate state is an undissociated ionic state which undergoes a radiationless transition to a repulsive ionic surface or redistributes vibrational or rotational energy such that energy in excess of a dissociation limit ends up in a particular vibrational mode or rotational degree of freedom. The bound ionic state may be populated by direct ioniza- tion or autoionization. Although these two modes of predissociation are concep- tually distinct, it is often not possible to separate them experimentally. While peak-like structures in a parent ion PIE curve is a sure indication of autoionization, the same structure in a fragment ion PIE curve only means that a dis- crete state of the neutral was populated on the pathway to the ion fragment. The dissociation may have occurred by Case (1) or Case (2) predissociation. Whether predissociation occurs by Case (1) or Case (2), the fragments may fly apart with a considerable amount of vibrational or rotational energy since the intermediate state . may be produced with internal excitation.12'14'24 32 discussion that follows is adapted from Forst. In RRKM theory dissociation is treated as a vibrational phenomenon. That is, dissociation will occur when sufficient energy is available in a vibrational mode such that a bond is extended beyond some critical value, leading to rupture. The probability of dissociation is then the probability that at least some critical amount of energy, the energy just needed to rupture a particular bond, is found in that bond. The pri- mary assumption of RRKM theory is that this probability is purely statistical. It depends only on there being sufficient vibrational energy available among all the normal modes of the excited molecule, and the ratio of the number of ways that the vibrational energy of a molecule can be distributed to put at least the critical energy into the dissociating bond to the total number of ways that the vibrational energy can be distributed within the molecule. For the dissociation to behave statistically, vibrational energy must flow freely and randomly throughout the molecule. RRKM theory considers energy randomization on only a single potential energy surface. STMS makes a second important assump- tion. STMS assumes also that electronic energy is freely and randomly converted into internal energy of just one low-lying state, usually the ground state of the ion (although it need not be the ground state). The theoretical and experimental evidence for and against randomization of electronic energy is fully discussed by Forst48 and need not be repeated here; it is concluded that in the great majority of cases the 33 assumption appears valid. In any case, the utility of STMS lies in its ability to predict the general features of frag- mentation of polyatomic molecular ions. STMS implies that only Case (2),described above, pre- dissociation, will be important. A dissociation is considered to be a unimolecular reaction of a parent ion which is formed with sufficient energy to enable the rupture of a bond. Fur- thermore, time is required for the energy to flow into the rupturable bond or bonds. Thus, STMS emphasizes the temporal aspects of PIMS. The interaction region (the ion source) of a photoionization mass spectrometer (or any mass spectrometer) is located a fixed distance from the mass analyzer. Any parent ion that fragments in the time it takes to travel from the interaction region to the mass spectrometer will be detec- ted as a fragment ion. The intensity of a fragment ion will depend on the number of parent ions formed with energy above the critical energy, the rate constant for the fragmentation channel, and the flight time from the interaction region to the mass analyzer. Some types of mass analyzers, e.g., mag— netic sector and time-of—flight instruments, may allow one to observe fragmentations that occur in the analyzer. The rate constant (k(E)) for one channel of fragmentation is given by:12'48 k (E) = 0 E < E , (II-5) 34 = aG*(E-Eo) HN(E) for E 2 EO , k(E) where k(E) is the rate constant at a total excitation energy E (with respect to the ground state of the ion), N(E) is the density of states at energy E, and G*(E-Eo) is the density of states of the activated complex (i.e., essentially the parent ion minus the fragmenting bond) integrated from the critical energy, E to the excitation energy E. The constant a is o' of the order of magnitude unity and takes into account degen- erate fragmentation channels due to the symmetry properties of the dissociation, and h is Planck's constant. Note that N(E) is essentially the number of ways that energy E can be distributed in the molecular ion and G*(E-EO) is the sum of all the ways that energy E can be distributed so as to leave at least energy E0 in the dissociating bond. Potential energy surfaces for dissociating ions rarely have a reverse energy of activation. That is, a fragmenting molecular ion usually does not pass over a hump in a potential energy surface, but only over a ledge. This means that,in general,fragments produced at the critical energy will be formed with no internal or translational kinetic energy. If a fragment ion is produced at a detectable rate at the criti- cal energy (E0), then the appearance potential (AP) of the fragment (the minimum photon energy at which the fragment is detected, assuming the neutral is in its ground electronic, vibrational and rotational state) will be at a photon energy equal to the ionization potential of the neutral plus the thermochemical threshold for the dissociation. Under these 35 conditions an accurate heat of formation can be calculated for the fragment ion. For example, if AB + hv + A+ + B + e- and AP = Aern, then AHf(A+) = AP(A+) + AHf(AB) - AHf(B). Under what conditions will it be possible for the experi- mental AP of a fragment ion to correspond to the thermochemical threshold? There are several points to note. First, in a PIMS experiment the PIE is measured at some fixed time after the ionization (usually 10.5-10-6 5.9). Second, the rate constant will have a non-zero value at the critical energy given approximately by:12'51 1 k(EO) :h—NTEET . (II-6) Third, in general,it is found that k(E) is a monotonically, rapidly-increasing function of E above E0.12'51 Assume for the moment that parent ions are formed with a suitable distribution of energy greater than the critical energy for a dissociation such that the dissociation is not limited by this factor. Three cases can be distinguished. Case (1). k(EO) is sufficiently large that all or nearly all parent ions formed with excitation energy E 2 Eo will be dissociated within the flight time to the mass analyzer. The rate at which the fragment ions arrive at the mass analyzer is then the rate at which the parent ions are produced with internal energy greater than E0. The dissociation rate constant will have no effect on the shape of the fragment ion PIE. In this case an accurate heat of formation of Case (2): Case (3): 36 the fragment ion can be determined from the AP. k(EO) is not large enough for a detectable number of fragment ions to be formed during the flight time. Only at energies above E0 would the fragment ion be detected, and the PIE would depend on k(E) and the rate at which the parent ions are formed with internal energy greater than BC. In general as the photon energy (Ep) is reduced from Ep > E0 toward E0 the fragment ion PIE asymptotically approaches zero. The apparent AP would lie at an energy greater than the critical energy and would serve as only an upper limit to the thermochemi- cal threshold. The shift of the experimental AP above the critical energy due to a low fragmentation rate constant is often referred to as the kinetic shift. k(EO) is sufficiently large for a detectable number of fragment ions to be produced at E0, but this process competes with one or more other fragmentation channels for which E0 is lower. Since fragmentation rate constants in- crease rapidly with energy, the rates for channels of lower critical energy may well be sufficiently fast at the critical energy of a higher energy channel that all parent ions dis- sociate by one of the lower energy pathways 37 and no parent ions survive sufficiently long to dissociate by the newly-available pathway. As EF is increased above E0 for this higher energy pathway the rate constant will increase such that it can compete with those for lower energy channels, and the fragment will be detected. The fragment ion PIE curve will be similar in shape to those of ions falling in Case (2) and the AP will yield only an upper limit for the thermochemical threshold. The shift from the thermochemical threshold of the measured AP due to competition with other fragmentation pathways is frequently referred to as the com- petitive shift. The quantity N(E) in equations 11-5 and 11-6 is a rapidly increasing function of E48 and therefore k(EO) (equation II-6) will be smaller the higher the critical energy and the more likely it is that the corresponding fragmentation channel will fall into Case (2). The function G*(E-Eo) is also a rapidly increasing function of E (G*(E-EO) increases more rapidly than N(E)48), and at a given energy a fragmentation channel with a higher critical energy will in general have a smaller rate constant than a dissociation with a lower critical energy. This results because N(E) will be the same for both channels whereas G*(E-EO) will be smaller for the higher critical energy channel due to the smaller energy range for the integration which determines G*(E-Eo). The lowest energy fragmentation 38 process is the one most likely to yield an accurate fragment ion heat of formation. The APs of higher energy fragmentation pathways are almost always shifted above the thermochemical threshold due to kinetic and (more generally) competitive shifts; they are useful only for determining upper limits for the heat of formation of the fragment ions involved.12 To this point in the discussion, little has been said about the rate at which the parent ions are produced. The rate at which a fragment is detected can never exceed the rate at which parent ions are produced with internal energy above the critical energy for the fragmentation process. (The rate at which a fragment is detected, however, may be greater than the rate at which the parent ion is detected if the frag- mentation is fast and more parent ions are produced with internal energies above the relevant critical energy than below it.) All fragment ion intensities are eventually limited by this rate. Insufficient production of parent ions with internal energy equal to the critical energy for a frag- mentation channel could, like the kinetic effects, delay the detection of a fragment ion until well above its thermochemical threshold. Often this possibility can be eliminated from con- sideration since many fragment APs are observed in regions where states of the parent ion are known to be populated by direct ionization and autoionization. However, when a fragment ion AP coincides with the threshold of an upper state of the parent ion this source of error should be considered. Methane and ethane are good examples of smaller polyatomic 39 molecules that fragment as predicted by STMS. The PIE curves of methane and its fragment ions can be found in Reference 52 and those of ethane in Reference 53 . In order to make a comparison between the theoretical predictions of STMS and the results of a PIMS experiment one would need to calculate the rate constants for each fragmentation channel and to know the distribution of internal energies in which the parent ions are formed at a given photon energy. The calculation of the rate constants is fairly straightforward and can be accom’ 12'48 The more plished with modest computational effort. severe problem is the need to know the distribution of internal energies with which the parent ions are formed. If direct ionization is the predominate process and if the direct ioniza- tion cross section approximates a step function, then an estimation of this distribution can be made by taking the first derivative with respect to photon energy of the total (parent 50 plus fragments) PIE. This approach was used by Chupka on methane52 and ethane,53 and the results showed reasonable agreement with theory. Most of the differences could be attributed to contributions to the PIE by autoionization. The agreement between theory and experiment was later confirmed by Stockbauer54 in a threshold electron-photoion coincidence experiment, where only fragments produced from parent ions of known internal energy are detected. D. Complementary Experimental Techniques Since the results of a PIMS experiment depend on the properties of the electronic and vibrational states of ions 40 and the properties of neutral states above the lowest ioniza- tion potential of a molecule, other techniques which provide information about these states often prove useful and comple- mentary. Results from two of these techniques, photoelectron spectroscopy (PBS) and electron impact energy loss spectroscopy (EIELS), have been invaluable for interpreting the acetonitrile PIMS data and will be referenced often in the discussion of acetonitrile in Chapter IV. A brief description of these two techniques follows. 1. Photoelectron Spectroscopy PES, like PIMS, utilizes ionization by photons to investigate ions. In PBS the kinetic energy of the ejected photoelectron is measured, but there is no attempt to observe the ions directly. A gaseous sample is irradiated with fixed- energy, monochromatic VUV light; most often, 21.21 eV photons produced from an electric discharge in helium are utilized, but other rare gas resonance lines have been employed. The kinetic energies of the photoelectrons are measured with an electron spectrometer and a plot is made of the intensity of the electrons as a function of the difference in energy between the monochromatic photons and the photoelectrons. Since energy and momentum are conserved, this difference is the excitation energy of the product ion with respect to the neutral molecule. The photoelectron spectrum displays the distribution of ionic states that are populated upon ioniza- tion. Photoelectron spectra are usually obtained at a photon 41 energy where autoionization does not significantly contribute to the photoionization cross section, and since direct ioniza- tion cross sections vary little with energy above threshold, PES is useful for judging the contribution of direct ionization to a PIE curve. Indeed, if autoionization did not contribute to the total PIE, the first derivative with respect to photon energy of the total PIE would closely approximate the photo- electron spectrum. The vibrational frequencies of an ion, as well as relative Franck-Condon factors for the transitions, may often be deter- mined directly from the vibrational structure in a photoelec- tron spectrum. These in turn are useful for determining changes in geometry between the neutral molecule and the ion. On the other hand, the photoelectron spectrum reveals little or no information about fragmentation of ions. Sometimes fragmentation may be inferred from broadening of or the absence of vibrational structure in a photoelectron spectrum; however this type of evidence is uncertain. Photoelectron spectro- scopists most often rely on PIMS for direct evidence of frag- mentation. More details of PBS are given in References 21, 22, and 55. 2. Electron Impact Energy Loss Spectroscopy Ordinary optical absorption spectra in the energy range of the ionization continuum are often useful for locating Rydberg states which might autoionize. However, because of the experimental difficulties of working in the VUV,very limited data are available. EIEL spectra provide much the same 42 information without many of the experimental constraints of VUV absorption spectroscopy. EIELS is an inelastic electron scattering technique in which a gaseous sample is irradiated with a beam of mono- energetic and fixed-energy electrons, and the kinetic energies of the scattered electrons are measured with an electron spectrometer. In a fashion similar to that used for PBS, the electron intensity of the scattered electrons is plotted as a function of the initial electron energy minus the scat- tered electron energy. The incident monoenergetic electrons induce transitions in the neutral sample molecules and will leave the scattering complex with an energy equal to the incident energy minus the transition energy. The EIEL spectrum will display peaks at the electron energy differences corresponding to the induced transitions. EIEL.spectra are most useful for locating Rydberg states which may autoionize and thus be observed in a PIE curve, however, they may display more transitions to Rydberg states than can be observed in an optical experiment, due to the relaxed selection rules of electron-induced transitions. The advantage over PIMS for observing transitions to Rydberg states is that transitions to states below the lowest IP of a , molecule can be observed. These transitions are often help- ful for assigning autoionization structure in a PIE curve for which a limited number of Rydberg series members are observed through autoionization and for which lower members of the series are located below the IP. EIEL spectra, however, 43 provide little information about ionic states since the Rydberg series join smoothly onto the ionization continuum. More details of EIELS can be found in References 20, 56 and 57. CHAPTER III EXPERIMENTAL APPARATUS AND PROCEDURES A. Introduction The purpose of this chapter is threefold: to present a brief description of the MSU PIMS instrument, to document the experimental conditions under which the data presented in Chapter IV were recorded, and to discuss experimental problems which were encountered in the course of this work. The instru- ment has been described in considerable detail elsewhere,19’58 and the brief overview is included here for completeness and to facilitate the discussion in this chapter. A diagram of the apparatus is shown in Figure 111-1. The instrumental problems encountered during the course of this investigation were innumerable, many of them due to lack of experience with the instrument, whose construction was completed just prior to the start of the acetonitrile experiments. Many of these problems have been resolved; however, some are likely to occur again,and it is hoped that the following discussion will be of help to future users of the instrument. B. The Instrument 1. Light Source The light source is a Hinterregger-type windowless discharge lamp with water-cooled anode and cathode. The dis- charge tube is a 25 cm long x 4 mm I.D. water-jacketed quartz capillary to which the discharge gas is admitted at the anode 44 45 P4 05 P5 '1 15x #4 if — i .‘L mi 1 51 0:90:12 H T [E BA IT L .l 1‘ "‘ *I Figure III-l. BA-baffle; EN-entrance slit; EX—exit slit; GR-grating; IS-ion source; IT—ion transducer; LP-lamp; Pl-first differential pumping port; P2-second differential pump- ing port; P3—monochromator pumping port; P4-sample chamber pumping port; PS-quadrupole chamber pumping port; PT—photon transducer; QP—quadrupole; QS-quadrupole support; LE-Ion Lense. 46 after it has passed through a molecular sieve trap immersed in liquid nitrogen. A small differential pressure is main- tained down the length of the discharge by pumping on the cathode end of the lamp; this stabilizes the discharge and helps eliminate sputtering of the cathode onto the mono- chromator entrance slits. The cathode vacuum pump is throttled with a needle valve. The rare gas continua can be produced by a high-power pulsed discharge in the pure gas. The MSU instrument utilizes a home-built, single vacuum tube, high-power switching circuit to pulse the output of a high voltage d.c. power supply. The switching circuit is in turn driven by a Cober A model 605P high-power pulse generator. A schematic diagram of the switching circuit is shown in Figure III-2. The hydrogen many-line pseudo continuum is produced by a d.c. discharge in pure hydrogen at a pressure of 2-4 torr. The d.c. high voltage power supply used in the switching cir- cuit, from which it is disconnected in this application, powers the hydrogen discharge. A 15009, 1500 W resistor limits the current drawn from the supply. The ionization potential of acetonitrile falls in a region of overlap between the helium Hopfield continuum, hereafter referred to as the helium continuum, and the hydro- gen pseudo continuum; however, neither light source is very intense in the overlap region. The output of both lamps is shown in Figure III-3. Preliminary data were taken with both lamps and the helium continuum was found to be the better 47 ——'> to Lamp 8170 from 006.! 250 _ 0.05 T 1.5 Mb A (TTTTI____.‘ | 0-20 KVDC — 0.22 1» I L t: 1 % 2'47 K S: 200 ‘> 1 ‘> 1 0.25 0.25 JL II jr H to Innrl:cll *——— “opts-1564 ( >—-vv—-‘Vvv-—JVW—41 2:470x 3'470K I “I NY 4.5 K + 2 K V N4007 IN 4007 i :4 P a. 707 VAC “14 VAC 1.1.1 17111 1 l lll_lvlvl Control Grid Screen 208 VAC 200 VAC Figure III-2. Rare Gas Lamp High Power Switching Circuit. 48 RELATIVE INTENSITY LLllL . YfVYIYYVYITIYYjIYTY‘Ifi’f‘TfifiIfiVYV I V 500 600 700 800 900 1000 1100 I200 HAVELENC‘UI . MOST!” 5 E l g 8 9|” "I 2 '1'». v1 . ( '1“ 4‘)" HAVILENGTH . ”GSTIOIS Figure III-3. Spectral Distribution of the Helium Continuum (top) and Hydrogen Pseudo- continuum (bottom) Lamps. 49 light source for the acetonitrile experiments; all PIES presented in this thesis were taken with that source. The hydrogen continuum with its large variation in intensity as a function of wavelength made it impossible to tell if struc- ture in the PIEs was real or an artifact caused by scattered light.* The smooth intensity variation of the wavelength output from the helium source, plus the few sharp lines from impurities in the lamp gas, made it relatively easy to dis- tinguish real structure from artifacts and to properly correct for scattered light. However, to minimize artifacts a great deal of effort went into reducing impurities to a minimum by always using a clean gas trap and a leak-free lamp gas plumbing system. The intensity of the helium continuum is controlled by several variables: helium pressure and purity, pulse width and frequency, d.c. high voltage, Cober pulse generator peak output voltage, and the condition of the lamp electrodes. The highest helium pressure possible, which is limited by the differential pumping between the lamp and monochromator and by the pumping speed of the monochromator pump, produces the highest intensity. Under current instrumental conditions the maximum pressure is about 70 torr with lOOu entrance slits, and about 130 torr with 500 entrance slits. The pulse frequency, pulse width and Cober high voltage * A discussion of such artifacts appears in Reference 21, p.319. 50 are adjustments made on the Cober pulse generator. The capabilities of the Cober and the limitations and require- ments of the switching circuit constrain these parameters. The best set of parameters are a pulse width of 0.4203 a frequency of 33.3 kHz and a Cober high voltage setting of 1.3 kV d.c. A narrower pulse width would allow a higher fre- quency to be used and would increase the lamp intensity. However, the pulse width is limited to 0.42us by the turn-on time of the switching circuit. The Cober's d.c. high voltage is set (1.3 kV) for a peak output voltage of 1 kV, which should not be exceeded since doing so may cause damage to the switching tube. With the above setting on the Cober and 70 torr of helium, the maximum intensity is attained with the switched high voltage at 10 kV d.c. At this voltage a current of 90-100 mA are drawn from the supply when the lamp gas is pure and the lamp electrodes are clean. Increasing the voltage above 10 kV increases the current drawn from the supply, but has no effect on the lamp intensity. A current from the high voltage power supply of less than 90 mA at 10 kV is a fairly good indicator of a dirty lamp cathode or impure lamp gas, both of which decrease the inten- sity of the lamp. After several months of operation the surface of the cathode on which the discharge occurs becomes blackened by deposits of unknown composition. The cathode is easily restored by removing the deposits with emery cloth and steel wool or by sand blasting, after which the cathode should be thoroughly cleaned with l,l,l-trichlorethane fol- lowed by methanol to remove any oil and debris. Impurities, primarily air, in the discharge gas not only cause a large number of very intense emission lines to be superimposed on the continuum, but also limit its intensity. The impurity level is most easily determined by recording a lamp output spectrum. Excessive impurities, indicated not only by many intense emission lines but also by absorption lines of nitrogen, are usually due to a leak in the lamp plumbing system. When the lamp plumbing is adequately leak- free it can be pumped down to 100 and the pressure will remain below about 3000 when the closed system stands overnight. Another source of impurities can be the compressed helium cylinders and the lamp gas manifold. When helium tanks are changed the lamp gas manifold should be thoroughly pumped out to remove any air that was admitted during cylinder change- over. Once the helium pressure is restored the manifold poses no problem, since it is considerably above atmospheric pressure. The quality of the helium gas varies from tank to tank, even though it has a stated purity of 99.99% (Airco, Inc.). Usually the molecular sieve trap will clean the helium sufficiently, and its use can increase the intensity of the lamp by as much as a factor of two. If one is working in a region where the continuum intensity is low and intense emission lines are a problem, a tank that is sufficiently free of impurities can usually be found by trying several tanks. Higher purity helium would alleviate this problem 52 but probably is not worth the extra cost since helium of reasonable purity is always available in the chemistry depart- ment, and a good clean trap usually provides sufficient cleanliness. The helium 584 A line, (182p)1P + (152)18 transition, is often useful for acquiring mass spectra since it can be produced with high spectral purity and intensity. It was used as the ionization source for the mass spectra presented in this thesis. The 584 A line is generated by a low pressure d.c. discharge. Maximum intensity is obtained with helium pressures just less than 1 torr and the maximum d.c. power supply current available. The high voltage d.c. power supply from the switching circuit, with the addition of a current- limiting resistor, is used to power the helium line lamp and the maximum current available from the supply is 800 mA. However, for reliable operation a current of 750 mA should be used; at 750 mA the voltage is 1500 V. The lamp and its power supplies were the biggest instru- mental bottleneck of the acetonitrile project. Most of the problems were with high voltage devices. Future operators of the instrument should carefully note that all of the lamp circuitry is potentially lethal even when it is unplugged! Strict precautions must be taken to insure that all devices are disconnected from their powerlines and all capacitors are discharged before checking or troubleshooting the system. If one lacks experience with high voltage devices, help should be obtained from an experienced person whenever there is a 53 problem. Never work on the high voltage circuits alone, and be sure that someone who knows what to do in case of electro- cution is present. The vacuum tube switching circuit is prone to two problems: arcing and breakdown of the rectifier diodes in the screen power supply. The arcing problem, now that all sharp points have been eliminated, is caused by accumulation of dust on the circuit shown in Figure III-2. The circuit is enclosed in a Plexiglas box, but air is forced through the box to cool the tube. Even though the air entering the Plexiglas box is filtered, after several years of operation enough dust can accumulate to cause problems. On the only circuit board in the Plexiglas box is an opto- isolator -- part of the interlock system -- which insures that the screen supply is not on unless the control grid supply is also on. Arcing causes the opto-isolator to turn off, which in turn (through an interlock circuit) shuts off the screen power supply, the Cober pulse generator and the d.c. high voltage power supply. The problem is symptomized by the screen, Cober and d.c. high voltage shutting off as the d.c. high voltage is turned up. Arcing can be diagnosed as the problem by removing the tube and disconnecting the Cober at its terminating resistor, as well as the secondaries of the screen, control grid, and filament transformers. Once this has been done the d.c. high voltage can be turned up to 10-15 kV and if arcing is the problem arcs will be observed at several 54 locations on the circuit. One should look especially closely at the circuit board containing the opto-isolator. The arcs are very dim and can be seen only in absolute darkness; it takes several minutes for the eyes to acclimate to a point where the arcs can be observed. If arcs are confirmed to be the problem the circuit must be disassembled and all the com- ponents thoroughly cleaned. Methanol is a suitable cleaning solvent; acetone is not recommended because it dissolves the paint and labels on the components. Breakdown of the screen's rectifier diodes can occur when there is input from the Cober while the d.c. high voltage is off. This has happened in a variety of situations and causes current from the filament to be collected by the screen instead of at the tube anode; the screen's rectifier diodes are damaged by this current. The 2000 resistor in series with the recti- fier and the screen was added and the primaries of the trans- formers were fused to protect the diodes, but they are still damaged occasionally. When the rectifier diodes have been ruined the fuses on the screen transformer primary will blow immediately when the screen is turned on. Damaged diodes are most easily confirmed as the problem by removing the circuit board on which they are located from the Plexiglas box and measuring the forward and reverse resistance of the diodes. This is most easily done with an ordinary ohmmeter, but the meter must use a high enough voltage to measure resistance to make a good diode conduct in the forward direction. For an ohmmeter which uses 1.5 V to measure resistances on the x 100 55 scale the forward resistance of a good diode should be about 2509 and the reverse resistance infinity. If any one diode in the bridge does not conform to these standards all diodes in the bridge should be replaced. (It is wise to have a reserve set on hand.) When the diodes are replaced the solder connections must be well beaded to eliminate any sharp points which can cause corona which in turn can cause the opto-iso- lator to shut off. The Cober pulse generator is quite reliable, but since it is primarily a vacuum tube circuit its performance degrades with extended use. If it is suspected that the Cober is not performing up to the standards outlined in its manual, its performance may be checked under a dummy load. The output is most easily and safely monitored with an oscilloscope at the a 1000 output connector on the front panel. The load should be a 2500, 250 W noninductive resistor. The rise and fall time of the output pulse should be checked as should the peak shape and peak height as a function of the d.c. high voltage and the repetition rate. If the Cober is not operating up to specification one of its vacuum tubes is probably weak. A tube in good condition will operate properly with the filament voltage reduced to 90% of its normal value whereas a weak tube will not. A Cinch-Jones two-pronged connector has been spliced into the primary side of the filament transformer, allowing one to insert a variac with which the filament voltage can be reduced. To test the tubes one first makes sure that the high voltage 56 is off and then observes the wave forms with an oscilloscope at the output of each tube under normal filament voltage. The schematics in the manual indicate the location of the test points and the correct wave forms. The observations are then repeated with the filament voltage reduced to 90% of its normal value. If the wave forms change the tube responsible for the change can be isolated and should be replaced. All the tube circuits are potentially lethal and since this testing process requires attaching an oscilloscope probe to the test points, the power should always be disconnected when the probe is moved. Again, it is also important that someone else be present in the room when these tests are performed. 2. Monochromator The monochromator is a McPherson model 225, l-meter, near-normal-incidence instrument. The dispersing element is a concave, magnesium fluoride overcoated aluminum grating, ruled with 1200 lines/mm and blazed at 1200 A. The grating has a reciprocal dispersion of 8.4 A per mm. Three fixed entrance slits: 100m, 500m, and 1000m, and four fixed exit slits: 100m, 50pm, lOOum, and 3000m, are mounted on moveable slit plates. The slits can be changed without breaking vacuum via knobs located outside of the vacuum chamber. The wavelength is driven with a stepping motor which can be operated manually or with a computer. After several months of use the diffraction grating accumu- lates a thin coating of diffusion pump oil, which significantly reduces its reflectivity in the vacuum ultraviolet and 57 increases scattered light in the ultraviolet and visible regions of the spectrum. This condition is most easily detected by observing the scattered light intensity in the 100-500 A range, by using the helium continuum as the light source and a sodium salicylate light detector. The helium continuum produces no light in this region, and with a clean grating no significant amount of scattered light will be ob- served. However, if the grating is coated with oil a large hump in the light intensity will be observed in this region, with a maximum at about 250 A. The scattered intensity varies depending on the amount of oil on the grating and its near 250 A can exceed the true intensity maximum near 800 A Oil on the grating will be polymerized by VUV light and therefore the grating should be promptly cleaned whenever scattered light is observed. The grating must be removed from the monochromator and its mask taken off. To remove the oil, the grating is first rinsed with Freon ll (CFC13), fol- lowed by a high purity methanol rinse. The methanol rinsing is best done by simultaneously applying the methanol with a squeeze bottle and drying the grating with low pressure, clean, compressed air, working from top to bottom. Great care must be taken not to touch the ruled surface of the grating. When the grating is reinstalled in the monochromator it will have to be realigned and focused, by following the procedure in the monochromator manual. 58 3. Ionization Region The ion source is a cubical, stainless steel box, one inch per side, located one centimeter beyond the exit slit of the monochromator. It has two rectangular apertures on opposite sides to enable transmission of the photon beam; on the horizontal perpendicular to the optic axis is a one-fourth inch diameter ion exit hole. The ion aperture is covered with a fine wire grid to minimize penetration into the source of electric fields from ion focusing lenses located just outside. An electrically-insulated, stainless steel plate, the repeller, is positioned inside the source, across from the ion exit aperture. An adjustable voltage is applied to the repeller to accelerate the ions from the source. The sample is admitted through a tube located behind the repeller and the sample pressure is measured through a tube in the top of the ion source. The sample pressure is controlled with a Granville-Phillips model 203 leak valve and measured with a Datametrics model lOl4-A barocell capacitance manometer. The manometer and leak valve are outside of the high vacuum. 4. Mass Spectrometer and Ion Optics The ions are guided from the ion source and focused onto the quadrupole mass filter with four electrostatic aperture lenses. The quadrupole mass filter is an Extranuclear model 324-9 with 1.9 cm diameter x 22 cm long rods and is powered by a model 311 power supply, equipped with a model E high-Q head. The ion lenses were supplied with the mass filter. The transmitted mass is controlled by a dial on the front 59 panel of the quadrupole power supply or by a voltage applied to a connector on the back of the power supply. When measur- ing a PIE curve the mass filter is used in the fixed mass mode, the mass being selected with the front panel dial. However, to obtain a mass spectrum the transmitted mass is scanned by applying a voltage ramp to the connector on the back panel. Part of this thesis research included the design and construction of a digital ramp generator that allows manual or computer control of the transmitted mass. Details of the circuit are given in Appendix A. 5. Ion and Photon Transducers The ion transducer, located on-axis at the exit hole of the quadrupole, is a Channeltron continuous-dynode electron multiplier supplied by Galileo Electro-optics. Most of the acetonitrile data were collected with a model CEM 4028 Channel- tron, but unfortunately in the course of modifications it was broken; it has been replaced with a CEM 4816 Channeltron The CEM 4816 has the advantage of being linear up to higher count rates; it has a lower gain than the CEM 4028, although the dif- ference in gain is very small. The CEM 4816 has been used for both positive and negative ion detection and the circuits used for each mode are shown in Figure III-4. The Channeltron output is measured either as direct current or pulse—counted. In the former mode the output is amplified with a Keithley model 417 electrometer. However, since the direct current method has several disadvantages 6O .mcoH m>Hummmz psm m>HuHmom new mEmuomflo .ocu_m 058 otoc‘ U0)! nl \ 3.32 922 .853 U0): ml 0 £2.30 \ 0‘0! :0. 02.30.. 3. 3.202 039:0 v means: loses Enos souuaosomso .enHHH ouomam .oco.m _ 2‘68 0 9... {9.1 / 3x0“ \ U0): 0+ 3.100- \ QSOn— a, O— €2.34 .oco.m 0‘01 co. 03.0002 61 compared to pulse counting when the output current is small,59 it was used only for setting up an experiment. All data were collected with the pulse counting technique, where the output pulses are amplified and discriminated with a circuit developed at MSU,60 and the counts are accumulated with a computer- interfaced counter.19 In the negative-ion mode pulse counting must always be used due to the need for a high voltage block- ing capacitor on the Channeltron output. The photon transducer is a sodium salicylate phosphor whose emission intensity is measured with an RCA 8850 photo— multiplier tube. Sodium salicylate is superior to ordinary (bare) photomultipliers because it possesses a nearly constant quantum yield throughout the helium continuum. The sodium salicylate emission band does not shift with the exciting wavelength and the emission intensity can be accurately measured with a standard photomultiplier. The output current of the photomultiplier is amplified with a Keithley 1800 current amplifier, and the output voltage is digitized with a voltage-to-frequency converter and a computer-interfaced counter.19 6. Vacuum System The monochromator is isolated from the discharge lamp with two stages of differential pumping. The first stage is a 300 l/sec four-inch diffusion injector pump. The mono- chromator chamber is pumped with a six-inch, 2400 l/sec dif- fusion pump, as is the sample region. The quadrupole region is pumped with a six-inch 1800 l/sec diffusion pump. 62 The pumping speeds quoted are for untrapped pumps; however, all four diffusion pumps are trapped with Freon-cooled baffles which reduce the pumping speeds somewhat. All the pumps can be isolated from the experimental vacuum chamber with pneumatically-operated gate valves. 7. Data Collection Once an experiment is set up, data collection and storage are controlled with a PDP 8/M or PDP 8/I minicomputer. The data are collected by using a variable integration time technique which permits recording data of the desired quality in an optimum amount of time.59 The monochromator wavelength is stepped in preset intervals, and at each wavelength ion and photon counts are accumulated until the datum has the desired signal-to-noise ratio. Photon and ion count rates, wavelengths and integration times are stored on a floppy or hard disk. The course of the experiment is continuously monitored via a computer-interfaced, stepped, stripchart recorder on which the real-time PIE is plotted. Periodic measurements of "light" and "dark" photon and ion count rates are made at a reference wavelength so that the data can be corrected for sample pressure and instrumental drift. Approxi- mate corrections for stray light are made by measuring the light intensity at two wavelengths where the helium continuum does not emit. Final corrections of the data for sample pressure, instrumental drift and stray light, as well as the final plot, are done with the MSU Chemistry Department's Cemcomgraph computer facility.61 63 C. Experimental Details 1. Samples Reagent grade protonated acetonitrile (CH3CN) was obtained from the Aldrich Chemical Co. and acetonitrile-d3 (CD3CN) from Stohler Isotope Chemicals. The deuterated com- pound had a stated atom purity of 99.5% D. Air was removed from the samples through the freeze-pump-thaw technique; the samples were frozen with a dry ice-methanol bath. No further purification was done on the protonated compound; however, the deuterated material was contaminated with sufficient quantities of D O and HDO that OD+ was observed in the CD+ PIE. For col- 2 3 lection of the CD; sieves and calcium hydride. PIE the sample was dried with molecular The samples were contained in pyrex bulbs which were con- nected to the leak valve via a glass-to-metal seal and a short length of copper tubing. The temperature of the room in which the instrument is located varies by as much as 50 F over a period of several hours. To minimize the effects of temperature variation on the sample vapor pressure the sample containers were placed in a half-liter Dewar flask filled with water. The data were collected with sample pressures between 4.5 - 5.5x10"4 torr, and the pressure varied by less than 10% throughout an experiment. CN+) Proton transfer from the acetonitrile parent ion (CH3 to neutral acetonitrile to form CH3CNH+ is an exothermic reaction with a large cross section at low relative kinetic 62 O l O I + o energies. The relative IntenSIty of CH CNH With respect 3 64 to CH CN+ is strongly dependent on the conditions in the ion 3 source. With small potential differences between the repeller and the ion source (less than 1 volt) and a sample pressure of 5 x 10"4 torr the intensity of CH3CNH+ is nearly half the intensity of CH3CN+. Because first-order processes were of primary interest and the proton transfer reaction significantly reduced the intensity of the parent ion, the reaction was sup- pressed by using a rather large potential difference between the repeller and the ion source (:10 V). This reduces the residence time of the ions in the ion source and therefore decreases the chance of a reactive interaction with the neutral compound. For the work reported here the repeller was main- tained at +13.8 V and the ion source at +3.8 V. The remaining ion lenses -- the extractor, lenses 1, 2 and 3, arranged in that order as the ions pass from the source to the quadrupole-- were set at -18.4 V, -44.8 V, —6.12 V and -5.4 V, respectively. This combination reduced the intensity of CH3CNH+ to less than 19% of the CH CN+ intensity, maximized the CH CN+ signal 3 and provided good mass spectrum peak shapes. 3 The mass filter was always set to the minimum resolution possible to just prevent overlap of ions that differed in mass by l u. Monochromator entrance and exit slits of 100 um, which give 0.84 A fwhm band pass, were used for all the PIE measure- ments reported here, except for the threshold regions of the parent ions; these were also measured with 500m entrance and exit slits, providing a 0.42 A fwhm band pass. Data recorded 65 with lOOum slits were taken in steps of 0.25 A, whereas 0.15 A steps were used with 500m slits. Integration and experiment times varied considerably depending on the intensity of the ions and photons in the region of interest, the slit width, the size of the wavelength step, and the desired quality of the data. An attempt was always made to record the data with a signal-to-noise ratio of at least 100. For the PIE curves reported here, this re- quired experiment times ranging from 12-48 hours. CHAPTER IV RESULTS AND DISCUSSION Photoionization of Acetonitrile A. Previous Studies Prior to the investigation reported here, studies of photo- ionization of acetonitrile had been reported by Watanabe, et al.,63 Nicholson,4 and Dibeler and Liston.65 The data obtained in references 63 and 64 were recorded without mass analysis and therefore only the total ionization efficiency could be measured. In both papers it was correctly assumed that the observed threshold was that of the parent ion; in neither paper were PIEs reported. Dibeler and Liston used mass anal- ysis and measured the ionization potential of acetonitrile as +0 2 2' ever, again no PIES were published. The results of these well as appearance potentials of CH CN+, CHCN+ and CH how— three publications are gathered in Table IV-5 together with the results of this investigation, and will be discussed along with the data reported here. B. The Photoelectron Spectrum The photoelectron spectrum (PBS) of acetonitrile has been reported.55'66’67 The PES of CH3CN from reference 55 is reproduced in Figure IV-l. The conclusions are as follows. The first band, with its origin at 12.21 eV in Figure IV-l, results from the removal of an electron from the C s N n-orbitals, 66 67 L000— ol22l ,l3 l4 Count sec" 3 Figure IV-l. Photoelectron Spectrum of CH (from Reference 55.) 3CN 68 as indicated by the vibrational structure of the band. The observed vibrational modes and their frequencies in the parent ion are compared to those of the neutral in Table IV—l. Three modes are excited in the ground state of the ion: the C N stretch (02), the symmetric CH deformation (v3), and the C—C 3 stretch (04). This assignment is supported by the CD CN spec- 3 trum. When compared to the ground state neutral molecule, the frequencies of the C _ N stretch and the C—C stretch decrease in the ion, while the CH3 symmetric deformation frequency in- creases. This indicates that the orbital from which the elec- tron is removed is C—N and C—C bonding, but C—H antibonding, thus a n-orbital of the C E N bond. This n-orbital has E symmetry in the C 68,69 3v point group and can be labelled as the 3e orbital. The second band, with its origin at 13.14 eV, has very little vibrational structure. The band has been assigned to the removal of an electron from the nitrogen lone pair, and the single vibrational excitation observed to the CH symmetric 3 deformation (v ). Since the nitrogen lone pair participates 3 little in the bonding, loss of one of the electrons should have little effect on the structure of the molecule, as is indicated by the paucity of vibrational excitation in the photoelectron band. The nitrogen lone pair can be labelled as Sal. The assignment of the remaining PE structure, in the region from 15-21 eV, is considerably less certain, but it 55 must result from ionization of C—C and C—H 0 bonds. Turner identifies the vibrational structure at the head of the band Table IV-l. Vibrational Modes and Frequencies of CH 3 compared to the Frequencies of the Neutral. 69 3 CN+ and CD CN+ Excited in the Photoelectron Spectra Frequency (cm- Frequency (cm- State Vibrational + a + of ion mode exc1ted CH3CN CH3CN CD3CN CD3CN i 02 c N stretch 2010 2267 1990 2278 v3 symm. CH3 def. 1430 1385 1070 1110 04 c—c stretch 810 920 810 831 E 03 symm. CH3 def. 1290 1385 970 1110 E 03 symm. CH3 def. 1440 1385 ——b 1110 v4 c—c stretch 860 920 ——b 831 aVibrational frequency in the ground electronic state of the neutral; taken from reference 88. bnot determined. 70 as arising from the CH3 symmetric deformation (03) and the C—C stretch (04); the changes in these frequencies from the neutral molecule suggests that the orbital ionized is C—H antibonding and C—C bonding. Frost, et a1.,66 does not even attempt an analysis. However, it is clear that in this region there are two bands: one centered at about 16 eV, the other at 17.5 eV, and that ionization from orbitals that are involved with C—C and C—H bonding is implicated in both cases. The absence of a strong transition at the origin indicates that at least some bond lengths have changed considerably. (The band centered at about 20 eV in Figure IV-l may be an instru- mental artifact.55) It can be concluded that there are two well defined states and at least two poorly defined states of the acetonitrile parent ion that will be accessible by direct ionization in a PIMS experiment. The ground state of the ion has several bond lengths and vibrational frequencies that differ from those of the neutral molecule, while the vibrational frequen- cies and bond lengths of the first excited state of the ion are probably very similar to those of the neutral molecule. The PES ionization potentials reported for CH3CN in ref- erences 55,56, and 67 are listed in Table IV-5, along with the photoionization data. Although these PES investigations A indicated that the spectrum of CD3CN was recorded, only Lake and Thompson67 reported IPs for CD3CN. 71 C. The Photoionization Mass Spectra of CH3CN and CDBCN The 584 A (21.2 eV) photoionization mass spectra (PMS) of CH3CN and CD3CN,recorded in the present study, are shown in Figure IV-2. Relative intensities and assignments are presen- ted in Table IV-2. The electron impact mass spectrum (EIMS) of CH3CN 62’70’71 shows slightly more fragmentation than the PMS, but the overall agreement between the two spectra is good, and the small differences can be attributed to the higher ionizing energy (50-80 eV) used in the El experiments. Most of the assignments of the PMS and EIMS are straight- forward even though the deuterated compound was slightly proton contaminated; however, some of the assignments deserve comments. Gray62 suggested that m/e=15 from CH3CN might con- tain some contribution from NH+ since a small amount of NH: (m/e=16) is detected. Therefore the peaks assigned as CD: and CD; (m/e=l6,l8) might also have some contribution from ND+ and NDE, respectively. However, thermochemical calculations show the thresholds for NH+ and NH: to be considerably above those for CH: and CH3, probably not contribute significantly to the 21 eV PMS. Indeed, so that NH+(ND+) and NH:(ND;) would no peak is observed at m/e=16. The high resolution EIMS‘70 of CH3CN shows that m/e=26 and 27 are multiple peaks (m/e=26 is 60% CN+ and 40% C2H+, and m/e=27 is 85% HCN+ and 15% C H+ at the unstated, but probably 2 3 >50 eV ionizing energy). However, this need not be of concern in this investigation, since the intensity of these ions is 72 c 03c N 1 ._ 11 10 2'0 31) 70 5'0 Jo 230 m /0 . CH3CN 1. _ 1 H0 Ab JP 40 9% 6b 75 '"V5 Figure IV-2. 584 A (21.2 eV) Photoionization Mass Spectra of CH3CN and CD3CN. (Quantitative Relative intensities should not be measured from these spectra.) 73 Table IV—2. CH3CN and CD3CN Photoionization Mass Spectra. (Ionization energy = 21.2 eV.) CH3CN CD3CN intensity Intensity Relative Relative to CH3CN+ to CD CN+ m/e (m/e=4l) Assignment (m/e= 4) Assignment 14 18.2 CH2+ 15 4.4 CH3+ <1 CHD+ 16 27.6 CD2+ l7 <1 CHD2+ + 18 6.8 CD3 + + 26 <1 CN 'C2H2 + + < 27 1 HCN ,C2H3 + + + + + 28 5.2 HZCN ,N2 <1 DCN ,N2 ,c202 + + 30 8.1 DZCN ,c203 38 1.2 C2N+ 1.2 CZN+ 39 23.0 HC2N+ + + + 40 78.7 HZCZN 22.7 DCZN ,(HZCZN ) 41 100 H c N+ 2 3 HDC N+(H c N+) 3 2 ° 2 3 2 + + + 42 13.3 H4C2N 80.4 DZCZN ,(H4C2N ) + > 43 6.0 HDZCZN + 44 100 D3C2N + 45 3.2 HD3C2N 46 19.0 D c N+ 4 2 74 too low below 21 eV for a PIE curve to be measured. The peak at m/e=28 in the PMS of CH CN does contain some contribution 3 from N3, but it is quite small as can be ascertained by com- parison with the CD3CN PMS. However, the N; PIE curve is quite distinct and N; did interfere with the HZCN+ PIE curve. (See section IV-E5.) The small differences in the CH3CN and CD CN PMSs are 3 mostly attributable to a difference in the sample pressure (if contributions from the protonated impurity in the CD3CN sample are neglected). The CH3CN PMS was recorded with a sample pressure of 4.9 x 10.4 torr and the CD3CN with a pres- 4 sure of 5.2 x 10- torr. This 6% descrepancy was uninten- tional, but it serves to demonstrate the pressure dependence of the reaction of the parent ion with the neutral sample to produce CH3CNH+ (CD3CND+). Due to the higher pressure of CD CN, the relative intensity of the reaction product is 3 greater in the CD CN PMS. Thus more of the parent ions 3 initially formed are detected as P+2 in the CD3CN PMS which accounts for most of the higher relative intensities of the fragment ions in the CD3CN PMS compared to the CH3CN PMS. CH3CN+, CHZCN+, CHCN+, HZCN+, CH; and CH3, as well as their deuterated analogs, are produced with sufficient inten- - sity to enable PIE curves to be recorded. Pressure-depen- dence studies show that all of these are primary ions. The absolute intensity of CH3CN+ and CD3CN+ in the mass spectra are about 3500 ions/sec, which is more than twice the 75 intensity that they are produced at the maximum (~800 A, 15.5 eV) of the helium continuum. A search was made for negative ions and CN- and CHZCN- ions were detected; however, their intensity was much too low for PIE curves to be recorded. D. Parent Ion PIE Curves 1. General Observations The PIES of CH3CN+ and CD3CN+ are presented in Figure IV-3. On the scale of the figure the two PIEs should be, and are, identical. There is a distinct threshold at about 1015 A (12.2 eV), followed by an autoionizing Rydberg series converg- ing to the first excited state of the ion. From the photo- electron spectrum, the threshold of the first excited state should be about 944 8 (13.1 eV) and the jump in the PIE at approximately 944 A is the threshold for this state. The dis- tinctness of the feature at 944 A indicates that the Rydberg states converging to the first excited state of the ion are strongly predissociating into neutrals, as well as auto- ionizing, since no ion fragments are observed in this region. If the Rydberg states were not predissociating the Rydberg series would join smoothly onto the direct ionization continuum of the first excited state.9 Acetonitrile is known to pre- dissociate into neutral fragments at energies considerably less than 13 eV.70 The region between 944 A (13.1 eV) and 820 A (15.1 eV) corresponds to a blank region in the PES spectrum, and thus additions to the PIE from direct ionization to new states are .m oeoaucoc ”nosuso med +zomoo poo +zommo .mu>H ousoam «689685 Eozmom>§ one. 80. 03 com can con 0mm can one 80 p—nhEP—nh-P—bF-prh-p—thP—nbblrhppbbPPPPn—Phth 76 0111: a 3 mm m.. on.’ J m M +20 Io n m 3.1. a. zomoo + q q q — q 4 q u u d d N d — q q u d u - a q: —- 11:4..-—«Judd-q:qq—uq-u-udd-—-uandH4-d—Jd-uq-d-du-qq-u--quq up n.» I. m.» Q. Nu m. 2. ON 36> cocoon: GEE 77 not expected. Nevertheless, as the photon energy rises fol- lowing the threshold of the first excited state, the PIE increases. This could be due to an increase in the direct ionization cross section to the first excited state of the ion, but it is most probably due to autoionization. The evidence for autoionization is the onset of fragmentation in this region. Therefore states above the first excited state of the ion are populated following photon absorption, but they must be popu- lated by autoionization since the photoelectron spectrum indi- cates that there are no ionic states in this region available for direct ionization. Moreover, in the CH3CN+ PIE there is some structure in the 885-800 A range which can only be attri- buted to autoionization. The leveling-off of the PIE at about 885 A (14.0 eV) correlates well with the fragmentation thresh- old for the first fragment, CHZCN+, and the decrease in the PIE from 885 A to 820 A (15.1 eV) is due to cessation of the autoionizing Rydberg series caused by fragmentation. That is, the parent ion states populated by autoionization in this region fragment and are not detected as parent ions. This is demonstrated in Figure IV—4 in which the dashed line is the total ionization PIE, the sum of parent and fragment ions. The total PIE actually increases. From the PES spectrum, the threshold for the second excited state of the ion is at approximately 820 A (15.1 eV). However, there is no discontinuity in the PIE near 820 A. This is due to a combination of two factors. First, there are Rydberg states converging to this ionic state and the autoionization A.mHo Hou0u 0:» mouooaoca mama omnmooc .zommo mo mHm Hobos .eu>H owsmam «meotmocé 1823.05; omo. ooo. omm ooo omo ooo own ooh ono ooo p—-up»—+L-p—LLbDP-pTP—anP—hppP—b-h-bnhbrlrb-b! 78 OI (snun 001119111) 31.1 q-q-(dw.-qqfi—udfiq-q4-q-fidlfiq—qfiuq_d1dqd.l—-idd114#q44 79 may join smoothly onto the continuum and hide the threshold. Second, and probably more important, is the likelihood that all, or nearly all, of the parent ions formed in the second excited state are dissociated into ion fragments on the time scale of the PIMS experiment. No discontinuity would be ex- pected because no parent ions formed in this state would be detected as P+. It is a general observation for polyatomic molecules that most, if not all, ions formed with internal energies above the first fragmentation threshold dissociate.l3’l4 The first fragmentation threshold for acetonitrile is at approximately 885 A (14.0 eV), well below the threshold of the second excited state. In acetonitrile some ions formed with internal energies just above the first fragmentation threshold do not dissociate, otherwise the autoionization beyond 944 A would not be observed in the CH3CN+ PIE. How- ever, the second excited state is more than 1 eV beyond the onset of fragmentation and is, most surely, nearly completely dissociated. The region beyond 820 A in the parent ion PIE, to 600 A (20.7 eV), then, would show no evidence of ionic states above the second excited state. The PIE in this region thus reflects the cross sections for direct ionization to the ground and first excited states of the ion. As would be expected in this case, the PIE slowly and smoothly decreases with increasing photon energy.22 Upon close examination, however, there are some small, but very important, differences in the PIE curves for CH3CN+ and 80 CD3CN+. In the following sections a more detailed examination of the autoionization structure and the threshold regions of these curves is presented. 2. Autoionization Structure: 1040 A - 940 A The CH3CN+ and CD3CN+ PIEs between 1040 A (11.9 eV) and 940 A (13.2 eV) are shown in Figures IV-5 and IV-6. Mem— bers of two Rydberg series are readily apparent and are indi- cated on the horizontal line near the top of the figures. Only two members of the weaker series are resolved; this is due to the broad line widths of the series members and not to instrumental resolution. Fridth72 has reported the electron impact energy loss (EIEL) spectrum of acetonitrile and has also observed these two series. In the following pages the assignment of the series type and the series limit are dis— cussed. The PES provides an initial estimate of the series limit. The location of the first excited state of the ion is known from the PES. From inspection of Figures IV-5 and IV-6 it can be seen that the series are converging to this state, and therefore the Rydberg series arise from excitation of the 5al orbital of the neutral molecule. The quantum defect, which is a primary consideration in the assignment of a Rydberg _ transition,can then be calculated by using the photoelectron ionization potential of the 5al orbital in the Rydberg equation: Ehv = IP _ R . (II-4) (n - 0)2 However, another option is to fit a series to the Rydberg 81 m .m cecanoem H0>uso mHm +zo mo .mu>H ounces «62.8.8.5 Eozmdés coo. omo. ooo. coo ooo owo pub-prPPPb-pppbbPP—pr--ppu—ppupphhpp—pnpup-pub——-~b-pPP_-npnpprPPh—ppppbhp—bb-PEPnpb—bP-Lb-pr 'I + ’0 a”... )I .. .\) .. 0‘ X ) r. x ... x. n\ ’2‘. co to!) on: m o a o oo.«. ... - _ _ 7 q F P . ..r .. 01: n o .o. I... u. 30 «Wu—fid-q—qqufiqqd-‘—uqdu—q-acqdde—dq11—1-quJuddfidfi—uu-u—qdqfid oN. Nd.» VN. ods ow; own... «.2 36> 8:89 SEE (swan 00111941) 31d 82 m .N oooalovm uw>HDU mHm +20 DU .oI>H wusmwm «motocross Eozmd>§ 0V0. ONO» coo. 0mm coo ova burn-pub?pbPP-hhh—pphbnPPPLppb-pb-beFP-ppppbp—hppupphbbl—Fp-pnnh-h—nnphb-Ppbl—bPh-bbppb—p-bnbbtnhh .o lc/(I +zomoo o<¢l y’ooo \o ..... ........ ...... ‘ Q o 0'09 f 0‘. T 00‘ I of a 3K) In?!» In a... m o k a can. 2.. 1— — p A p F — — -P .0 .03: m 0 00/0 U . JO dud‘dqfiqd-fiqfififiJl-q—qqq-—11-u—u1W.—qudddqqqd-qdqq—q-11-1141-d-1d od. «N— Y? ad. ad. ofi. NM. 36> cotton» SEE (swan Mommv) 3Id 83 equation with the ionization potential and the quantum defect as adjustable parameters. Very accurate ionization potentials can be determined by this method.73 The more intense of the two series observed in the CH3CN+ and CD3CN+ PIE curves was fit to the Rydberg equation by means of KINFIT 4, a general non-linear, least-squares curve-fitting computer program.72 The series limit and the quantum defect were taken as adjustable parameters. The quantum defect was assumed to be constant as a function of n. Although this is not strictly correct, the error that results for n 2 5 for a molecule composed of second row atoms is negligible.73 Fridth's EIEL data72 were also fit for comparison with the photoionization data. The results of the fits are given in Table IV-3 along with the experimental locations of the ob- served series members. First consider the series limit. The photoelectron ioniza- tion potentials reported for the 5al orbital are 13.11 eV66 and 13.14 ev.55'67 Although no error limits were published, the PES spectra were probably recorded on instruments with about 50 meV resolution, as will be shown when the lowest ionization potential is discussed. With 50 meV resolution peak positions can usually be measured to within about 10 meV. Fridth's EIEL data were also collected with an electron energy analyzer having about 50 meV resolution. The PIE data were obtained with a photon bandwidth of 0.84 A; at 980 A this corresponds to an energy bandwidth of 11 meV. It was possible to locate the autoionization peak ct . an Ml CU 1 PH a 1:” 84 Table IV-3. Rydberg Series Converging to the First Excited State of CH3CN+ and CD3CN+. CH3CN CD3CN This work Fridth72 This work Series a b limit(eV) 13.133 i .004 13.135 1 .006 13.136 1 .004 ~6(nso) .93 i .04 1.01 r .004 .92 i .04 Experimental energies (eV) and 6 calculated from series limits derived from Fits n energy(eV) 6 energy(eV) 6 energy(eV) 6 n50 3 9.713 1.01b 4 11.594 1.03 5 12.295 0.98 6 12.601 0.94 12.596 0.98 12.606 0.93 7 12.775 0.84 12.759 0.99 12.775 0.86 8 12.864 0.89 12.854 1.03 12.868 0.88 9 12.924 0.93 12.92 1.05 12.925 0.97 10 12.964 1.03 11 12 13.020 1.03 13.023 1.03 ndo 12.180 0.23 12.542 0.22C 12.534 0.24 12.550 0.182 6 12.724 0.23 12.706 0.37 0 -_ 12.732 aErrors are linear estimates of the standard deviation. .200 bResults of this investigator's fit of Fridth's po series. CQuantum defects for the ndo series were calculated from limits of nso series. 85 centers to within r 0.5 A or t 6 meV at 980 A. Furthermore, the photoionization data have a built-in calibration — the atomic emission lines in the light source output — which the photoelectron and electron impact data lack. If the fit is reliable the photoionization data should provide a more accurate value for the 5al ionization potential. The series limits obtained from the fit to the photoionization data are 13.133 1 0.004 eV for CH CN and 13.136 : 0.004 eV for CD CN. 3 3 The estimates of the standard deviation indicate that the fit is quite good. However, because the Rydberg series fit is an indirect method it could be argued that the small difference between the Rydberg series limit and the photoelectron ioni- zation potential is not significant. However, the PIES present a very fortunate situation. If the step at 944 A is indeed the threshold for ionization from the 5a1 orbital, then an independent measure of this orbital ionization energy is available. Generally, when a step cor- responding to the threshold for direct ionization to a vibronic state of an ion is observed, the point on the step correspond— ing to the peak center in a resolved photoelectron line, the ionization potential for the vibronic state, will be at the inflection point or midpoint of the rise of the step.75 This assumes that autoionization does not significantly contribute to the step. Because the autoionization intensity in CH3CN+ and CD3CN+ is low in this region, and because there is a fairly sharp break at the foot of the step, autoionization probably does not significantly contribute in this case, and the step 86 should yield an accurate ionization potential. The photo— electron ionization potentials for the 5a1 orbital (13.14 eV and 13.11 eV) are marked on the PIEs in Figures IV-4 and IV-5. Also marked are the limits derived from the Rydberg series fits. The Rydberg series limits are closer to the midpoints of the steps than are the photoelectron results. Consider- ing the smaller error limits of the photoionization data, the series limits derived from the fits, and the step at the limit, it is concluded that the photoionization result is indeed a more accurate determination of the 5al nitrogen lone pair orbital ionization potential. The difference between the CH3CN+ and CD3CN+ limits could be real, due to a difference in the change of the zero point energy between the neutral and the ion in the isotopic mole- cules. However, since the difference is small — within the error limits — and since 13.133 eV appears to be a better estimate of the center of the step rise in the CD3CN+ PIE, it can be concluded that CH3CN and CD3CN have the same 5al ionization potential: 13.133 1 0.005 eV. It is worth noting that some members of the more intense series are missing in the PIE curves, based on the quantum defect and the series limit. For example, the n=5 member should provide an intense peak at 1005 A, but no distinct peak is observed there. Probably an unbound potential surface of the neutral molecule crosses the n=5 Rydberg series surface, allowing it to be nearly completely dissociated. This member is clearly observed in Fridth's72 EIEL spectrum, and there are 87 remnants of it in the PIE. The n=10 member of CH3CN+, and n=10 and 11 of CD3CN+ are also missing, presumably for the same reason. The n=4 member of the less intense series was also observed by Fridth;72 it is located below the first ioni- zation potential and therefore would not contribute to the PIE. Consider next the series assignment. Remembering that s-type Rydberg series have quantum defects of about 1.0, p-type 28’35 it would be reason— of about 0.6, and d-type of near zero, able to assume that the more intense series is an s—type with 6:0.92 and the less intense series a d-type with 6:0.2. On the basis of the photoionization data alone, little more can be said. However, Fridth72 was able to observe several series members converging to the second ionization potential in the spectroscopic range inaccessible to PIMS (below the first ionization potential), as well as some members above the first ionization potential. Moreover, several series converging to the first ionization potential were observed. With the addi- tion of these data, more definitive assignments are possible. 68,76 Two optical studies in the vacuum ultraviolet and another EIEL investigation77 of CH3CN have been reported, but in all three publications either the data were not recorded to high enough energy or were obtained with inadequate reso- lution to be of much use. Fridth has assigned all the prominent features in the EIEL spectrum,72 including the two series observed in this study. The series corresponding to the weaker autoionization in the PIES was designated do with 6:0.23, and this assignment is 88 in good agreement with the photoionization results. However, the more intense series was assigned as pa with 6:0.97. AS pointed out by Fridth, this assignment is somewhat questionable and this investigator believes that it would be more correctly assigned as an So series. Fridth assigned the series in CH3CN by analogy with HCN. The EIEL Spectrum of HCN has been reported by Fridth and Asbrink,78 and they observed two series corresponding to exci- tation of the nitrogen lone pair. These two series were assigned as so (6:0.91 for the higher energy members of the series) and do(6=0.4). Later, however, Asbrink, Fridth and 79 Lindholm published results of a HAM/3 calculation of HCN. (HAM/3 is a semi-empirical technique for calculating valence 80) From the results of transitions, developed by Lindholm. the calculation it was concluded that some structure in the EIEL spectrum of HCN, which had originally been interpreted as Rydberg transitions, was really due to valence transitions. The Rydberg transitions then needed to be reassigned. As a result only one series, corresponding to excitation of the nitrogen lone pair, was identified. It was assigned as po with 6=0.80.79 It was suggested that the quantum defect was larger than is typical for a pa series due to penetration of the p Rydberg orbitals into the H—C—N bonding orbitals. For CH3CN, Fridth reported three series, all correspond- ing to excitation of the nitrogen lone pair and converging to the second ionization potential, and four series corresponding to excitation from the C E N n-bond and converging to the first 89 ionization potential. The upper series were assigned as p0 (6:0.97), pn (o=0.60), and do (6:0.23); the lower series as so (6:0.97), p0 (6:0.76), pr (6:0.61) and d0 (6:0.23).72 Two points need to be made. First, a very important ob- servation in the spectroscopy of Rydberg states is that members of the same types of series, converging to different states of an ion, have very nearly the same term values.33 (A term value is the difference between the series limit and the energy of the transition to the Rydberg state, i.e., R/(n-6)2 in equation II-4). Thus, a po series converging to the first excited state of an ion should have nearly the same quantum defect as a po series converging to the ground ionic state, and likewise for other series. The implication is that the molecular orbital from which the electron is excited has very little effect on the energy of the Rydberg orbital;33 this is a reasonable conclusion when one considers the large spatial extent of the Rydberg orbitals. Second, when the EIEL data for CH CN are compared to the 3 photoionization energies (as in Table IV-3), it is noted that the EIEL values are systematically lower than corresponding photoionization values by an average of 10 meV. Considering that the photoionization data were taken with higher resolu- _ tion and, most importantly, were calibrated with several atomic emission lines in the lamp output, the error most likely lies in the EIEL data. This type of systematic error is common for electron spectrometers due to changes in contact potentials, caused by contamination of electrodes in the spectrometer.21’22 90 Although this error is small, it has a significant effect on the quantum defects calculated for series members with large n. However, for series members with small n, the EIEL quantum defects should approach the photoionization values. For example, from the Rydberg equation and an assumed IP of 13.00eV, for n=8 a transition energy of 12.758 eV gives 6=0.500, while transition energy of 12.748 eV gives 6=0.652. For n=4, a transition energy of 11.889 eV yields 6=0.500 whereas ll.879<§J gives 6=0.516. The quantum defects calculated from Fridth's data continue to differ from the photoionization values as n decreases. It appears that some of the lower members of Fridth's p0 series may have been assigned incorrectly. This is further exemplified by the limit calculated from the data in reference 72, which agrees very well with the photoioniza- tion limit, yet should be about 10 meV lower. The above ideas are represented schematically in Figure IV-7. Figure IV-Ladepicts the transition energies from the ground state nitrogen lone pair orbital to Rydberg orbitals as measured and assigned by Fridth.72 Also included are n+w* transitions assigned in that work. In Figure IV-7b are hypo- thetical transition energies to Rydberg states converging to the second ionization limit, calculated from the quantum defects reported in reference 72 for series converging to the first ionization potential of CH3CN. The photoionization data are shown in Figure IV—7c; the dotted lines were calculated from the photoionization quantum defects. 91 I 3.0 l2.5 '— Energy (eV) 17> <3 I ".5— 990 8pc 790 660 Spa 560 So? 590 460 4pn 1n>n’ Thrfl' 490 7hbfl' n-brr‘ (o) (b) 080 9808‘“,- do 7pc 660 Spn Spa 650 560 Spn Spa 550 41:10- 4px 490 450 300 7px 990' Si Odo 600 Odo- ------- 40: (c) ".00 Figure IV—7. Transition Energy Diagram for Excitation from the 58 (a) EIEL transition energies from Reference 72; (b) Hypothetical Rydberg transitions cal- culated from the quantum defects of Reference 72 for series converging to the ground state of the ion; (C) Photoionization data of this work. Orbital of CH3CN. 92 From Figure IV-7a and IV-7b it is easily seen that Fridth's p0 series converging to the second ionization potential is in- consistent with the series converging to the first ionization potential. The small discrepancy between the calculated series and the photoionization series is not unreasonable (see reference 33). Moreover, it seems that the highest energy transition labeled w-n* by Fridth would be better assigned as 4pc, and the third n-n* component is more likely 450. These changes would make the EIEL results self-consistent as well as consistent with the photoionization results. The series with 6=0.92 is therefore assigned as so. Figures IV-5 and IV-6 and Table IV-3 have been labeled accordingly. 3. Autoionization: 900-800 A PIE curves for CH CN+ and CD CN+ from 800 to 900 A 3 3 are shown in Figure IV-8. The structure in CH3CN+, which as noted earlier is weak, but prominent, would appear to be auto- ionization; yet the CD3CN+ PIE in this region is smooth and structureless. No optical or EIEL spectra have been pub- lished of this region with which to compare the photoioniza- tion data. In general, one would not expect so much difference between a protonated compound and its deuterated analog. If autoionization structure is observed in one, it Should also be observed in the other, although any vibrational structure may be different. This leads one to wonder if the structure in the CH3CN+ PIE curve is an artifact. The PIES in Figure IV-8 are reproducible and therefore it is unlikely that a difference of this type is an instrumental artifact; the 93 .m oomuoom unoiso med +zomoo o «mEotnaci EoEd>§ 00m 0mm com Ovm owm E——.p—P#~_pp——ht—p:_—-_-:b_p_:__:Pb_—_p__~____—:p—bF-____:::_—_—PPP_—P_——: m cm +20 $0 00% __:_»—»—~_:»b—_ M. \. +20 00 1.....r... :1. .. . .. . ... v .7... .... 51...... .. ‘ a. h ’0. I o... o(. .. s natal . a. a .o I If . shba o‘- 00 \ a}... r) . .... . n ... ...;s a ....»J . 20 I0 Hang. .. .o ((1.4..‘05 + aunt”... . .JJ’AI...¢..... . ... . . livax.” . ...: o..:.. o .. \...oka‘.o‘ . ..x. .. .0. in: «om—.64.". (a. one .o- o o b 0.. ‘0 u a . A6‘a a o .0, fl .. O). a o. . u. . . I o'... C 0!... art... . As .... . ...)... ...v: I). (6.. .... .v. 1... v . Bill . . n ...5 ca. 5. o . . .. w .‘ .Q C.- o ... ....r. 92.2.. callus. .. .... v.4 . v . . . .o. Cd ... ... .. .. . .0 a < “to o . a . . \ I v 4.4.1....» . r a...»" 51%.“ a... . . \a . of. . n o . :0 o o o. o .52.... . . .. I1. . ..v...a«.. . . ..o. . ...xk... ....f .. . ....r. s... . . . . .......... ...ut.....¢..w.n...n.... .....u..........,+..... .3. .. . . . 9.... ......,......a..... L 0'. ‘. “(3.. ... 361...... . I s. .. .LA/LsAk .O’. .- nus-0%. a it... n __ 'I .1 '1'. \ \ a’{“ at" C" .0 O.- " 3.2:. \- "e l N '——-"‘t:—__._—__—_' .5 +62. n o n; o to: n o h . J+flzobnc r. o -qq—qqaq———_—_16_««—d_qfi—__qd—_~«_a—_fia.—H~_—_«q.—_—__—~—__—_—d_—_————_—_——:fi__::___d4 Omen,“ @041 CN.: 06.: 09: 0%.: 00d: ONOA 06.05 tics, ......-:.._,.:C 34:21.1 .m1>H musoflh ( 521W? 41001947) 31d 94 only parameter which is different for the two PIE'S is the mass setting. + . . The m/e of CH CN 15 41; the most intense or second most 3 intense peak in the mass spectra of many longer-chain ali- 71 However, the possibility phatic nitriles falls at m/e=4l. that the structure in the m/e=4l PIE is caused by another aliphatic nitrile is unlikely for three reasons. First, there is no evidence in the mass spectrum of the acetonitrile used in this investigation of longer chain aliphatic nitriles. Second, it is unlikely that a longer chain aliphatic nitrile would introduce autoionization structure which is so sharp as the observed structure. The photoelectron spectra of longer chain nitriles81 show only broad and featureless bands for states to which Rydberg series in this region would converge. Rydberg states, being very similar in structure to the ionic states to which they converge,33 would exhibit autoionization which is also broad and featureless. Third, the same auto- ionization structure is also observed in the CHZCN+ fragment PIE at m/e=40. Other aliphatic nitriles also have a fragment at m/e=40, but its intensity is so low (3-5% of m/e=4l)71 that it could not account for the structure in the CHZCN+ PIE. (Consistent with the observations for the deuterated parent, this autoionization is also absent in the PIE curve for CDZCN+J Another possible source of the structure in the 800-900 A range is reaction with, or Penning ionization by, excited argon. This would have to arise from argon in the background air. This might seem to be a very remote possibility, but 95 (1) much of the structure corresponds with optically popu- latable argon states;82 (2) with a background air pressure of 10..5 torr, 107 argon atoms are in the light beam at any one time (argon is 0.033% of air) and the structure represents variations of only about 30 ion counts/s on a background of about 1500 counts/s; and (3) Ar+ is known to react with mol- 83 Interference ecular hydrogen to produce ArH+ (m/e=4l). from ArH+ was also considered in the ion cyclotron resonance ion-molecule reaction study of acetonitrile by Gray.62 There are three reactions that could account for ions at m/e=40 and 41: (1) Ar* + CH CN + ArH+ + CHZCN 3 (2) Ar* + CH3CN + Ar + CH3CN+ (3) Ar* + CH3CN + Ar + CHZCN+ + H. Ar+ is not responsible because the structure lies below the ionization potential of argon. Reaction (1) alone would not account for structure in the CHZCN+ PIE, and reactions (2) and (3) should be just as probable for the deuterated compound as for the protonated compound. Thus, argon would not seem to be the cause of the structure. However, to be absolutely sure, the PIE at m/e=4l was measured for a 1:5 argon: acetoni- trile mixture. The argon pressure was l.0><10—4 torr, a 3Xl04 fold increase over the highest possible background pressure. There were no significant changes in the CH3CN+ PIE. The structure in question can thus be assigned to autoionization from Rydberg series converging to the second excited state of the acetonitrile parent ion. 96 .mm mocmuommu Eonmm l ..IIIIIIIIIIIIJIIIII!‘ In-nl.l||l.liIlll|lIbul.II1l no.a n @ mommm>m OH.H mma.ma mo.H mmo.ma no.0 mHo.mH mm.o Hmm.vH m mH.H oma.mH vo.a mHo.mH mo.H ovm.va mo.H omw.va m 0H.H mao.ma Ho.H Hmm.va mo.H mom.va NN.H mmh.va n mo.a mom.va mm.o mmn.va mo.H who.va mo.H mom.v~ o wo.H vmm.v~ mo.a mav.va HH.H Hmm.va VH.H mam.va m m amuwcm @ >mumcm @ xmuocm e xmuwcm a pan A>ov moflmuocm Hmucoeflnmmxm v>H.m>H 9H 9H ocsoum mmumum Hmsoflpmunw> av.ma om.mH mm.mH ma.ma A>ov uflEflH mowumm m +20 m mu mo mamum omufloxm ocoomm onu.ou mcfimum>cou mowuom muwnozm .vI>H OHQMB 97 From Turner's55 ionization potential and vibrational assignments for the second excited state of CH CN+, four 3 series converging to four different vibrational states can be identified. The series are shown in Figure IV-8 on the bars near the top of the figure and are listed in Table IV—4. From the quantum defects the series is assigned as an So type. The same Rydberg states are expected to autoionize in CD3CN; however if the vibrational frequencies are sufficiently closer in CD3CN the autoionization structure may be so over- lapped that the series are unidentifiable. Indeed the vibra- tional structure of the second excited state in the photo- electron spectrum of CD3CN is much less distinct and more over- lapped than in CH3CN;66 the same should be expected of the Rydberg series converging to this state. 4. The Threshold Region: 1000-1040 A In order to precisely determine the adiabatic ioniza- tion potentials, and to resolve as much structure as possible, PIES of the parent ions in the threshold region were acquired with a monochromator bandpass of 0.42 A (5 meV at 1010 A); they are presented in Figure IV-9. Recall from Chapter II that in the absence of autoionization a PIE would be a series of steps, each step being the threshold for a transition to a - new state of the ion. The threshold region of CH CN+, for 3 which there are two well defined steps at 1016 and 1012 A, demonstrates this nicely. Judging from the high intensity and location of the first peak of the photoelectron spectrum 98 JllllIllJlJJlilIJJJJ—lJllllllllll ’\. f.\o ‘3 ° ‘ Oa ' 0‘ 400...: O ? O‘.‘~.. 5’. . '-:. 00. \ I“ 0. 0":’: u. .2 I. 0.. . ... ‘ °.°'.:. ‘. + 1,} ' . °. CH3CN .2 O 5 ’1 1.. b .° 3.00 E 5: “E 3: ° 3 Q : .. é a '< \a .0 2 E ’: - 0. l a + - . CD3CN 1 E o \. ‘ s . I I a”. ... o~‘00. — O \ \. ° ne'. 0“. co "-"‘.~° «3.435. — 0 TerjjijTjIIIUIjjITjIl'TIIIjIIIIr 1000 1010 1020 1030 WAVELENGTH (Angstroms) Figure IV-9. CH3CN+ and CD3CN+ PIE Curves: 1000-1040 A. n: 99 of CH3CN, there is little doubt that the first step is the threshold for the transition to the ground vibrational state of the ground electronic state of the ion (the adiabatic ionization potential). The second step must be a vibronic transition to an excited vibrational state of the ion, and is a transition which has not been resolved from the 0+0 transi- tion in the photoelectron spectrum. It is only because Rydberg states converging to excited vibrational states of the ion are strongly predissociated that these steps are so distinct. The third very small step at 1010 8 is in all probability due to autoionization, as will be discussed below. The CD CN+ PIE in this region is significantly different 3 from that of CH3CN+. There is only one poorly defined step at 1012 g, followed by an increase in intensity which, like the third step in the CH CN... PIE, can be attributed to auto- 3 ionization. Close examination shows also that the initial step in the CD3CN+ PIE is at a slightly shorter wavelength than the corresponding rise in the CH3CN+ PIE. It appears that the first ionization potential of CD3CN is higher than that of CH3CN. Indeed, one photoelectron study67 has reported a difference in the first ionization potential of CH3CN and CD3CN, in the same direction indicated by this work. The adiabatic ionization potential is unambiguously de- fined as the difference in energy between the neutral mole- cule in its electronic, vibrational and rotational ground state and the ion in the lowest vibrational and rotational level of a particular electronic state.22 Since it is certain, by 100 comparison with the photoelectron spectrum, that the initial step of the acetonitrile PIE is the transition from the ground vibrational state of the neutral to the ground vibrational state of the ion, the problem remains to determine the loca- tion of the rotationless transition on the step. This requires an approximation since the resolution of the instru- ment is far from sufficient to resolve rotational structure. The 5 meV bandpass is equivalent to 40 cm—l. Before discussing the rotational structure of the steps, the effect of the instru- ment function should be considered. The effects of the instrument function have been discussed in the literature,84 and the results are presented in Figure IV-lO. A step function is assumed for the photoionization cross section for a single transition from the neutral to the ion and is convolved with a triangular slit function, which 85 The result is has been shown to be a good approximation. a sigmoidal curve whose inflection point is at the true threshold and for which a line drawn tangent to the inflection point intercepts the energy axis at 6/2 (6 is the slit width) below the true threshold. Application of this result to the threshold PIES shows that the slit width accounts for about half the breadth of the rise of the steps. The remainder, then must be due to the rotational envelope within the vibronic transition. The rotational structure within a vibronic photoionization transition has not been well characterized experimentally or theoretically. In only a very few cases has some rotational 101 PIE Figure IV-lO. The Effects of a Triangular-Slit Function on a Step Function. (a) Triangular-Slit Function; (b) Triangular—Slit Function convoluted with a Step Function. 102 structure been experimentally resolved (the photoelectron spectra of H2,86 HF and DF87), and the theoretical deriva- tion of the applicable selection rules is complicated by coupling of the ejected electron's angular momentum, which can take on several values, with the total angular momentum of the ion.22 However, the selection rules for bent ABz—type molecules have been worked out,22 and it is clear that rota- tional transitions in P, Q and R branches are allowed. It is likely that at least P- and R-branch, and perhaps even Q-branch, transitions will be allowed in acetonitrile. Because the PIEs were measured at room temperature, the thermal distribution of populated rotational states of the neutral molecule is available for ionization. Therefore a photoionizing vibronic transition will be accompanied by a rotational envelope. An approximation is needed for the tran- sition probabilities within the rotational envelope which in- cludes the thermal distribution of rotational states of the neutral. The transition probability then can be convolved with the slit function and a step function to obtain a model PIE which could be used to fit the experimental data in order to arrive at an accurate ionization potential. The model chosen to approximate the transition probability ‘ within the rotational envelope, convoluted with the slit function was a Gaussian distribution centered at the ioniza- tion potential. The choice of a Gaussian distribution is easily checked by fitting the data, and it will be shown that 103 it is a good approximation to the transition probability. The assumption that the rotationless transition is at the center of the rotational envelope is somewhat less certain. However, a survey of ordinary absorption spectra24 supports this assumption within the limits of resolution of these experiments. Moreover, this is the same assumption that is used in photoelectron spectroscopy, where the maximum of the lowest energy vibronic peak of an electronic band is assigned as the ionization potential. For a Gaussian transition probability within the rotational envelope, the PIE (Ep), the photoionization efficiency at photon energy E ,is calculated by integrating the Gaussian P from -0 to E E _ _ 2 I p e b(B IP) dE, (IV-l) PIE 03) = a p 0 where E is the excitation energy above the ground state of the neutral, IP is the ionization potential, 9 determines the width of the step and 3 represents the height of the step. An example of this model is given in Figure IV-ll; the ionization potential is at the inflection point on the rise of the step. To test this model and to locate the inflection points as accurately as possible, the thresholds were fit by means of 74 '_ KINFIT 4 with a, b, and IP chosen as adjustable parameters. The results of the fits, which demonstrate that a Gaussian is indeed a reasonable approximation, are presented in Figure IV-12. For CH3CN+ the second step was also included in the fit by integrating the sum of two Gaussians, one for each step, Energy (0) 1 El Energy (b) Figure IV-ll. Integration of a Gaussian Transition Proba— bility. (a) Gaussian function; (b) The integrated Gaussian function. 105 Jlllllllllllllllllllllljllllllllll PIE (Arbitrary Units) l I l I I I I ' l l I l l l‘ 'I ' I 1 I ‘7“ l I I 1 I I I I T 1000 1010 1020 1030 WAVELENGTH (Angstroms) Figure IV-12. The Pit of the CH3CN+ and CD3CN+ Thresholds. 106 with different parameters a, b, and IP. In this way the funda- mental vibrational frequency of the excited vibrations can be obtained. The ionization potentials arising from the fits are indi- cated with arrows in Figure IV-12, and are listed and compared with values reported by other investigators in Table IV—S. In general the agreement is good, but merits some comment. For CH3CN note that all the values, except those of references 65 and 66, are higher than the ionization potential determined in this work. Probably the reason is that the vibronic transi- tion corresponding to the second step of the PIE was not resolved, which would cause one to arrive at a higher ioniza- tion potential. This is definitely the case for the photo- electron results for which spectra werepublished, and although neither instrumental resolution nor PIE curves were reported, this was probably the case with the previous photo- ionization results too, judging from other publications of the investigators. The ionization potential of reference 66 is obviously too low. The limits of error reported for this work are realistic. The PIE curves can be calibrated to better than i 0.1 R with atomic emission lines in the light source output, and differ- ences in the ionization potential of i 0.3 3 were clearly discernible in the quality of the fit. The propagated sum of these two errors is i 3.6 meV at the ionization potential. The error limits reported in Table IV-S, r 5 meV, reflect the uncertainty that the ionization potential is at the center of 107 Table IV-S. First Ionization Potentials of CH3CN and CD3CN. Reported Value (eV) Methoda Reference CH3CN 12.194 i 0.005 PIMS This work 12.19 i 0.01 PIMS 65 12.205 i 0.004 PI 64 12.22 i 0.01 PI 63 12.20 : 0.01 PE 67 12.21 PE 55 12.12 PE 66 CD3CN 12.235 : 0.005 PIMS This work 12.23 i 0.01 PE 67 % aPIMS-—Photoionization mass spectrometry; PI-Photoionization without mass analyses; PE-Photoelectron spectroscopy. 108 the Gaussian transition probability assumed for the fit. The ionization potential of CD CN is indeed higher than 3 the ionization potential of CH3CN by approximately 41 meV. This difference agrees with the results of reference 67, the only published ionization potential of CD3CN, and is further substantiated by a shift to higher energy in CD3CN (compared to CH3CN) of Rydberg series converging to the ground state of the ion.72 This isotope shift must be the result of a differ- ence in the difference of zero-point energies between the neutral and the ion in the two molecules. 1 The second step in the CH CN+ PIE is 53 meV (427 cm- ) 3 above the ground vibrational state of the ion. From a com- parison of this energy difference to the vibrational frequencies of the neutral molecule, the most likely assignment for the step is the transition from the ground vibrational state of the neutral to the vibrational state of the ion with a single quantum of excitation in the CCN bend (v8). The vibration V8 has a fundamental frequency of 362 cm.1 88 in the neutral and is the lowest frequency normal mode in acetonitrile. The only other remotely possible assignment would be the transition to v=l of the next lowest frequency vibrational mode of the ion, the CC stretch (v4), for which the frequency in the neutral is 920 cm-1.88 It is extremely unlikely that the CC stretching frequency in the ground electronic state of the ion would be less than half that in the neutral molecule.22 The ratio v8(CH3CN):v8(CD CN) in the neutral molecule is 3 109 1.094, and from this ratio one would predict V8 in CD3CN+ to be approximately 390 cm”1 (48 meV). This would put the mid- point of a step in the CD3CN+ PIE for the threshold for a single excitation of V8 at 1009.4 3. However, the midpoint of the second step in the CD3CN+ PIE curve is at 1010.3 fl, the same wavelength as the third step in the CH3CN+ PIE. If the second step in the CD3CN+ PIE is the vibronic threshold for the single excitation of v8 it would require the frequency of V8 in CD3CN+ to be 295 cm_l, which is much smaller than the predicted value. The autoionization in this region is from Rydberg states converging to the first excited state of the ion and is not isotope dependent. Because the second step in the CD CN+ PIE is at the same wavelength as the third 3 . + step in the CH3CN PIE, and because assigning it as the vibronic threshold for the single excitation of v8 would re- quire an unreasonably low frequency for V8 in CD3CN+, the second step in the CD3CN+ PIE and the third in the CH3CN+ PIE are most reasonably attributed to autoionization. The vibronic threshold for the single excitation of v8 in CD3CN+ is most likely hidden under the autoionization as a consequence of CD3CN's higher ionization potential. The transition in CH3CN+ to the first excited level of v8, .the CCN bend, has important consequences for the structure of the ground electronic state of the ion. In general, photo— ionization transitions are adequately described by the Frank- 21,22 Condon principle. The Franck-Condon principle assumes that the electronic and nuclear motions of a molecule are not 110 coupled, and therefore a transition probability can be resolved into electronic, vibrational and rotational factors. The intensity of a vibrational band in an electronically allowed transition is then proportional to the vibrational factor, commonly called the Franck-Condon factor, 2, where I wv and wv are the vibrational wavefunctions of the lower and 24 For the Franck-Condon factor to upper states respectively. be non-zero the product (0; 0;) must be symmetric with respect to all symmetry operations common to the final and initial electronic states; that is, F(w;) x F(w;) = I (totally symmetric). point group;88 its photoelectrcn CH CN belongs to the C 55,56,57 3 spectrum 3v as well as the Walsh diagram of CHBXZ type predict that the ground electronic state of CH CN+ 3 is also C3v' The CCN bend is an "e" vibration88 and therefore molecules89 the first excited vibrational level (v=l) belongs to the E representation. The ground vibrational level (v=0), as always, belongs to the totally symmetric representation Al. The product E x A = E and the vibronic transition from the ground 1 vibrational level of the neutral to the first excited vibra- tional level of V8 in the ion is not allowed within the assump- tions of the Franck-Condon principle. It appears that the Franck-Condon principle is not a good approximation for transi- tions to the ground state of CH3CN+ However, this is not completely surprising. The ground electronic state of CH3CN+, which results from the removal of 111 a n electron from the CEN bond, is orbitally doubly degenerate (a 2E state) and is therefore subject to Jahn-Teller distortion. The Jahn-Teller theorem states that if a non-linear molecule (not Cmv) has an orbitally degenerate electronic state when the nuclei are in a symmetrical configuration, then the molecule is unstable with respect to at least one asymmetric displace- ment of the nuclei that may lift the orbital degeneracy.2 In other words, there can be a coupling of the electronic motion with at least one of the non-totally symmetric vibra- tions such that the vibronic levels which were degenerate by 24 ' way of the orbital degeneracy are split. The "E" vibrations of an E electronic state of the C 21,22,24 3v molecule are Jahn—Teller active. That is, they may interact with an E electronic state to split its degeneracy. If vibronic coupling of V8 with the 2E state of CH3CN+ is invoked, the transition in question is allowed. To demon- strate this the vibronic wavefunction wev and the dipole 24 The probability of a l 2 transition moment MeV must be used. '2 fl 0 o 0 I I ' M = A > Vlbronlc tranSltlon is given by I ev I