Investigation Into the Radical and Ionic Species That Dictate Chemistry in Our Atmosphere and Space

Investigation into the Radical and Ionic Species that Dictate Chemistry in our Atmosphere and Space

Callan Michael Wilcox

A thesis in fulﬁlment of the requirements for the degree of Doctor of Philosophy

School of Chemistry Faculty of Science March 2018 i

PLEASE TYPE THE UNIVERSITY OF NEW SOUTH WALES Thesis/Dissertation Sheet

Surname or Family name: Wilcox

First name: Callan Other name/s: Michael

Abbreviation for degree as given in the University calendar:

School: Chemistry Faculty: Science

Title: Investigation into the Radical and Ionic Species that Dictate Chemistry in our Atmosphere and Space

Abstract 350 words maximum: (PLEASE TYPE)

This thesis embodies two focuses, pertaining to the observation and identification of atmospherically relevant radicals involved in mechanisms with global ramifications on tropospheric air quality, and an investigation into potential radical and ionic carriers hypothesised to exist in the interstellar medium.

Isomer specific observations were made of two rotamers the ortho-hydroxycylohexadienyl radical, a radical involved in the atmospheric degradation of benzene. These spectra were untangled by hole burning spectroscopy and assigned by aid of TD-DFT theory.

A detailed study of the effects of methyl substitution on the allyl radical ensued, and the spectra of various electronic transitions of the 1,1-dimethylallyl and 1,2-dimethylallyl were reported, in addition to their ionisation potentials. These radicals were confirmed as products formed from the hydrogen addition to isoprene reaction by comparison with their spectra. A series of efforts were made towards elucidating spectroscopic information on the hydroxy isoprenyl class of radicals, generally assumed to be a major intermediate in the formation of secondary organic aerosol from isoprene within the atmosphere. Mass spectra are reported for these radicals, following the discharge of appropriate precursors. Suggestions for the direction of future studies are discussed.

Spectra are reported for the resonantly stabilised 9-hydroanthracenyl and 9-deuteroanthracenyl radicals and their cations, as part of our efforts toward resolving the polycyclic aromatic hydrocarbon hypothesis to account for interstellar absorption features such as the diffuse interstellar bands. Spectra of the cations 9-hydroanthracyllium, and 9-deuteroanthracyllium, were recorded by triple resonance depletion spectroscopy. These spectra were assigned by the aid of TD-DFT calculations, and the modelling of 1-D effective potentials. While not matching any diffuse interstellar features, their spectroscopy is interesting from a fundamental perspective.

Declaration relating to disposition of project thesis/dissertation

I hereby grant to the University of New South Wales or its agents the right to archive and to make available my thesis or dissertation in whole or in part in the University libraries in all forms of media, now or here after known, subject to the provisions of the Copyright Act 1968. I retain all property rights, such as patent rights. I also retain the right to use in future works (such as articles or books) all or part of this thesis or dissertation.

I also authorise University Microfilms to use the 350 word abstract of my thesis in Dissertation Abstracts International (this is applicable to doctoral theses only).

…………………………………………………………… ……………………………………..……………… ……….……………………...…….… Signature Witness Signature Date

The University recognises that there may be exceptional circumstances requiring restrictions on copying or conditions on use. Requests for restriction for a period of up to 2 years must be made in writing. Requests for a longer period of restriction may be considered in exceptional circumstances and require the approval of the Dean of Graduate Research.

FOR OFFICE USE ONLY Date of completion of requirements for Award: Copyright and Authenticity Statements

‘I hereby grant the University of New South Wales or its agents the right to archive and to make available my thesis or dissertation in whole or part in the University libraries in all forms of media, now or here after known, subject to the provisions of the Copyright Act 1968. I retain all proprietary rights, such as patent rights. I also retain the right to use in future works (such as articles or books) all or part of this thesis or dissertation. I also authorise University Microfilms to use the 350 word abstract of my thesis in Dissertation Abstract International (this is applicable to doctoral theses only). I have either used no substantial portions of copyright material in my thesis or I have obtained permission to use copyright material; where permission has not been granted I have applied/will apply for a partial restriction of the digital copy of my thesis or dissertation.'

Signed ……………………………………………......

Date ……………………………………………......

AUTHENTICITY STATEMENT

‘I certify that the Library deposit digital copy is a direct equivalent of the final officially approved version of my thesis. No emendation of content has occurred and if there are any minor variations in formatting, they are the result of the conversion to digital format.’

Signed ……………………………………………......

Date ……………………………………………......

ii Declaration of Originality

ORIGINALITY STATEMENT

‘I hereby declare that this submission is my own work and to the best of my knowledge it contains no materials previously published or written by another person, or substantial proportions of material which have been accepted for the award of any other degree or diploma at UNSW or any other educational institution, except where due acknowledgement is made in the thesis. Any contribution made to the research by others, with whom I have worked at UNSW or elsewhere, is explicitly acknowledged in the thesis. I also declare that the intellectual content of this thesis is the product of my own work, except to the extent that assistance from others in the project's design and conception or in style, presentation and linguistic expression is acknowledged.’

Signed ……………………………………………......

Date ……………………………………………......

iii Abstract

This thesis is split into two parts, the ﬁrst part of which contains an investigation into radical species that have important roles in both urban and pristine atmospheric mechanisms, representing the primary focus of this thesis. The second part of this thesis is concerned with the identiﬁcation and spectroscopy of radicals and cations that are candidates for absorption features observed in the interstellar medium.

Spectroscopic methods are utilised in order to identify and characterise the electronic and vibrational transitions of these molecules. Most extensively used throughout this work is the resonance enhanced multiphoton ionisation (REMPI) technique. Various REMPI schemes are employed to measure the mass, electronic transitions, excited state lifetimes and ionisation energies of various radicals and cations. To aid in assignment of these spectra, we rely on a wide variety of theoretical calculations, including ab initio electronic structure theory, density functional theory and a variety of eﬀective potentials.

Within urban environments, formation of aerosol and consequently photochemical smog owes chieﬂy to the oxidation of aromatics, in particular benzene, initiated by the addition of a hydroxyl radical, forming hydroxycylohexadienyl radicals. Two rotamers of the ortho-hydroxycyclohexadienyl isomer are identiﬁed following a discharge of phenol and water vapor. The complex D1 ← D0 spectrum is separated into two spectra, owing to each rotamer, by hole burning spectroscopy.

−1 The electronic origins of the anti and syn rotamers D1 states lie at 19182 cm and 19200 cm−1 respectively. The rich vibrational structure of each isomer is assigned by comparison with time-dependent density functional theory computations. Duschin-

iv sky matrices are created to map the change in character of modes upon excitation, and are used assign the symmetry of each rotamer in the D1 state.

The dominant biogenic volatile organic compound released into the atmosphere is isoprene. The mechanism for its atmospheric loss is also initiated by hydroxyl addition to one of the two double bonds, forming a resonantly stabilised hydroxy isoprenyl radicals with an allylic resonance. As such, a study was undertaken to understand the eﬀects of methyl substitution on allyl, with a goal to understanding the spectroscopy of the hydroxy isoprenyl radicals. Following calibration studies on the Rydberg transitions of allyl and 2-methyl allyl, we report various electronic transitions of the 1,1-dimethylallyl and 1,2-dimethylallyl radicals. The D1 and D2 origins of the 1,1-dimethylallyl radical are observed at 24058 cm−1 and 38329 cm−1 respectively. The vibrational structure of the D1 state is dominated by methyl rotor transitions, coupling to the electronic and vibronic modes of the radical, which are assigned with aid from computational algorithms. Transitions to the D2 and −1 D3 states of the 1,2-dimethylallyl radical are also observed, at 38501.2 cm and ∼39600 cm−1 respectively. Finally, hydrogen atom addition to isoprene within a discharge environment was studied. The spectra in the visible and UV were identi- ﬁed to contain signatures of both the 1,1- and 1,2-dimethylallyl radicals, in addition to other peaks, which were tentatively assigned as owing to inner addition products.

We report considerable eﬀorts towards elucidating spectroscopic information on the hydroxy isoprenyl radicals. A discharge dependent enhancement of the hydroxy isoprenyl mass signal was reported, however despite the employment of various techniques, no spectrum was obtained. The experimental approach is discussed, and recommendations are presented for future work.

The focus of the second part of this thesis changes to the investigation of radical and cationic polycyclic aromatic hydrocarbons. These species are postulated to account for interstellar absorption features such as the unidentified infrared bands, and the diffuse interstellar bands, a series of more than 500 bands of which only one carrier has been identified.

v The electronic transitions are reported for the hydrogen and deuterium addition to anthracene, forming the 9-hydroanthracenyl and 9-deuteroanthracenyl radicals. The D1 ← D0 transitions of these radicals are observed, with origins of 19115 cm−1 and 19112 cm−1 respectively. An inversion coordinate describes the change in geometry incurred upon electronic excitation, and progressions of this mode are observed to build upon the majority of other vibrational modes in these spectra. Additional features are observed in the 9-deuteroanthracenyl spectrum, which arise due to the eﬀects of variances in the zero-point energy of the molecule, arising from the relative location of the deuterium atom, removing degeneracies experienced by the 9-hydroanthracenyl radical. The ionisation potentials of these radicals were determined to be 6.3392 ±0.0010 eV and 6.3405 ±0.0005 eV respectively.

The 9-hydroanthracyllium, and 9-deuteroanthracyllium cations were prepared by the threshold ionisation of the respective radicals, ensuring no vibrational quanta were populated. The electronic spectroscopy of these radicals was probed via triple resonance depletion techniques. The rich vibronic spectra of the D1 ← D0 transitions of these cations was assigned based on comparison with TD-DFT calculations. This same rich vibronic spectra precluded these cations from being DIB carriers, which are reserved for origin dominated species. Nevertheless, the isomer speciﬁc techniques used to generate and investigate these cations can be extended to other DIB candidates in future studies.

vi Acknowledgements

First and foremost, I would like to thank my supervisor, Professor Scott Kable. The passion with which you communicate science is infectious. I am very grateful to have completed my studies within the research group that you have culminated over the years. The time and patience that you have aﬀorded me throughout this tenure and writing process will always be remembered. I cannot thank you enough for the freedom you allowed me to have during my research. To my co-supervisor Professor Timothy Schmidt, I want to thank you for turning me to the world of physical chemistry all those years ago. You have always made the time when I needed it, and I have enjoyed our spectroscopic discussions on rotors and blobology. Your handle on quantum chemistry never ceases to amaze.

To Dr. Olha Krechkivska, thank you for not giving up on me all those years ago! I will always remember the ﬁrst year of my tenure as an exciting time full of new spectroscopy, and admit that I have often missed those times when we were told to ‘stop gathering data so fast’. Thank you for all of your assistance and patience throughout the years. To Dr. Klaas Nauta, the man everyone turns to when something isn’t working. Your hands-on approach to solving problems, and your attitude toward spectroscopy of never being certain until you are certain beyond any doubt are qualities that I have learned a lot from. I am seriously indebted to you for all the time and eﬀort you spent reviewing my thesis. I would not have been able to complete this thesis without the guidance from both of you.

A huge thank you to all of the members of the Mo-pho group who have made my tenure so enjoyable. To Kelvin and his infectious laugh, you made my time in lab a much more delightful experience than it otherwise would have been, ‘with great power, comes great REMPI!’ To Vineeth and Alireza, your banter got me through the days, and I am thankful to have made such great friends, you were

vii both great drinking partners and I enjoyed our conference trips. To Aaron, I’ll never forget that Taiwan trip, ﬁlled with late night drinks and adventures, you’ve been a great person to get to know. To Blair, I’m glad to be leaving the lab in such capable and humorous hands. I’d also like to thank past and current members of the group, Joe, Alain, Gerry, Mitch, Kieran, and Yu for your friendship, assistance and for making work such a great place.

I’d like to thank Professor Jason Harper, and Nick Konstandaras for allowing me to use their synthetic lab, and their assistance in the the synthesis of my precursor compounds. I would also like to thank Dr. Joe Gallagher again for his assistance with my organic chemistry. To Professor Yuan-Pern Lee, thank you for providing me with the opportunity to visit and work in your lab in Taiwan, it was an incredible experience. Thank you also to Dr. Karolina Haupa, for all your assistance in teaching me matrix spectroscopy into the twilight hours of each day, and for showing me around Taiwan.

To Mum and Dad, thank you for your unwavering support across my years at university. Your love and kindness has provided the foundation for my pursuit of happiness in my career and to complete such a monumental task as this, and I appreciate everything you have done for me across the years. To Eve, thank you for being such a tremendously awesome sister, it was your love for the people of the world that inspired me to appreciate it more wholly.

A great word of thanks to Bruce and James, my lifelong friends who have always a steady rock for me to rely on in times of stress. Thank you for providing escape throughout this tenure, and for sticking by me all these years despite my late lifestyle as a hermit.

To my Angie and Max. I never could have completed this work without your continual support and love. You have both been incredible throughout this whole process, lifting me up when I was overwhelmed, listening to my constant rants about science, and bringing so much love and happiness to my life. I can’t wait to see where the future takes our little family, but I feel assured that with the two of you by my side, we can do anything.

viii The miles mean more if you have travelled them one step at a time and felt the ground change beneath your feet.

MARK LAWRENCE Contents

I Free Radicals in Our Atmosphere xxi

1 Introduction 1 1.1 Chemistry of the Troposphere ...... 1 1.2 Formation of Secondary Organic Aerosol (SOA) ...... 4 1.2.1 Alkanes ...... 6 1.2.2 Alkenes ...... 7 1.2.3 Alkynes ...... 10 1.3 Oxidation of Biogenic Volatile Organic Compounds (VOCs) . . . . . 11 1.3.1 Isoprene Derived SOA ...... 11 1.3.2 Diurnal Oxidation of Isoprene ...... 12 1.4 Oxidation of Anthropogenic VOCs ...... 16 1.5 Thesis Focus and Outline ...... 17

2 Experimental Theory, Apparatus and Techniques 19 2.1 Experimental Theory ...... 20 2.1.1 The Vacuum Setup and Molecular Beam Methods ...... 20 2.1.2 Generation of Radicals ...... 22 2.1.3 Resonance-Enhanced Mutiphoton Ionization (REMPI) Time- of-Flight Mass Spectroscopy (TOF-MS) ...... 24 2.1.4 Matrix Isolation Spectroscopy (MIS) ...... 28 2.2 Apparatus and Applications ...... 31 2.2.1 REMPI coupled ToF-MS Experimental Apparatus ...... 31 2.2.2 Matrix Spectroscopy ...... 34

3 The Hydroxycyclohexadienyl (CHD-OH) Radical 38 3.1 Introduction ...... 38 3.2 Experimental ...... 41 3.3 Theory ...... 42 3.3.1 Duschinsky Mixing ...... 46

x 3.4 Results and Discussion ...... 51 3.4.1 The Syn Conformer ...... 54 3.4.2 The Anti Conformer ...... 58 3.5 Conclusion ...... 63

4 The Methylated Allyl Chromophore 64 4.1 Introduction ...... 64 4.2 Calibration Studies and Method of Approach ...... 70 4.2.1 Radical Precursor Selection ...... 70 4.2.2 The Allyl Radical ...... 70 4.2.3 The Methyl Allyl Radical ...... 73 4.2.4 Implications ...... 75 4.3 Investigation of The 1,1 - Dimethyl Allyl Radical ...... 78 4.3.1 Experimental and Methods of Generation ...... 79 4.3.2 Theory ...... 82 4.3.3 Results and Discussion ...... 85 4.3.4 Concluding Remarks ...... 99 4.4 Investigation of The 1,2 - Dimethyl Allyl Radical ...... 101 4.4.1 Experimental ...... 102 4.4.2 Theory ...... 104 4.4.3 Results and Discussion ...... 107 4.4.4 Conclusion ...... 111 4.5 Hydrogen Addition to Isoprene ...... 112 4.5.1 Experimental ...... 113 4.5.2 Results and Discussion ...... 114 4.5.3 Conclusion ...... 118

5 Eﬀorts towards the Hydroxylation of Isoprene 120 5.1 Introduction ...... 120 5.2 Calibration Studies on Indene ...... 122 5.2.1 Hydrogen and Deuterium Addition to Indene ...... 123 5.2.2 Hydroxyl and OD Addition to Indene ...... 125 5.2.3 Chemical Implications ...... 127 5.3 Observation of Hydroxy Isoprenyl Radicals ...... 128 5.3.1 Precursor Selection ...... 128 5.3.2 Observation of 4-OH ...... 130 5.3.3 Observation of 1-OH and 2-OH ...... 134

xi 5.3.4 Conclusion and Future Work ...... 136 5.4 Matrix Studies ...... 137 5.4.1 Experimental ...... 138 5.4.2 Theory ...... 140 5.4.3 Results and Discussion ...... 144 5.4.4 Concluding Remarks ...... 148

6 Epilogue 149 6.1 Conclusions of Part I ...... 149

II Cations in Space 154

7 Introduction 155 7.1 Composition of the Interstellar Medium ...... 155 7.2 The Diﬀuse Interstellar Bands ...... 156 7.2.1 Polycyclic Aromatic Hydorcarbons ...... 158 7.3 Protonated Species in the Interstellar Medium ...... 159 7.4 Focus and Outlines of Part II ...... 160

8 The Hydrogenation and Deuteration of Anthracene 161 8.1 Introduction ...... 161 8.2 Experimental ...... 163 8.3 Results and Theory ...... 165 8.4 Discussion ...... 174 8.5 Concluding Remarks ...... 193 8.6 Appendix to Chapter 8 ...... 194

9 The Protonation and Deuteronation of Anthracene 197 9.1 Introduction ...... 197 9.2 Experimental ...... 200 9.3 Results and Discussion ...... 201 9.4 Concluding Remarks ...... 214 9.5 Appendix to Chapter 9 ...... 216

10 Epilogue 218 10.1 Conclusions of Part II ...... 218

xii List of Figures

1.1 Generic reaction scheme for alkanes within the troposphere...... 6 1.2 Generic reaction scheme for (conjugated) alkenes within the troposphere...... 7 1.3 Generic reaction scheme for alkynes within the troposphere...... 10 1.4 The oxidation pathways of isoprene, initiated by the hyrdoxyl radical, towards the formation of secondary organic aerosol. Figure adapted from references [69] and [26] ...... 13

2.1 Illustration of the pulsed discharge nozzle used in these experiments. . 23 2.2 Jablonski diagram of the various REMPI schemes employed in this work...... 26 2.3 The ‘cage eﬀect’ of a hard matrix preventing spacial separation, and enforcing recombination of dissociated products...... 30 2.4 The REMPI ToF-MS chamber used in these experiments. Repro- duced with permission from Troy.[123] ...... 32 2.5 Illustration of the matrix setup utilised in these experiments...... 35

3.1 Possible ring-breaking mechanism from the ipso-hydroxycyclohexadienyl radical. A scheme of ketene formation from an initial OH-addition to Benzene yielding 1i. [132] The dotted line emphasises that the hydrogen migration pathway from 1i to 1o is energetically unfavourable.[133] 40 3.2 Proﬁle and top down views of ground and excited state anti and syn rotamers, showcasing the geometry shift and the near-planar nature of the excited syn geometry. These geometries were optimised at the (TD) – B3LYP/6-311+G(d,p) level of theory...... 42 3.3 Duschinsky matrices for the anti isomer (left) and the syn isomer (right). Both the ground and excited state modes are labelled according to the Mulliken convention assuming a C1 point group. Matrix elements are shown in a linear greyscale, where white represents 0 and black represents 1...... 47 3.4 Duschinsky matrix for the syn isomer. The ground state has been labelled according to traditional Mulliken convention. The excited state axis exhibits both the original Mulliken convention (as though the excited state were C1), and the ﬁnal state assigned by Cs symmetry after diagonalization. Matrix elements are shown in a linear greyscale, where white represents 0 and black represents 1...... 49

xiii 3.5 Duschinsky matrix for the anti isomer. The ground state has been labelled according to traditional Mulliken convention. The excited state axis exhibits both the original Mulliken convention labelling and the new numbering after the matrix was diagonalised. Matrix elements are shown in a linear greyscale, where white represents 0 and black represents 1...... 51 3.6 (Upper) REMPI spectrum of ortho-hydroxy-CHD radical. (Lower) Hole burning (depletion) spectrum measured while monitoring the REMPI signal 18219 cm˘1. The spectrum is assigned to the syn conformer. Vibrational assignments are indicated in blue font for a0 vibrations and black font for a00 vibrations...... 52 3.7 (Upper) REMPI spectrum of ortho-hydroxy-CHD radical. (Lower) Hole burning (depletion) spectrum measured while monitoring the REMPI signal 18200 cm˘1. The spectrum is assigned to the anti conformer. Vibrational assignments are indicated in red font for fundamental vibrations and black font for overtones and combination bands. The ‘?’ symbol denotes a depleted band that has no corresponding absorption line in REMPI experiments; any coincidence is merely an artefact of the compacted scale...... 59

4.1 Structures of the substituted allyl radicals investigated in our ‘ground up’ approach towards the hydroxy isoprenyl radicals ...... 65 4.2 The central C-C bond break of 1,5-hexadiene, induced via discharge, to form two allyl radicals...... 71 2 4.3 The C( B2), 3py Rydberg transition of the allyl radical, recorded by 1+1 REMPI. Inset adapted from Gasser et al. [158] ...... 72 4.4 The formation of the 2-methyl allyl radical, by the discharge of β- methyl allyl chloride, cleaving the C-Cl bond...... 73 2 4.5 The 2-methyl allyl B(1 A1) state, (3s Rydberg transition), observed by 1+1 REMPI. Inset adapted from Gasser et al. [170] ...... 75 4.6 The formation of the 1,1 dimethyl allyl radical as a discharge product of 3,3-dimehthyl allyl bromide and geraniol. The co-fragment of the geraniol dissociation is a hydroxy isoprenyl radical, 4-OH ...... 79 4.7 Structural details of the D0 and D1 states of the 1,1-dimethylallyl radical ...... 82 4.8 REMPI spectrum of the D1 state of the 1,1-dimethylallyl radical . . . 85 4.9 The scaled displacement vectors for the lowest frequency modes of the 1,1-dimethylallyl radical ...... 87 4.10 Lobe representations of vibrational wavefunctions for the methyl rotor 89 4.11 Right: Calculated ground and excited state 1-D potential energy surfaces for the methyl rotor 1 with a lower barrier. Right: Spec- tral predictions and assignments for the methyl rotor coupling to the electronic excitation...... 91 4.12 Right: Calculated ground and excited state 1-D potential energy surfaces for the methyl rotor 2 with a higher barrier. Right: Spec- tral predictions and assignments for the methyl rotor coupling to the electronic excitation...... 92

xiv 4.13 Assigned REMPI spectrum of the D1 state of the 1,1-dimethylallyl radical ...... 93 4.14 Lifetime scan of the origin peak of the 1,1-dimethyallyl radical, fit to the convolution of a gaussian and an exponential decay function . . . 95 4.15 Ionisation potential step scan of the 1,1-dimethylallyl radical. Left: Signal of the m/z 69 trace with (black) and without (red) the resonance laser. Right: Difference between the black and red traces as a function of the combined excitation and ionisation energies. . . . . 96 4.16 REMPI spectrum of the D2 state of the 1,1-dimethylallyl radical . . . 98 4.17 The formation of the 1,2-dimethylallyl radical, following the dissociation of the C-Br bond in the (2E)-1-bromo-2-methylbut-2-ene precursor102 4.18 Geometric details of the D0 and D1 states of the 1,2-dimethylallyl radical ...... 105 4.19 Higher excited states of the 1,2-dimethylallyl radical in the UV region 108 4.20 Photoionisation efficiency curves for the 1,2-dimethylally radical . . . 110 4.21 Plausible addition sites for H atom to isoprene and the respective radicals produced ...... 112 4.22 Upper Trace: REMPI spectrum of the isoprene + H addition products in the region likely to contain the D1 transitions of the radicals. Lower Trace: The D1 electronic spectrum of the 1,1-dimethyallyl radical...... 114 4.23 Middle Trace: The UV spectrum of the m/z 69 trace corresponding to the isoprene + H addition products Upper Trace: Reference spectrum of the D2 and D3 states of the 1,2-dimethylallyl radical. Lower Trace: Reference spectrum of the D2 state of the 1,1-dimethylallyl radical. The identities of the peaks marked with red asterisks are proposed in the text at the end of this section...... 116

5.1 Isoprene addition sites for the hydroxyl radical, and the corresponding OH-adducts. The two product distributions are described in the text. 121 5.2 The excitation spectrum of the m/z 117 and 118 traces, corresponding to H and D addition to indene. Spectrum of the 1-indanyl radical reproduced from Troy et al. [189] ...... 124 5.3 The excitation spectrum of the m/z 133 and 134 traces, corresponding to OH and OD addition to indene. Spectrum of the 2-hydroxy-indan- 1-yl radical reproduced from Troy et al. [134] ...... 126 5.4 The synthetic and atmospheric routes to the formation of the hydroxy isoprenyl radicals ...... 129 5.5 Discharge dependent enhancement of the m/z 85 carrier, 4-OH ... 131 5.6 Left: Mass traces from varying laser sources Right: Comparison between the 2 laser signal mass trace, and the sum of the two individual mass traces ...... 133 5.7 Discharge dependent enhancement of the m/z 85 carrier, containing 1-OH and 2-OH ...... 135 5.8 Photolysis of (E)-4-bromo-3-methyl-2-buten-1-ol, yielding 4-OH and a bromine atom...... 139 5.9 Geometric conﬁgurations and relative energies of the D1 states of the 4p rotomers...... 140

xv 5.10 Geometric details for the D0 state of the 4-OH radical ...... 141 5.11 Scaled D0 vibrational frequencies for the precursor (iii) and the radical (4-OH)...... 143 5.12 Upper Trace: Deposition (8 hours). Middle Trace: Photolysis (3 minutes, 248 nm). Lower Trace: Difference (Photolysis - Deposi- tion). Blue line indicates peaks which have depleted in intensity after irradiation. Green line indicates peaks which have formed following irradiation...... 145 5.13 Results of the secondary photolysis at 405 nm and 365 nm. The purple trace has been adapted from the difference trace in figure 5.12. The green and blue lines indicate peaks which have increased and decreased in intensity following 405 nm and 365 nm irradiation, respectively...... 147

8.1 Top: REMPI spectrum of the m/z 179 trace, the H-An radical. Bot- tom: REMPI spectrum of the m/z 180 trace, the D-An radical. . . . 165 8.2 Numbering convention for the Anthracene substituents, unique addition locations shown in bold...... 166 8.3 Photo Ionisation Efficiency curves for the for the first 2 peaks in the spectrum of the H addition species, labelled as Ha and Hb, and the first 5 peaks in the spectrum of the D addition species, labelled as Da -De. The energy represents the combined sum of the excitation and ionisation laser energies. Bracketed values indicate the number of scans averaged for each case. The central line represents the determined ionisation energy, and the grey band represents the error for this value. Additional onsets are seen for Ha and Dd, (see text). . . . 169 8.4 Hole-burning (depletion) spectrum measured whilst monitoring the REMPI signal on the D addition origin (m/z 180) at 19112 cm−1... 173 8.5 Left: Geometric details for the ground state of the H addition isomer, indicating bond lengths, and specific angles. The profile view emphasises the out-of-plane pucker of the sp2 hybridised carbon. Right: Geometric details for the excited state of the H addition isomer, indicating bond lengths, and specific angles. The profile view emphasises the increased out-of-plane pucker of the sp2 hybridised carbon. . . . . 175 8.6 The inversion ‘butterfly’ coordinate of the anthracene ring. Left: The 9H-An isotopologue. Right: Depiction of the equatorial and axial configurations of the 9D-An isotopologue ...... 177 8.7 Excited state double well potentials fit to the 9H-An and 9D-An REMPI spectra. The first 200 cm−1 of the corresponding REMPI spectrum are shown, with the simulated line positions drawn in red. . 179 8.8 Top: PES for the 1-D ν37 inversion coordinate of 9H-An, including the ZPEs. Bottom: PES for the 1-D ν37 inversion coordinate of 9D- An, including the ZPEs. All energies were calculated relative to the lowest energy configuration in the ground state...... 183 8.9 REMPI spectrum of the 9-hydroanthracenyl radical. Assignments of the first instance of a0 and a00 modes are made with progressions of 2 and 4 quanta of ν37 following the legend inset. Unidentified bands are marked with a star...... 187

xvi 8.10 Scaled displacement vectors representing the active modes present in the 9-hydroanthracenyl spectrum...... 188 8.11 REMPI spectrum of the 9-deuteroanthracenyl radical. Assignments of the first instance of a’ and a" modes are made with coupled progressions of 2 and 4 quanta of ν37 following the legend inset. Bands attributed to excitation of HD9Deq-An are marked in red, whilst those coming from the 9Dax-An side of the double well are marked in blue. 191 8.12 Lifetime scan of the D1 origin of 9-hydroanthracenyl, fit to a first order exponential decay function...... 192

9.1 Top: Triple resonance depletion spectrum of the m/z 179 trace, representing the 9H-An+ cation. Bottom Left: Triple resonance depletion spectrum of the m/z 180 trace, representing the 9D-An+ cation. 202 9.2 Ground and excited state structural parameters for the 9H-An+ cation. The symmetry reduces from C2v to Cs upon excitation, as only the plane of symmetry in the plane of the anthracene rings remain. . . . . 203 9.3 Assignment of the excitation spectrum of the 9H-An+ cation. The top spectrum was adapted from reference [141], as are the mode labels to be consistent with our labelling scheme...... 205 9.4 Assignment of the excitation spectrum of the 9D-An+ cation...... 210

xvii List of Tables

3.1 Ground and excited state frequency calculations for the syn conformer of the hydroxy-CHD radical, at the (TD)-B3LYP/6-311+G(d,p) level of theory. Frequencies are labelled according to the Cs point group of the ground state, where the excited state has been reordered according to the Duschinsky matrix, see text. Scaled frequencies (x0.9688) are included for comparison...... 44 3.2 Ground and excited state frequency calculations for the anti conformer of the hydroxy-CHD radical, at the (TD)-B3LYP/6-311+G(d,p) level of theory. Frequencies are labelled according to the Cs point group of the ground state, where the excited state has been reordered according to the Duschinsky matrix, see text. Scaled frequencies (x0.9688) are included for comparison...... 45 3.3 Syn-hydroxy-CHD assignments and comparison with theory, including the mean absolute deviation (MAD) for each symmetry a0, a00 and the combined. MAD is calculated per total number of modes, and the combined deviation of ν23ν33ν34 is split three ways. All units are in wavenumbers (cm˘1). The theoretical values for the a00 modes are not strictly from the TD-B3LYP output, see text...... 55 3.4 Comparison between single quanta observed/determined frequencies and TD-B3LYP theory, including the mean absolute deviation (MAD) for both the a0 and a00 modes. All units are in wavenumbers (cm−1). 57 3.5 Anti assignments and comparison to theory, including the mean absolute deviation (MAD) for the fundamental assignments. All units are in wavenumbers (cm˘1)...... 60

4.1 Adiabatic and Vertical Ionisation Energies for the 1- and 2- methyl allyl radicals, adapted from reference [168]...... 67 4.2 List of frequencies for the 1,1-dimethylallyl radical, calculated at the M06-2X/6-311+G(d,p) level of theory. Modes ν34 and ν35, indicated with an asterisk (*), are the two methyl torsion modes and are treated separately below...... 84 4.3 Predicted and observed ionisation potentials of the 1,1-dimethylallyl radical...... 96 4.4 Predicted transition energies for the D1−3 electronic states of the 1,2- dimethylallyl radical, along with their associated oscillator strengths . 106 4.5 Theoretical ionisation potentials of the 1,2-dimethylallyl radical. . . . 106

xviii 4.6 List of observed peaks and their relative frequencies in the D2 state of the 1,2-dimethylallyl radical. Intensities are labelled as strong (s), medium (m), weak (w) and very weak (vw). All units are in wavenumbers (cm−1)...... 109

5.1 List of frequencies for the D0 state of Precursor (iii) and Radical 4- OH calculated at the b3pw91/aug-cc-pVTZ level of theory. All values are in wavenumbers (cm−1). Harmonic frequencies, scaled by 0.9646, have been included for comparison with experiment. [199] ...... 142 5.2 Peak positions observed to decay (precursor) and form (radical) in the matrix. All units are in wavenumbers (cm−1)...... 146

8.1 Energies of formation of the 1H-An and 2H-An conformational isomers relative to the lowest energy 9H-An isomer. All energies are in kJ mol−1 ...... 167 8.2 Zero point energy corrected adiabatic excitation energies for the 1H- An, 2H-An and 9H-An conformational isomers ...... 167 8.3 Adiabatic ionisation energies calculated for the 1H-An, 2H-An and 9H-An confirmational isomers, ZPE corrected values are also shown. All units are in eV...... 170 8.4 Excited state frequencies (<1200 cm−1) for the 9H-An radical, and the 9Deq-An and 9Dax-An configurations in the Cs symmetry point group. The C2v character representations are also shown for the 9H- An isomer, with a calculated negative ν37 frequency. Scaled frequencies (0.97) and vibrational symmetries are included for comparison with experiment. The Mulliken convention for mode labelling is applied to the 9H-An Cs configuration, and others were rearranged to keep the label consistent with molecular motions. Calculations were carried out at the TD-B3LYP/6-311+G(d,p) level of theory. All units are in wavenumbers (cm−1)...... 181 8.5 9-hydroanthracenyl assignments and comparison to theory, including the mean absolute deviation (MAD). The daggers (†) indicate modes not calculated by DFT, and the asterisks (*) represent tentative assignments, (see text). All units are in wavenumbers (cm−1)...... 186 8.6 9-deuteroanthracenyl assignments and comparison to theory, for both the 9Deq-An and 9Dax-An configurations, including the mean absolute deviation (MAD) for each. All units are in wavenumbers (cm−1). . . . 190 8.7 Vibrational frequencies for the D1 electronic state of the 9H-An radical194 8.8 Vibrational frequencies for the D1 electronic state of the 9Deq-An radical195 8.9 Vibrational frequencies for the D1 electronic state of the 9Dax-An radical ...... 196

xix 9.1 List of frequencies (under 1200cm−1) for the 9H-An+ and 9D-An+ cations, calculated with the TD-DFT B3LYP/6-311+G(d,p) level of theory. Symmetry labels are determined for the 9H-An+ cation following Mulliken convention. Mode labels were rearranged for the 9D-An+ cation to retain the same label for the same molecular motions. Frequencies were scaled by 0.991 and 0.998 for the 9H-An+ and 9D-An+ cations respectively to ﬁt with experiment. All units are in wavenumbers (cm−1)...... 204 9.2 9H-An+ assignments and comparison to theory. Unique modes are represented in blue and are used to calculate the the mean absolute deviation (MAD). All units are in wavenumbers (cm−1)...... 207 9.3 9D-An+ assignments and comparison to theory. Unique modes are represented in blue and are used to calculate the the mean absolute deviation (MAD). All units are in wavenumbers (cm−1)...... 211 + 9.4 Vibrational frequencies for the D1 electronic state of the 9H-An ... 216 + 9.5 Vibrational frequencies for the D1 electronic state of the 9H-An ... 217

xx Part I

Free Radicals in Our Atmosphere Chapter 1

Introduction

1.1 Chemistry of the Troposphere

The Earth’s atmosphere, is a complex and evolving system. As the population and industrialisation of our civilisation continues to rise, so too does our impact on the world around us. Without proper understanding and consideration of the interplay between chemical and physical processes within our atmosphere, we risk further contamination and other adverse eﬀects on the very air we rely on for life. The continued warming of our climate by an associated increase in the carbon load of the atmosphere, and the damage caused to the Earth’s protective ozone layer by the class of chloroﬂuorocarbons serve to illustrate this point.[1, 2, 3] To preserve and ensure a clean atmosphere for future generations, a more complete understanding of the processes governing atmospheric cycles, and of potential adverse human impacts upon these cycles, is needed.

Our understanding of atmospheric processes has drastically increased alongside advancements in experimental and theoretical techniques, computational modelling capabilities and the rich interplay between these ﬁelds. Field measurements of atmospheric species and their (relative) abundances provide the basis for this knowledge. These data are compiled, and as new experimentals are observed, they are implemented in an ever evolving model of our atmosphere, consisting of thousands

1 of chemical species, reaction pathways, kinetics, photochemistry, sources and sinks, lifetimes, mixing and transport. [4]

The majority of photochemical, kinetic and thermochemical data in the models have been collected for closed shell species, which are easier to identify due to their stable nature, allowing their concentrations to build up in the atmosphere. The transformation between these species however, is initiated by, and involves intermediates that are, primarily free radicals. Free radicals are a transient species due to their reactive nature and, as a consequence, are very short lived. This makes their identiﬁcation challenging under atmospheric conditions, where collisions are abundant; indeed the majority of intermediate radicals have never been identiﬁed. As such, these reaction pathways are often retro-developed, beginning with the observation of compounds emitted into the atmosphere via a variety of processes, and by the end product analysis of chamber experiments which are seeded with these known compounds, under a variety of conditions to account for the diverse nature of the atmosphere.

Theoretical calculations and structure-activity relationships are often used where experimental and observational data are incomplete. To improve the accuracy of these models, and to account for any shortfalls (which will be identiﬁed later), greater eﬀort must be placed towards identifying and characterising the energetics, stability and photochemistry of these reactive intermediates. This vital information will bring validity to aspects of models, whilst at the same time uncovering new reaction pathways.

The atmospheric chemistry and cycles of particulate matter (PM), comprised largely of organic aerosol (OA), is one such field lacking in sufficient experimental data. Aerosols within the atmosphere have been linked to radiative forcing (influencing the Earth’s radiation budget), of both solar and terrestrial radiation, by absorption and scattering of light. [5, 6] OA are known to affect visibility and have a direct impact on air quality. They influence cloud formation, affecting the local and regional climate, by providing cloud condensation nuclei (CCN) for cloud droplets. [7, 8] Further, OA have been linked to damaging effects on the human cardiovascular and respiratory systems. [9, 10, 11] Despite the far reaching implica-

2 tions and importance of aerosols, the mechanistic details of their formation process remain relatively unknown.[12, 13, 14] This results in a large variance in the output of atmospheric models.

There are also large uncertainties in the measurements of OAs which arise due to their chemical complexity.[15, 16] Aerosol consists of liquid or solid particles suspended in air, and contributes significantly to the fine particle content of the atmosphere. In the troposphere, organic aerosol comprises a significant (20-90%) component of the total aerosol mass.[17, 18, 19] The chemical composition of OA varies extensively, and hence aerosols have a wide range of physical and chemical properties.

Aerosols are classed into primary and secondary. Primary aerosols are emitted directly into the atmosphere, by volcanic activity, sea spray, wind-borne dust, release of soil/rock debris, emission of biomass burning smoke, and additionally from plant origin, such as fungal spores, pollen, plant debris and microbial particles. [20] Secondary aerosols refer to the class of aerosols formed within the atmosphere, most predominantly from the successive oxidation of volatile organic compound (VOC’s), of both biogenic and anthropogenic origin. [21, 22] SOA contributes between 50-85% of the total OA mass, which depends on the proximity to source, season, environment, etc., and this broad range exempliﬁes the challenge that experimental theoretical chemists face. [23, 24, 25]

As VOC’s condense, via a gas-particle transfer, or by further oxidation of ﬁrst-generation reaction products, there is an appreciable uncertainty over the qualitative and quantitative formation of SOA. [21, 23, 26] Indeed, by a ‘bottom- up’ approach, where experimental data taken from smog chamber experiments is extrapolated to a global scale, biogenic SOA production is estimated to produce between 12 - 70 Tg yr−1. [18] This scheme was improved by Hallquist et al., to include the total SOA budget, yielding some 50 - 90 Tg yr−1.[23] Four independent estimates by Goldstein and Galbally however, place this range between 140 - 910 Tg yr−1.[16] They reason that the underestimation by smog chamber experiments was in part due to their time constraints, truncating SOA production before all products were able to reach their ﬁnal state.

3 Goldstein et al. propose that the 104 to 105 diﬀerent VOC compounds observed within the atmosphere could represent only a small portion of those present, especially considering these numbers do not include isotopes, complexes, free radicals and intermediates.[16]

1.2 Formation of Secondary Organic Aerosol (SOA)

Approximately 1100 Tg of biogenic volatile organic compounds (BVOCs, vapor pressure ≥ 10−5 atm) are released annually into the atmosphere. The composition of this loading is predominantly isoprene, (2-methyl-1,3-butadiene, C5H8), contributing to −1 almost half (∼535 Tg yr ) of the total budget; and its monoterpene (C10H16, 162 −1 Tg yr , ∼15%), and sesquiterpene (C15H24, 3%) derivatives. Apart from carbon monoxide (CO, 81.6 Tg yr−1), other key BVOCs include methanol, ethanol, acetalde- hyde, acetone, α-pinene, β-pinene, t-β-ocimene, limonene, ethene and propene. [27] The major source of BVOCs are tropical trees, and despite only covering ∼18% of the global land mass, produce an astonishing ∼80% of the terpene derivatives, and ∼50% of all other BVOCs. [27]

The emission of anthropogenic VOCs (AVOCs) varies between 98 - 158 Tg yr−1, with an average estimated loading of 127 Tg yr−1. [28] Though this loading is roughly an order of magnitude lower than the BVOCs, AVOCs are notably more important in urban areas.[29] These emissions are comprised of ∼40% alkanes, ∼10% alkenes and ∼20% aromatics. The remaining ∼30% consists of oxygenated and unidentiﬁed compounds. [30, 31]

The night and day time chemistry of VOCs vary in accordance with the type and concentration of available radical initiators. During the day time, VOCs are prone to photolytic degradation, and are chieﬂy initiated by abstraction or addition reactions with the hydroxyl radical (OH). Overnight, nitrate (NO3) radical chemistry dominates. Reaction with ozone (O3) is also an important sink for VOCs, and in marine and costal regions, by reaction with Cl. [31]

4 Biogenic emissions are characteristically diurnal, considering that the emissions maxima of most tropical and temperate trees peaks at ∼40◦C, and often show a linear increase in emission to the intensity of light.[32, 33] Emissions of VOCs in urban areas correspond to human activity, and also peak in the day, dominated by morning and afternoon traﬃc rush hours. [34, 35] As such, the work in this thesis focusses on the important diurnal chemistry of VOCs, initiated primarily through reaction with the hydroxyl radical.

In the diurnal cycle, the dominant production route of the hydroxyl radical begins with the photolytic degradation of ozone, equation 1.1, which typically requires λ ≤ 310 nm. The highly energetic O(1D) can deactivate by collisions with ambient N2 or O2, or abstract a hydrogen atom from water vapor, equation 1.2, to generate two hydroxyl radicals (OH). [36, 37]

1 O3 + hν → O2 + O( D) (1.1)

1 O( D) + H2O → 2OH (1.2)

In urban areas, with elevated NO2 concentrations, formation of HONO has been observed overnight. [38] As the sun rises, HONO is photolysed into NO and OH radicals, equation 1.3, which kick starts the oxidation schemes of urban VOCs.

HONO + hν → OH + NO (1.3)

The generally accepted atmospheric fate of VOCs is to either decrease in volatility through successive oxidation resulting in the formation of SOA, decompose into eventual CO2 and H2O products, or to be removed from the atmosphere by wet/dry deposition. In the following sections, the general oxidation schemes for the classes of alkanes, alkenes (including cyclic and aromatic species) and alkynes will be described, and their propensity to form SOA recognised.

5 1.2.1 Alkanes

Alkanes are present in significant concentrations in urban areas, generally released from motor vehicle exhaust, considering they comprise a large portion of gasoline and diesel fuels. In these populated areas, as much as half of the ambient non- methane VOC’s are alkanes. [39, 40] The atmospheric fate of all alkanes within the troposphere is similar. Abstraction of a H atom is initiated by either OH, NO3 or Cl radicals, with the former being dominant. [41, 42] A simplified reaction scheme is shown in figure 1.1. Here R represents a generic alkane hydrocarbon.

HOO ROOH

O2 OH NO H abstraction by O 2 + O2 RH R ROO RO decomposition 1,5 - H isomerisation

H2O NO 2

RONO 2 +NO

Figure 1.1: Generic reaction scheme for alkanes within the troposphere.

Following H atom abstraction, the radical R· adds O2 exclusively, to form an alkyl peroxy radical, ROO·.[43] From here, the chemistry depends upon the relative concentrations of NO, NO2, HO2, RO2 and during the night time, NO3.

The dominant pathways are shown in ﬁgure 1.1; for small alkyl peroxy radicals, with 1 or 2 carbon atoms, an oxygen atom is removed by NO, forming NO2 and an alkoxy radical RO·.[41] For small alkoxy radicals, abstraction of a H atom by O2 can occur, yielding a carbonyl species and HO2, or isomerisation can occur by intramolecular 1,5-hydrogen shifts, however their predominant fate is to eventually decompose into products of CO2 and H2O. [43, 44]

For larger alkyl peroxy radicals, the same abstraction of an oxygen atom by NO can occur, yielding the alkoxy radical and NO2, however with increased temperature and pressure, NO is observed to add directly to the radical site, forming an alkyl nitrate (RONO2).[45] Alternatively, a hydrogen atom can be scavenged from

HO2 to form a hydroperoxide (ROOH) and O2.[41]

6 The formation of secondary organic aerosol, originating from cyclic, linear and branched alkanes has been the focus of a number of recent studies. [46, 47, 48, 49, 50] These and other studies imply that some of products containing more functional groups of alkane oxidation can condense to form SOA, especially cyclic alkanes, and those with larger carbon frameworks (C12 -C30), which are still suﬃ- ciently volatile. [29, 51, 52] The quantity of atmospheric SOA produced from alkanes however, remains undeﬁned. [18]

1.2.2 Alkenes

Alkenes are the dominant species of VOCs released into the atmosphere. The initial step in their chemistry diﬀers from alkane chemistry, due to their electron rich double bond. Alkenes react with OH predominantly, however also react with O3,

NO3 and Cl. [37] The hydroxyl initiated reaction can proceed through abstraction of hydrogen (minor pathway, ∼5-10%), however the electrophilic addition of OH across a double bond is the most dominant outcome, see top of ﬁgure 1.2. Here R and R0 represent saturated hydrocarbons.

HOO R(OH)R’OOH

NO H abstraction by O 2 + OH + O2 R R’ R(OH) R’ R(OH) R’OO R(OH) R’O decomposition 1,5 - H isomerisation NO 2

R(OH)R’ONO 2 +NO

Conjugated Alkene: β-

+ OH + O2 R CH CH R’ R(OH) CH CH R’ R(OH) CH(OO) CH R’ 1 2 3 4 similar ROO chemistry; second C = C bond allows for further OH addition + O2 R(OH) CH CH R’ R(OH) CH CH R’(OO) δ-

Figure 1.2: Generic reaction scheme for (conjugated) alkenes within the troposphere.

Molecular oxygen rapidly (and solely) adds to the radical carbon of the hydroxy alkyl radical, forming a hydroxy alkyl peroxy intermediate. The reaction pathway then follows that for alkanes, where NO can either add (minor pathway), to form a hydroxyalkyl nitrate, following rearrangement of the ONO terminal group to

7 NO2, or abstract an oxygen atom (major pathway), forming an alkoxy alkyl radical and NO2. Alternatively, the hydroxy alkyl peroxy radical can either lose a hydrogen atom to reaction with O2, followed by a decomposition into two carbonyl products, or abstract a hydrogen group from HO2.[37, 53, 54]

The alkoxy alkyl radical proceeds as for the alkane pathway, by H abstraction to form a carbonyl, decomposition (which is the most important pathway for alkenes with C ≤ 4, [55]), or intramolecularly isomerise, followed by further reactions with O2/NO. A more in-depth review on the OH initiated chemistry of alkanes and non-conjugated alkenes is presented by Ziemann. [54]

For conjugated alkenes, see bottom of ﬁgure 1.2, the dominant addition site of the hydroxyl radical is at the least substituted carbon 1- and 4- position, yielding a resonance stabilised β-hydroxy alkyl radical, with an allyl chromophore.

Prior to O2 addition, the radical site can shift to the δ- position, which complicates the following chemistry by increasing the number of isomers in each step. The overarching chemistry is still similar however, whereafter O2 adds to the β- or δ- radical site, forming the relative hydroxy alkyl peroxy radicals. Whilst the next steps are similar to the unconjugated alkenes, the presence of an additional double bond allows for further oxidative chemistry to occur, increasing the functionality and subsequently lowering the volatility of the species.[37, 53] As a result, conjugated alkenes have a larger propensity to form more highly oxidated species, increasing their role in the formation of SOA.

Cyclic and Aromatic Alkenes

Cyclic alkenes have both anthropogenic and biogenic origins. Once released into the atmosphere, oxidation of aromatic species is initiated primarily by reaction with the hydroxyl radical, similar to the reactions of linear and branched alkenes.[41] Though hydrogen can be abstracted by OH to form an alkyl type radical R·, hydroxyl addition across a double bond is the most atmospherically relevant reaction. [29,

39] Whilst reaction of cyclic alkenes radical product with NO2 is possible, under tropospheric conditions the O2 addition prevails. [56] The rate of reaction with

8 NO2 increases however, for polycyclic aromatic hydrocarbons (PAHs). [57] The subsequent chemistry of the peroxy radical is similar to that for alkenes, dominated by reactions with NOx and O2, and unimolecular decomposition.

For cyclic alkenes with multiple double bonds, further reaction with O2 is common, yielding products with lower vapor pressure, and increasing the chance of aerosol adsorption. Upon the aromatic limit, the propensity to form SOA increases, as degradation proceeds through ring opening, yielding multiple sites where functional groups are formed. [57, 58] H-atom abstraction by OH reemerges as a possible pathway for aromatics and in particular methyl substituted aromatics, though these pathways account for less than 10%.[59] The reaction pathways of many aromatic- OH adducts have been investigated under tropospheric conditions, however many reaction details and intermediates have not yet been identified, and indeed ∼50% of the reaction products are still not yet quantified. [37] Despite this, aromatic hydrocarbons have been identified as the key precursors to SOA formation in urban areas. [60]

9 1.2.3 Alkynes

Alkyne chemistry is relatively simple in comparison, where the degradation mechanism, see ﬁgure 1.3, proceeds almost solely through hydroxyl addition across the triple bond. [53]

+ OH + O2 R R’ R(OH) R’ R(OH) R’OO OH NO OR R’O R(OH) R’O R(OH) R’O

H2O NO 2

Figure 1.3: Generic reaction scheme for alkynes within the troposphere.

By addition of an O2 molecule, a hydroxy peroxy adduct is formed. Follow- ing alkene chemistry, the peroxy is reduced to a phenoxy functional group following oxygen atom abstraction by NO. A rearrangement of the radical site to the C(OH) carbon follows, whereafter H abstraction by O2 yields a dicarbonyl compound.

The simplest alkyne, acetylene (C2H2), was observed to form glyoxal (CHO- CHO) as a dominant product in both photochemical and dark studies. [61] The ability of glyoxal to form SOA is reported, and the authors suggest other α-dicarbonyls as potential candidates for SOA precursors.

Chemical Implications

As described in the above reaction schemes, the formation of (highly) oxygenated products can occur for each of the alkane, alkene and alkyne classes, initiated by the addition and abstraction reactions of the hydroxy radical. For small molecules, with a backbone of 1-4 carbons, the degradation mechanism is favoured, yielding smaller, volatile organics and carbonyls. [37, 53] Larger molecules however, especially those with (multiple) double bonds, are able to continually oxidise in the atmosphere, forming multiple generations of oxidised products. As species are continually oxidised, their volatility decreases, to the point where they undergo particle- or aqueous-phase accretion reactions, yielding SOA.[26]

10 The relative importance of a species, in the grand scheme of SOA formation, not only relies on its ability to form SOA, but on its relative abundance in the atmosphere. Consider isoprene, with a global emission of some 535 Tg yr−1. Even if the yield of SOA from isoprene oxidation is only ∼1%, then this results in a contribution of 5.35 Tg yr−1 of particulate matter, which is larger than even the emissions of most species.[26]

1.3 Oxidation of Biogenic Volatile Organic Com- pounds (VOCs)

1.3.1 Isoprene Derived SOA

Of the total annual non-methane VOC emission budget, one key species (and its derivatives), isoprene (2-methyl-1,3-butadiene), accounts for an astonishing 68% (50% contribution from isoprene, 15% from monoterpenes and 3% from sesquiter- penes). [27] Isoprene is released into the atmosphere by a broad distribution of ﬂora, especially from trees, ferns and mosses. [62] In chloroplasts, the mevalonic acid pathway is responsible for the production of isoprene.[32, 63, 64] The intermediate, isopentenyl pyrophosphate is isomerised into dimethylallyl pyrophosphate (DMAPP). The degradation of DMAPP generates pyrophosphate and isoprene products. Prenyltransferase of the isopentyl pyrophosphate and DMAPP generates geranyl pyrophosphate, which through monoterpene cyclases, generates monoterpene products such as myrcene and limonene. Isoprene is released immediately from the plants leaves, as there is no mechanism by which to store it. As this process is driven by light, it accounts for the diurnal release of isoprene.

It was not until recently that isoprene was recognised as a precursor to the formation of secondary organic aerosol. [65] Claeys et al. identiﬁed two tetrols, 2-methylthreitol and 2-methylerythritol, in the natural aerosols above the Amazon rainforest. These tetrols, containing the same C5 carbon framework as isoprene, were identiﬁed as products of the hydroxyl initiated oxidation of isoprene, with an

11 estimated annual production of some ∼2 Tg yr−1. This study reinvigorated eﬀorts of the scientiﬁc community towards a more complete understanding of the oxidation pathways of isoprene.

Following this study, the quantity of SOA formed from isoprene was modelled by various approaches.[66, 67, 68] The amount of isoprene-derived SOA has been predicted to be ∼6 - 30 Tg yr−1. [26] This contribution accounts for up to 30% of the global particulate organic matter budget. [66] Wennberg et al. state that despite the signiﬁcant progress in this oxidation mechanism, the current understanding of isoprene chemistry is insuﬃcient to be correctly implemented into atmospheric chemical transport models. [69] As a result, global models simulating air quality and climate feedback inherit this kinetic and mechanistic uncertainty.

1.3.2 Diurnal Oxidation of Isoprene

The atmospheric lifetime of isoprene is only ∼1.5 hours, due primarily to reaction with the key daytime oxidant, OH. [70] This reaction prevents the accumulation of isoprene in the troposphere, restricting its steady state concentration to ∼30 ppb near the crown region of pristine forests in the Amazon. This value drops oﬀ to ∼10 ppb at 10 - 20 m above the forest region. [71] Whilst the short lifetime of isoprene limits the transport from its source, the inﬂuence of this (highly emitted) species on the oxidative capacity of the troposphere is translated into other key atmospheric cycles with far reaching implications.[72]

The hydroxy radical can add to any of the four positions described in ﬁgure 1.4, forming a hydroxy alkyl intermediate. The abstraction channel of a methyl hydrogen, despite yielding a radical with an allylic resonance, has been calculated to be of negligible importance under tropospheric conditions.[73]

12 OH + Isoprene 2 4 (cis/trans) 1 3

hydroxy isoprenyl radicals HO

HO OH

OH +O 1-OH 2-OH 3-OH 4-OH +O 2 (cis/trans) (cis/trans) 2 0.67 (0.63) 0.02 (0.00) 0.02 (0.00) 0.29 (0.37)

+O 2 +O 2 O 2 O2 OH OH OH O2 O2 O2 O O O2 OH 2 OH 2 OH OH

hydroxy peroxy isoprenyl radicals

NO HO 2 13 RO 2

O O ONO2 OOH OH

OH HO O OH OH

hydroxynitrate methyl vinyl ketone methacrolein hydroxy carbonyl diol hydroxy hydroperoxide

first generation products generation first (and isomers) and formaldehyde and formaldehyde (and isomers) (and isomers) (and isomers)

+OH water soluble compounds: glyoxal, glycoaldehyde second generation compounds particle phase accretion reactions +OH methylglyoxal, hydroxyacetone isoprene tetrol oligomeric species (esters, etc.) +OH 2-methylglyceric acid multifunctional acids, nitrates, organic sulfates pyruic acid, glyoxylic acid, polyols, etc. oxalic axid

Figure 1.4: The oxidation pathways of isoprene, initiated by the hyrdoxyl radical, towards the formation of secondary organic aerosol. Figure adapted from references [69] and [26] Addition to the terminal sites is preferred, yielding a resonantly stabilised radical intermediate with an allylic chromophore. This stabilisation is reﬂected in calculations of the branching ratios between isomers 1-OH, 2-OH, 3-OH and 4-OH, of 0.67:0.02:0.02:0.29 respectively. These ratios were calculated in a two- transition state model by Greenwald et al. and are the current IUPAC recommended values. [74] These values have recently been revised on the basis of experimental studies by Teng et al. who infer that the inner addition channels, corresponding to isomers 2 and 3, produce less than 1% of the second generation products and so should be ignored in atmospheric models until further experimental evidence comes forward to the contrary. They oﬀer a revised branching ratio of the OH addition to isoprene of 0.63:0.00:0.00:0.37.[69, 75]

Upon addition, thermal energy from the hydroxyl radical is suﬃcient to overcome the rotational barrier between the cis and trans forms of isoprene. Peeters et al. has calculated the cis:trans ratio of the ﬁrst generation isomers 1-OH and 4-OH to be 0.46:0.54 and 0.69:0.31 respectively. [76]

The hydroxy alkyl species then exclusively adds molecular oxygen at the radical site, which, as described in section 1.2.2, can shift between the β- or δ- positions for a conjugated alkene. This forms a variety of 8 hydroxy alkyl peroxy radicals. The reversibility, and hence interconversion between these peroxy species has been conﬁrmed experimentally by Crounse et al. and Teng et al. [75, 77]

As described in ﬁgure 1.4, the two sets of three peroxy radicals, formed from O2 addition to the 1-OH and 4-OH radicals, are unable to interconvert with one another. As such, current models (which group the 8 peroxy radicals together) are likely misrepresenting the ratio of end products formed by these distinct groups, directly eﬀecting the ratio of methacrolein:methyl vinyl ketone yields.

Within the troposphere, the hydroxy alkyl peroxy radicals are able to react with NO, NO2, HO2, undergo (1,5) and (1,6) hydrogen shifts, forming a plethora of ﬁrst generation products. The most important of these, are methacrolein (MACR), methyl vinyl ketone (MVK), formaldehyde, various diols, hydroxy hydroperoxides (HPALDS), hydroxy nitrates, C5 and C4 hydroxycarbonyls, and C5 carbonyls. [21, 26] The reader is referred to an excellent recent review by Wennberg et al. which

14 provides further details on the up to date yields and photochemistry of these ﬁrst generation products.[69]

Improving the Accuracy of Atmospheric Models

The ability to correctly model the steady state concentration of the hydroxyl radical can act as a proxy for our understanding of the local chemistry. In a review of various ﬁeld and laboratory studies on the concentration of OH, a signiﬁcant error in the models was observed, which correlates with the concentration of isoprene.[78] Over pristine rainforests, often characterised by low NOx (NOx = NO, NO2 and NO3), and where there is a considerably larger amount of volatile isoprene, the observations of [OH] are often more than a magnitude greater than models would predict. This indicates that isoprene has a major role in the recycling of OH, further implying that the current oxidation mechanisms of isoprene, such as the updated LIM1 mechanism, whilst considerably detailed, must be incomplete.[79]

Considering that 85% of the reactive fate of isoprene begins with hydroxyl addition, and that the steady state concentration of OH cannot be accurately modelled in regions of high isoprene concentration, there is no surprise that the global SOA contribution from isoprene has such a broad range. [72] Despite the various experiments conducted to help elucidate details of this reaction mechanism, not once has any of the 6 intermediate hydroxy alkyl or 8 hydroxyl alkyl peroxy radicals ever been directly observed.

In the only isomer speciﬁc studies reported to date, conducted by Green- wald et al. and Ghosh et al. in the North laboratory, key reaction paths were identiﬁed for the inner addition products, 2-OH and 3-OH, and isomerism of the

δ-hydroxy alkoxy radical, formed from O2 addition to 1-OH, was observed through isotopic labelling experiments. [80, 81] Even if the reactions of the inner addition products, and the peroxy isomerisms only accounted for a minute portion (1 - 2%) of the total reaction pathway, the reaction products represent the same carbon balance as the total emission of key anthropogenic VOCs. Again in these studies however, none of the radical intermediates were directly observed, and the reac-

15 tion paths were identiﬁed due to retro-analysis of the end products. This begs the question; by probing the spectroscopy, structure, energetics and distribution of the key radical intermediates directly, what additional information could be found and implemented into chemical mechanisms to improve the accuracy of isoprene derived SOA? By identifying spectroscopic signatures for these intermediates, in particular their ionisation energies, these radical species and their relative abundances could be identiﬁed and probed in situ, which could provide experimental constraints for the branching ratios and reaction kinetics of the initial stages of isoprene oxidation.

1.4 Oxidation of Anthropogenic VOCs

Anthropogenic VOC emissions only represent ∼10% of the total VOC budget, however they play a signiﬁcant role in the tropospheric chemistry of urban areas. [29] These emissions are known to aﬀect and enhance the low altitude production of ozone and SOA, contributing to photochemical smog. [23, 82, 83, 84, 85, 86] The most important class of anthropogenic emissions, in regards to SOA formation, are the aromatic species. They have been modelled to account for ∼80% of the total anthropogenic SOA mass, followed by alkenes (16%) and alkanes (4%). [87, 88] The total SOA produced from aromatics is around 3.5 Tg yr−1, however this estimate lies within a range of some 2 - 12 Tg yr−1. [85]

The oxidation of aromatic species is outlined in section 1.2.2. As for other alkenes, the hydroxyl radical is prone to electrophilic addition at one of the unsaturated carbon sites, forming an OH adduct. Addition of molecular oxygen can then occur at multiple positions, as the radical site can shift around the odd-membered chromophore. Further reaction by NO and O2 can lead to ring breaking, and decomposition of the species. Under atmospheric conditions, with low NO2, Ziemann et al. measured high yields of dicarbonyl products, (46% - 64 %) from various substituted OH-adducts, often proceeding through bi-radical intermediates. Many of the intermediates and reaction pathways here have yet to be identified, resulting in a 20 - 40% quantifiable discrepancy between the initial and final carbon balance. [29]

16 In a recent study by Henze et al., benzene was identiﬁed to be the most important of these aromatic species, contributing to the same quantity of SOA as both toluene and xylene combined.[85] Interaction with the hydroxyl radical dominates the atmospheric loss mechanisms of benzene. This reaction is temperature dependent, and the addition reaction is most dominant under atmospheric conditions. Under elevated temperatures, for example in combustion, the abstraction pathway dominates. [82, 89, 90, 91, 92] Benzene represents a key AVOC, by its emission ﬂux, prototypical structure and propensity to form SOA. Considering its substantial contribution to the anthropogenic SOA budget, that itself encompasses a wide range, further revision and elucidation of benzene’s oxidation mechanism is clearly warranted.

1.5 Thesis Focus and Outline

By the turn of the present century, the global SOA budget is expected to rise by 36%. Increased biogenic and anthropogenic emissions account for 26% and 7% of this budget respectively. [66] Given that our current state-of-the-art models predict the annual formation of SOA to lie anywhere between 50 - 910 Tg yr−1, it is clear that our understanding of this formation process in both prisine and urban environments must be considerably improved to be able to predict and understand the eﬀects of the growing SOA budget on the health of our planet and its occupants. [16, 23]

The formation of aerosol is initiated in both urban and pristine regions primarily by interactions with the radical species OH, NOx,O3 and Cl. The majority of plant and anthropogenic emissions occur during the diurnal cycle, respectively corresponding to increased photolytic and human activity. The key oxidant of VOCs during this cycle is the hydroxyl radical. Considering that the majority of reactions proceed by addition or abstraction by the hydroxyl radical, followed by molecular oxygen addition, it is surprising that the vast majority of ﬁrst and second generation radical products have never been observed. Indeed, the majority of the reaction schemes and pathways considered in the formation of SOA products are based on

17 structure-activity relationships, end-product retro-analysis, and theoretical calculations. These assumptions accumulate into magnitudinous errors on a global scale.

The studies presented in this work aim to identify and characterise key radical intermediates in the OH-initiated oxidation of the most important biogenic and anthropogenic VOCs related to SOA formation, isoprene and benzene. These intermediates will be probed by various spectroscopic arrangements, providing mass, isomer and even rotamer selectivity. The experimental apparatus and techniques used to probe the transient intermediates in this study will be detailed in chapter 2.

In chapter 3, we examine a postulated intermediate involved in the atmospheric degradation of benzene. The D1 ← D0 vibronic spectroscopy of two rotamers of the ortho-hydroxycyclohexadienyl radical is explored over the region 18150 - 19250 cm−1. The observed spectral features are analysed in comparison with quantum theoretical calculations, to infer the excited state geometries of the species.

The dominant radicals produced from the OH initiated oxidation of isoprene contain allylic chromophores, see section 1.3.2. In chapter 4, we explore the eﬀect of methyl substituants on the allylic framework. Precursor selection, synthesis and experimental technique ensures the isomer selectivity of these methylated allyls, which is necessary to unravelling the complex excitation spectra of the multiple dimethyl allyl radicals formed from hydrogen addition to isoprene.

Chapter 5 presents the efforts made towards identifying species formed from the hydroxyl addition to isoprene. Various synthetic and experimental approaches are employed in order to detect the isoprene OH-adducts for the first time. In chapter 6, concluding remarks are made in regards to these efforts, and recommendations are made to direct future experimental and theoretical work.

The second part of this thesis follows, concerned with the identiﬁcation and spectroscopy of radical and ionic species relevant to astrophysical processes.

18 Chapter 2

Experimental Theory, Apparatus and Techniques

Short lived, radical species have often been challenging to investigate due to their transient and reactive nature. These molecules dictate chemical process all throughout our atmosphere and the universe and so an understanding of their photochemistry is vital. Within our atmosphere, these species incur billions of collisions per second with other ambient species, and so often evade detection.

The vacuum techniques used within this study are eﬀective in their isolation of these reactive, but stable, intermediates.[93] Once isolated, these radicals are probed with photons with wavelengths from the IR to the deep UV regions of the EM spectrum. This photochemistry can be correlated to natural phenomenon, where light is sourced from the sun and other stars. In this way the terrestrial and extra terrestrial photochemistry of radicals and ions can be explored.

The majority of studies conducted within this thesis utilise resonance- enhanced multiphoton ionisation (REMPI) techniques, whilst accompanying studies were conducted abroad employing matrix isolation (MI) spectroscopy.

19 2.1 Experimental Theory

2.1.1 The Vacuum Setup and Molecular Beam Methods

In order to properly simulate an atmospheric environment, correct ambient concentrations of molecular and atomic species must be accounted for. The temporal propagation of this system allows for an accurate representation and determination of products at chosen intervals. When used in tandem with computational modelling packages, this data can be used to infer plausible reaction pathways. The core of the chemistry however, driven primarily by reactive intermediates, occurs on a timescale undetectable in such instrumentation. Due to the proximity of surrounding molecules, the mean free path of these radicals is on the order of 0.066 µm.[53] This makes studying them, before their populations are quenched by further reactions, a very challenging task.

To study these isolated molecules they must be isolated from one another either by an inert barrier or be spatially separated. Two common experimental setups involve separating these molecules by freezing them in place within an unre- active framework or by studying them in a vacuum. The ﬁrst of these methods is called matrix isolation spectroscopy and will be discussed in section 2.1.4. Whilst matrix spectroscopy has been an incredibly useful technique and has been utilised for decades, the matrix itself perturbs the energetic characteristics of the frozen molecules through van der Waals interactions, namely London dispersion forces.[94] This renders their spectra unable to be quantitatively compared to other data, such as the astronomical data on the diﬀuse interstellar bands, however provides a good reference point for future studies of these reactive and/or short lived molecules. Take

+ for instance the observation of C60 in neon and argon matrices by Fulara et al. [95], which provided the motive for further gas phase studies to be undertaken on this species, completed some 22 years later by Campbell et al. of the same group, and positively assigned as the ﬁrst spectral carrier of a DIB.[96]

20 Molecules isolated from one another in a vacuum incur no such perturbation. To ensure the continuity of vacuum, molecules are introduced through a nozzle in a pulsed manner with conditions reﬂective of the load of the vacuum pump(s) available.[97] The static pressures of our chambers can reach as low as 10−7 Torr, whilst operating pressures are usually on the order of 10−6 Torr.

As the high pressure gas is pulsed into the vacuum chamber, an adiabatic free jet expansion occurs. If the difference in pressure between the two regions is large enough, then the gas accelerates to (local) supersonic speeds at the exit of the nozzle.[93] The expanding gas contains a mixture of a buffer gas, usually argon or helium, and some ∼0.1-1% of organic precursor. The seed ratio of the precursor can be controlled through temperature regulation. An increase in the temperature of the gas pre-expansion can also have effects on the molecular beam. A hotter premix can result in a less molecules being able to pass through the nozzle with each pulse, reducing the required pump load.[98]

The energy of the molecules pre expansion is equipartitioned into their internal and external (kinetic) energy. Upon expansion, collisions allow a net transfer of the internal energy to translational (due to the change in pressure). This translational energy is then transferred to the buffer gas through continual collisions. As the gas expansion propagates, the mean free path of the molecules increases to the point where the gas can no longer be considered a continuous medium, rendering the internal degrees of freedom ‘frozen’ from further partitioning.[98, 99] The hotter products continue to move further from the centre of the gas, and the colder part is skimmed. This renders the temperature of the molecular beam to the order of 10’s of Kelvin.[93] This cooling of the internal degrees of freedom greatly simplifies spectroscopy experiments conducted on these molecules, as the general population is now predominantly in a ground, vibrationless state. The use of a supersonic jet- cooled molecular beam, in addition to the low pressures afforded by the diffusion and turbomolecular vacuum pumps, allow the recorded spectra to be more readily compared to the astronomical spectra of molecules, radicals and ions within the interstellar medium.

21 2.1.2 Generation of Radicals

Generation of radicals can occur through various means such as pyrolysis, photolysis, or electrical discharge. The radicals in this study were created using the latter two methods, utilising a number of precursors, both purchased and synthesised. The majority of radicals were created utilising an electric discharge, pulsed across the molecular expansion, a technique that was first developed by Schlachta et al.[100] This pulsed discharge nozzle (PDN) is based on that described by Ohshima and Endo, and has been used in our laboratory for number of years.[101] Pyrolysis has been used in the past, however usually results in a hotter molecular beam temperature distribution and is only used for molecules which are significantly challenging to get into the gas phase, such as solids samples. Photolysis has also been used with mixed success, by aligning a laser onto the nozzle orifice. Photolysis allows for the breaking of certain bonds, however removes the possibility of a successive step in chemistry, as the gas will propagate with fewer to no collisions after this point. Hence we utilise a discharge, to impart energy to our molecular beam, allow some interesting chemistry to occur in the region contained by the PDN, followed by the expansion after the important chemistry has taken place. This is especially important where we are searching for products generated by addition, requiring collisions between species.

The setup used to generate these radicals can be seen in ﬁgure 2.1 below. Here a carrier gas of argon is passed through a container housing the precursor sample, soaked in cotton wool or held between glass wool. This container can be heated or cooled (if external) to allow appropriate seeding ratios. More volatile samples are often cooled to extend their lifetime and reduce unwanted clustering, to which high seed ratios are prone. Once the vapour is collected, the gas mixture is sequentially pulsed into the chamber by a Series 9 General Valve pulsed solenoid valve. Any heating is done in a graduated fashion, keeping the nozzle at a higher temperature to reduce condensation of products and malfunction of the nozzle.

Before it expands, the gas passes through two electrodes, separated from one another and the nozzle by polytetraﬂuoroethylene (PTFE) spacers. The outer

22 HV

Argon buffer sample series 9 pulsed PTFE discharge skimmer molecular gas inlet container solenoid valve spacers electrodes beam

Figure 2.1: Illustration of the pulsed discharge nozzle used in these experiments. electrode is commonly pulsed with a negative voltage of 1.2 - 2.1 kV which arcs across the gas mixture to the inner, ground, electrode, imparting part of this energy into the buﬀer gas and precursors.

A multitude of black box chemistry ensues, most commonly related to the breaking of bonds within the precursor, forming atomic, molecular, ionic and radical fragments. The more stable a product, the more likely it is to be formed in this harsh environment. Radical intermediates which are resonantly stabilised are commonly seen. These smaller fragments are able to add to or abstract from, other precursor molecules, forming second generation products. The exact chemistry is unknown, but can be greatly inﬂuenced by a number of factors, including choice of precursor, the timing, duration, voltage and resistance of the discharge, which will be discussed later.

The choice of argon here is used as more than just a buﬀer gas. Metastable argon has a similar energy to the bond dissociation energy of water. Hence, if the premix is passed through a sample of water vapour prior to the discharge, then H and OH radical products are readily formed from collisions with metastable argon.[102] In this way, a clean generation of these radicals can be allowed to react (addition or abstraction) with the precursors and precursor discharge products. This has been observed for water concentrations as low as 100 ppm.[103]

With such a large number of products generated within the discharge, there exists the need for very eﬀective separation techniques. The most distinguishable feature of the generated products is their masses. However there could be multi-

23 ple isomeric matches for each mass. Furthermore, of each isomer, multiple forms (cis/trans, anti/syn, E/Z) can be created. Appropriate choice of precursor greatly simplifies this problem, which when combined with chemical calculations and an experimental setup affording isomeric resolution, allows spectra to be confirmed to an exact carrier.

2.1.3 Resonance-Enhanced Mutiphoton Ionization (REMPI) Time-of-Flight Mass Spectroscopy (TOF-MS)

Single photon transitions are governed by the selection rules regarding the conser- vation of their orbital angular momentum, as such only excited states with change of ±1 can be measured in this way. Two photon absorption however, allows one to probe excited states which altercate the orbital angular momentum by 0 or ±2, for instance where an excited state has the same symmetry as the ground state. This process proceeds through a virtual state, allowed by the imaginary part of the molecules third-order nonlinear susceptibility tensor.[104]

Resonance-enhanced multiphoton ionisation is a very useful technique used to probe species with low molecular absorption cross sections. This technique makes use of a vastly increased transition probability wherever an incident photon is of the same electronic, vibrational or rotational energy level of an atom or molecule. Vari- ations in REMPI schemes allow one to selectively excite and then ionise a molecule via its intermediate vibrational and electronic levels. These processes are described in figure 2.2. Briefly, once the radical of interest has been generated in a molecular beam, it is intersected by two counter prorogating laser beams with photons of wavelengths ν1 and ν2, temporally separated by some time t. Neither of these photons have enough energy to ionise the molecule individually, however when ν1 is tuned to some resonant state of the molecule, then it can be subsequently ionised by ν2. The intersection point is within a static electric field gradient (EFG). Once ions are formed, they are excited up the length of a Wiley-McLaren type time-of- flight mass spectrometer.[105] Here the ions impact a multichannel plate at a time proportional to the square root of their mass. This static EFG is able to be ma-

24 nipulated by orthogonal ‘steering voltages’ to sharpen the mass peaks and yield an effective resolution of ∼ m/z 1 (mass per charge unit). In comparison to fluorescence spectroscopy, where there is no mass resolution, REMPI schemes allows one to accurately separate out absorption profiles into a mass dependence. Where species carry the same m/z, hole burning spectroscopy techniques are employed, described shortly.

To describe a multi-photon event, different approaches can be taken, such as by defining boundary conditions for Bloch and kinetic equations,[106, 107] or by perturbation theory.[108] The transition probability W (f, i) for a two photon transition between some initial state wavefunction Ψi and some final state wavefunction

Ψf can be described by;

2 hΨ |µ | Ψ i hΨ |µ | Ψ i 2 f fk k k ki i W (f, i) ∝ I (2.1) ∆Eik − hνr where Ψk represents the wavefunction of the intermediate state k, h is Planck’s constant, µ is the dipole moment, and ∆Eik is the energy diﬀerence between the initial and intermediate state. W (f, i) scales with In for a generalised n-photon process and is here I2, ﬁnally the wavelength of the incident photon is denoted by

νr.

A resonance-enhancement of the transition can be seen as the wavelength tends towards hνr = ∆Eik. Though this would appear to cause a mathematical divergence, in a physical sense the intermediate states are not described by a delta function and have some nonzero linewidth. This linewidth is directly related to the lifetime of the state, and the dephasing process associated with the rate of phase loss between the initial and resonant states of the transition. As such, the transition divergence is avoided. The lifetime of this state depends upon its energetic landscape, where one must consider phenomenon such as inter system crossing, internal conversion, pre-dissociation and conical intersections.

A Jablonski diagram of the various REMPI Schemes used within this thesis is displayed in Figure 2.2. This type of diagram represents the electronic states in

25 a bold line, and vibrational levels of that state in the thinner lines above. Each arrow represents a photon of a speciﬁc wavelength, the blue ν1 are usually within the visible region for the ﬁrst excited state and UV for higher excited states and

Rydberg transitions; whilst the purple ν2 are always UV photons. The wavelength of the green ν3 photons depend on the experiment and are described for each case.

The radicals are described by a D0 ground state, and a D(1−n) excited state. The first case (i) represents a typical 1 + 10 REMPI scheme, where the 0 indicates a photon different wavelength. Every time that ν1 is equal to a vibronic band in the excited state, the molecule is ionised by ν2, accelerated up the time-of-flight mass spectrometer and a signal is recorded. As such, the excited state is mapped out by scanning ν1, with a fixed ν2. Case (ii) represents a typical 2 + 2 REMPI scheme, to observe states where excitation requires an even number of photons to conserve orbital angular momentum.

Ionisation S + + + + 0 Continuum

D1-x

D 0 (i) (ii) (iii) (iv)

Ionisation S + S + + 0 1 Continuum

+ S0

D1-x

D1-x t ~ 100ns

D0 D0 (v) (vi) (vii)

Figure 2.2: Jablonski diagram of the various REMPI schemes employed in this work

26 For case (iii), ν1 is fixed and ν2 is scanned to find the onset of the ionisation potential of the molecule. As the molecule will only be ionised when ν2 is of sufficient energy, the ionisation potential can often be reported to an accuracy of ∼1 meV. The influence of the electric field on the ionisation energy of these molecules has been reported and is described by the following equation;

√ IE = IE0 − k V (2.2)

where IE is the experimentally observed ionisation energy, IE0 is the ﬁeld-free ionisation energy, V is the voltage of the static electric ﬁeld and k is an experimentally derived constant. This correction is on the order of 7 - 8 meV for a EFG of some ∼200 V.[109] The ionisation energy can also be obtained, as seen in case (iv), from the non-resonant absorption of multiple photons ν1. Without a tight focussing regime, the likelihood of this absorption is negligible. This method of determining the ionisation energy is often much noisier, contributing to a larger associated error in the measurement, see section 4.4.

It is also possible to measure the lifetime of the excited state, as seen in case (v). Here, the ionisation laser is temporally scanned and the signal proﬁle is recorded. The signal represents a convolution of the two gaussian laser proﬁles and an exponential decay function. Fitting this exponential decay allows one to extract the lifetime of the intermediate state.

One of the current challenges of spectroscopy is to accurately determine the vibronic structure of cold cations. Case (vi) describes the production of cold cations

0 from a regular 1 + 1 REMPI scheme, see chapter 9. Importantly here, ν2 is chosen to threshold ionise the neutral species, ensuring the cation is prepared in a cold, vibrationless ground state. We then scan a third photon, ν3, promoting the cation from S0 to S1. A high energy output laser is chosen for the production of ν3, which when combined with a tightly focussed beam proﬁle, generates a high photon ﬂux at the molecular beam, enabling multiple photons to be absorbed above S1. As a result, the cation promptly dissociates, and a reduced mass signal is observed in the radical (parent) mass channel for each time ν3 is resonant. One could measure the

27 fragmentation ions resultant from this dissociation as an indirect way to probe for the cation’s S1 ← S0 transition, however these mass channels are usually populated by fragments produced in the discharge, and are not reliably steady. Furthermore, it is more eﬀective to measure the total depletion of the one radical channel, than to measure a smaller increase in one of the various fragmentation peaks, which will be closer to the experiments level of noise. The black dotted line in the ﬁgure represents that following excitation and the absorption of further ν3 photons, the molecule will likely dissociate, or at least not return to the same S0 vibrationless ground state.

Finally, case (vii) depicts hole burning (HB) spectroscopy, a technique used to separate out a spectrum into it’s isomeric dependence. We begin by setting up a 1 + 10 REMPI experiment on a certain transition in the spectrum. Hole burning is then used to determine which other peaks in the spectrum hail from the same isomer. A third photon, ν3 is scanned around 100 ns prior to the REMPI experiment.

Each time ν3 excites a transition of the same isomer, a depletion is observed in the

REMPI signal. If ν3 excites a mode relating to another isomer, the REMPI signal will remain unaﬀected. This can be done as many times as there are isomers in a spectral region to pull it apart entirely. This technique is very precise and can even be applied to rotational isomers (rotomers), as seen in chapter 3.

By the utilisation of the above techniques, spectral lines are able to be accurately assigned according to a mass (TOF-MS) and isomeric form (case (vii)).

2.1.4 Matrix Isolation Spectroscopy (MIS)

The matrix isolation technique has been developed extensively over the later half of the 20th century, since its inception by Pimentel and Porter, and is an important technique for studying fundamental processes of atoms, molecules and more notably free radicals and other transient/reactive species.[110, 111, 112]

A matrix is formed by the condensation of a seeded gas onto a cold substrate, typically gold plated copper. The substrate is cooled to a few Kelvin by the use of a closed circuit helium refrigerator. The gas, termed ‘host’, can be a range of

28 molecules or compounds; most notably rare gases are employed for their chemical inertness, and their transparency in the IR region. The molecules of interest to study, termed ‘guest’, can be co-deposited into the matrix in a variety of methods depending on its vapour pressure and reactivity. If the molecules vapour pressure is suﬃcient, a premix of the host and guest can be made to variable concentrations by external temperature control. If the vapour pressure is too low, laser ablation can be performed, or the sample placed in a hot oven and its vapor picked up by the host gas. Once deposited, the matrix can maintain its shape and properties protracted periods of time.

An advantage of MIS is that the matrix can be formed simultaneously whilst conducting FTIR absorption experiments. This allows for real time precise control over the depth and quality of the matrix to ensure the guest is not saturated and suﬃcient signal can be recorded. As the majority of sample is directly deposited onto the substrate, much smaller quantities of guest molecules are required when comparing to pulsed molecular beam experiments such as REMPI. This allows for the study of relatively expensive isotopologues and synthetically prepared samples.

Due to the very low operating temperatures of MIS, reactive or transient species can be effectively isolated and cooled to low internal degrees of freedom. This ‘cage effect’ induced by the frozen matrix incurs advantages and disadvantages.[113] As seen in figure 2.3, these guest molecules are held in a chemically inert environment, where no further internal or bimolecular reactions can occur. This allows the matrix to retain its structure and composition for long periods of time. Photolysing species in such an environment allows for an effective isomerisation into multiple conformers of the guest.[114, 115] As the radical species are predominantly unable to separate from one another through space, trapped in the same ‘cage’, they re- combine, in a manner reflective of the relative energies of the conformers. To study radicals however, they are often formed prior to the deposition, such as through photolysis or pyrolysis of relevant precursors.

In comparison to the gas phase, the matrix hampers the internal rotation of the guest, (especially for larger molecules), resulting in a reduced rotational proﬁle, and thus a sharper signal/noise spectrum can be obtained.[113, 116] Furthermore,

29 Figure 2.3: The ‘cage eﬀect’ of a hard matrix preventing spacial separation, and enforcing recombination of dissociated products. the temperature of the matrix removes any hot band transitions that may arise in the gas phase. Typically, inert gases only perturb the true absorption features of molecules by ∼1%.[117, 118] This makes MIS great preliminary experiments for those conducted in the gas phase. Considering the relative ease of MIS experiments, this makes them a very eﬀective tool when combined with gas phase studies, especially in the interest of comparing with astrologically important species where line position and shape is crucial.

In recent years, para-hydrogen (p-H2) has been successfully employed as a matrix medium due to its interesting properties as a quantum solid. As a quantum solid, the amplitude of the zero-point lattice vibrations of p-H2 are a substantial fraction of the spacing between the nearest p-H2 molecules. As a result, the matrix is said to be ’soft’, resulting in diminishing cage eﬀects and reduced inhomogeneous broadening.[119, 120] The ’soft’ nature also allows for internal rotations of the guest molecule.[121]

30 The crystal defects in a quantum solid are self-repairing, due to the allowed movement of the p-H2 molecules. This makes the surrounding environment for a guest homogeneous, overcoming entropic barriers even at 0K.[120] Unlike the cage effects as seen in figure 2.3, a p-H2 matrix allows vacant sites, where guests, and their photolysed products, may move. As such, proper precursor design, involving photoliable C-X halogen bonds, enable one to create a matrix containing a sole configurational isomer. As it is impossible convert o-H2 to p-H2 in unity, residual o-H2 will remain in the matrix.[122] To prevent any o-H2 from gaining vibrational quanta and being able to transfer this to the guest, a 2.4µm filter is employed.

Once the matrix has been formed, and the guests photolysed, radicals and other products must be identified. This is done through a procedure known as ’grouping’, where the spectral lines are labelled in the same group if they exhibit the same behaviour to external perturbations, such as annealing or secondary photolysis, or that they maintain the same relative intensity to one another upon varying experimental conditions, such as different sample concentration, or different deposition temperatures. In secondary photolysis, the matrix is subjected to a variety of low intensity wavelengths, from LED’s, lamps (e.g. Mg, Sn) and UV irradiation from power mitigated laser sources. From this, the spectral lines can be grouped by their response to each wavelength, indicating they all belong to the same species. After the grouping procedure has been completed, each group is then assigned a spectral carrier by comparison with high level theory and calculations.

2.2 Apparatus and Applications

2.2.1 REMPI coupled ToF-MS Experimental Apparatus

All REMPI experiments were conducted using the apparatus in figure 2.4. This apparatus was constructed in house by Dr. Kokkin and consists of two differentially pumped vacuum chambers which are connected by a 2mm skimmer. Vacuum is maintained in the larger source chamber (left) by a VARIAN VHS-250 oil vapour diffusion pump backed by an ALCA-TEL 2033 rotary vane pump. The smaller

31 extraction chamber (right in ﬁgure), connected to the time-of-ﬂight tube, is pumped by a PFEIFFER BALZERS TPH330 turbomolecular pump, in turn backed by an ALCATEL 2012A rotary vane pump. As before mentioned, the static pressures of these chambers reaches ∼10−7 Torr, whilst operating pressures are usually on the order of 10−6 Torr for the source and extraction chambers. These pressures were adjusted by altering the opening time of the nozzle and the backing pressure of the premix gas. Pressures were measured during the pump down procedure by two VARIAN 843 Vacuum Ionization Gauges. Once the pressures were suitably below 10−4 torr, they were measured by use of two DUNIWAY T-KF40 ionisation bulbs and read by digital GRANVILLE-PHILLIPS 330 Ionization Guage Controllers.

MCP

time-of-flight tube

PDN skimmer extraction grids

nozzle translator

carrier gas electrical feed throughs lasers (hν1 + hν2) molecular beam

to turbo pump

to diffussion pump

Figure 2.4: The REMPI ToF-MS chamber used in these experiments. Reproduced with permission from Troy.[123]

A ﬂange is connected to the source chamber which contains the majority of electrical and gas feed throughs for the experiment. Electrical feed throughs are available to connect heated wiring to the gas lines, sample container and nozzle head.

32 Temperatures were maintained through the use of K-type thermocouples attached directly to the container and nozzle, and were viewed external to the chamber on in-house built digital readouts. Samples were able to be safely heated to ∼110◦C without damaging the nozzle components.

The extraction chamber contains 3 fused silica-windows to allow lasers to enter the chamber. These can be replaced by MgF2 windows when vacuum ultraviolet light is required for experiments. The top of the extraction chamber is coupled to a Wiley-McLaren type flight tube, manufactured by R. M. JORDAN. This tube is orthogonal to the molecular beam and measures 1.5 m. At the top of the flight tube, a dual micro-channel plate (MCP) is mounted, and connected to a high voltage power supply, feeding the unit with up to 4.5 kV. Within the chamber exist two extraction grids, which are held at a static voltage of around +2.1 to +2.5 kV, with a difference of around 150-250 V.

The REMPI experiments, as described in ﬁgure 2.2, were conducted in the following manner. A carrier gas of argon (4 - 7 bar) was passed through a sample container (internal or external) containing the precursor and in some cases additional (heavy) water. Temperatures were regulated so that this would typically result in a seeded gas mix of ∼0.1-1% organic (and ∼1% water). The sample was then expanded into the source chamber through a PARKER HANNIFIN GENERAL VALVE SERIES 9 pulsed solenoid nozzle, controlled by a PARKER HANNIFIN GENERAL VALVE IOTA ONE nozzle driver. This setup was previously shown in 2.1. The discharge was then struck for ∼80-300 µs to coincide with the later part of the expanding gas. The discharge voltage varied between 1.2 kV and 2.1 kV, through a 2 - 25kΩ ballast resistor. Upon expansion the discharge products cool to the order of 10’s of Kelvin, and pass through a 2 mm skimmer into the extraction chamber.

Once in the extraction chamber, the beam is subjected to as many as three counter propagating laser pulses. Two types of Nd:YAG pumped dye lasers were used in varying arrangements for the various REMPI schemes, a QUANTEL TDL90 and a SIRAH COBRA-STRETCH, which were pumped by the second or third harmonic of a QUANTEL Nd:YAG (YG980) laser and a QUANTA-RAY (BEAMLOK) Nd:YAG

33 laser respectively. EXCITON type dyes such as Exalite 411/417/428, Coumarin 440/460/480/503/540a, Fluorescein 27, and Rhodamine 590/610/640 (dissolved in para-dioxane or ethanol as required) were used to generate photons across the visible range. The spectral bandwidth of these lasers is on the order of 0.1 cm−1. Tunable ionsation radiation was generated by the use of frequency doubling the output of one of the dye lasers, by use of a variety of barium borate (BBO) crystals. Fixed wavelength ionisation photons (266 nm) were provided by frequency doubling the the second harmonic of one of the Nd:YAG pump lasers. The UV radiation was separated (when required) from the fundamental wavelength by use of a Pellin-Broca setup. Experiments were run at either 10 Hz or 20 Hz, dictated by experiments requiring the TDL90, limited to a 10 Hz repetition rate. Absolute laser frequencies were calibrated with a COHERENT WAVEMASTER wavemeter, which periodically calibrated itself (internally) by comparison with the spectral lines of neon gas.

Once ionised, cations were accelerated up the time-of-ﬂight tube toward the MCP. Signal was collected by a LECROY 9354TM digital storage sampling oscilloscope and processed real time using software generated by Dr. Tyler Troy in LabView version 7.1 as part of his doctoral thesis.[123]

The timings of all events were controlled by a STANFORD RESEARCH SYSTEM DG 535 and two HIGHLAND TECHNOLOGY T560 digital delay gen- erators (DDGs). The STANFORD DDG was triggered internally and acted as a master clock. Custom built software, again by Dr. Tyler Troy, was written to edit the timings of the firing of the laser flash lamps and Q-switches, the firing time of the nozzle and discharge, (and the duration of the latter) and data acquisition cycle of the oscilloscope. All other conditions were permuted manually.

2.2.2 Matrix Spectroscopy

Matrix experiments were conducted during a month long trip to Professor Y. P. Lee’s laboratory at the National Chiao Tung University in Taiwan. Two types of experiments were conducted, the ﬁrst involved the deposition and photolysis of prepared precursors, the second involving an electron impact ionisation coupled

34 deposition of isoprene (not included in this work). The experimental chambers differed only in the availability of the electron gun, and a simplified schematic is shown in figure 2.5.

germanium MCT mid window

IR beam 2.4 um lter

vertex 80v seeded matrix

gold plated copper

KBr window electron gun 250 V, 20 μA

3.3 Kelvin 1-3 e-6 torr deposition <1 e-5 torr static (300K) para hydrogen

ne mesh grid precursor secondary photolysis laser/LED

Figure 2.5: Illustration of the matrix setup utilised in these experiments.

Normal H2 gas (99.9999%, Scott Specialty Gasses) was passed through a cold trap (liq. N2, 77 K) to remove any unwanted species from the gas lines. It was then passed through a p-H2 converter which consisted of a copper cell filled with an iron (III) catalyst (Aldrich) and cooled using a closed-cycle refrigerator (ADVANCED RESEARCH SYSTEMS, DE204AF). The temperature is a key component in controlling the efficiency of the conversion, and was set to 12.9 K. This resulted in a population of less than 100 ppm o-H2. [124] Lower populations are obtainable, but at a large time cost as the flow rate of H2 over the converter must −1 be reduced. Further, if the flow rate is reduced to below 600 mbar h , then H2 will freeze in the pipes.

A gold plated copper substrate was used as both a medium to deposit the matrix upon, and a mirror to reﬂect incident infrared radiation to a detector. The temperature of the substrate was maintained at 3.2 − 3.3K throughout deposition and experiments, by a closed-cycle helium cryostat refrigerator (SUMITOMO

35 HEAVY INDUSTRIES, RDK-415). [124] Infrared absorption spectra can be obtained for the range of 450 − 5000cm−1 by a Fourier-transform infrared (FTIR) spectrometer (BRUKER, VERTEX 80V); however in practice, the range below 500cm−1 is very noisy, and the spectrum above ∼4160 cm−1 is truncated by the use of a 2.4µm filter. This band pass filter is introduced to prevent vibrationally exciting o-H2 impurities, which are able to react with/quench radical populations. The FTIR spectrometer is equipped with a KBr beam splitter and a mid-IR mercury cadmium telluride (MCT) detector that is continually cooled by liquid nitrogen to 77K. Each spectrum was recorded as an average of 200 scans with a resolution of 0.25cm−1. As seen in figure 2.5, the path of the IR beam is reflected by 90◦ off the gold plated surface, and as such the IR beam passes through twice the amount of sample.

A gaseous mixture of the precursor seeded in p-H2 is made to a ratio of 1 : 1000 − 1 : 10000 and is deposited over a period of 3-8 hours at a ﬂow rate of 11−12 mmol h−1. During deposition, spectra were taken hourly to ensure the proper formation of the matrix and that the seed ratio is suﬃcient. It is recommended to deposit a thicker matrix with a smaller seed ratio to stop the matrix evaporating upon (secondary) photolysis. A low concentration also limits the likelihood of clustering and secondary reactions. As the process to generate and study a matrix can often take a full day, (with a 7 hour period before a new experiment can be started to allow the detector to slowly return to room temperature), spending an extra couple of hours to ensure a good matrix has been formed is extremely recommended.

Where a halogen containing molecule was used as the precursor, the 248 nm output from a COHERENT COMPexPro 50/102f excimer laser was used as a photolysis source. This source had a minimum output of 40 mJ pulse−1 and so it was attenuated to around 1−3 mJ pulse−1 by a variety of steel meshes. The sample is photolysed minute by minute and a spectrum is recorded at each point to ensure that precursor populations are decreasing and new lines/species are formed. The matrix is also visually inspected between each scan to ensure transparency.

For both types of experiments, a range of secondary photolysis sources are employed to group the new spectral features, as described earlier. Sources consisted

36 of light-emitting diodes of wavelengths 520, 445, 405, 380, 370 and 365 ±6 nm (4 − 5.8 W), mercury and zinc lamps for wavelengths 308, 254 and 214 nm using band pass ﬁlters, and a full spectrum low pressure mercury lamp. The matrix was subjected to each secondary wavelength for 1 − 10 minutes depending on the matrix response.

37 Chapter 3

The Hydroxycyclohexadienyl (CHD-OH) Radical

3.1 Introduction

Aromatic compounds are released to the atmosphere via many pathways, both anthropogenic (e.g. motor vehicle exhaust or evaporation of gasoline products) and natural (e.g. emission of volatile oils from plants). [39, 92] Aromatics can comprise a large fraction of fuels, (the Australian standard allows for up to 30% aromatics by weight), and they account for as much as 20-30% of the urban volatile organic compound budget to the atmosphere.

The atmospheric oxidation of aromatics, such as benzene and its methylated derivatives, leads to enhanced tropospheric ozone concentrations and photochemical smog. [82] Their ability to form secondary organic aerosols (SOA) is an expanding area of recent study, and predicted to generate signiﬁcant quantities of SOA within urban environment. [23, 83, 84, 85, 86] These aerosols have been docu- mented to have damaging health eﬀects, especially on cardiovascular and respiratory systems.[23]

The reactivity of benzene is relatively low compared to other substituted aromatics, and hence its atmospheric half-life is long, lasting up to several weeks

38 before reaction.[83] Degradation is chieﬂy initiated by reaction with the hydroxyl radical, which proceeds through two key mechanisms depending on temperature; [82, 89, 90, 91]

C6H6 + OH → C6H6OH (1)

C6H6 + OH → C6H5 + H2O (2)

For atmospherically relevant temperatures, direct addition dominates, see reaction (1), forming a C6H6OH adduct, ipso-hydroxycyclohexadienyl radical (i- hydroxy-CHD). The rest of the degradation mechanism is being continually improved. [125] A number of mechanistic studies have been performed on this radical, with intermediate products and reaction channels showing a clear dependence on ambient NOx concentrations. [83, 84, 86, 126, 127] Subsequent reactions involve removal of hydrogen by molecular oxygen abstraction, resulting in formation of phenol

+ HO2. [82, 89, 127, 128, 129] The ring breaking and propagating mechanisms of further oxidation have been studied, and believed to involve the formation of bicyclic species, via O2 addition across the phenol aromatic ring.[92, 130]

For temperatures above ∼325 K, the hydroxy-CHD radical becomes unstable and readily decays back to reactants (benzene + OH or phenol + H), while for even higher temperatures (combustion-relevant, T > 500 K) the reaction almost exclusively follows a hydrogen abstraction mechanism, reaction (2), forming phenyl and H2O.[131] This abstraction channel, accounting for only ∼5% tropospheric benzene oxidation is comparably well understood [59] and its intermediates have been successfully observed and identiﬁed in an argon matrix by Mardyukov et al. [132]

Despite clear progress in the understanding of the degradation pathways of benzene and other alkyl substituted aromatics, under diﬀerent temperature and varied ambient molecular concentrations, top-down and bottom-up approaches to SOA formations vary by an order of magnitude in their carbon balance. [23, 84]

In their matrix study of the phenyl-water complex, Mardyukov et al. observed the formation of ketene products and proposed a new mechanism, shown in ﬁgure 3.1.[132] As discussed above, in the troposphere the i-hydroxy-CHD (1i)

39 radical is able to form products (2), usually via O2 abstraction. Whilst a complex between phenol and H atom is proposed to be unlikely, it was trapped in their lattice and generation of (3Z)-1,3,5-Hexatrien-1-one, (3), was observed. Mardyukov stated that their mechanism remains speculative, as the proposed intermediates were not detected. The intermolecular hydrogen migration from ipso (1i) to form the ortho (1o) radical seems unlikely, leaving the formation of (1o) owing to the addition of H to phenol.[133]

2 OH O

+OH+OH -H

+H O H H

C -H

OH H H OH OH H -H 3

H 1i 1o

Figure 3.1: Possible ring-breaking mechanism from the ipso-hydroxycyclohexadienyl radical. A scheme of ketene formation from an initial OH-addition to Benzene yielding 1i. [132] The dotted line emphasises that the hydrogen migration pathway from 1i to 1o is energetically unfavourable.[133]

A recent paper by our group has detailed how hydrogen addition to phenol is ortho directed.[133] An absorption spectrum taken of the ﬁrst excited state revealed a complicated spectrum, containing numerous peaks over an extended range. An unexpectedly small spacing between the ﬁrst two peaks of only 18 cm˘1 led us to believe that the spectrum contained more than a single compound. Further investigation determined that these species were isomers of the same conformer, the anti and syn conformers of (1o). No other H addition isomers; ipso, (1i), meta (1m) or para (1p) were detected. Lifetime measurements and ionisations potentials were also discussed in this paper.

40 In the current chapter we examine the detailed spectroscopy of these ortho- hydroxy-CHD isomers and evaluate the theoretical methods and approximations used to assign their spectra.

3.2 Experimental

A description of the experimental apparatus can be found in chapter 2. Details speciﬁc to these experiments only are described below.

Argon (6 - 7 bar) was bubbled through a sample of phenol (Aldrich, ≥99%) and H2O, resulting in a seeded gas mix of ∼0.1-1% organic and ∼1% water. The sample was expanded into a diﬀerentially-pumped vacuum chamber through a pulsed discharge nozzle. The discharge (1.6 kV, 25 kΩ) is struck for ∼80 - 130 µs to coincide with the later part of expanding sample. A discharge in argon, with trace amounts of water has been shown to be an eﬀective way of producing H and OH radicals; occurring through collisions with metastable argon.[134] The discharge products were cooled to ∼10 K in a supersonic expansion, and the coldest part of this expansion was passed through a 2 mm skimmer.

The molecular beam entered the ionization chamber, where the molecules were examined using resonance enhanced multiphoton ionisation (REMPI) spectroscopy, which, here, utilised up to three counter-propagating laser beams. For normal REMPI experiments, the excitation laser, near 550 nm, utilised Coumarin 540A dye. The ionisation laser, at 230 nm, was the frequency doubled output of a dye laser using Coumarin 460 dye. For hole burning experiments, a third laser was introduced around 100 ns before the REMPI lasers, and resonant with a speciﬁc transition (dye laser with Coumarin 540A dye).

The resulting cations are orthogonally accelerated down the length of a Wiley-McLaren-type time-of-ﬂight mass spectrometer (ToF-MS), detected by a tandem multi-channel plate (MCP) and viewed by an oscilloscope. Custom-written LabView software was used to record the spectra. A wavemeter was used to cali- brate laser wavelengths.

41 3.3 Theory

The geometries of syn and anti conformers of hydroxy-CHD in the ground electronic state were calculated using density functional theory (DFT) at B3LYP/6- 311+G(d,p) level of theory due to its relatively low cost and reasonable accuracy. All calculations were computed utilizing the GAUSSIAN 16 software.[135] The ground state geometries for both anti and syn conformers have Cs symmetry at this level of theory, as shown in ﬁgure 3.2. syn anti ground state excited state ground state excited state

Figure 3.2: Proﬁle and top down views of ground and excited state anti and syn rotamers, showcasing the geometry shift and the near-planar nature of the excited syn geometry. These geometries were optimised at the (TD) – B3LYP/6-311+G(d,p) level of theory.

Time-dependent density functional theory (TD-DFT) was used for calculating the structure and harmonic vibrational frequencies in the excited states (B3LYP/6-311+G(d,p)). Upon excitation for both conformers, the functionalised part of the framework is prone to puckering, and torsion of the hydroxyl group occurs, resulting in formal C1 symmetry. However, the magnitudes of these changes are quite diﬀerent for each conformer. The results of these calculations can be seen in ﬁgure 3.2.

For the anti-conformer, a signiﬁcant geometry change occurs involving the hydroxyl group. The oxygen bends out of plane by ∼16◦ and the respective H atom increases its dihedral angle to the framework by ∼6.5◦. The ring-carbon bound to

42 this hydroxyl group also puckers out-of-plane. The vibrational modes that contain these prominent out of plane distortions are likely to show the dominant Franck- Condon activity. Minor out-of-plane distortions of some ring-CH hydrogens also occur as do small in-plane distortions of CC bonds around the ring also occur.

The syn conformer, on the other hand, undergoes much less distortion. Excited state calculations carried out without diﬀuse functions (6-311G(d,p)) yielded a Cs excited state, however upon addition of this diﬀusion function, the excited state became C1 geometry. Imposing a symmetry constraint of Cs geometry yielded a single imaginary frequency, corresponding to the CO carbon puckering out of plane. We also note that the sp3 hybridised carbon and OH groups rotate and subtly distort to oppose this puckering. The excited state is thus assigned by the software as C1.

Frequencies calculated for both the ground and excited states of the syn and anti isomers can be seen in tables 3.1 and 3.2 respectively. Ground state frequencies have been scaled by a factor of 0.9688, suggested by Merrick et al., as very tight convergence criteria were applied to the geometry optimization.[136] The excited state was scaled by the same factor. The largest diﬀerence between the conformers lies in the frequencies of the lowest energy modes, which is helpful in distinguishing between the isomers. The anti and syn electronic transitions are both of a double excitation nature; where one electron moves from the HOMO (π) into the SOMO(n), and another from the SOMO(n) into the LUMO (π*).

43 Table 3.1: Ground and excited state frequency calculations for the syn conformer of the hydroxy- CHD radical, at the (TD)-B3LYP/6-311+G(d,p) level of theory. Frequencies are labelled according to the Cs point group of the ground state, where the excited state has been reordered according to the Duschinsky matrix, see text. Scaled frequencies (x0.9688) are included for comparison.

syn Ground State Excited State

Mode # Symmetry Frequency Scaled Frequency Frequency Scaled Frequency Mode Description

1 a0 3836 3716 3809 3691 OH stretch 2 a0 3199 3100 3165 3067 CH stretch 3 a0 3179 3080 3206 3106 CH stretch 4 a0 3176 3077 3218 3118 CH stretch 5 a0 3157 3059 3191 3092 CH stretch 6 a0 2896 2806 2922 2831 CH2 sym stretch 7 a0 1635 1584 1583 1533 C=C sym stretch 8 a0 1552 1504 1448 1403 C=C asym stretch 9 a0 1451 1406 1394 1351 CH2 scissors 10 a0 1441 1396 1432 1387 CH ip bend 11 a0 1435 1391 1377 1334 CH ip bend 12 a0 1362 1320 1333 1291 CH ip bend 13 a0 1312 1271 1543 1495 CH2 ip bend 14 a0 1238 1200 1235 1196 CO stretch 15 a0 1193 1156 1174 1138 OH ip bend 16 a0 1184 1147 1160 1124 HC=CH ip bend 17 a0 1125 1090 1135 1100 CH2 torsion 18 a0 1006 975 1019 987 ring C=C stretch 19 a0 974 944 961 931 ip ring deformation 20 a0 933 904 929 900 ring breathe 21 a0 780 756 631 611 ip ring deformation 22 a0 586 567 759 735 ip ring deformation 23 a0 497 481 539 522 ip ring deformation 24 a0 392 379 382 371 CO ip bend 25 a00 2880 2791 2949 2858 CH2 asym stretch 26 a00 1183 1146 1104 1069 CH2 twist 27 a00 956 927 718 696 CH oop bend 28 a00 936 907 795 771 CH2 oop rock 29 a00 894 866 944 915 CH oop bend 30 a00 743 719 517 501 CH oop bend 31 a00 661 640 476 461 CH oop bend 32 a00 516 500 468 453 oop ring deformation 33 a00 443 429 205 198 oop ring deformation 34 a00 285 277 148 143 oop ring deformation 35 a00 156 151 324 314 CO oop bend 36 a00 119 116 91 88 CH2 oop rock

ip = in-plane; oop = out-of-plane Table 3.2: Ground and excited state frequency calculations for the anti conformer of the hydroxy- CHD radical, at the (TD)-B3LYP/6-311+G(d,p) level of theory. Frequencies are labelled according to the Cs point group of the ground state, where the excited state has been reordered according to the Duschinsky matrix, see text. Scaled frequencies (x0.9688) are included for comparison.

anti Ground State Excited State

Mode # Symmetry Frequency Scaled Frequency Frequency Scaled Frequency Mode Description

1 a0 3811 3692 3627 3514 OH stretch 2 a0 3198 3098 3152 3054 CH stretch 3 a0 3176 3078 3200 3100 CH stretch 4 a0 3157 3059 3175 3077 CH stretch 5 a0 3138 3040 3181 3082 CH stretch 6 a0 2934 2842 2951 2859 CH2 sym stretch 7 a0 1624 1573 1549 1501 C=C sym stretch 8 a0 1545 1497 1433 1389 C=C asym stretch 9 a0 1461 1415 1450 1405 CH ip bend 10 a0 1449 1404 1359 1317 CH2 scissors 11 a0 1434 1389 1399 1355 CH ip bend 12 a0 1364 1321 1308 1267 CH ip bend 13 a0 1316 1275 1575 1526 CH2 ip bend 14 a0 1257 1218 1228 1190 CO stretch 15 a0 1182 1145 1183 1146 HC=CH ip bend 16 a0 1174 1138 1157 1121 OH ip bend 17 a0 1127 1092 1143 1108 HC-CH ip bend 18 a0 1010 978 1024 992 ring C-C stretch 19 a0 975 945 974 943 ip ring deformation 20 a0 929 900 930 901 ip ring deformation 21 a0 783 759 764 740 ip ring deformation 22 a0 586 568 549 532 ip ring deformation 23 a0 500 484 481 466 ip ring deformation 24 a0 397 384 392 380 CO ip bend 25 a00 2925 2834 2994 2901 CH2 asym stretch 26 a00 1183 1147 1097 1063 CH2 twist 27 a00 956 926 759 735 CH oop bend 28 a00 937 908 793 768 CH2 oop rock 29 a00 863 836 579 561 CH oop bend 30 a00 736 713 948 919 CH oop bend 31 a00 657 637 572 554 CH oop bend 32 a00 517 501 460 445 oop ring deformation 33 a00 457 443 299 289 oop ring deformation 34 a00 403 391 350 339 OH torsion 35 a00 279 270 178 173 CO oop bend 36 a00 131 127 50 48 oop ring deformation

ip = in-plane; oop = out-of-plane 3.3.1 Duschinsky Mixing

The reader should notice that the ground state frequencies in tables 3.1 and 3.2 are labelled according to the Mulliken convention. For a molecule with Cs symmetry, we ﬁrst label the a0 modes in order of decreasing frequency, followed by the a00 modes in a similar fashion. The excited state frequencies however appear to be more random in order, and this section will be used to clarify.

Upon excitation, the geometry of the syn and anti configurations distort, as seen in figure 3.2. Hence, the frequencies of each mode will be different, impacting the order. In labelling the excited state modes by Mulliken convention also, the labels become inconsistent and are consequentially not useful for spectroscopy. In order to better label the excited state modes, Duschinsky matrices were constructed by using the FCLab II Software.[137] We have previously used such matrices to assign excited state modes of the 1-phenylpropargyl radical and found they worked favourably.[138]

A Duschinsky matrix is constructed by taking scalar product of the mass- corrected displacement of each normal mode in the ground state with each of those in the excited state. This develops a 3N-6 × 3N-6 matrix where all grid elements represent the overlap integral between modes in the ground and excited states. The matrix elements are coloured according to this scalar product, where 0 is represented by white, 1 by black, and a linear greyscale between the two. In the case where the geometry change is minimal, and all modes in the excited state directly correlate to those in the ground state, and a diagonal matrix would be generated.

The necessity for a new labelling convention becomes apparent when we consider how the modes change in frequency, and hence order, upon excitation, as shown in ﬁgure 3.3. Here, the modes in the ground and excited state are labelled in the C1 Mulliken convention, (decreasing frequency order), straight from a GAUS- SIAN output.

46 anti syn

1.0

0.8

0.6

ν’’ 18 ν’’ 16

47 0.4 ν’ 19 ν’ 19

0.2

0.0

Figure 3.3: Duschinsky matrices for the anti isomer (left) and the syn isomer (right). Both the ground and excited state modes are labelled according to the

Mulliken convention assuming a C1 point group. Matrix elements are shown in a linear greyscale, where white represents 0 and black represents 1. In ﬁgure 3.3, there exists a large number of oﬀ-diagonal elements. As a result, the excited state labels become inconsistent with the ground state for the

0 majority of modes. For example, for the anti isomer, the excited state mode ν19 00 has a near perfect overlap with ground state mode ν16, indicated by the blue lines, inferring that these two diﬀerent labels represent the same molecular motion, but are of a diﬀerent relative frequency order. This is also apparent for the syn isomer, where

0 00 ν19 has a near perfect overlap with ν18. Other than this direct mode relabelling, the 00 00 0 purple boxes emphasise a binary mode mixing (for both isomers) of ν6 , ν7 → ν7, 0 ν6, indicating the labels of these modes should be swapped for the excited state. Further, the modes within the green box represent an extensive scrambling, where no ground state mode label is correctly represented by its corresponding excited state label.

To rectify these labelling inconsistencies, and to label similar vibrations with the same label, we have rearranged the excited state frequency order, to best diagonalise the matrix. This maps modes with similar character onto the same label and makes the spectral labelling much more consistent. In order to prevent confusion, when talking about the original C1 excited state labels, the preﬁx ‘or- ’ will be used. Before the excited state is dealt with, the ground state modes are reordered according to Mulliken convention for a molecule with Cs symmetry. Boxes emphasising the ground state symmetry labels have been drawn on the matrices, which will be discussed later.

For each matrix, the excited state column headings list both the original unordered (C1) labelling from figure 3.3, and the final labelling after diagonalization. This labelling convention ensures that when discussing a vibrational mode in the excited state, we can draw direct comparison with the same (or most similar) mode in the ground state. It should be noted that non-zero off-diagonal elements do not represent a coupling between these two modes upon electronic excitation. Rather, they signify the excited state normal mode contains character of more than one ground state normal mode.

The reordered Duschinsky matrix for the syn isomer is displayed in ﬁgure

3.4. A quick comparison between the original (C1) and ﬁnal (Cs) mode labels

48 emphasises the need for a reordered matrix, as only three mode labels, (ν1, ν3

0 0 and ν36), have been retained. Two modes, ν25 and ν32, are calculated to contain significant character of ground state modes with differing symmetries, which will be discussed later. Apart from these two modes, the symmetry of the remainder of modes is preserved upon excitation. From this plot, we can see that the majority of excited state modes contain character from more than one ground state mode. In some circumstances two excited state modes were calculated to contain a large component of the same ground state mode. For the syn isomer, as seen in figure 3.4, this occurred three times. syn

1 0.8 0.6 0.4 0.2 0

Cs Final

C1 Original

Figure 3.4: Duschinsky matrix for the syn isomer. The ground state has been labelled according to traditional Mulliken convention. The excited state axis exhibits both the original Mulliken convention (as though the excited state were C1), and the ﬁnal state assigned by Cs symmetry after diagonalization. Matrix elements are shown in a linear greyscale, where white represents 0 and black represents 1.

49 0 0 Excited state modes or-ν18 and or-ν16 originally had their largest matrix 00 element mapping to ground state mode ν17, shown by the green dotted lines in figure 0 00 3.4; or-ν18 however had a larger element and was thus matched with ν17. We then 0 00 considered the next highest unassigned element of or-ν16 which was that of ν15, and 0 after confirmation upon visual inspection, assigned accordingly. Modes or-ν13 and 0 or-ν12 underwent a similar review, indicated by the blue dotted lines, resulting in 0 00 0 00 0 mode or-ν12 being assigned to ν9 , and or-ν13 to mode ν11. The original modes or-ν9 0 00 and or-ν8 also both contained a dominant character from ν7 , indicated by the purple 0 0 dotted lines. As or-ν8 contained the largest character, mode or-ν9 was reassigned by rd 00 00 its 3 highest overlap with ν13, as the 2nd highest (ν8 ) already had a well-defined diagonal element. In cases like this, the label loses meaning with respect to ground state modes and becomes just a label.

The reordered Duschinsky matrix for the anti isomer is displayed in ﬁgure 3.5 below.

Similar to the syn matrix, only four labels remained unmodiﬁed following the rotation of the anti matrix, (here ν1, ν4, ν35 and ν36), providing further evidence for the need to reorder the excited state modes. Upon excitation, 8 of the 12 modes

0 0 00 between ν25 - ν36, which were of a symmetry in the ground state, are calculated to contain character of ground state modes with both a0 and a00 symmetries. This is signiﬁcantly more mixing than observed for the syn isomer and will be discussed later in reference to the excited state symmetries of both isomers.

The anti-isomer, also contained three elements with non-unique ground

0 states, which can be seen in ﬁgure 3.5. Mode or-ν26 contains almost even character 00 00 0 of ν21 (45%) and ν27 (43%), indicated by the green dotted lines. Original mode or-ν25 00 00 has a 41% overlap with ν21, but only 25% with mode ν27. As such, to maximise the 0 0 00 00 diagonal, modes or-ν26 and or-ν25 were assigned to modes ν27 and ν21 accordingly. 0 0 00 00 Modes or-ν22 and or-ν27 were paired respectively with modes ν30 and ν29 in a similar 0 0 00 fashion, indicated by the purple lines. Finally, or-ν18 and or-ν16 were assigned as ν17 00 and ν15 by their overlap elements, indicated by the blue lines.

50 anti

1 0.8 0.6 0.4 0.2 0

C1 Final

C1 Original

Figure 3.5: Duschinsky matrix for the anti isomer. The ground state has been labelled according to traditional Mulliken convention. The excited state axis exhibits both the original Mulliken convention labelling and the new numbering after the matrix was diagonalised. Matrix elements are shown in a linear greyscale, where white represents 0 and black represents 1.

3.4 Results and Discussion

Figure 3.6 (upper trace) shows a REMPI spectrum of the m/z 95 compound created from the discharge of phenol and water seeded in argon. The excitation laser wavelength was scanned between 18150 and 19250 cm−1, whilst the ionization laser was ﬁxed near 43500 cm−1. This spectrum has been assigned previously to both the syn and anti conformers of the hydroxy-CHD radical, with the electronic origin of the anti conformer at 18200 cm−1 and the syn conformer at 18218 cm−1. [133]

Hole burning spectroscopy was employed to untangle the contributions of the two conformers. The excitation laser was tuned to the 18200 cm−1 transition for the anti isomer, and a third laser, timed ∼100 ns prior to the REMPI lasers,

51 wavenumber (cm-1)

18200 18400 18600 1880019000 19200

H O H

'' 1 1 2 1 1 2 1 1 1 1 1 a vibrations 34 036 0 34 0 33 034 0 33 0 34 035 0 23 033 034 0

0 ' 1 1 1 2 1 00 a vibrations 24 0 23 0 22 0 24 0 18 0

0 100200 300 400 500 600 700 800 900 1000 relative wavenumber (cm-1)

Figure 3.6: (Upper) REMPI spectrum of ortho-hydroxy-CHD radical. (Lower) Hole burning (depletion) spectrum measured while monitoring the REMPI signal 18219 cm˘1. The spectrum is assigned to the syn conformer. Vibrational assignments are indicated in blue font for a0 vibrations and black font for a00 vibrations. was scanned across the same spectral range. All transitions with a ground state in common with the 18200 cm−1 peak appear as a depletion in the REMPI signal. The hole burning experiment was repeated for the peak at 18218 cm−1 for the syn isomer. These two experiments give rise to the depletion traces, reflected at the bottom, in figures 3.6 and 3.7, (figure 3.7 is shown later in the chapter).

In assigning the spectra, we give priority to peaks in the REMPI spectrum; the signal-to-noise is better, and intensities are more reliable. In all cases, peaks in the REMPI spectrum have counterparts in the depletion spectrum. However there are instances where apparent peaks appear in the depletion spectrum with no REMPI counterpart, which might arise from other experimental artefacts. Any

52 peak that appears solely in one of the hole burning experiments is labelled ‘?’ and not assigned.

The hole burning experiment attributed just 12 peaks (including the origin) of the m/z 95 spectrum to the syn conformer. This is far fewer than were attributed to the anti conformer (the remaining lines, justified in figure 3.7), which begs the question - what is different between the two excited states?

As stated in the theory section, the symmetry of the excited syn conformer was sensitive to the inclusion of a diﬀuse functional in the calculations. By an evaluation of the Duschinsky matrices, in ﬁgures 3.4 and 3.5, we can investigate the preservation of symmetry upon excitation, to explore the possibility that a higher symmetry is responsible for fewer peaks, due to symmetry selection rules.

As mentioned previously, the syn conformer undergoes a very small geometric distortion upon excitation, when calculated at the 6-311+G(d,p) level, breaking its plane of symmetry. In ﬁgure 3.4, boxes of symmetry are drawn to help us understand how well the symmetry is preserved upon excitation. For the syn conformer, we can see that almost every element lies within the two boxes, indicating that they retain the same symmetric character of their respective ground state mode. There are a couple of signiﬁcant non-zero matrix elements that require further evaluation.

00 00 The ground state symmetric (ν6 ) and asymmetric (ν25) CH2 scissoring motions show signiﬁcant mixing. As the molecule is excited, local modes are formed; and though the vectors retain their symmetry/asymmetry, only one hydrogen shows signiﬁcant displacement. This makes the two modes appear to be almost identical. With this in mind they are labelled by the symmetry/asymmetry of their vectors, even if the magnitude of displacement for the two hydrogens is uneven. The formation of local modes which break the symmetry of the molecule has been observed previously by Kemp et al.. In their studies on meta-disubstituted benzenes, they observed that upon increasing the mass of one substituent, the symmetric and asymmetric stretches of the C-X bonds (where X represents the substituent) eventually became local modes, where only one stretching mode was active, and the other near stationary, and vice versa. [139]

53 0 Another, smaller, anomaly is mode ν32, which contains character of both 00 0 00 the a mode ν32 (60%) and a mode ν23 (19%). Upon visual inspection, this mode is clearly dominated by the asymmetric out-of-plane ring distortion, along with large amplitudes of CH/CH2/OH wag. The mode retains some small character of an in plane elongation/contraction of the ring, and may excite in a single quanta.

As such, the excited state of the syn-conformer preserves the vast majority of vibrational mode symmetry. If indeed the molecule is of Cs symmetry, then it will become subject to the corresponding Cs selection rules. That is, only transitions of 0 overall a symmetry are allowed. However if the molecule is of C1 symmetry, as the TD-DFT study initially indicated, then single quantum transitions of out-of-plane modes will be observed.

The Duschinsky matrix for the anti-isomer, (ﬁgure 3.5), shows more extensive mixing between the a0 and a00 modes upon electronic excitation. This agrees with the larger displacement observed in the excited state geometry. We attribute the larger number of peaks in the anti spectrum to arise from the lowering of symmetry from Cs to C1 and the consequent relaxation of symmetry selection rules. In

C1 symmetry, transitions involving all vibrations are allowed.

3.4.1 The Syn Conformer

As discussed above, by an evaluation of the Duschinsky matrix, the structure of the excited state of the syn conformer is eﬀectively of Cs symmetry. We therefore start by assigning the vibrational structure in the spectrum in the Cs point group, where a0 modes are Franck-Condon active and a00 modes appear only as even overtones and combinations. If any a00 modes are observed in single quantum, then the symmetry of the molecule is lowered to C1. The calculated and scaled frequencies for the syn isomer, along with their respective symmetries are displayed in table 3.1. The

0 −1 electronic origin for this transition, 00, is observed at 18218 cm , and this transition was used as the ‘burn’ transition for the depletion spectrum, shown in the lower panel of ﬁgure 3.6. The assignments are displayed in table 3.3.

54 Table 3.3: Syn-hydroxy-CHD assignments and comparison with theory, including the mean absolute deviation (MAD) for each symmetry a0, a00 and the combined. MAD is calculated per total number of modes, and the combined deviation of ν23ν33ν34 is split three ways. All units are in wavenumbers (cm˘1). The theoretical values for the a00 modes are not strictly from the TD-B3LYP output, see text.

Observed Assignment Symmetry Label Theory ∆T −O

00 00 215 ν34ν36 a ⊗ a 231 16 00 274 2ν34 2a 286 12 00 00 356 ν33ν34 a ⊗ a 348 -8 0 373 ν24 a 371 -2 00 421 2ν33 2a 396 -25 00 00 461 ν34ν35 a ⊗ a 451 -10 0 522 ν23 a 522 0 0 731 ν22 a 735 4 0 734 2ν24 2a 742 8 0 00 00 873 ν23 ν33 ν34 a ⊗ a ⊗ a 878 5 0 988 ν18 a 987 -1

We begin our assignment with the a0 modes. The lowest frequency a0 mode,

0 −1 ν24, is the in-plane C-C-O bend, calculated at 371 cm , which is an excellent match with a medium-strong transition displaced 373 cm−1 from the origin transition of the syn conformer and therefore assigned as such. We can then assign the next lowest a0

0 −1 frequency, ν23, predicted and observed at 522 cm . There is a doublet at 731 and −1 0 734 cm . One of these is likely to involve the ring-breathing mode, ν22, calculated to have an excited state frequency of 735 cm−1. The other band likely contains the

0 overtone of the previously assigned ν24. In the harmonic limit, this overtone would appear at twice its fundamental, 742 cm−1. This overtone is uncharacteristically strong for an origin dominated spectra, and likely forms a Fermi resonance with

0 mode ν22. If the overtone is indeed anharmonic, this reduction in the frequency is −1 0 reasonable. We therefore assign the lower frequency component at 731 cm to ν22 −1 0 and 734 cm to 2ν24, although the equal intensity of the two components suggests almost complete mixing of the two wavefunctions. Finally, the ring breathe and CH

55 0 −1 −1 bend of ν18, predicted at 987 cm is observed just 1 cm higher. Single quantum 0 0 transitions in modes ν23 - ν21 are not observed.

0 0 The ν36 - ν33 asymmetric, out-of-plane modes are all involved in the change in geometry from the ground to excited state, and many exhibit Franck-Condon activity. However, none of these modes are observed nor assigned to a single quantum transition. This reaﬃrms our evaluation from the Duschinsky matrix analysis, which reinforces that the excited state has Cs symmetry, as the transitions of the syn conformer adhere to the corresponding Cs selection rules.

The next step is to determine the low frequency a00 overtones, in order to derive their single quantum frequency for assignments of combination bands. These revised frequencies are then used when determining the position of combination bands, rather than the original TD-B3LYP values. The band observed at 274 cm−1 is

0 assigned to the overtone of ν34. Assuming that the mode is harmonic, the frequency 0 −1 −1 of a the single quantum of ν34 is determined to be 137 cm , some 6 cm below the theoretically predicted value. The most appropriate assignment for the band

−1 at 421 cm is the overtone of ν33, representing a large amplitude out of plane ring distortion. Again, this implies that the single quantum ν33 has a frequency of ∼211 cm−1, which is signiﬁcantly larger than the theoretically predicted value of 198 cm−1.

We now turn to the other low frequency unassigned bands, at 215 cm−1, 356 cm−1 and 461 cm−1. The ﬁrst of these, at 215 cm−1 can only be assigned to

0 0 −1 the combination of ν36 and ν34, calculated to be 16 cm higher in frequency, at 231 −1 0 −1 cm . As the single quantum of ν34 was determined to be 137 cm , this infers that 0 −1 −1 ν36 has been over-predicted by 10 cm , and is in fact closer to 78 cm .

The band observed 356 cm−1 from the origin does not assign well. The

0 0 closest assignment is the combination of ν34 and ν33. These modes have experimentally derived single quanta of 137 cm−1 and 211 cm−1 respectively, which sum to 348 cm−1, indicating this combination is very anharmonic. The closest other

0 0 0 assignment however, is the combination of ν35 and ν36, where the frequency of ν35 has been determined to be 78 cm−1. This combination is predicted at 392 cm−1, 36 cm−1 higher in energy than the observed band, and is considered an unreasonable

56 −1 0 assignment. Finally, the band at 461 cm is assigned to the combination of ν34 and 0 0 −1 ν35, predicted at 451. As ν34 has been determined to be 137 cm , this implies that 0 −1 −1 the single quantum of ν35 is 324 cm , 10 cm higher than calculated.

The only band remaining unassigned in the spectrum resides at +873 cm−1 from the origin transition. This band does not assign to any binary combinations,

0 0 0 −1 −1 however it assigns well to the ν23 progression on ν33ν34 at 878 cm , only 5 cm higher than observed. This is the only band where multiple quanta of out of plane modes (ν33 and ν34) are coupled to an in plane mode (ν23), involving a ring contraction and elongation.

Overall there is an excellent agreement between the experimentally observed a0 modes and the scaled theoretical values, with a mean absolute deviation (MAD) of only 1.8 cm−1, see table 3.4.

Table 3.4: Comparison between single quanta observed/determined frequencies and TD-B3LYP theory, including the mean absolute deviation (MAD) for both the a0 and a00 modes. All units are in wavenumbers (cm−1)

0 00 a mode Observed TD-B3LYP ∆T −O a mode Observed TD-B3LYP ∆T −O

ν24 373 371 -2 ν36 88 78 10

ν23 522 522 0 ν35 324 314 -10

ν22 731 734 4 ν34 137 143 6

ν18 988 987 -1 ν33 210.5 198 -12.5

MAD 1.8 MAD 9.6

The a00 modes however are not very well predicted by this level of theory, with a MAD of 9.6 cm−1. The shortfalls of TD-B3LYP in calculating out-of-plane modes accurately were also observed by Krechkivska et al. during their studies on protonated naphthalene. They compared their level of theory with the RI-CC2 method used by Alata et al. for the same cation and found the latter signiﬁcantly superior in this case.[140, 141]

Reviewing the discrepancy between the out-of-plane combinations and the observed bands reveals that the a00 modes appear to be very anharmonic. This spectroscopic evidence of anharmonicity agrees with the implications of the theoretical

57 study, where different levels of theory and basis set size would converge to either planar and non-planar configurations, implying a very flat potential. The out-of-plane potential is likely anharmonic, or even of a pseudo-double minimum potential. In- troduction of a quartic term in the potential gives rise to alternating spacings in the vibrational coordinates. This is intimated in the experimental data, however there are not enough observed transitions to create an empirical fit. The construction of a double-minimum potential to fit anharmonic spectral features is explored further in chapter 8.

The ability of theory to successfully account for every peak within the hole burnt spectrum ensures that this spectrum has been correctly assigned as the syn conformer of the ortho complex.

3.4.2 The Anti Conformer

The TD-DFT calculations, and an evaluation of the Duschinsky matrix for the excited state of the anti-conformer, indicate C1 symmetry. With no symmetry, single quantum transitions of all in- and out-of-plane modes are allowed. The electronic

0 −1 origin for this transition, 00, is observed at 18200 cm , and this transition was used as the ‘burn’ transition for the depletion spectrum, shown in the lower trace of ﬁgure 3.7. The recorded spectrum spans ∼1000 cm−1 and the calculated and scaled frequencies are displayed in table 3.2.

We begin our assignment by assigning the single quantum fundamental transitions, identified in figure 3.7 in the red font. The first band observed in the spectrum is at 168 cm−1, which we assign to the single quantum out-of-plane CO

0 −1 −1 bend and CH2 rock, ν35, predicted at 173 cm . The peaks observed at 281 cm −1 0 0 and 326 cm are then assigned as the out-of-plane modes ν33 and ν34 respectively, calculated at 289 cm−1 and 339 cm−1. These (and all) assignment for the anti isomer are displayed in table 3.5.

The peak observed at 226 cm−1 must now be discussed for purposes of future assignment, however it cannot be accounted for by a single quantum transition.

58 wavenumber (cm-1 ) 18200 18400 18600 18800 19000 19200

H O H

?? ? ? ? 2 2 34 0 24 0 1 1 1 1 1 1 1 1 2 2 34 0 36 0 32 0 36 0 22 0 36 0 32 0 33 0 32 0 23 0

1 1 2 1 1 1 1 1 1 1 1 1 1 1 1 1 1 35 0 36 0 35 0 36 0 23 0 36 0 24 0 35 0 36 0 24 0 33 0 27 0 36 0 22 0 34 0 27 0 33 0

1 1 1 1 1 1 1 1 1 1 1 1 35 0 33 0 34 0 24 0 32 0 23 0 22 0 27 0 21 0 20 0 19 0 18 0

0 00

0 100 200 300 400 500 600 700 800 900 1000 relative wavenumber (cm-1 )

Figure 3.7: (Upper) REMPI spectrum of ortho-hydroxy-CHD radical. (Lower) Hole burning (depletion) spectrum measured while monitoring the REMPI signal 18200 cm˘1. The spectrum is assigned to the anti conformer. Vibrational assignments are indicated in red font for fundamental vibrations and black font for overtones and combination bands. The ‘?’ symbol denotes a depleted band that has no corresponding absorption line in REMPI experiments; any coincidence is merely an artefact of the compacted scale.

0 0 0 This peak can only be assigned to the combination of modes ν35 and ν36. As ν35 −1 0 has been assigned to the band at 168 cm , this infers that the frequency of ν36 is −1 −1 0 ∼58 cm , some 10 cm higher than theory. Although ν36 is not observed in single 0 00 1 quantum, it readily builds upon various other a and a transitions, including 350, 1 1 1 1 1 340, 320, 230, 220 and 270, discussed shortly.

A doublet is then observed at 368 cm−1 and 372 cm−1. The only single

0 quantum band calculated in this region is ν24, the lowest frequency near-in-plane

59 Table 3.5: Anti assignments and comparison to theory, including the mean absolute deviation (MAD) for the fundamental assignments. All units are in wavenumbers (cm˘1).

Observed Fundamental Assignments Combinations & Overtones Theory ∆T −O

168 ν35 173 5

226 ν35ν36 221 -5

281 ν33 289 7

326 ν34 339 13

368 ν24 380 12

372 ν34ν36 384 12

403 2ν35ν36 394 -9

442 ν32 445 3

467 ν23 466 -1

513 ν32ν36 500 -13

522 ν23ν36 525 3

533 ν22 532 -1

579 ν22ν36 591 12

599 ν24ν35ν36 594 -5

655 ν24ν33 649 -6

657 2ν34 652 -5

723 ν32ν33 723 0

733 ν27 735 2

740 ν21 740 0

742 2ν24 736 -6

793 ν27ν36 791 -2

860 ν34ν22 859 -1

879 2ν32 884 5

888 ν20 901 13

939 ν19 943 4

940 2ν23 934 -6

1003 ν18 992 -11

1017 ν27ν33 1014 -3

MAD (fundamentals) 6.0

C-C-O bend, predicted at 380 cm−1. Though we would usually assign this mode to the peak observed at 372 cm−1 based on proximity, the only other feasible as-

0 0 −1 signment for this region is the combination of ν34 and ν36, at 384 cm based on earlier assignments. To assign the peak at 368 cm−1 to this combination would be

0 unreasonable, and so we have assigned it to the single quantum ν24. This leaves the −1 0 0 peak at 372 cm to be assigned to the combination of ν34 and ν36.

60 −1 0 The medium intensity peak at 442 cm is assigned to ν32, predicted at 445 −1 0 0 0 0 cm . The modes ν35, ν34, ν33 and ν32 all contained a largest overlap with ground state modes of a00 character. Thus, the observation of these transitions in single quantum validate the theoretical calculation of an excited state with C1 symmetry.

−1 0 The following peak, observed at 467 cm is assigned to ν23, predicted at −1 0 −1 −1 466 cm . ν22 is then observed at 533 cm , similarly 1 cm away from theory. 0 0 −1 −1 Modes ν27 and ν21 are assigned to the peaks observed at 733 cm and 740 cm , based on their agreement with calculations of 735 cm−1 and 740 cm−1 respectively. The latter of these two bands appears to have a shoulder, at 742 cm−1, which can

0 −1 be assigned to the overtone of ν24, which should lie at 736 cm . This assignment is made based on the intensity of the fundamental, and by assuming the mode is anharmonic to account for the positive spacing between successive quanta.

Finally, three peaks, observed at 888 cm−1, 939 cm−1 and 1003 cm−1 are

0 0 0 assigned to the fundamental modes ν20, ν19 and ν18 respectively, based on their agreement with theory. The band at 939 cm−1 appears to have a shoulder, which

0 assigns well to the overtone of ν23.

We can now return to the remaining assignments of combination and overtones, keeping in mind that the ‘theory’ column for these assignments is based upon the observed single quantum frequencies, (unless they were not observed in single quantum, in which case the TD-B3LYP value is used). We have already observed

−1 −1 0 that the bands at 226 cm and 372 cm contain ν36 built onto the fundamental 0 0 0 modes ν35 and ν34, and so we begin by searching for progressions of ν36 built upon higher frequency modes. Indeed, the peaks at 513 cm−1 and 522 cm−1 are reasonable

0 0 0 candidates for the progression of ν36 built upon modes ν32 and ν23 respectively. We 0 0 0 0 also observe ν36 built upon modes ν22 and ν27. Finally, the combination of ν36 and 0 −1 ν27 is observed at 793 cm .

−1 0 The band at 403 cm assigns best to a second quantum of ν35 built onto 0 0 −1 the previously discussed combination band ν35ν36 at 226 cm . The C-C-O near-in- 0 0 0 plane bend, ν24 is observed to build on top of the combination band ν35ν36, at 599 cm−1. A peak and shoulder at 657 cm−1 and 655 cm−1 are then observed. These

61 0 0 0 are assigned as the overtone of ν34 and the combination of ν24 and ν33 based on comparison with their fundamentals, and to minimise the error of each assignment.

−1 0 The peak at 860 cm is attributed to the binary combination of ν22 and 0 −1 0 ν34, whose fundamentals sum to 859 cm . The overtone of ν34 is assigned to the peak at 879 cm−1, some 5 cm−1 red shifted from the harmonic overtone expected at 884 cm−1. Finally, the peak at 1017cm−1 is assigned to the combination of modes

0 0 ν33 built onto ν27.

The mean absolute diﬀerence for the fundamental assignments is 6 cm−1, including all in- and out-of-plane modes, which is much smaller than the MAD of 9.6 cm−1 observed for the syn rotamer. As the out-of-plane distortion is more prominent in the anti rotamer, this would result in deeper potential wells for the out-of-plane modes, generally resulting in a more harmonic behaviour. The maximum ∆T −O deviation is 13 cm−1, which is comparable to the ∼12.5 cm−1 maximum, observed for the syn rotamer. With all peaks accounted for, we believe that this spectrum can be successfully assigned as the anti-conformer of the ortho complex.

62 3.5 Conclusion

A REMPI scheme has been recorded for the hydrogen addition to the ortho position of phenol. Two rotamers were uncovered, where the hydroxyl group is situated syn

−1 and anti to the ortho CH2. The spectra span some ∼1000 cm and through the use of hole burning spectroscopy, was able to be separated into an isomeric dependence. The origin transition lies at 18200 cm−1 for the anti-conformer, and 18218 cm−1 for the syn conformer. All peaks recorded within the REMPI scan were able to be assigned to one of the two rotamers. In their ground state, both conformers are of Cs symmetry. Upon excitation, the anti-conformer undergoes a geometric change to C1, whilst the syn conformer retains an almost planar Cs nature and the spectra were assigned accordingly. This change in geometry and symmetry is validated by an evaluation of the Duschinsky matrices for each isomer’s excitation. The MAD for the anti fundamental assignments was 6.0 cm−1. For the syn rotamer, the a0 and a00 assignments were made with MADs of 1.8 and 9.6 cm−1 respectively. Whilst the relatively cheap TD-B3LYP/6-311+G(d,p) level of theory was successful in calculating the anti and in-plane modes for the syn isomer, a revised level of theory should be implemented for the out-of-plane modes of the syn isomer.

63 Chapter 4

The Methylated Allyl Chromophore

4.1 Introduction

The hydroxyl addition to isoprene can occur at one of four places as described in chapter 1. The preferential addition at the terminal sites results in the formation of two resonantly stabilised radicals. This resonance is represented by electron rearrangement within the allylic chromophore, where the radical site shifts between the β- and δ- position. Our laboratory had never before studied radicals with this most fundamental resonance. Consequently we chose a ‘ground up’ approach, see ﬁgure 4.1, beginning our investigation with a review of the well-studied allyl radical, whose photoexcitation pathways have recently been revisited. Sequential methylation, and ﬁnally hydroxylation, will yield a radical moiety identical to products of the direct hydroxyl addition to isoprene.

The allyl radical is the simplest π-conjugated resonantly stabilised radical and one of the best understood. Its relatively high stability has led to its observation as an intermediate in tropospheric processes,[142] the combustion mechanisms of hydrocarbons, and the formation of benzene and larger aromatics.[143, 144] It has been observed in propane-, butane- and acetylene-rich ﬂames and is involved

64 OH

2-methylallyl 1,2-dimethylallyl 4-hydroxyisoprenyl (2-ma) (1,2-dma) (4-OH)

allyl

1-methylallyl 1,1-dimethylallyl 1-hydroxyisoprenyl (1-ma) (1,1-dma) (1-OH)

Figure 4.1: Structures of the substituted allyl radicals investigated in our ‘ground up’ approach towards the hydroxy isoprenyl radicals in soot formation.[145, 146] The chemistry of small hydrocarbon radicals are also important in planetary atmospheres such as Saturn’s moon, Titan,[147] and interstellar clouds.[148] Its universal abundance has promoted it as a reference molecule for investigations of radical dynamics and theoretical benchmarking. [149]

˜ The allyl radical, as seen in figure 4.1, has C2v symmetry in its ground, X 2 ˜ A2 state. Excitation to the A state involves the two CH2 groups rotating out-of- plane to form a dihedral angle of ∼40◦. The CC bond length is also seen to increase from 1.3869 Å to ∼1.47 Å.[150, 151] These significant geometrical changes are attributed to the mixed excitation of this transition, whereby electrons are promoted from bonding π to non bonding (n) orbitals, and from non bonding to anti bonding ˜ 2 π* orbitals. The electronic excitation to the A B1 state was first reported in flash photolysis experiments by Currie and Ramsay, who measured a weak and diffuse set of bands over the range ∼370 - 410 nm and assigned the origin transition to 408.3 nm. [152] Later, Maier et al. remeasured the absorption profile of the allyl radical in an argon matrix at 10 K, identifying a structured absorption with a band origin at 408.5 nm.[153] These assignments were confirmed by cavity ring down measurements by Tonokura and Koshi, reporting the absorption cross section of this transition as (2.0 ± 0.4) × 10−19 cm2 molecule−1.[154] The broadened nature of these bands was originally attributed to predissociation or isomerization.

65 Higher excited states of the allyl radical were ﬁrst detected by Callaer and Lee, in the UV region of 260-220 nm, and were assigned as Rydberg transitions

2 to B1 states.[155] Hudgens and Dulcey expanded these studies, recording the ﬁrst REMPI schemes of the allyl radical.[156] Multiphoton excitation allowed them to

2 observe the vibrationless 3s A1 state, forbidden in 1-photon excitation from the 2 2 ground A2 state. Single photon excitation to vibrationally populated 3s A1 states however, was recorded later by Blush et al., who concluded that it was a result

2 of strong vibronic coupling with a nearby B1 state.[157] The most comprehensive study of the Rydberg states was conducted by later members of the same group. Gasser et al. remeasured the Rydberg transitions, and was able to (for the ﬁrst ˜ 2 ˜ 2 ˜ 2 ˜ time) energetically order them as B A1 (3s), C B2 (3py), D A1 (3pz), and E 2 B1 (3px).[158] They conﬁrmed the earlier studies of Blush et al., by measuring the depletion spectrum of the short lived E˜ state, whose large oscillator strength gives rise to the vibronic coupling. The 3d, and four ns states (n=4,6-8) were observed by Wu et al., and from the convergence of this series, determined an adiabatic ionisation energy of 8.138 ± 0.002 eV.[159] This value is consistent with observations made by Houle et al. and Fischer et al. who report ionisation energies of 8.13 ± 0.02 and 8.13 ± 0.01 eV from conventional and threshold photoelectron experiments respectively. [160, 161]

Increasing oscillator strengths were calculated for higher lying Rydberg transitions, however in experiments, REMPI signal on these transitions broadened and decreased rapidly after the 240 nm mark. This indicated a fast, non radiative decay after photoexcitation. As such, Fischer and Chen conducted picosecond time- resolved photoelectron experiments on the UV excitations of the allyl radical.[149] They found an initial internal conversion to the A˜ state occurred, within ∼20 ps. From here, one of two things can happen, the allyl radical can proceed via conical intersections to the ground state, which follow conrotatory or disrotatory pathways of the terminal CH2 groups (92%), or the radical can cyclise in a predominantly disrotatory fashion to form cyclopropyl (8%), whereafter it can ring open again to allyl on the ground state surface. These numerous photochemical deactivation pathways have led to a short calculated lifetime of the A˜ state of ∼0.6 ps. [151, 162] Following the production of hot ground state allyl radicals, unimolecular dissociation

66 of a H atom from allyl can lead to the co-products allene, propyne and to a lesser extent cyclopropene. Dissociation of hydrogen at the central carbon is the most likely outcome, creating the co-fragment allene, conﬁrmed by photofragment doppler studies of partially deuterated allyl. Methyl products have also been observed, with co-fragments acetylene and vinylidene; the former produced by as many as three pathways.[149, 151, 163, 164, 165, 166] Song et al. indicate that increasing excitation energies makes the formation of acetylene a signiﬁcant pathway.[167].

Castiglioni et al. made use of the primary photodissociation pathway, to be able to measure the A˜ state by conducting photo-fragment excitation (PHOFEX) spectroscopy using the hydrogen photoproduct. That is, they monitored hydrogen production as a product of excitation wavelength over the range 380 - 420 nm to produce an action spectrum of the A˜ state.[151]

By methylating the allyl radical sequentially, to form 1- or 2-methyl allyl (1- ma, 2-ma), we can understand the effect of the position of the methyl substituents on the excited state spectroscopy. The 1- and 2- methyl allyl radicals were first observed in the gas phase by Callaer and Lee. [155] They generated the respective radicals by flash photolysis of multiple precursors. The 1-ma radical was seen to absorb strongly in the range ∼226-238 nm, with 8 broad and diffuse bands identified. The 2-ma radical showed comparatively sharper absorptions, over the larger, ∼223-258 nm range, however toward the higher energy end, the spectrum became successively broader and more complex. The vertical and adiabatic ionisation potentials of the two radicals were measured via photoelectron spectroscopy by Schultz et al. and the results are summarised in table 4.1.[168] The effect of methylation on the adiabatic ionisation energies (I.E.’s) is considerable when compared to the value of 8.138 ± 0.002 eV obtained for the allyl radical. [159]

Table 4.1: Adiabatic and Vertical Ionisation Energies for the 1- and 2- methyl allyl radicals, adapted from reference [168].

1-methyl allyl 2-methyl allyl

Adiabatic I.E. 7.49 ± 0.02 eV 7.90 ± 0.02 eV Vertical I.E. 7.67 ± 0.02 eV 7.95 ± 0.02 eV

67 Hudgens and Dulcey reported the ﬁrst REMPI scheme for the 2-ma radical, and conﬁrmed the adiabatic IE value of 7.90 eV. [156] They observed that the origin

2 of the 3s A1 Rydberg state is red shifted by 22 nm compared to the allyl radical. C.-C. Chen et al. report an extensive study on the various 1+1 and 2+2 REMPI schemes of the 2-ma radical. They attribute the broadness of some peaks to arise due to intensity borrowing from a nearby electronic state of the same character.[169] Gasser et al. re-examined the UV absorption of the 2-ma radical and by the aid of selective deuteration and greater resolution, were able to more extensively assign vibrational spectra of the excited states. They utilised UV-UV depletion techniques to reveal strong, broad and continuous absorptions of short lived 3p Rydberg transitions between 41750 and 42450 cm−1, and infered a very short lifetime of these excited states. [170]

Ro¨der et al. built on earlier studies of the photodissociation dynamics of the 2-methyl allyl by Gasser et al. and Herterich et al. and provided a most comprehensive analysis. [170, 171, 172] They excited the 2-ma radical with 236 nm light, and noted that the radicals quickly populate all three of the 3p Rydberg states. These states rapidly depopulate to the 3s Rydberg state within the instrument response time of their femtosecond apparatus. Over the course of 100’s of fs, the 3s state will then decay to the nπ A˜ state. [171] The A˜ state then proceeds (via cycli- sation to the methyl cyclopropyl radical) to the ground state where by unimolecular dissociation of a C-H bond, forms products methylene cyclopropane + H. Alterna- tively, reaction pathways have been calculated indicating possible isomerisation to 1-ma, which then likely dissociates to form buta-1,3-diene. [173]

Herterich et al. determined that the lifetimes of the higher excited states of 2-ma were on average 30-50% shorter than those of the allyl radical. They attribute this to the increased density of states available due to the methyl substituent, referring to Fermi’s Golden rule. [172] As a result, one could predict an even shorter lifetime for higher excited states of dimethyl substituents.

Prompted by their earlier theoretical studies on the dissociation of 1-ma to likely form buta-1,3-diene, Gasser et al. ﬁrst reported the A˜ ← X˜ transitions of the 1-methyl allyl radical through REMPI and PHOFEX spectroscopy.[173, 174]

68 Three bands were observed from the REMPI experiments, and the first of these at 23990 cm−1 is assigned to the origin of the (Z) conformer, and the second peak at 24090 cm−1 to the origin of the (E) conformer. Further peaks, though not assigned, are identified through PHOFEX spectroscopy of the hydrogen photoproduct, and they attributed a short lifetime to the A˜ state responsible for the truncated REMPI spectrum. They observed that following photoexcitation, nonradiative decay to the ground state retains sufficient internal energy for fast E/Z isomerisation. Both the E and Z isomers then favourably dissociate to form H atom and buta-1,3-diene products, confirmed by deuterated studies. These studies by Gasser on the 1-ma and 2-ma radicals indicate that the methyl rotor is not a structural motif responsible for non-statistical behaviour in deactivation pathways, as exhibited by several alkyl radicals, such as ethyl radical.[170, 174]

Further methylation, yielding 1,1- and 1,2- methyl allyl, is not only a pro- gressive step towards the end goal, (see ﬁgure 4.1), but also represents two of the possible isomers formed from the hydrogen addition to isoprene. The spectroscopy of these dimethyl allyl radicals has not yet been investigated in solution or gas phase. The former, also known as ‘prenyl’, is a well known and versatile compound in organic and biological chemistry. In synthesis it can be utilised as a protecting group, of both sulfonamides, and the hydroxy group.[175, 176] The isoprenoid lipids farsenyl and geranyl-geranyl utilise their prenyl functional group to modify eukary- otic proteins. ‘Prenylation’ involves the addition of these hydrophobic lipids to or near to the C-terminus of proteins. This addition aﬀects the activity of a variety of protein-protein interactions, crucial in biological regulation. [177, 178]

69 4.2 Calibration Studies and Method of Approach

4.2.1 Radical Precursor Selection

Appropriate choice of a precursor is vital to ensure that the required product is formed in reasonable concentrations to study within the highly energetic discharge environment. Products can be formed in a discharge through either the addition of molecular or atomic species, or through the dissociation of bonds within the precursor. Products formed through addition reactions are able to form a multitude of isomeric products. These competing pathways reduce the population of the sought after product. Though these isomers can often be distinguished through hole burning spectroscopy, as seen in chapter 3, this introduces an unnecessary complication in the identiﬁcation and assignment.

Rather, we begin with a commercially available (or synthetic) precursor, such that dissociation of the weakest bond produces a single isomer. In this way, we are able to investigate the proposed products one at a time and identify their spectral features. We can then return to the addition pathway, produce all of the products available, and determine (by comparison with our earlier experimental reference data) new species.

4.2.2 The Allyl Radical

The detection of the allyl radical represents the starting point for this calibration study. Gasser et al. measured the 3s and 3p Rydberg states of the allyl radical over the range 39300-45000 cm−1 by 1+1 and 2+2 REMPI spectroscopy, combined with UV/UV depletion.[158] They created the allyl radical by either ﬂash pyrolysis of 1,5-hexadiene, or by the photolysis of allyl iodide. As this is simply a calibration experiment, we attempted to recreate the conditions necessary to produce the allyl radical in our setup and measure part of the 1+1 REMPI spectrum, without further investigation.

70 We chose to use 1,5-hexadiene as a precursor, as dissociation favours the production of two resonantly stabilised allyl radicals, see ﬁgure 4.2. It also has a substantially larger vapour pressure than allyl iodide, making for a richer molecular beam. 2

Figure 4.2: The central C-C bond break of 1,5-hexadiene, induced via discharge, to form two allyl radicals.

The experimental apparatus has been previously described, and here we only mention details explicit to this experiment. Argon (6.5 bar) was bubbled through a sample of 1,5-hexadiene (Aldrich, 97%, used without further puriﬁcation). The sample was expanded into a diﬀerentially-pumped vacuum chamber through a pulsed discharge nozzle. As the sample expanded, a discharge of 2090 V across 15 kΩ ballast resistance was struck for 150 µs.

The gas expansion was skimmed by a 2 mm skimmer, and the cold molecular beam was orthogonally intersected by laser pulses to excite and then ionise the radical. The output of a SIRAH COBRA-STRETCH Nd:YAG pumped dye laser, circulating the Coumarin 480 dye, was frequency doubled, providing excitation and ionisation photons for the 1+1 REMPI scheme over the wavelength 41500 - 42300 cm−1. The combined energy of the two photons was 10.29 - 10.49 eV, well exceeding the ionisation energy of 8.13 eV. The spectrum was taken by recording the m/z 41 channel as a function of the excitation laser wavelength. The experiment was run at 10Hz.

−1 2 The first 700 cm of the C( B2), 3py Rydberg transition of the allyl radical was recorded and is shown in figure 4.3. The reference spectrum, adapted from Gasser, is depicted in the same figure.[158] The spectra exhibit similar intensity patterns and resolution. The origin transition and the vibrational modes ν7 and ν9 are clearly observed.

The most recent observation of the D1 ← D0 transition of the allyl radical was made by Castiglioni et al., where PHOFEX spectroscopy was conducted on the hydrogen atom product, formed from the dissociation of allyl into allene and H

71 1 1 2 0 C( B2) 00 70 90

this work

41600 41700 41800 41900 42000 42100 42200

1 1 2 0 C( B2) 00 70 90 adapted from Gasser et al.

40000 41000 42000 43000 44000 Wavenumbers (cm -1 )

2 Figure 4.3: The C( B2), 3py Rydberg transition of the allyl radical, recorded by 1+1 REMPI. Inset adapted from Gasser et al. [158] atom following photoexcitation.[151] This experiment was brieﬂy attempted within our laboratory, however the discharge created a background signal for the hydrogen atom (m/z 1) more than an order of magnitude larger than the signal observed for the

2 C( B2) Rydberg transition. We considered using photolysis rather than discharge to generate allyl radical, however even with the discharge oﬀ, there was still too much background signal on the m/z 1 channel to conduct PHOFEX spectroscopy on the hydrogen atom.

72 4.2.3 The Methyl Allyl Radical

Following observation of the allyl radical, we conducted experiments to observe the Rydberg transition of the 2-methyl allyl radical. The higher excited states of both the 1- and 2-ma radicals have been reported previously, however it is the latter that has been studied in more detail with more clearly defined peak positions and assignments. Gasser et al. conducted REMPI experiments on the 2-methyl allyl radical over the range 38200-43300 cm−1, and reported observation of the various vibronic features of the 3s Rydberg state by both 1+1 and 2+2 REMPI.[170] UV- UV depletion techniques were used to observe the broad and featureless 3p Rydberg states. The result of ab initio calculations on the 3p states indicated that only the 3px and 3py states had significant oscillator strength. The broad and diffuse character of these higher lying Rydberg transitions was justified by Ro¨der et al. who remarked that they would rapidly decay to the 3s state (within femtoseconds),[171] making observation by nanosecond REMPI techniques near impossible.

Gasser et al. generated the 2-ma radicals by photolysis of a series of purchased and synthesised precursors; 3-iodo-2-methylprop-1-ene, 3-bromo-2-methylprop- 1-ene, and 3-methylbut-3-enyl nitrite. [170] They had previously used jet ﬂash pyrolysis of the precursors 2-methylallyl-iodide and 3-methylbut-3-enyl nitrite when studying the dissociation dynamics of this radical, and found that the nitrate containing precursor generated more signal, by an order of magnitude. [173]

For our experiments, β-methyl allyl chloride (3-chloro-2-methyl-1-propene) was chosen as a suitable precursor, as it has a vapor pressure ∼4 times that of 2- methylallyl-iodide, and so should provide a richer molecular beam, and hence signal more comparable to the nitrogen containing compounds employed by Gasser. The dissociation of β-methyl allyl chloride by discharge is shown in ﬁgure 4.4.

+ Cl Cl

Figure 4.4: The formation of the 2-methyl allyl radical, by the discharge of β-methyl allyl chloride, cleaving the C-Cl bond.

73 The experimental setup reﬂects that of the allyl radical, and only details explicit to this experiment are mentioned. The pressure of argon was increased to 7 bar and was bubbled through a sample of 3-chloro-2-methyl-1-propene, (Alrdich, 90%, used without further puriﬁcation). The discharge conditions were also adjusted, and a voltage of 1890 V, across 15 kΩ ballast resistance was applied to the expanding sample for 150 µs.

The output of a SIRAH COBRA-STRETCH Nd:YAG dye laser, circulating the Coumarin 503 dye, was frequency doubled, providing photons in the range 38700- 41400 cm−1 to be used for part of the 1+1 REMPI scheme of the 3s Rydberg transition. The spectrum was taken by recording the m/z 55 peak as a function of excitation wavelength. The origin of this transition, at 38342 cm−1, is only accessible via a 2-photon excitation. As such, our calibration records the most intense features available to 1-photon excitation, beginning with ν25. The results are displayed in ﬁgure 4.5, alongside the reference adapted from Gasser et al. [170]

The spectra of the discharged 3-chloro-2-methyl-1-propene can be clearly identified as the 2-ma radical by comparison with Gasser et al. No additional features are observed, and the level of noise has been reduced, which helps with peak position identification around the 40500-41000cm−1 region. The origin of the 3s Rydberg state was also briefly observed (not shown) via 2+2 REMPI. This was conducted by removal of the doubling crystal and scanning over the region around 38342 cm−1, with a 300mm focusing lens. Our signal to noise ratio was much higher than that recorded by Gasser et al. [170]

74 x 0.5

this work

1 1 1 25 0 18 0 19 0

adapted from Gasser et al.

1 B 250

39000 39500 40000 40500 41000 Wavenumbers (cm -1 )

2 Figure 4.5: The 2-methyl allyl B(1 A1) state, (3s Rydberg transition), observed by 1+1 REMPI. Inset adapted from Gasser et al. [170]

4.2.4 Implications

The various photo-excitation pathways of the allyl and the 1- and 2-methyl allyl radicals provide a foundation of what to expect upon further methylation.

The D1 ← D0 transition of the allyl radical is not amenable to nanosecond REMPI experiments in the gas phase, and required the use of PHOFEX spectroscopy for detection. [151] The short lifetime of the upper state was attributed to conical intersections to the ground state, which were accessed by the con- and dis-rotary motions of the terminal CH2 groups. The hot ground state allyl radical promptly dissociates to preferentially form allene and hydrogen atom products. The D1 ← D0

75 transition of the 2-methyl allyl radical has not yet been observed by any technique, though it potentially may be detected in the same way as the allyl radical, either by PHOFEX spectroscopy, or in a matrix. [153] The D1 ← D0 transition of the 1-methyl allyl radical however, has been observed in the gas phase by 1+1 REMPI. [174] Gasser et al. measured 3 peaks via REMPI spectroscopy, and attribute the ﬁrst two of these to the origin transitions of the (Z) and (E) stereo-isomers, validated by ab initio. calculations. Further vibrational structure was uncovered by PHOFEX spectroscopy, again on a hydrogen atom photo-product, though this remains unassigned. The truncated REMPI spectrum is attributed to the short lifetime of this state.

Various higher excited states have been reported for each of the allyl, 1-ma and 2-ma radicals. Allyl has naturally had the most attention, and multiple Rydberg transitions have been identiﬁed and energetically ordered by 1+1 and 2+2 REMPI. The B˜, C˜, and D˜ states are observed to have well deﬁned vibrational structure. The E˜ state was observed through depletion spectroscopy, and becomes broad and featureless towards higher energy.[158] This is consistent with the matrix studies of Maier et al. who observed a strong, broad absorption beginning at ∼230 nm and peaking at 213 nm.[153]

The 3s Rydberg transition of the 2-ma radical can be observed through a combination 1+1 and 2+2 spectroscopy. [170] The 3p Rydberg states can also be observed by 1+1 REMPI; it is broad and continuous, and the peak intensity drops quickly toward higher energy. A UV-UV depletion experiment reveals more of the 3p Rydberg states over the range 41750 cm−1 to 42450 cm−1, two of which are calculated to have a high oscillator strength. Again, a short lifetime is attributed to the higher excited states to account for the broad and diffuse spectral features. The UV spectrum of the 1-ma radical was reported by Callaer and Lee, beginning at 42039 cm−1. Eight bands are identified over the following ∼2000 cm−1. They are described as broad and diffuse, and identification was impeded by a competing continuous absorption. [155]

In our calibration experiments, the required carbon-carbon and carbon- halogen bonds were able to be eﬀectively broken in the discharge to generate the

76 required allyl and 2-methyl allyl radicals. The Rydberg states of these radicals were observed by the REMPI technique, with similar resolution and signal to noise ratios as the spectra previously reported.

Of the three radicals, only the 1-methyl allyl radical has an A˜ state that can be observed by REMPI spectroscopy. If substitution at this position is responsible ˜ for the increased lifetime of the A state, then we may be able to observe the D1 ←

D0 transition for radicals with similar structural motifs. With regard to the higher excited states, methylation at the 2- position was observed to red shift the origin of the 3s Rydberg state by 22 nm compared to the unsubstituted allyl radical. Further, all three radicals were observed to have continuous, broad absorptions towards higher energy Rydberg transitions. This spectral feature could well be present for the dimethyl allyl radicals. Each of the three radicals are also prone to dissociation on the hot ground state following photoexcitation, predominantly by dissociating a hydrogen atom. This evaluation of the allyl and methyl allyl radicals concludes the calibration part of this chapter, and with this information in mind, we proceed to a second methyl substitution and the investigation of the 1,1- and 1,2- dimethyl allyl radicals.

77 4.3 Investigation of The 1,1 - Dimethyl Allyl Rad- ical

By continuing with our ground up approach, the next class of radicals to investigate are the 1,1- and 1,2-dimethyl allyl radicals (1,1-dma and 1,2-dma), shown earlier in ﬁgure 4.1. These radicals are identical with two of the products formed from the H-addition to isoprene. Addition of the hydrogen atom to isoprene can occur at one of four places across the double bonds. Just like addition of OH to isoprene, we expect that addition preferentially occurs at the terminal sites, forming resonance stabilised radical intermediates 1,1-dma and 1,2-dma. The chromophores of these radicals are identical to those of their corresponding radicals formed from the OH addition to isoprene.

The only diﬀerence between the di-methyl allyl radicals, and the products of OH addition to isoprene, are a substitution of hydrogen atom for hydroxyl. By a comparison with similarly substituted species; the cyclohexadienyl radical (C6H7), formed from H-addition to benzene has an electronic D1 ← D0 origin at 18191.5 −1 cm .[179] In chapter 3 we measured the D1 ← D0 origin of the two rotomers formed from H-addition to phenol, which corresponded to energies of 18200 cm−1 and 18218 cm−1 for the anti and syn rotomers respectively. The shift of only 8.5 cm−1 and 26.5 cm−1 suggests that the H- and OH- addition to isoprene products may have similar D1 ← D0 electronic origins.

The electonic spectroscopy of the 1,1-dma and 1,2-dma radicals will therefore be investigated, followed by H addition to isoprene experiments, described in section 4.5, which could reveal further insight and potentially new species. This same experimental route will then be followed for the OH addition to isoprene, in chapter 5. The following section reports our ﬁndings on the 1,1-dimethyl allyl racial. Section 4.4 will discuss the electronic spectroscopy of the 1,2-dimethyl allyl radical.

78 4.3.1 Experimental and Methods of Generation

As discussed previously, our choice of precursor is vital to ensure that formation of the required conformer occurs within the discharge. The precursors selected to produce the 1,1-dma radical are 3,3-dimethylallyl bromide and geraniol, and the expected discharge products are shown in ﬁgure 4.6.

+ Br Br

3,3-dimethyl allyl bromide 1,1-dimethyl allyl

OH OH 4-OH +

geraniol 1,1-dma

Figure 4.6: The formation of the 1,1 dimethyl allyl radical as a discharge product of 3,3-dimehthyl allyl bromide and geraniol. The co-fragment of the geraniol dissociation is a hydroxy isoprenyl radical, 4-OH

Dissociation of the 3,3-dimethylallyl bromide is expected to yield the required 1,1-dma and bromine atom as a co-fragment. We have found in the past that optimal production of the halogen leaving group usually requires very harsh discharge conditions, (high voltage, low resistance, long duration), to ensure that all carbon-halogen bonds are broken. This does not necessarily correspond to optimal production of the required radical, where populations decrease if the discharge is harsh. It does however, provide a good starting point for generation of radicals within the discharge, as REMPI schemes are well known for atomic halogens.

The precursor geraniol is an ideal candidate. If we are to assume the molecule fragments in the way drawn in ﬁgure 4.6, then not only do we produce the 1,1-dma radical, but the co-fragment is a postulated product formed from the

79 OH addition to isoprene. Hence, we will begin with the precursor 3,3-dimethylallyl bromide, by creating optimal conditions for bromine atom production within the discharge, and then varying the conditions whilst searching for 1,1-dma. Following this experiment, we change to geraniol as a precursor, ﬁnd the optimal conditions for 1,1-dma and then look for the m/z 85 cofragment, the 4-OH radial.

The experimental details and apparatus have been previously described in chapter 2, and here only details explicit to this experiment are listed. The samples 3,3-dimethylallyl bromide and geraniol were both used as precursors for the formation of the 1,1-dma radical. The experimental details will be listen in full for the case of 3,3-dimethylallylbromide, as conditions were near equivalent for the two precursors. The only diﬀerence is that geraniol has a lower vapor pressure at room temperature, requiring the sample container to be heated up to 100◦C, with the nozzle heated to 110◦C to prevent condensation.

Argon, (6-7 bar), was passed through the sample container to collect the vapors of 3,3-dimethylallyl bromide. As the mixture was supersonically pulsed into the vacuum chamber, a discharge of 2090 V, with a 25 kΩ ballast resistor, was struck for 260 µs to coincide with the later part of the expanding gas. The discharged gas was skimmed by a 2 mm skimmer, to form a cold molecular beam, and proceeded into a second chamber. Guided by the red shift of the 2-ma radical’s B˜ state compared to the allyl radical, a 1+1 REMPI experiment was conducted over the region 247 nm - 280 nm. The excitation and ionisation photons were produced from the doubled output of a SIRAH COBRA-STRETCH Nd:YAG dye laser, circulating the Coumarin 503 and 540a dyes. The excitation photons were scanned in wavelength and the signal response was recorded for the m/z 69 channel.

The D0 ← D1 transition of the 1,1-dma radical was probed by 1+2 and 1+10 REMPI. For the 1+2 scheme, a 250 mm focusing lens was used, focussed on the molecular beam. The dye was changed to Exalite 428 to provide excitation photons for both schemes (and ionisation photons for the former). The ionisation wavelength in the 1+10 scheme was 266 nm, the fourth harmonic of a BRILLIANT B, Nd:YAG type laser.

80 The lifetime of the A˜ state was recorded by ﬁxing the timing of the 266 nm ionisation laser, and temporally scanning the excitation laser pulse, set to the origin transition of the A˜ state. The lifetime is recorded as the convolution between the two gaussian laser pulses, and an exponential decay function, which is used to calculate the lifetime of the excited state.

The ionisation potential of the 1,1-dma radical was determined by fixing the excitation on the first band in the spectrum, and then scanning the ionising laser in wavelength. The ionising wavelength of ∼300 nm was determined to photodissociate the parent, providing fluctuating background signal if scanned in frequency. To circumvent the large background signal fluctuations, we stepped the wavelength of the laser in 0.05 nm steps, and recorded the average signal at each point twice, once with the excitation laser providing resonant ionisation, and once without. The difference between these two traces was used to generate a photo ionisation efficiency (PIE) curve.

81 4.3.2 Theory

The ground, D0 state of the 1,1-dma radical was calculated to be of either C1 or Cs symmetry depending upon the treatment of the methyl rotors. Calculations were run with GAUSSIAN16 [135] at the M06-2X and B3LYP levels of theory, and the structural details are displayed in the top half of ﬁgure 4.7. Whilst the carbon framework of the molecule remains planar upon changes to the theoretical method used, the torsional angles of the methyl groups vary by restrictions on the symmetry of the molecule and the convergence criterion. This was also seen to be the case for the 2-methylallyl radical, where the authors report both calculated symmetries for the electronic states.[170]

1.095

1.086 1.095

1.496 115.95 123.48 1.080 122.73 116.80 128.09 1.391 1.378 1.095 1.079 1.497 120.47 1.095 120.58 1.086 112.05 115.60 116.31 1.089

D Ground State 0

1.101

46.53 1.093 1.095 154.11 1.493 1.086 121.01 162.47 119.26 120.06 118.00 123.70

1.407 1.094 1.087 1.498 1.429 120.70 1.100 118.37 1.095 111.25 116.41 119.87 1.094

side-on profile top-down

D Excited State 1

Figure 4.7: Structural details of the D0 and D1 states of the 1,1-dimethylallyl radical

82 The ﬁrst excited, D1 state of the 1,1-dma radical was calculated at various TD-DFT and ab initio levels of theory over consecutive years. The only method that converged the molecule to a non-transition state structure was the TD-DFT

M06-2X method, and the geometric details of the D1 state are shown in the bottom half of ﬁgure 4.7. The TD-DFT calculations in this and the following section 4.4 were completed in a collaboration with a fellow PhD candidate in computational chemistry, now Dr. Miranda Shaw. All other methods yielded multiple (2-4) imaginary frequencies, which involved the methyl rotor torsional modes, and an out of plane CH2 twist. A similar CH2 twist, (in both con- and dis-rotary fashions), was seen to be responsible for the short lifetime of the allyl radical A˜ state, as it coupled this state of the molecule to the ground state via conical intersections.[162] Further- more, this same motion was calculated to be the largest geometric change in the excitation of the 1-ma conformers, which likely lead to their short A˜ state lifetimes. [174]

Large distortions of geometry are calculated upon excitation. The carbon framework of the molecule bends out of plane by 17.53◦ to the CH carbon, and

◦ a further 8.36 to the terminal CH2 carbon. The CH2 group also rotates in a conrotatory fashion, to form a dihedral angle of 46.53 degrees with the methyl carbons. This is a similar rotation to the CH2 group of the allyl radical upon excitation, where a ∼40◦ dihedral angle is formed.[151] The two C-C bonds within in the chromophore increase from 1.391 Å and 1.378 Å, to 1.429 Å and 1.407 Å respectively, which indicates the promotion of electrons within the chromophore from bonding to non bonding, and/or non bonding to anti bonding orbitals. Finally, the two CH3 groups are calculated to rotate. The angle of rotation depends upon the reference point in the respective ground state chosen, as the angle varies if symmetry is not restricted. These methyl groups are calculated to rotate by 15.87◦ for the top

◦ rotor and 15.64 for the side rotor compared to a ground state with a Cs plane of symmetry.

The ﬁrst excited state is calculated, at the time-dependent M06-2X/6- 311+G(d,p) level of theory, to have a ZPE corrected adiabatic transition energy of 25091 cm−1. As a reference point, the A˜ state of the allyl radical was calculated at

83 the same level of theory to have an origin at 25121 cm−1, which is 629 cm−1 higher than the observed value of 24492 cm−1 (408.3 nm). As a reminder, the A˜ states of the 1-ma radicals were observed at 23980 cm−1 and 24080 cm−1 for the (Z) and (E) conformers respectively.

As convergence of the D2 state of the 1,1-dma radical has yet to be accom- plished, an approximate value for the D2 ← D0 transition energy was calculated, again at the TD-DFT M06-2X/6-311+G(d,p) level of theory. This was done by adding the diﬀerence of the D2 and D1 vertical energies to the calculated transition of the D1 state. This transition was calculated have an energy of ∼249.7 nm. As frequencies could not be calculated, (due to failed convergence), this energy has not been corrected by the ZPE of the ground and excited states.

The full set of calculated harmonic frequencies for the A˜ state of the 1,1- dma radical are displayed in table 4.2 below. These calculations were performed at the same TD-DFT M06-2X/6-311+G(d,p) level of theory.

Table 4.2: List of frequencies for the 1,1-dimethylallyl radical, calculated at the M06-2X/6-

311+G(d,p) level of theory. Modes ν34 and ν35, indicated with an asterisk (*), are the two methyl torsion modes and are treated separately below.

Mode # Frequency Mode # Frequency Mode # Frequency Mode # Frequency

1 3229.8 10 2436.8 19 1293.2 28 612.9 2 3156.8 11 1510.9 20 1223.6 29 535.5 3 3118.8 12 1484.5 21 1080.0 30 440.7 4 3115.1 13 1468.0 22 1056.9 31 370.5 5 3080.2 14 1455.1 23 999.5 32 275.8 6 3071.1 15 1430.1 24 961.4 33 257.3 7 3059.1 16 1410.3 25 940.8 34 222.3* 8 3001.9 17 1398.7 26 872.1 35 175.7* 9 2991.7 18 1356.3 27 774.1 36 137.5

The anharmonic frequencies were also calculated, though are not displayed as they included one imaginary frequency. The calculation of an imaginary anharmonic frequency does not imply that this is a transition state; rather that the modes may be subject to large anharmonicity and hence deviation is expected from the calculated harmonic values.

84 4.3.3 Results and Discussion

The D1 ← D0 transition of 1,1-dma

The D1 ← D0 transition of the 1,1-dimethyl allyl radical was recorded by monitoring the m/z 69 channel as a function of excitation wavelength, whilst conducting a 1+10 REMPI experiment over the region of 23900 cm−1 to 24400 cm−1, with the results displayed in ﬁgure 4.8. All of the spectral features were observed whether the precursor 3,3-dimethylallyl bromide or geraniol was used. Unlike the 1-ma radical, there are no geometric isomers for the 1,1-dma, and so all of the peaks are attributed to the one spectral carrier.

0 00

hot bands

Figure 4.8: REMPI spectrum of the D1 state of the 1,1-dimethylallyl radical

The spectrum spans some ∼300 cm−1, with more than 10 clearly deﬁned peaks present within this region. After ∼24270 cm−1, the spectrum is abruptly truncated, and no further peaks were identiﬁed over the following 500 cm−1. The intensities of the peaks below 24058 cm−1 varied the most between scans and chang- ing discharge conditions, and were labelled as hot bands. As such, the peak at 24058

85 cm−1 was assigned as the origin transition, and the spectrum spans ∼220 cm−1 from this point.

The spectrum is similar in nature to that observed for the 1-ma radicals.[174] There, only 3 peaks were identified by REMPI experiments, after which point the short lifetime of the excited state meant that the higher energy modes were unable to be observed by REMPI spectroscopy. Gasser et al. were able to measure more of the vibrational spectrum through PHOFEX spectroscopy of a hydrogen atom photoproduct. This experiment was briefly attempted for the 1,1-dma radical, however the discharge required to dissociate the radicals also produced a fluctuating background of hydrogen atoms with two orders of magnitude more signal than the 1,1-dma radical. Photolysis was suggested, however even without the discharge, the number of nascent hydrogen atoms meant that PHOFEX spectroscopy was not an option within our instrument.

Referring to table 4.2, only ν36 - ν34, and tentatively ν33, are calculated to appear in the first 220 cm−1 above the origin. The fundamentals, combinations and overtones of these modes are unable to account for the number of spectral features observed in this small range. Recently our group has measured the spectrum of the ortho-methyl-cyclohexadienyl species formed from the hydrogen atom addition to toluene. [133] The origin region similarly contains a multitude of peaks which could not be assigned as vibrational modes. The peaks are assigned as torsional progressions of the methyl rotor, coupled to the electronic transition, and the irreg- ular spectral spacings and intensities are similar to the spectrum observed for the 1,1-dma radical. This assignment was confirmed by a series of partially/completely deuterated experiments. This deuterated study also confirmed the origin of the hydrogen atom as exclusively the water vapor, rather than in combination of the toluene discharge products. The completely deuterated toluene (C7D8) showed no sign of D addition when H2O was seeded into the molecular beam, and the +D radical was only formed when the H2O was substituted for D2O. Likewise in the toluene + D2O experiments, only D addition was observed.

The scaled displacement vectors for the ν36 - ν31 modes are shown in ﬁgure

4.9. The lowest frequency mode, ν36, describes the out-of-plane CH2 twist, and the

86 displacement of the CH and CH2 carbons out of plane. This mode encompasses the geometrical distortions of the D1 ← D0 transition and will be observed with signiﬁcant intensity. Modes ν35 and ν34 represent the side and top methyl rotations respectively. Mode ν33 contains some methyl rotation, coupled with a framework twist and CH2 rotation.

36 35 34

33 32 31

Figure 4.9: The scaled displacement vectors for the lowest frequency modes of the 1,1-dimethylallyl radical

The two methyl rotors coupling to the electronic transition could well account for the numerous peaks in the origin region of the spectrum, and so a theoretical investigation was undertaken. The methyl rotor formalisms have been described previously, and the reader is referred to an excellent review by Spangler.[180] Here we brieﬂy describe the treatment of the rotors and the variables used to ﬁt the spectrum.

The energy levels of the methyl rotors in the 1,1-dma radical lie between the limiting cases of a free and an internally rigid rotor. For a free rotor, the internal rotation follows the formalisms of a particle in a ring. Here, the eigenvalues are given by E = m2F, where F is the internal rotation constant , and the energy levels are given by the quantum number m, where m = 0, ±1, ±2, ... Here the sign of m relates to the degenerate rotations of the methyl group, in clockwise and anticlockwise directions. As the rotor is part of the larger, 1,1-dma radical, a

87 hindering potential exists. As the three hydrogens within the rotor are equivalent, the potential is periodic, and as the 1,1-dma radical is not symmetric in its excited state, there are three minima. The potential, V , is described by a cosine expansion;

V V V V = 3 [1 − cos(3φ)] + 6 [1 − cos(6φ)] + 9 [1 − cos(9φ)] 2 2 2

The V3 term here represents the number of minima, and the depth of the potential. The higher order terms help to shape the well, and only the V6 term is required here for an accurate ﬁt. Upon excitation, the internal angle of rotation of the methyl group is represented by φ. If there is no phase change upon excitation, then the potential V = 0, and likewise a maximal potential arises for phase changes near 60◦.

In constructing a V3 potential by perturbing the particle-in-a-ring basis set, this perturbation has the effect of coupling free rotor wavefunctions that differ in quantum number by 3, with a coupling term of V3/2.[181] The energy levels are defined as having symmetries of a and e type, where m = ±3n,(n is an integer), have a symmetry and m = 3n ± 1 have e symmetry. A lobe representation of the vibrational wavefunctions of the a and e type symmetries are described below in figure 4.10.

This ﬁgure can help to clarify selection rules which govern the methyl rotor transitions. The three ovals represent the hydrogens within the methyl rotor, and are shaded in red or grey to represent the sign of the wavefunction. Here only a → a, and e → e type transitions are allowed, governed by the overlap of the two states. The a levels are all rotationally symmetric, and are divided into the subclasses of a1 and a2 depending on their symmetry/asymmetry respectively about the axis described by the dotted lines in ﬁgure 4.10.

The 1,1-dma radical has two methyl rotors, both of which lie near to one another and are likely to be coupled. We have not yet accounted for this coupling, and the two methyl rotors have been ﬁt independently of one another to the spectrum. The internal rotation constant is inversely related to the reduced mass of the

88 0a 1 1e

3a 2 2e

Figure 4.10: Lobe representations of vibrational wavefunctions for the methyl rotor molecule and the methyl rotor. In the limit where the mass of the molecule is much greater than the mass of the methyl rotor, this reduced mass tends toward that of methane. As a result, the internal rotation constant F tends towards 5.4 cm−1. For

−1 the 1,1-dma radical a value of F = 5.5 cm yielded the best fit. The V3, V6 and φ parameter space was also explored to find the best fit with experiment. The same ground state fit was used for each rotor, which was determined by the hot band transitions.

Two different excited state potentials were fit for the two rotors. The potentials, along with the associated energy levels are displayed on the right hand sides of figures 4.11 and 4.12. To account for the multitude of observed lines, one rotor was fit with a low potential barrier, of only 78 cm−1, which helped to account for the lower frequency structure. Here the potential was fit with the parameters

0 −1 0 −1 ◦ V3 = 80 cm , V6 = -35 cm and a phase shift of 69 . A second rotor was ﬁt with a larger barrier, of 248 cm−1 which was used to ﬁt the latter part of the observed

0 −1 0 −1 ◦ spectrum. The terms V3 = 247 cm , V6 = -31 cm and a phase shift of 69 were used for this ﬁt. The same ground state was used for both rotors, where the

0 −1 0 −1 potential was ﬁt with V3 = 187 cm and V6 = 34 cm .

89 The ab initio tortional angle changes for the two rotors were 15.87◦ for the top rotor and 15.64◦ for the side rotor. Using these ab initio numbers as ﬁtting parameters did not represent the observed intensities by any measure. Rather, an angle of 69◦ was required to produce similar intensities to the observed spectrum. The intensity pattern produced by phase change of 69◦ is equivalent to one by 51◦ which could also be the case. The calculated values of ∼16◦ were of course dependent upon a ground state geometry of Cs symmetry, and would be larger if the molecule had not been restricted, accounting for part of this discrepancy.

90 300 Excited State

250

6a 1'

)

-1 6a ' 200 2

5e' 150

4e' 100 3a 1'

3a 2' Relative Energy (cm Energy Relative 50 2e' 1e'

0 0a 1'

0 50 100 150 200 250 300 350

’ Torsional Angle

2 300

91 Ground State

2e’’ 2e’ 2e’’ 5e"

0a!” 6a 0a!”

1e” 5e’ 1e”

2e’’ 1e’ 2e’’

) 4e"

-1 1e” 4e’ 1e” 200 3a "

’ 1

0a!” 0a!” 0a!’ & 1e” & 1e” 1e’ 3a 2" 2e"

1e’’ 2e’ 1e’’ 100

0a!” 0a!” 3a

0a 1"

Relative Energy (cm Energy Relative & 1e"

0 50 100 150 200 250 300 350 Wavenumbers realtive to origin (cm-1 ) Torsional Angle

Figure 4.11: Right: Calculated ground and excited state 1-D potential energy surfaces for the methyl rotor 1 with a lower barrier. Right: Spectral predictions and assignments for the methyl rotor coupling to the electronic excitation. 300 Excited State 5e' 250

) 4e' -1 200 3a 1'

150 3a 2' 2e' 100

Relative Energy (cm Energy Relative 0a ' 50 1 & 1e'

0 50 100 150 200 250 300 350 Torsional Angle 300

92 Ground State

2e” 4e’ 2e” 5e"

2e” 5e’ 2e”

2e’’ 1e’ 2e’’

) 4e"

-1

’ 200

2 1e” 5e’ 1e” 3a 1"

3a 2" 0a!” 3a!’ 0a!” 2e"

0a!” 0a!” 0a!’ & 1e” & 1e” 1e’

& 1e” & 1e” 2e’ 0a!” 0a!” 3a 100

0a 1"

Relative Energy (cm Energy Relative & 1e"

0 50 100 150 200 250 300 350 Wavenumbers realtive to origin (cm-1 ) Torsional Angle

Figure 4.12: Right: Calculated ground and excited state 1-D potential energy surfaces for the methyl rotor 2 with a higher barrier. Right: Spectral predictions and assignments for the methyl rotor coupling to the electronic excitation. The majority of spectral peak positions and intensities have been accounted for by the methyl rotor structure of the two rotors. A few small peaks are predicted, but not observed, and are likely in the noise of the spectrum. There exist a couple of mismatched intensities and peak positions, such as the latter peaks within ﬁgure 4.12, however we are still yet to understand how a coupling of the rotors will aﬀect the potential surfaces. The calculated peaks beyond ∼225 cm−1 are not observed due to the expected short lifetime of the A˜ state, but are shown for future reference if more of the vibrational structure is uncovered.

1 2 36 0 36 0

0 00

predissociation

Relative Wavenumbers (cm-1 )

Figure 4.13: Assigned REMPI spectrum of the D1 state of the 1,1-dimethylallyl radical

The calculated frequencies for the excited state are listed above in table 4.2.

The two modes, ν35 and ν34 correspond to the motion of the two methyl rotors, and are accounted for by their torsional progressions coupling to the electronic excitation.

−1 The remaining modes calculated within this region are ν36 at 137 cm and ν33 at 257 cm−1. The assigned spectrum is shown in ﬁgure 4.13.

93 The largest peak, at 120 cm−1 could not be ﬁt in any sensible way with the other peaks in this region, in both intensity and peak position. This peak is assigned to mode ν36, representing the CH2 out of plane twist, coupled with the out of plane CCC bending modes, calculated to be 137 cm−1. The methyl rotor progressions are also built onto this mode, see the stick spectrum in ﬁgure 4.13. The CH2 twisting mode is likely to perturb the potential of the nearby methyl rotor, which would in turn perturb the other methyl rotor. As such, the potentials of these rotors would change when coupling to ν36, which accounts for the shift in the coupled rotor’s line positions.

The only remaining band in the spectrum is observed at 201 cm−1. The

−1 next lowest frequency mode is ν33, calculated to appear at 257 cm , which is simply too far from the observed value. The results of the anharmonic frequency calculations suggests that the modes are very anharmonic and likely deviate from the harmonic frequencies. As such, we assign the band at 201 cm−1 to the overtone of

−1 ν36 based on the observed value of ν36 at 120 cm . Evidently, care needs to be taken with the coupled methyl rotors, and future work may provide further information on the way these rotors couple to one another, to aid in a revisited assignment.

The ab inito calculations predicted the frequencies of the two methyl rotors,

−1 −1 ν35 and ν34 at 175.7 cm and 222.3 cm respectively. Mode ν35 represents the torsion of the chain/side rotor at the α position, which is further from the terminal

CH2 group, and thus closer to a free rotor. Mode ν34 on the other hand, represents the top rotor, at the β position and additionally contains a partial twist of the CH2 group, see ﬁgure 4.9. Thus its rotor potential will be more inﬂuenced by this steric interaction, leading to higher barriers. As a result, we can assign the low barrier potential, exhibited in 4.11 (blue), to ν35, and the potential with the greater barrier, exhibited in 4.12 to ν34.

To understand the origin of the truncated spectrum, the lifetime of the A˜ state was recorded, by fixing the scanning the excitation laser temporally, whilst fixing the timing of the ionisation laser. The results of this scan are shown in figure 4.14.

94 Figure 4.14: Lifetime scan of the origin peak of the 1,1-dimethyallyl radical, ﬁt to the convolution of a gaussian and an exponential decay function

The profile of the scan was fit to the convolution of a Gaussian and an exponential decay function. As the baseline of the scan is different on either side of the peak, various fits were taken, which gave an expected lifetime in the range of ∼2-4 ns. The most accurate of these, (largest R2 value), is displayed above in figure 4.14, which estimates a lifetime of 2.3 ns, and we attribute an error of 1.5 ns to account for the range of values which yielded appropriate fits. This lifetime is towards the lower limit of what our 2-colour REMPI experiment can typically measure (> 0.5 ns), and if peaks towards higher energy had a significantly shorter lifetime, this identifies a potential reason why we could not observe any further vibrational structure with this technique.

The ionisation potential of the 1,1-dma radical was also determined, guided by an electron impact ionisation study conducted by Lossing and Traeger, where an ionisation potential of 7.13 ± 0.2 eV was recorded. [182] We sought to improve the accuracy of this measurement, by fixing the excitation wavelength on the first peak of the spectrum, and scanning the ionisation wavelength. The results of this scan are shown in figure 4.15. Originally, the peak at 23968 cm−1 was thought to be the origin, and the 1 + 10 REMPI scheme accessed this intermediate state to record the ionisation potential. As such, a value of 90 cm−1 was added to the final IP to account for this hot band discrepancy.

The parent was seen to absorb in the region scanned by the ionisation laser, and as a result a lot of ﬂuctuating, non-resonant, background signal was observed in the m/z 69 trace, owing to the dissociation of the parent molecule. To circumvent

95 on httecluain ea ocneg nacnitn daai adverti- (and adiabatic consistent a on converge to began calculations the that found tde nte12dardcl n h eut r ipae ntbe4.3. table in displayed are future results the for and benchmark radical, a 1,2-dma as the theory on of studies level B3LYP/6-311+G(d,p) the at calculated 7.13 of [182] IP Traeger. the and with the consistent 7.118 of is of energy which potential combined recorded, ionisation the An of which photons. 4.15, function ionisation figure a and of as excitation side The signal right the ionised graph. in resonantly the shown the in is shown displays traces bars red and error black and the point of data three difference the measured excitation was give point the to Each averaged with red). and interval, in times each (shown at without taken and black), was in signal (shown pulse 69 m/z intervals the small of in average stepped point was laser ionisation energies. the ionisation this, and excitation combined the of function laser. a resonance as the traces (red) red and without radical. black and 1,1-dimethylallyl (black) the with of trace scan 69 step m/z potential Ionisation 4.15: Figure oiainPtnil7007.031 7.010 Potential Ionisation Signal (mV) 10 12 14 16 18 20 P orce .5 7.067 7.053 Corrected ZPE 6 8 al .:Peitdadosre oiainptnil fte11dmtyallradical. 1,1-dimethylallyl the of potentials ionisation observed and Predicted 4.3: Table 0.0299.50 300.00 rycnutdavs td ndffrnl ie ol ai es and sets, basis Pople sized differently on study vast a conducted Troy h etcladaibtcinsto oetaso h ,-m aia were radical 1,1-dma the of potentials ionisation adiabatic and vertical The Ionisation Wavelength (nmvac) 9.5292 9.0287 298.50 298.75 299.00 299.25 299.75 daai etclRfrne[8]Observed [182] Reference Vertical Adiabatic 96

Signal Difference (mV) -1 0 1 2 3 4 5 6 7 ± .1 .2 7.125 7.120 7.115 7.13 . Vdtrie yLossing by determined eV 0.2 Ionisation Potential(eV) ± Right: ∼ . 7.1184 0.2 .501n n 1000 a and nm 0.05-0.1 ieec ewe the between Diﬀerence Left: ± .0 Vwas eV 0.002 7.130 inlo the of Signal ± 0.001 7.135 cal) ionisation potential by the time they reached the 6-311+G(d,p) size. [123] He then calculated the adiabatic (and vertical) ionisation potentials for various radicals with a 6-311++G(d,p) basis, (which are near-equivalent to calculations using a 6/3-11+G(d,p) basis), and determined a mean deviation of -0.14 eV (-0.08 eV) compared to experiment. An assumption was made that the ZPE of the ground state of the neutral and cation would be equivalent, and so the calculations without ZPE corrections have been included for comparison. The 1,1-dma radical is calculated to have an adiabatic (and vertical) IP of 7.010 eV (7.031 eV), which lies 0.108 eV (0.088 eV) below the observed values, and are consistent with the discrepancies determined by Troy.

The D2 ← D0 transition

Following the theoretical prediction, the calibration experiments, and the observation of the higher excited states of the allyl and 2-ma radicals, the D2 ← D0 transition of the 1,1-dma radical was sought in the region 247 nm - 280 nm. The results of the 1+1 REMPI experiment are displayed in figure 4.16. This spectrum was obtained from each of the precursors 3,3-dimethyl allyl bromide and geraniol. The first peak in the spectrum is observed at 38329 cm−1, (260.90 nm), and assigned as the origin of this electronic transition. The spectrum spans more than 2000 cm−1, and is expected to continue. The spectral features are seen to broaden significantly towards higher energy, where peaks are identified on top of a larger, more diffuse underlying structure.

The peaks have a full width at half maximum (FWHM) in the range of 12-25 cm−1, depending where the baseline is taken. Low power scans were taken to confirm that the width of these peaks were not power-broadened. To estimate the lower bound on the lifetime of this excited state, we take larger FWHM of 25 cm−1 and calculate a lifetime of >0.2 ps. This calculation assumes no rotational profile/broadening, which are likely to be evident, increasing the lifetime. Furthermore, this experiment was conducted utilising a 1+1 REMPI scheme, with a focussed laser profile, and so the ∼0.5 ns limit discussed earlier (for a 2 photon, 2 colour experiment) is not the case here. A strong oscillator strength of f = 0.1639 was predicted

97 Wavenumbers (cm-1 )

Figure 4.16: REMPI spectrum of the D2 state of the 1,1-dimethylallyl radical for this transition, which is more than a magnitude larger than that of the A˜ state, where f = 0.0115. This is a known feature of radicals with alternant π systems, where a mixed excitation of electrons from the π orbital into the non-bonding n orbital, and the n into the anti bonding π* orbital results in two near degenerate excited state conﬁgurations. This conﬁguration interaction lowers the energy and oscillator strength of one electronic state, here the D1 state, and raises the energy and oscillator strength of the corresponding state, here the D2 state. [183, 184]

Whilst calculations for the D2 ← D0 transition have yet to converge, the sharp and intense peaks at the beginning of the spectrum appear similar to the methyl rotor structure observed within the origin region of the D1 state. Various attempts were made to yield a higher resolution, including variations in the scan rate, wavelength step size, averaging, laser power and precursor selection, but to no avail. Though direct assignments of the spectral features have not been made, the data can be used as a reference spectrum when conducting the hydrogen addition to isoprene experiments, which will likely form a multitude of conﬁgurational isomers.

98 Upon excitation to the Rydberg states in the allyl and methyl allyl radicals, a fast internal conversion to the A˜ state was observed. Without the ability to conduct PHOFEX experiments on the predicted hydrogen photo products, this phenomenon cannot be explored by our experimental setup, but should be considered by future experimentalists.

Once the 1,1-dma radical had been identiﬁed, the cofragment of the geraniol precursor, the hydroxyl isoprenyl radical 4-OH, was sought after. The m/z 85 channel was monitored whilst employing the same REMPI schemes used to identify the D1 and D2 transition of the 1,1-dma radical. In searching for the D1 transition, the ionisation laser was replaced by a higher energy 205 nm source, produced from the tripled output of a SIRAH COBRA-STRETCH Nd:YAG dye laser, circulating a mixture of the Rhodamine 610 and 640 dyes. No signal response was observed for the m/z 85 channel, indicating that the required energy of the discharge, used to produce the 1,1-dma fragment, may have also cleaved the C-OH bond in 4-OH. This result is justiﬁed in part by the electron ionisation mass spectrum of geraniol, where the dominant fragments recorded are m/z 41 (allyl) and 69 (1,1-dma). The intensity of the m/z 85 fragment is less than 5% of the intensity recorded for the m/z 69 fragment. [185]

4.3.4 Concluding Remarks

Observation of the D1 and D2 electronic states of the 1,1-dimethylallyl radical have been made for the first time, by the use of REMPI spectroscopy. Identification of the m/z 69 species as the 1,1-dma radical was verified by the use of both 3,3- dimethylallyl bromide and geraniol as precursors, and the consistency of observed values with theoretical predictions.

−1 The D1 ← D0 transition was observed to have an origin at 24058 cm .

Upon excitation, the symmetry of the molecule lowered from Cs to C1, as the carbon framework puckered out-of-plane, assisting the out-of-plane rotation of the terminal

CH2 group. The origin region of the D1 state was observed to have rich structure. This structure could not be wholly assigned by the few vibrational modes predicted

99 in this region. Rather, the coupling of the methyl rotors to the electronic excitation was explored, to account for the additional peaks. Indeed, the majority of the structure was able to be assigned to the various a → a, and e → e type transitions of the two internal methyl rotors. One of the methyl groups was assigned as having a low barrier (of 78 cm−1 to rotation, and accounted for many of the lower energy peaks. The other methyl group was assigned to have a larger barrier (of 247 cm−1), which was used to fit the higher energy end of the spectrum. The two methyl rotors were fit with a torsional angle change of ∼51/69◦, which was larger than the ∼16◦ predicted by (TD)-DFT theory. The three remaining bands in the spectrum were assigned to vibrational modes of the radical. The truncated spectrum appeared similar to that observed for the 1-ma radical, and we similarly attributed it to a short lifetime of the D1 state, which was determined to be 2.3 ± 1.5 ns. [174] The ionisation potential was also determined to be 7.1184 ±0.001 eV, revising the value of 7.13 ±0.02 eV determined by Lossing and Traeger. [182] The measured IP was consistent within the mean error of the B3LYP/6-311+G(d,p) level of theory, and was calculated as a benchmark for the 1,2-dma radical. Hole burning spectroscopy was attempted on the D1 state briefly, to uncover more of the spectrum, however the signal was unable to be depleted. This was also the case for hole burning experiments conducted within our group on the H + toluene radicals.

The D2 ← D0 transition was also observed, through a 1+1 REMPI scheme. The electronic origin was determined to be 38329 cm−1. The spectrum features broad and diﬀuse features over a ∼2000 cm−1 range. A lower limit on the lifetime of this state was determined to be ∼0.2 ps, by a measurement of the largest FWHM value. The photochemistry of this excited state should be explored in future experiments, to determine whether it follows a deactivation to the D1 state, as seen for the allyl and methyl allyl radicals.

100 4.4 Investigation of The 1,2 - Dimethyl Allyl Rad- ical

Continuing with our study on the doubly methylated allyl radicals, the 1,2-dimethylallyl radical (1,2-dma) will be generated in our discharge environment from a suitable precursor. This will complete our isomer speciﬁc study on the radicals formed from the H-atom addition to the terminal sites of isoprene.

A truncated spectrum of the A˜ ← X˜ state of the 1-ma radical was observed by Gasser et al. [174] We reported a similarly truncated A˜ state for the 1,1-dma radical in the previous section, where the two methyl substituents are at the α position on the allyl radical. By methylating the allyl radical the at β position, forming the 2-ma radical, the A˜ state was unable to be observed by REMPI spectroscopy. As such, we expect that methylation at the α position of the 2-ma radical, (generating the 1,2-dma radical), will likely also produce an unobservable A˜ state by REMPI techniques.

Various higher excited (3s and 3p Rydberg) states of the singly methyl- substituted 2-ma radical were observed by Gasser at al. by 1+1, 2+2 and UV- UV depletion REMPI schemes. The B˜ state contained rich, intense and sharp vibrational structure, spanning some 3400 cm−1. Toward the latter part of the spectrum, a low intensity, continuous and broad structure emerged. This region was investigated by UV-UV depletion, and the authors identiﬁed a very strong absorption with a short lifetime, that they attributed to the 3pz and 3px states, both calculated to carry large oscillator strengths. [170]

101 4.4.1 Experimental

Without the commercial availability of any appropriate precursors for the 1,2-dma radical, the compound (2E)-1-bromo-2-methylbut-2-ene was synthesised with the aid of Associate Professor Jason Harper and PhD student Nick Konstandaras, following the procedure of Lorenz and Kalesse. [186] The sample was synthesised on multiple occasions with varying yields and purities. The tandem mass and spectral resolution of our technique means that impurities do not aﬀect our experiment, provided that the impurities do not react with or degrade the sample over time. The dissociation scheme of (2E)-1-bromo-2-methylbut-2-ene to form the 1,2-dma radical is shown in ﬁgure 4.17.

+ Br

Figure 4.17: The formation of the 1,2-dimethylallyl radical, following the dissociation of the C-Br bond in the (2E)-1-bromo-2-methylbut-2-ene precursor

The spectroscopic experiments conducted here have been described earlier, and only details pertaining to this experiment are described. Argon (7-7.5 bar) was passed through the sample container, to collect the vapor of (2E)-1-bromo-2- methylbut-2-ene. The gas mixture was pulsed into a vacuum chamber, whereupon a discharge of 1500-2090 V was struck, with 6-25 kΩ ballast resistance, for 100-300 µs. The discharge was timed to strike at the latter part of the expanding gas, which was then skimmed by a 2 mm skimmer into an extraction chamber. The cold molecular beam was probed by various REMPI schemes to search for the ﬁrst (D1) and higher electronic states.

The A˜ state of the allyl and 2-ma radicals have not yet been observed by REMPI spectroscopy, and so our search encompassed the range of the allyl and 1,1-dma excited state, and proceeded toward lower energy in search of vibrational structure or an origin transition.

102 0 In attempt to observe the D0 to D1 transition, a 1+1 REMPI scheme was ventured over the region 404-435 nm. The output a SIRAH COBRA-STRETCH dye laser, circulating the Exalite 411 and 428 dyes was used as an excitation source. Ionisation photons of 266 nm were provided by the fourth harmonic of a BRILLIANT B, Nd:YAG laser. As the ﬁrst signal on the 1,1-dma radical was observed by a 1+2 REMPI scheme, utilising a 250 mm focussing lens, the same scheme was used here in our search for the D1 state of 1,2-dma. Finally, a higher energy ionisation source of ∼205 nm was also employed, where photons were generated by the tripled output of a second dye laser, circulating a mixture of the Rhodamine 610 and 640 dyes. The signal on m/z 69 was recorded as a function of excitation wavelength.

The higher excited states of the 1,2-dma radical were sought after in the same region where the D2 state of the 1,1-dma radical was observed. The excitation and ionisation source for the 1+1 REMPI scheme was the doubled output of a SIRAH COBRA-STRETCH Nd:YAG pumped dye laser, circulating the Coumarin 503 dye. A 2+2 REMPI experiment was also conducted, by removal of the doubling crystal, and the installation of a 200 mm focussing lens, with the focus on the molecular beam. Rather than proceeding through a REMPI scheme to ionisation, this focussing resulted in the observation of a non-resonant 3 photon ionisation curve, and so the ionisation potential of the 1,2-dma radical was recorded accordingly.

103 4.4.2 Theory

A theoretical investigation was undertaken to help explain the features of the observed electronic states. These results were obtained following experimentation due to the diﬃculties faced with the computation of the internal rotation of the methyl rotors, similar to the computational challenges faced with the 1,1-dma radical, and will be used to analyse the data below and direct future studies. These calculations were completed in collaboration with fellow PhD candidate, now Dr. Miranda Shaw.

As with the 1,1-dma radical, the symmetry of the ground state of the 1,2- dma radical depends upon the treatment of the methyl rotors. Geometric details of the D0 and D1 states of the 1,2-dma radical are presented below in ﬁgure 4.18. These calculations were performed at the (TD)-DFT M06-2X/6-311+G(d,p) level of theory, with the GAUSSIAN16 package.[135]

The carbon framework is planar in the ground state and the symmetry of the state converges to Cs. As for the 1,1-dma radical, the CH2 group remains in the plane of the molecule. Upon excitation, this group rotates anticlockwise by 55.59◦. Further, the C-C bonds within the allyl chromophore extend in length from 1.380 Å and 1.395 Å to 1.409 Å and 1.408 Å respectively. This is consistent with the promotion of electrons from bonding (π) to non-bonding (n), and/or from non- bonding to anti-bonding (π*) orbitals. Upon excitation, the C-C-C angle of the allyl chromophore narrows by ∼4◦. Furthermore, the top methyl rotor at the β position rotates by 38.17◦. This rotation keeps the relative torsional positions of the two methyl rotors similar to their conﬁguration in the ground state.

The adiabatic excitation energy for the D1 ← D0 transition was calculated at the (TD)-DFT M06-2X/6-311+G(d,p) level of theory, and the result (including the zero-point corrected energy) is shown in table 4.4.

The adiabatic excitation was calculated in the following way;

E = E(TD−DFT ) + ∆ESCF + ∆EZPE

104 1.093

1.090 1.093

1.513 119.31 120.11 1.090 112.45

1.096 121.51 125.93 1.380 1.395 1.492

1.082 1.096 120.58 1.086 1.083 117.66 120.83 117.27 116.80

D Ground State 0

1.098

1.092 1.091

1.489 136.74 120.98 122.12 1.093 111.04 55.59 120.03 121.83 1.409 1.408 1.504 1.092 1.089 116.84 1.098 38.17 119.07 118.31 120.13 117.48 1.087 1.085

side-on profile top-down

D Excited State 1

Figure 4.18: Geometric details of the D0 and D1 states of the 1,2-dimethylallyl radical where the diﬀerence in the ground and excited state ZPE’s and self-consistent ﬁeld (SCF) energies are added to the excitation energy taken from the TD-DFT output. As with the 1,1-dma radical, calculations for the higher excited states of the 1,2-dma radical did not converge, and as such ZPE corrections are not available. The D2 and D3 energies are approximate values, calculated by replacing the ∆EZPE term with the D(2/3) -D1 energy from the TD-DFT output. As the geometry was not converged for the D2 or D3 states, they represent an approximated vertical excitation energy rather than adiabatic, and so we can expect an overestimation compared to the observed excitation energy. The oscillator strength for the D3 transition is predicted to be ∼5× that of the D2 transition, and due to their similar excitation energies, the spectral features may overlap. For the 1,1-dma radical, the origin of

105 Table 4.4: Predicted transition energies for the D1−3 electronic states of the 1,2-dimethylallyl radical, along with their associated oscillator strengths

D1 (Adiabatic) D2 (Vertical) D3 (Vertical)

22441.9 cm−1 42274.3 cm−1 45016.6 cm−1 Excitation Energy (445.60 nm) (236.55 nm) (222.14 nm) 22637.7 cm−1 ZPE Corrected (441.74 nm)

Oscillator Strength (f) 0.0134 0.0262 0.0955

−1 −1 the D2 ← D0 transition was observed at 38,329 cm , 1719 cm lower than the −1 calculated approximate D2 energy of 40048 cm , (249.7 nm).

The vertical and adiabatic ionisation energies were calculated at the B3LYP/6- 311+G(d,p) level of theory for consistency and comparability with other calculations in this work, and are displayed in table 4.5.

Table 4.5: Theoretical ionisation potentials of the 1,2-dimethylallyl radical.

Adiabatic Vertical Reference [182]

Ionisation Potential 7.154 7.258 7.40 ZPE Corrected 7.203 7.297

Lossing and Traeger predict an IP of 7.40 for the 1,2-dma radical, based ˙ on the IP of the CH2=CHCHCH3 radical (1-ma) of 7.55 eV and an expected -0.15 eV shift for β-substitution of CH3. [182] The IP of the 1,1-dma radical was observed 0.108 eV higher in energy than the calculated, non-ZPE corrected, adiabatic IP at this level of theory. Based on our calculations, the IP of the 1,2-dma radical should be observed around 7.262 eV, 0.138 eV lower in energy than the prediction by Lossing and Traeger.

106 4.4.3 Results and Discussion

The D1 ← D0 transition

The D1 ← D0 transition of the 1,2-dma radical was sought after in a similar region to the transitions of the allyl, 1-ma and 1,1-dma radicals, that is between ∼404- 435 nm. Discharge conditions were varied as described by the ranges within the experimental section, and the bromine atom discharge co-product was monitored to ensure the timings of the discharge pulse were correct. To conﬁrm the production of the 1,2-dma radical, the 435 nm output of the Nd:YAG pumped dye laser was frequency doubled, and focussed with a 200 mm focussing lens onto the molecular beam. Plentiful signal was observed on the m/z 69 peak, conﬁrming the presence of the radical, however, no spectrum was obtained for the D1 state. Additionally, the 3,3-dimethylallyl bromide precursor was later introduced alongside the (2E)- 1-bromo-2-methylbut-2-ene precursor, to ensure that the experimental setup was optimal. Indeed, the 1,1-dma spectrum was re-recorded within this region, however no additional peaks were observed.

Recent computational results have predicted a D1 origin to appear at 22637.7 cm−1 (441.74 nm), which is considerably red shifted compared to the allyl, 1-ma and 1,1-dma radicals. The D1 origin of the 1,1-dma radical was predicted at the same level of theory to be 25091 cm−1, and was observed some 1033 cm−1 to the red. If we apply this same shift to the 1,2-dma origin, then a revised search should be conducted in the region ∼21604.7 cm−1, (462.86 nm). Observation of the 1,2-dma radical by REMPI spectroscopy requires that the additional methylation at the α position increases the lifetime of this upper state, as the D1 state of the 2-ma radical has not been able to be observed by this technique due to the conical intersections between the D1 and D0 states, leading to dissociation on D0.

107 The D2/3 ← D0 transitions

Following the observation of the D2 state of the 1,1-dma radical, the same region −1 −1 (37000 cm - 41000 cm ) was explored for the D2 state of the 1,2-dma radical. The results of the 1+1 REMPI scan are displayed in ﬁgure 4.19. A series of sharp peaks, beginning with an origin at 38501.2 cm−1 span some ∼1000 cm−1 before a broad and diﬀuse structure emerges over the following ∼1500 cm−1, which is reminiscent of the higher energy region of the 1,1-dma D2 state.

* *

* * * * * * * * *

Wavelength (cm-1 )

Figure 4.19: Higher excited states of the 1,2-dimethylallyl radical in the UV region

The origin region around 38500 cm−1 is expanded in the inset to conﬁrm a FWHM value of 4.1 cm−1, indicating a minimum lifetime of 1.3 ps. The peaks over the following ∼1000 cm−1 have similar widths. As the vibrational structure is barely above the level of the baseline, the scans were repeated multiple times and the consistent peaks have been indicated by a red asterisk. As the D2 state is yet to

108 Table 4.6: List of observed peaks and their relative frequencies in the D2 state of the 1,2- dimethylallyl radical. Intensities are labelled as strong (s), medium (m), weak (w) and very weak (vw). All units are in wavenumbers (cm−1).

Frequency Relative Intensity Frequency Relative Intensity

38502 0 (s) 39247 745 (w) 38910 408 (m) 39326 824 (w) 38971 469 (s) 39358 856 (w) 39037 535 (vw) 39506 1004 (w) 39126 624 (m) 39536 1034 (m) 39191 689 (vw) converge, assignments for these peaks can not yet be made. The peaks have been identiﬁed and are presented as a list of frequencies relative to the origin in table 4.6.

The D2 and D3 states were estimated to have transition energies of 42274.3 cm−1 and 45016.6 cm−1, with associated oscillator strengths of 0.0262 and 0.0955 respectively. The spectrum in ﬁgure 4.19 appears to be split into two parts, separated by ∼1000 cm−1. We thus assign the sharp structure beginning at 38501.2 cm−1 to the D2 state, and the broad, intense, diﬀuse structure beginning at ∼39600 to the D3 −1 −1 state. Observation of the D2 and D3 states lie 3773 cm and 5417 cm lower than the predicted values. As a point of reference, the D2 state of the 1,1-dma radical was observed 1719 cm−1 lower in energy than the assigned origin. Whilst the range between these values varies, this method of estimating the D2 transition energy from the output of the D1 calculation appears reliable in over-predicting the energy, by an average deviation of 3636 cm−1. Of course, as geometry of these states were not optimised, we expect this over-prediction in the electronic transition energies.

Without being able to access a low lying intermediate step such as the

D1 state, the ionisation potential of the 1,2-dma radical was determined by a 3 photon non-resonant ionisation scheme. The photo-ionisation eﬃciency (PIE) curve, displayed in ﬁgure 4.20, was generated by recording the m/z 69 signal as a function of 3 photon wavelength.

109 2 photon energy (cm-1 ) Ionisation Potential (eV)

Figure 4.20: Photoionisation eﬃciency curves for the 1,2-dimethylally radical

On the left hand side of ﬁgure 4.20, the total 3 photon energy is described by the top x-axis. The PIE curve was obtained whilst searching for a 2+2 REMPI scheme within the D2 state, as the origin of the D2 state of the 2-ma radical is only accessible via a 2 photon excitation. As such, the bottom axis represents the energy sum of 2 photons, and we can clearly see that the PIE onset begins around

−1 −1 the 39000 cm mark. This is ∼500 cm higher in energy the D2 origin, and ∼500 −1 cm lower than the diﬀuse D3 state. This conﬁrms that the PIE curve is unique from the D2 and D3 structure observed via 1+1 REMPI.

The ionisation potential was determined to be 7.275 ±0.025 eV, by the intersection of the two lines fit to the baseline and the onset of the PIE curve, as can be seen in the (zoomed-in) right hand side of figure 4.20. The confidence interval of ±0.025 eV was determined by the variance of appropriate fits. Though the baseline appears to have a positive gradient in the right hand figure, by comparison with the left hand figure, we can see that this is a natural fluctuation of the baseline, and that the onset does not begin until around the 7.25 eV mark.

The determined IP of 7.275 ±0.025 eV is within error of the theoretically predicted value of 7.262 eV, conﬁrming our consistent use of the B3LYP/6- 311+G(d,p) level of theory with a consistent oﬀset from calculation. The recorded value is 0.125 eV lower than that suggested by Lossing and Traeger, indicating that

110 the β substitution of the methyl group lowers the IP by -0.275 eV, rather than -0.15 eV.

4.4.4 Conclusion

Two excited electronic states of the 1,2-dimethyl allyl radical have been recorded, and were assigned as the D2 and D3 states based on agreement with theoretical −1 predictions. The D2 state, recorded with an origin at 38501.2 cm , exhibited sharp vibrational structure, and 11 recurring peaks were identified in the spectrum spanning 1034 cm−1 from the origin. The latter part of the vibrational structure is entangled with onset of the strong, broad and diffuse structure of the D3 electronic state. This state, beginning around 39600 cm−1, was calculated and observed to carry a large oscillator strength. The continuous and diffuse nature makes identifying peak positions difficult, and the spectrum is expected to continue beyond 41000 cm−1.

Further theoretical work is required to converge these electronic conﬁgu- rations and provide frequencies with which to assign the sharp structure of the D2 state. The spectra of the D2 and D3 states is very similar those of the 2-ma radical, where sharp vibrational structure is observed for the 3s Rydberg state, followed by a continuous structure carrying large oscillator strength, conﬁrmed by UV-UV depletion techniques, and assigned to the 3p Rydberg states.[170]

The ionisation potential of the 1,2-dma radical was recorded via a 3 photon non-resonant ionisation scheme. The IP was determined to be 7.275 ±0.025 eV, consistent with the theoretically predicted value and oﬀset of the B3LYP/6-311+G(d,p) level of theory of 7.262 eV.

111 4.5 Hydrogen Addition to Isoprene

The four possible hydrogen addition sites to isoprene are shown in ﬁgure 4.21. Ad- dition at the terminal sites, yielding the resonantly stabilised 1,1-dma and 1,2-dma radicals, are the most likely outcomes, and the vibronic spectroscopy of these two addition products were reported in the previous two sections 4.3 and 4.4.

+ H 2 4 1 3

(1) 1,1-dma (2) (3) (4) 1,2-dma

Figure 4.21: Plausible addition sites for H atom to isoprene and the respective radicals produced

By the addition of hydrogen to isoprene, we can conﬁrm the production of these two radical species by comparison with their spectra. Any new features present within the spectra can be attributed to the inner addition products (3) and (4). The eﬀectiveness of this procedure as a way to identify multiple addition products from a single reaction will be assessed, with the goal toward experimental repetition for the hydroxyl addition to isoprene reaction.

The two regions of interest encompass the D1 transition observed for the

1,1-dma radical, and the region containing the D2 (and D3) electronic states of the 1,1-dma and 1,2-dma radicals.

112 4.5.1 Experimental

As described in chapter 3, the discharge of a gas mixture containing water is an eﬀective way of producing hydrogen atoms. Argon (7 bar) was bubbled through an external sample container of water to collect its vapour, before passing through a second sample container of isoprene (Aldrich, 99%, used without further puriﬁ- cation). The water was heated to ∼50◦C to increase the vapor pressure. The gas mixture was pulsed into the REMPI chamber, and the latter part of the expanding gas was struck by a discharge of 2090 V, ballasted by 9 kΩ resistance for 250 µs, to generate the hydrogen atoms for reaction. The expanding gas was skimmed by a 2 mm skimmer, and the molecular beam was interrogated by various REMPI schemes.

The region containing the D1 transition of the 1,1-dma radical was explored via 1+10 REMPI, over the range of 404-435 nm, in hope to ﬁnd additional peaks owing to the 1,2-dma radical, or the inner addition products 2 and 3. The excitation photons were provided by a SIRAH COBRA-STRETCH Nd:YAG pumped dye laser circulating the Exalite 411 and 417 dyes. Ionisation photons were provided by either the fourth harmonic of a BRILLIANT B, Nd:YAG type laser at 266 nm, or the tripled frequency output of a second SIRAH COBRA-STRETCH Nd:YAG pumped dye laser circulating a mixture of Rhodamine 610 and 640 dyes, providing a ∼205 nm ionisation source.

The UV region containing transitions to higher excited states of the 1,1- dma and 1,2-dma radicals, (37000 - 40500 cm−1), was explored via 1+1 REMPI schemes for the isoprene + H reaction. The dye was changed to Coumarin 503, and the output was frequency doubled to provide the UV photons for both excitation and ionisation. Ideally, the spectrum would contain features of not only the 1,1-dma and 1,2-dma radicals, but of all four isoprene + H addition products described by ﬁgure 4.21.

113 4.5.2 Results and Discussion

The First Excited State(s)

The spectrum generated for the m/z 69 trace, identiﬁed as products of the isoprene + H reaction, is displayed in the top of ﬁgure 4.22. The extended region over 23000 -

−1 24750 cm was explored via 1+1’ REMPI, and the D1 state of the 1,1-dma radical, lower trace in ﬁgure 4.21, has been included for direct comparison.

+ H

isoprene *

1,1-dma

Figure 4.22: Upper Trace: REMPI spectrum of the isoprene + H addition products in the region likely to contain the D1 transitions of the radicals. Lower Trace: The D1 electronic spectrum of the 1,1-dimethyallyl radical.

Clearly, the features of the m/z 69 trace can be identiﬁed as belonging to the 1,1-dma radical. The prominent features of the 1,1-dma spectrum, including the hot band transition at 23968 cm−1, are observed. The reduced signal to noise ratio in the upper trace precludes the observation of some of the smaller features in the spectrum. Further, the resolution is also lower, merging the two peaks observed

114 ∼24140 into one. This is due to the spectrum being saturated, evidenced by the relative intensities of the peaks to the band at 24178 cm−1. The peak at ∼24100 cm−1, identiﬁed with a red asterisk, was never observed more than the once, and was determined to be noise.

As no additional peaks were observed over this range, the D1 states of the products (2), (3) and 1,2-dma remain elusive. As the products (2), (3) do not contain allylic chromophores, or any π-conjugation, their D1 electronic states are predicted to lie higher in energy. Recent calculations have determined that the D1 state of the 1,2-dma is likely to be observed lower in energy, near to 21600 cm−1, and future experiments should explore a wider range to encompass these energy regions.

Water was substituted for heavy water (D2O) to observe the eﬀects of deuteration on the vibrational structure and methyl rotors of the 1,1-dma radical, however no spectrum was obtained for the m/z 70 trace, despite a protracted study. Addition of deuterium at the (1) position would cause an uneven potential in one of the methyl rotors, and shift the respective peak positions. This deuterated study was undertaken to conﬁrm which of the rotors belonged to which (high or low barrier) potential.

From a discharge containing water and isoprene, addition of of the H-atom to the (1) site of isoprene was conﬁrmed to occur, as the m/z 69 spectral trace matches the postulated 1,1-dma radical product.

Higher Excited State(s)

Following observation of the 1,1-dma radical, the discharge and timing conditions were retained, and the dye was replaced to access the UV region of interest between 37000 cm−1 and 41000 cm−1. The m/z 69 spectrum obtained from a discharge containing isoprene and water is displayed in the middle trace of ﬁgure 4.23. The

1,1-dma spectrum of the D2 state, (bottom trace), and the D2 and D3 states of the 1,2-dma radical, (top trace), have both been included for comparison.

115 1,2-dma

* * * + H

isoprene

1,1-dma

Wavenumbers (cm-1 )

Figure 4.23: Middle Trace: The UV spectrum of the m/z 69 trace corresponding to the isoprene

+ H addition products Upper Trace: Reference spectrum of the D2 and D3 states of the 1,2- dimethylallyl radical. Lower Trace: Reference spectrum of the D2 state of the 1,1-dimethylallyl radical. The identities of the peaks marked with red asterisks are proposed in the text at the end of this section.

Over the first ∼1200 cm−1 of the spectrum, a rich vibrational spectrum is observed. Peaks which correspond with features of the 1,2-dma D2 state are joined by a blue dotted line, and account for the majority of the intense spectral lines in the middle trace. It is interesting to note the signal-to-noise ratio of the 1,2-dma peaks within the isoprene + H trace, compared to the spectrum formed from the synthesised precursor, allowing confirmation of the assigned peak positions. The synthesised precursor and highly energetic discharge used to generate the spectrum in the top trace of figure 4.23 was considerably unstable, and between scans the intensity profile of the various bands would vary greatly, especially for longer scans (taking 1-2 hours). We suspect that the trace generated from the isoprene + H experiment contains a set of more accurate intensity profiles.

116 Peaks corresponding with features of the 1,1-dma D2 state are joined by the dotted green line. In the m/z 69 trace, the features belonging to the 1,2-dma radical are observed to be much stronger than those for the 1,1-dma radical. As a result, many of the broad features of the 1,1-dma radical appear as shoulders on peaks belonging to the 1,2-dma radical. Observation of the three intense bands between 38300 cm−1 - 38500 cm−1 is the most convincing argument for the presence of the 1,1-dma radical.

Beyond 39600 cm−1 the features of the isoprene + H trace appear to broaden, due to the onset of the D3 state of 1,2-dma and the broad features of 1,1-dma. The combination of both spectra account for the peaks present in the m/z 69 trace. Only three distinct features in the m/z 69 trace cannot be accounted for by the two reference spectra, identiﬁed with the red asterisks. The most notable of these is the peak at 38886 cm−1, and although it is similar in intensity and shape to other features of the 1,2-dma radical, no corresponding peak is observed. As such, we label these three peaks as potential candidates for the presence of the inner addition products (3) and (4).

The identification of the spectrum of the 1,2-dma radical in the m/z 69 trace confirms the postulated addition of hydrogen atom to the (4) position of isoprene. We can now conclusively state that the hydrogen atom adds to the terminal sites of isoprene, by comparison with the spectra of the 1,1-dma and 1,2-dma radicals. The inner addition sites are still plausible, and could account for the 3 unidentified peaks within the trace.

117 4.5.3 Conclusion

This chapter presented an analysis on the electronic spectroscopy of methyl-substituted allyl radicals, with the goal of deconvoluting the hydrogen addition to isoprene reaction, in preparation for the hydroxyl addition to isoprene reaction, presented in the following chapter. The calibration studies on allyl and 2-methylallyl provided the necessary experimental framework for the study of the higher electronic states of the dimethylallyls.

The 1,1-methylallyl radical was generated by a discharge of 3,3-dimethylallyl bromide and geraniol precursors seeded in argon. The D1 state of the 1,1-methylallyl was determined to be short lived, evidenced by a truncated spectrum spanning only ∼300 cm−1. The origin was observed at 24058 cm−1 with the majority of the spectral features assigned, with the aid of theoretical modelling, to the two methyl rotors coupling to the electronic transition.

The only vibrational band to appear in the spectrum was the CH2 twist and its first overtone. The methyl rotor structure was observed to couple to this vibration, albiet with different spacings reflecting the impact of the CH2 twist on the methyl rotor potentials. The D2 state was also observed, with an origin at 38329 cm−1, and consisted of a series of intense peaks, followed by a broad and continuous spectrum spanning ∼2000 cm−1 and expected to continue. Finally, the ionisation potential was revised to the value of 7.118 ± 0.002 eV, consistent with earlier findings. [182]

The compound (2E)-1-bromo-2-methylbut-2-ene was synthesised, as an appropriate precursor to the 1,2-dimethylallyl radical was not commercially available.

Similar to the 2-methylallyl radical, the D1 state of the 1,2-dimethylallyl was unable to be observed via REMPI spectroscopy. A revised search in the region near

−1 21600 cm will be conducted following recent computational ﬁndings. The D2 and

D3 electronic states were observed in the UV region, the former was reported to have an origin transition at 38501.2 cm−1 followed by sharp vibrational structure.

−1 The D3 state was observed some ∼1100 cm higher in energy, beginning around

118 39600 cm−1 and consisting of continuous and broad features carrying a large oscillator strength. The ionisation potential was determined to be 7.275 ±0.025 eV, consistent with theory.

Following the observation of the two dimethylallyl radicals, the hydrogen addition to isoprene reaction was explored. Spectra of the 1,1-dma and 1,2-dma radicals were both identiﬁed in the UV region of the m/z 69 trace, conﬁrming the addition reaction to the terminal sites of isoprene. Three additional peaks were observed in this region and were tentatively assigned to the inner addition products of the reaction. The D1 state of the 1,1-dma radical was also reproduced by this reaction, however deuteration experiments were unyielding. The ability to decouple the isoprene + hydrogen trace by the reference spectra of the individual isomers was a key experimental feat toward decoupling the hydroxyl addition to isoprene radical products in the following chapter, and the same experimental procedure will be followed.

In regards to the dimethylallyl radicals, there remains computational work to be done toward the convergence of the higher excited states, to properly assign the D2 state of the 1,2-dma radical. Further, the present assignment of the D1 state relies on the assumption of uncoupled methyl rotors, and may indeed change if the rotors are found to couple signiﬁcantly. The short lifetime of the D1 state remains a feature of all of the radicals explored in this chapter, arising from an out of plane rotation of a terminal CH2 group, enabling a rapid internal conversion to the ground state. Conical intersections for the 1,1-dma radical are hypothesised and should be explored by PHOFEX spectroscopy and high level theoretical studies in future experiments.

119 Chapter 5

Eﬀorts towards the Hydroxylation of Isoprene

5.1 Introduction

The VOC budget is dominated by one molecule, isoprene, with an emission ﬂux of 535 Tg yr−1. [27] The key loss mechanism for isoprene is reaction with the hydroxyl radical (∼85%).[72] The OH initiated oxidation of isoprene results in the production of tropospheric ozone [187, 188] and secondary organic aerosol [65]. Current studies have estimated that the global production of SOA lies somewhere within the range of 50 - 910 Tg yr−1, dominated by biogenic aerosol.[16, 18, 23] Considering the eﬀect that aerosols have on regional and global climate through radiative forcing, [5, 6] acting as cloud condensation nuclei,[7, 8] and their role in air quality, the chemistry underpinning models of SOA formation must clearly be improved.

The atmospheric lifetime of isoprene is only around 1.5 hours, due to its high reactivity with the hydroxyl radical. [70] In regions of high isoprene concentration, such as pristine forests, models cannot correctly account for the concentration and recycling of the hydroxyl radical, under-predicting [OH] by an order of magnitude.[78] Considering the relative abundance of isoprene, and its impact on SOA production, it is surprising that its most fundamental oxidation products, the

120 hydroxy isoprenyl radical adducts seen in ﬁgure 5.1, have never been detected. Here 1-OH represents both the cis and trans isomers.

Isoprene

2 4 (cis/trans) 1 3 + OH

HO OH

OH 1-OH 2-OH 3-OH 4-OH (cis/trans) (cis/trans) 0.67 (0.63) 0.02 (0.00) 0.02 (0.00) 0.29 (0.37)

Figure 5.1: Isoprene addition sites for the hydroxyl radical, and the corresponding OH-adducts. The two product distributions are described in the text.

The product distributions in ﬁgure 5.1 were theoretically determined by Greenwald et al. [74] and have recently been revised by Teng et al., shown in parentheses.[69, 75] Addition to the terminal sites is preferred, generating a resonantly stabilised allylic chromophore. The inner addition channels have been revised as negligible by Teng et al., however they suggest that further experimental work should be conducted on these channels as they may account for part of the carbon balance in SOA formation.[69, 75]

In recent efforts by the North group, the oxidation of isomer selective reaction intermediates, including the 2-OH, 3-OH, and 4-OH adducts, was conducted in the presence of O2 and NOx. [80, 81] By analysing the products proceeding through the 4-OH intermediate, the δ-hydroxyalkoxy channel was experimentally confirmed for the first time, including information on the branching ratio of this end product.[81] Similarly, for the inner addition products, Greenwald et al. confirmed that the initially formed β-hydroxy isoprenyl radical underwent a prompt rearrangement via a cyclic intermediate for form to an α-hydroxy isoprenyl radical.

They proposed a route to the formation of the ﬁrst generation C5 carbonyl species through this cyclic intermediate. [80] The insight gained from these isomer selective studies has aided in the mechanism of the hydroxyl initiated oxidation of isoprene.

121 In section 4.5, we demonstrated that from a discharge containing water vapor and isoprene, we were able to add hydrogen atoms to the double bonds of isoprene, forming the 1,1-dma and 1,2-dma radicals. In the current chapter, we begin with a series of calibration experiments on indene, comparing with studies conducted by Troy,[123] conﬁrming our ability to add hydrogen and hydroxyl radicals within the discharge region. These studies are extended by deuteration experiments. In order to observe the individual OH-isoprene radical adducts, precursor radicals containing the OH-moiety are synthesised and discharged, following on from our isomer selective studies conducted on the 1,1- and 1,2-dma radicals in sections 4.3 and 4.4. Finally, in order to investigate the ground state of these important OH- isoprene radical adducts, the relevant precursors were sent to Taiwan, where matrix isolation spectroscopy was performed.

5.2 Calibration Studies on Indene

The ability to add hydroxy radical to a molecular species was first demonstrated in our lab by Dr. Troy. In a discharge of indene (m/z 116), Troy measured the spectrum of a range of generated radicals, with m/z values of 115, 117 and 133. The first was identified as 1-phenylpropargyl, formed from by a C-C bond cleavage and H-loss. The m/z 117 carrier was identified as the 1-indanyl radical, generated by the addition of a hydrogen atom within the discharge.[189] The final species, with m/z 133 (+17), was later identified as the 2-hydroxy-indan-1-yl, formed by the addition of hydroxyl radical to indene. [134]

The hydroxyl radicals were determined to originate from trace water introduced into the sample container as it was prepared, and into the external argon gas lines during maintenance or experiment adjustments. From a discharge containing water vapor and either indene, 2-methylindene or 2-ethylindene, he reported the detection of species with +17 mass units, corresponding to the 2-hydroxy-indan-1-yl, 2- hydroxy-2-methyl-indan-1-yl and 2-ethyl-2-hydroxy-indan-1-yl radicals respectively. [123, 134]

122 As described in section 4.3.3, both the hydrogen and hydroxyl radicals involved in the addition reaction to indene were exclusively formed in the discharge from the interaction of metastable argon colliding with, and dissociating, the trace water within the experiment. As such, introduction of heavy water vapor (D2O) to the sample gas line will likely result in detection of m/z +2 and +18, corresponding to D and OD addition. The intention of these proof of principle experiments is to demonstrate eﬃcacy in H/D and OH/OD addition experiments in this thesis. We have not attempted vibronic assignment here, but leave as future work.

5.2.1 Hydrogen and Deuterium Addition to Indene

We have recently reported the addition of hydrogen atom to phenol, seen in chapter 3 and the hydrogen addition to isoprene, detailed in section 4.4, and so begin our calibration study by investigating the hydrogen addition to indene. To extend this study, D2O was added to the external H2O sample container, in hope of observing +D and +OD addition products.

The experimental setup has been previously described and here only details explicit to this experiment are mentioned. To generate the required radicals, argon (5.5 bar) was bubbled through a sample container of 1:2 H2O and D2O and then through an internal sample container of indene (Fluka, 95%, used without further puriﬁcation) heated to 51◦C to collect their vapours. As the gas mixture was expanded into a diﬀerentially pumped vacuum chamber, the latter part of it was struck by a voltage of 2090 V, with 9.09 kΩ ballast resistance, for 100 µs. The nozzle was heated to 55◦C to prevent re-condensation of indene. The supersonically cooled expansion was skimmed with a 2 mm skimmer to form a molecular beam, which was probed by 1+10 REMPI spectroscopy. Excitation photons were provided by the output of a SIRAH COBRA-STRETCH Nd:YAG pumped dye laser, circulating Coumarin 460. The doubled output of a QUANTEL TDL-90 dye laser, circulating Rhodamine 610, was used as an ionisation source (∼285 nm). The experiments were conducted at 10 Hz.

123 The origin of the D1 ← D0 transition of the 1-indanyl was determined to −1 be 21159 cm . By discharging the vapor of water (H2O and D2O) and indene, the first 1000 cm−1 of this spectrum has been retaken, by recording the m/z 117 trace as a function of excitation wave number, seen in figure 5.2. The m/z 118 trace, identified as the deuterated indanyl product, was also recorded. The inset of the 1-indanyl radical indicates the position of H/D addition. The spectrum taken by Troy has also been included for comparison.[123, 189]

Indene + D (m/z 118)

Indene + H (m/z 117)

H/D

1-indanyl adapted from Troy

-1000 100 200 300 400 500 600 700 800 900 1000 Relative Wavenumbers (cm-1 )

Figure 5.2: The excitation spectrum of the m/z 117 and 118 traces, corresponding to H and D addition to indene. Spectrum of the 1-indanyl radical reproduced from Troy et al. [189]

Immediately apparent between the m/z 117 traces is the increased intensity of the vibrational structure relative to the origin transitions. This is due to saturation of the stronger peaks. Although saturated, it has helped to identify more of the vibrational structure by lifting various peaks out of the noise, including the

124 conﬁrmation of a hot band at -40 cm−1. This will aid in future assignment of the

1-indanyl D1 spectrum.

Addition of a deuterium atom to indene generates the +2 trace represented by m/z 118. The trace has been shifted by 21159 cm−1 to line up the origin with the 1-indanyl spectrum, to allow a direct comparison with the vibrational structure. The true origin position lies 2 cm−1 higher in energy, at 21161 cm−1. The spectrum is assigned as the 1-indanyl isotopologue, formed from D addition at the 1-position based on the choice of precursor and the similarity between the m/z 117 and m/z 118 spectra. Further theoretical work is required to assign the excited states of the 1-indanyl isotopologues.

5.2.2 Hydroxyl and OD Addition to Indene

The m/z 133 trace, later identiﬁed as the 2-hydroxy-indan-1-yl, was also observed in a discharge of indene by Troy. [123, 189] By repeating this experiment, and introducing both H2O and D2O into the path of the carrier gas, our expectation was to reproduce the spectrum of the 2-hydroxy-indan-1-yl radical and observe the isotopologue formed from OD addition.

The experimental and discharge conditions required to generate the 2- hydroxy-indan-1-yl isotopologues were nearly identical to those used to form the 1-indanyl isotopologues, described in section 5.2.1. As the IP of 2-hydroxy-indan- 1-yl was determined to be higher than that of the 1-indanyl radical, the ionisation source was replaced by the fourth harmonic (266 nm) of a BRILLIANT B, Nd:YAG laser. [134]

Figure 5.3 exhibits the D1 excitation spectra recorded for the m/z 133 and m/z 134 carriers, corresponding to the isotopologues of the 2-hydroxy-indan-1-yl radicals. The spectrum taken by Troy is included for comparison, with an origin at 21364 cm−1, spanning some ∼750 cm−1 and containing rich vibrational structure.

Though some of the features of the m/z 133 trace are missing, (indicated by the red asterisks), the majority of the 2-hydroxy-indan-1-yl spectrum has been

125 Indene + OD (m/z 134)

Indene + OH (m/z 133)

* * * * * *

H/D

2-hydroxy-indan-1-yl adapted from Troy

0 50 100 150 200 250 300 350 400 450 500 550 600 650 700 Relative Wavenumbers (cm-1 )

Figure 5.3: The excitation spectrum of the m/z 133 and 134 traces, corresponding to OH and OD addition to indene. Spectrum of the 2-hydroxy-indan-1-yl radical reproduced from Troy et al. [134] successfully reproduced, and we can be conﬁdent in our observation of this radical species. The origin of the m/z 134 species was observed at 21363.5 cm−1, corresponding to a blue shift of 0.5 cm−1. The most distinct diﬀerence in vibrational structure between the two isotopologues is the 50 cm−1 red shift of the band at 246 cm−1 to 196 cm−1, indicated by the dotted line. Further, fewer lines are observed in the region between 375 cm−1 and 575 cm−1.

Troy calculated the barriers to the internal rotation of the C-OH group, and found two additional rotomer conﬁgurations lying 46 cm−1 and 320 cm−1 higher

126 in energy. He suggested that these configurations may be required to assign the complex spectrum. The absent features may very well indicate the absence of a rotomer configuration in our experiment. In chapter 3, two configurations of the hydroxycyclohexadienyl radical were observed with origins seperated by only 18 cm−1. Hole burning spectroscopy was demonstrated to be an effective tool in separating the spectrum, and the same experimental technique is advised for the complex 2- hydroxy-indan-1-yl spectrum.

By substituting the hydroxyl group by OD, the energy landscape of the internal rotor would change based on the increased moment of inertia of the OD group. As a result, there may be fewer configurations available to the the m/z 134 carrier, which could account for the numerous missing peaks when compared to the m/z 133 trace. Hole burning spectroscopy could be used again to confirm the number of configurational isomers present in this region.

5.2.3 Chemical Implications

Following the observation of the 2-hydroxy-indan-1-yl isotopologues, we attempted to repeat the spectrum to observe radicals formed from the OH/OD addition to benzene, toluene, naphthalene and 2-methylnaphthalene. Despite a range of experimental conditions, we were unable to identify any signatures of radicals with +17/+18 amu, corresponding to OH/OD addition. Unlike the aforementioned precursors, indene possesses a double bond external to the aromatic ring and it is at this site that OH addition occurs. The 2-methylindene and 2-ethylindene precursors both have this same property, and their corresponding OH addition products, the 2-hydroxy-2-methyl-indan-1-yl and 2-ethyl-2-hydroxy-indan-1-yl radicals, have both been observed. [134]

An investigation of the discharge products from water vapor and the precursor 1,3-cyclohexadiene was then carried out. The choice of 1,3-cyclohexadiene was two-fold, not only did the precursor contain a non-aromatic double bond, but addition of H/OH would yield a radical with an allylic chromophore. To recall, the primary deactivation pathway of the D1 state of the allyl radical involved con- and

127 disrotatory motions of the terminal CH2 groups, allowing the molecule to return to the D0 state through conical intersections.[162] In this case, the allylic chromophore would be embedded in a carbon ring, and rotation of the ring to facilitate deactivation would incur a large energetic penalty. Searches for the D1 state in the ∼400-440 nm region, and for higher excited states in the ∼240-280 nm region were conducted, however no spectral traces with +1/+17 amu were observed.

The 1,3,5-cycloheptatriene radical was also investigated in a discharge containing water vapor, as H/OH addition would form a 5-membered chromophore, similar to that observed for the hydroxycyclohexadienyl radical in chapter 3. Whilst a spectrum for the +1 carrier, assumed to be the hydrogen addition product, was recorded (not shown), a +17 trace was never identiﬁed. In conclusion, despite the observation of hydroxyl (and OD) addition to indene, and the previous work done on hydroxyl addition to substituted indenes by Troy, we have not been able to successfully observe the hydroxyl-addition adduct for any other precursor.

5.3 Observation of Hydroxy Isoprenyl Radicals

5.3.1 Precursor Selection

The discharge conditions required to generate the 1-indanyl and 2-hydroxy-indan-1- yl from a discharge of water vapor and indene were near identical. As determined by the experiments in section 4.5, multiple dimethylallyl products were observed from a discharge containing water vapour and isoprene, however no trace corresponding to hydroxyl addition to isoprene (m/z 85) was ever observed in these experiments.

An alternative means of producing the desired OH-adducted radicals is by following the approach used for the 1,1-dma and 1,2-dma radicals, and discharging an appropriate halogenated precursor already containing the OH moiety. As for 1,2- dma case, no commercial precursors were available and hence synthesis was required. Atmospheric addition of hydroxyl can occur at one of the four places described in the top right of ﬁgure 5.4. Addition is most likely to occur at sites 1 and 4, generating the

128 radicals 1-OH and 4-OH respectively. The dissociation schemes for the precursors, used to form the atmospherically relevant radicals are also shown.

I (+ I )

OH OH OH 2 4 1 3 1p : 2-iodo-2-methylbut-3-en-1-ol 1-OH

OH OH (+ I )

I OH

2p : 1-iodo-2-methylbut-3-en-2-ol 2-OH

(+ Br )

Br OH OH OH

4p : (E)-4-bromo-3-methyl-2-buten-1-ol 4-OH

Synthesised Precursor Hydroxy Isoprenyl Radicals Atmospheric Reaction

Figure 5.4: The synthetic and atmospheric routes to the formation of the hydroxy isoprenyl radicals

The synthetic route to form the precursors 2-iodo-2-methylbut-3-en-1-ol (1p) and (E)-4-bromo-3-methyl-2-buten-1-ol (4p) was chosen following discussions with Associate Professor Jason Harper, and the synthesis and identification of products was conducted in our lab under the supervision of Dr. Joe Gallagher. The 1p precursor was synthesised following the procedures of Ghosh et al., based on earlier work conducted by Snider et al. [81, 190]. The largest impurity in the sample was determined to be one of the reactants, acetonitrile. A fortunate result of the synthesis was that both the tertiary and primary alcohol precursors, 2p and 1p, were produced, in a 3 : 1 ratio respectively. Whilst our mass spectrometer cannot distin- guish between these products, they will have differing spectral signatures, which can be separated by techniques such as hole burning, see chapter 3. The 4p precursor was synthesised and purified according to Kuk et al. [191] Ethyl acetate and hexane, used as solvents during column chromatography, were identified as impurities in the final product.

129 5.3.2 Observation of 4-OH

To generate the 4-OH radical, both discharge and photolysis techniques were employed to cleave the C-Br bond within the 4p precursor. The range of experimental conﬁgurations utilised in attempt to observe a REMPI scheme for the 4-OH radical will be presented, along with a discharge dependent enhancement of the m/z 85 signal.

The experimental conﬁguration for discharge experiments has been described previously. To increase the seed ratio of the precursor, the backing pressure of argon was lowered (from a typical 5-7 bar) to as low as 3 bar. By passing the carrier gas through an internal sample container of the 4p precursor, the vapor was collected. The PDN apparatus has been described earlier in section 2.1.2, and here the discharge voltage was applied to arc from both the inner and outer electrodes (in seperate experiments), timed to strike various parts of the expanding gas. The voltage applied to the discharge ranged from 1000 - 2090 V, which was passed through a ballast resistor of 1-25.5 kΩ, and lasted for 100-320 µs. The lower voltage conditions were typically used when arcing from the inner electrode to maintain a steady discharge. The nozzle pulse was varied between 200 - 320 µs depending on the backing pressure, to maintain a consistent pulse. All experiments were run at 10 Hz.

The chromophore and structure of the 4-OH radical is analogous to the 2-ma radical investigated in section 4.4, with a terminal hydrogen substituted for a hydroxyl group. As discussed in section 4.3, a comparison between the electronic origins of the D1 state of the cyclohexadienyl radical and the hydroxycyclohexadienyl radical revealed a shift of only 8.5 cm−1.[133, 179] This suggests that the 4-OH radical and the 2-ma radical may similarly absorb in the same spectral region. Cal- culations on the excited states of the various hydroxy isoprenyl radicals encountered convergence failures similar to those found for the dimethyl allyls, relating to the internal rotors, and were inconclusive.

130 Signal corresponding to a m/z 85 carrier was first identified when an ionisation source of ∼210 nm light (1.2 mJ pulse−1) was focussed onto the molecular beam with a 200 mm focussing lens. The ionisation photons were provided by the doubled output of an Nd:YAG pumped dye laser, circulating the Exalite 428 dye. The signal on m/z 85 was observed to be enhanced by the discharge, as seen in figure 5.5. The oscilloscope was set to encompass the m/z 72 - 90 range. The red and black traces were taken with the discharge off and on respectively. The red trace indicates the background signal of each m/z carrier, non-resonantly ionised by the focussed UV light. Additional activity was observed in other mass channels, however without further spectroscopy, these species remain unidentified.

DC ON 4-OH DC OFF 15

5 Relative(mV) Intensity

Mass Per Charge Ratio (m/z)

Figure 5.5: Discharge dependent enhancement of the m/z 85 carrier, 4-OH

The continual production of bromine atoms (formed from the C-Br bond breaking) was monitored by periodically adjusting the photon wavelength to coincide with bromine atomic resonances on the m/z 79 and 81 channels in the 250 nm region. An appreciable increase in the m/z 85 trace is observed when the discharge is turned

131 on. Alternate UV sources were used for non-resonant ionisation; such as 266 nm radiation, obtained from the fourth harmonic of a BRILLIANT B, Nd:YAG laser, and the doubled output of an Nd:YAG pumped dye laser, circulating the Coumarin 480 and 503 dyes. Focussing the ionisation radiation was required to obtain non- resonant m/z 85 signal.

As ethyl acetate (mass 88.11) and hexane (mass 86.18) were both observed as impurities in the final product 4p, we must be certain that both species do not contribute to the m/z 85 signal, as the discharge is very capable of stripping hydrogen atom(s) from a molecule. The ionisation energies of ethyl acetate and hexane have been reported as 10.01 ±0.05 eV and 10.13 ±0.10 eV respectively. [192, 193] As the ionisation pulse of ∼210 nm light (∼5.904 eV) was focussed directly onto the molecular beam with sufficient energy, we expect that non-resonant 2 photon absorption can occur. The combined energy of two such photons is equivalent to ∼11.808 eV, considerably higher than the ionisation energies of ethyl acetate. With the discharge turned off, the m/z 86 and 88 channels should show significant signal if present within the molecular beam. Further, their corresponding signal should decrease when the discharge is turned on due to fragmentation. As these events do not correspond with those observed in the m/z 86 and 88 channels, we can be sure that the parent molecules are not present in sufficient concentration to account for the increased signal observed on the m/z 85 channel.

The formation of discharge dependent m/z 85 signal, and corresponding identiﬁcation of the atomic bromine co-product, formed from dissociation of the 4p precursor, presents an argument for the production of the 4-OH radical in the molecular beam.

Following observation of the 2-ma radical in the 38500 - 41000 cm−1 region, REMPI schemes for the 4-OH radical were sought in this same region. This region was explored with focussing regimes (tight and loose focussing) and without focussing. Experimental parameters, such as the discharge conditions and backing pressures, were varied within the ranges presented above. Though m/z 85 signal was identiﬁed to be discharge enhanced, no REMPI scheme was obtained for any higher electronic state.

132 The 435 - 404 nm (∼23000 - 24750 cm−1) region was also explored in hope of observing the D1 state of the 4-OH radical; as the allyl, 1-ma and 1,1-dma radicals

D1 states were observed in this region. As the 1,1-dma radical was ﬁrst observed by a 1+2 REMPI scheme, a 200 mm focussing lens was inserted into the laser path, focussed on the molecular beam. Additionally, 266 nm and (412-435)/2 nm radiation was utilised as alternative ionisation sources for 1+10 REMPI schemes. Despite the various laser setups and REMPI schemes employed, resonant enhancement of the m/z 85 trace was not identiﬁed in this range.

An alternative way of breaking the C-Br bond in the 4p precursor is by laser photolysis. In these experiments, the discharge was turned off, and a photolysis laser (of 266 nm) was aligned either down propagation axis of the molecular beam, or orthogonal to the molecular beam and counter-propogating with the excitation/ionisation laser. The photolysis laser was focussed with a 200 mm focussing lens, and was fired ∼50 ns prior to the excitation laser. The mass spectrum obtained from the photolysis pulse is represented by the blue trace in figure 5.6.

25 Photolysis Only (266 nm) 25 2 Laser Signal 2 Laser Signal Blue + Red Trace Excitation/Ionisation 20 Only (252 nm) 20

15 15

10 10

Relative(mV) Intensity 5 Relative(mV) Intensity 5

0 0

81 82 83 84 85 86 87 81 82 83 84 85 86 87 Mass Per Charge Ratio (m/z) Mass Per Charge Ratio (m/z)

Figure 5.6: Left: Mass traces from varying laser sources Right: Comparison between the 2 laser signal mass trace, and the sum of the two individual mass traces

The black trace in ﬁgure 5.6 represents the mass spectrum when both the photolysis and UV laser (providing excitation and ionisation photons at 252 nm) were turned on, showing an enhancement in the m/z 85 channel. Turning oﬀ the photolysis spectrum yielded the red trace.

The right hand side of ﬁgure 5.6 includes the black trace from the left side ﬁgure, accompanied by a purple trace representing the direct sum of the red and

133 blue traces. In the event of resonant enhancement, we would expect the black trace to be larger than the sum of the individual traces, but as the two traces are equal, we can conclude that the blue and red traces represent separate photolysis events. The UV region of 38500 - 41000 cm−1 was explored again for the m/z 85 carrier following photolysis, though no resonant spectrum was obtained across this range.

5.3.3 Observation of 1-OH and 2-OH

The 1p and 2p precursors, seen in ﬁgure 5.4, were synthesised in a ratio of 1 : 3. The crude sample was added directly to the sample container, and upon expansion into the chamber, a discharge was used to break the C-I bond. REMPI spectroscopy was used to search for the desired 1-OH and 2-OH radicals. The same experimental setup as the previous section was employed. The discharge voltage, ballast resistance and the width of the discharge pulse sampled a larger range of conditions however, and the discharge was applied to varying regions of the expanding sample, not just the latter part.

On the addition of hydrogen to isoprene, higher excited states of both the 1,1-dma and 1,2-dma were observed in the region of 38500 - 41000 cm−1. Additional peaks were observed within this spectral region, which were tentatively attributed to the inner addition products.

To initially observe non-resonant mass signatures, including the desired m/z 85, ionising radiation of ∼ 245 nm was focussed onto the molecular beam with a 250 mm focussing lens. Following detection of m/z 85, the region of 38500 - 41650 cm−1 was probed in search of the higher excited states of the 1-OH and 2-OH radicals, which are respectively analogous to the outer and inner addition products of the hydroxyl addition to isoprene.

The ionising radiation was later replaced by a 203 nm source. This wavelength proved most eﬀective in ionising the m/z 85 trace, and the discharge dependent mass trace, gated from m/z 78 - 99, is shown below in ﬁgure 5.7.

134 12 Discharge ON Discharge OFF 10

8 1-OH / 2-OH 6

2 Relative(mV) Intensity 0

-2

-4

Mass Per Charge Ratio (m/z)

Figure 5.7: Discharge dependent enhancement of the m/z 85 carrier, containing 1-OH and 2-OH

In ﬁgure 5.7 we can clearly see the discharge dependence of all mass peaks. As acetonitrile (m/z 41) was identiﬁed in the crude sample, we can tentatively attribute the m/z 82 carrier to the acetonitrile dimer. The co-product of 1-OH and 2-OH dissociation, iodine, was observed at m/z 127 with a similar discharge dependence. Despite an exhaustive search over the UV region, no resonant signal was observed for the m/z 85 trace. The sample (synthesised on multiple occasions) was depleted before a search in the 404-435 nm region could be undertaken.

The 1-OH radical is the hydroxylated analogue of the 1,1-dma radical, examined in section 4.3. Whilst substitution of the hydrogen for a hydroxyl group at the 1- position reduces the number of methyl rotors, it introduces another internal rotor, of the OH group as observed in chapter 3. We expect that the spectrum pertaining to the 1-OH radical will continue to exhibit methyl rotor coupling to the electronic transition. In addition, multiple rotomers of the OH group may form

135 in the cooled molecular beam, evidenced also by our calibration experiment of the 2-hydroxy-indan-1-yl radical, presented in section 5.2.2. [134]

Observation of a discharge dependent m/z 85 signal, formed from the dissociation of the 1p and 2p precursors, and the identiﬁcation of discharge dependent indene atoms presents an argument for their presence (either or both of the 1-OH and 2-OH radicals) in the molecular beam. However, spectroscopic characterization was not forthcoming.

5.3.4 Conclusion and Future Work

We have directly measured the class of hydroxy isoprenyl radicals for the ﬁrst time.

Enhanced signal was observed on the m/z 85 channel, (·C5H8OH), following the discharge of the precursors (E)-4-bromo-3-methyl-2-buten-1-ol, 2-iodo-2-methylbut- 3-en-1-ol and 1-iodo-2-methylbut-3-en-2-ol. Photolysis of the precursor 4p at 266 nm was also employed to cleave the C-Br bond. The colour and UV regions were probed by REMPI spectroscopy in search of the various electronic states of the hydroxy isoprenyl radicals, however despite all attempts, no resonance enhancement was obtained.

The mass spectral identification of the hydroxy isoprenyl radicals here pro- motes the use of our experimental setup as an effective way to form specific reaction intermediates in the gas phase. Investigation into the photo-excitation pathways of these class of radicals will provide valuable information for future experiments. Lately efforts have been made toward creating a vacuum-ultraviolet source which could 1-photon ionise the majority of species within the molecular beam. This would allow us to use hole burning as a way of indirectly observing short lived excited states. Early experiments have proven promising in the identification of 2-methylallyl, 1-photon ionised by 118 nm radiation, as a discharge product of 3- chloro-2-methyl-1-propene, in keeping with the calibration experiments performed in section 4.2.3.

136 5.4 Matrix Studies

The short lifetime of the excited states of the substituted allyls discussed in this work pose a signiﬁcant experimental challenge for nano-second REMPI experiments. As our primary goal is the detection of these radical intermediates, FTIR spectroscopy can be employed to identify these radicals by their vibrational spectra in a p-H2 matrix. The Lee group has proven their ability in their dissociation of carbon- halogen bonds by UV photolysis in this soft matrix, which allows for the spacial separation of the photolysis products. [194, 195] Our synthesised precursors, 1p, 2p and 4p, were indeed chosen on the basis that photolysis/discharge could break the relevant carbon-halogen bond, and yield the desired 1-OH, 2-OH and 4-OH radicals respectively.

The Lee group has also observed peroxy type radicals in a p-H2 matrix, such as iodomethylperoxy (syn-ICH2OO). [194] This was done by seeding O2 into the premix gas. After irradiation of the precursor I2CH2, molecular oxygen trapped within the matrix allowed for in situ O2 addition to I2CH2. The vibrational spectrum of syn-ICH2OO and residual I2CH2 was then recorded. After annealing, the residual

I2CH2 signal was observed to decrease, accompanied by an increase in syn-ICH2OO signal as O2 was allowed to move more freely through the matrix. Following our detection of the 1-OH, 2-OH and 4-OH radicals within the matrix, by the addition of O2 to the premix gas, the plan was to recreate the second stage of atmospheric oxidation, and generate a suite of hydroxy-isoprenyl-peroxy type radicals.

137 5.4.1 Experimental

Two samples were sent to Taiwan by air freight. The ﬁrst consisted of the 4p precursor, and the second contained the crude products of 1p and 2p. Unfortunately, the samples were held up in Taiwanese customs and were not received until ∼ 1 month after shipping. Upon acquiring the samples, thin-layer chromatography was conducted on the precursors. The precursors were identiﬁed, however a multitude of other spots were observed on the TLC plate, indicating that an unknown amount of the sample had decomposed during transit. TLC was conducted before each consecutive experiment, and additional/darker spots were observed each time.

As there was a limited time in which to conduct the experiments, experiments were then directed toward only one species. The more pure sample of 4p was chosen over the crude sample containing 1p and 2p.

The generation of p-H2 and the setup for matrix experiments have been previously described in section 2.2.2. The vapor of 4-bromo-3-methyl-2-buten-1-ol was collected and mixed with p-H2 in a ratio of 1:1000. The sample was deposited over a period of 8 hours, at a rate of 11-12 mmol h−1, onto a gold plated copper substrate, held at 3.2 K by a closed-cycle helium cryostat refrigerator. The concentration of sample within the matrix was monitored hourly with FTIR. During deposition, irradiation and secondary photolysis, spectra were recorded as an average of 200 scans, at a resolution of 0.25 cm−1. The IR beam path was purged with

N2 for ∼30 minutes to remove trace amounts of H2O, CO2 and other impurities.

To generate the 4-OH radical in the matrix, the 4p precursor was photolysed with 248 nm to cleave the C-Br bond, as shown in ﬁgure 5.8. The second harmonic (248 nm) from a COHERENT COMPexPro 50/102f laser was used as a photolysis source, and was run at 10 Hz.

The sample was irradiated minute by minute for a total of 3 minutes. After each minute the matrix was inspected visually (to ensure that it had not turned opaque or evaporated) and by FTIR spectroscopy. The matrix was then subjected to secondary photolysis in order to attribute carriers to the photolysed

138 photolysis + Br Br OH 248 nm OH

4p : (E)-4-bromo-3-methyl-2-buten-1-ol 4-OH

Figure 5.8: Photolysis of (E)-4-bromo-3-methyl-2-buten-1-ol, yielding 4-OH and a bromine atom. and generated spectral lines. Various LED’s provided photolysis wavelengths of 520 nm, 445 nm, 405 nm and 365 nm. Wavelengths of 308 nm and 214 nm were provided by a zinc lamp, and 254 nm by a mercury lamp, by using various bandpass ﬁlters. The matrix was photolysed by each source in turn, from low to high energy, for 1-5 minutes depending upon the signal response.

139 5.4.2 Theory

The relative energies of the various conﬁgurations of the 4p precursor were calculated using DFT theory, and the results are shown in ﬁgure 5.9. The B3PW91/aug-cc- pVTZ method and basis were chosen they treat halogenated molecules well, and to remain consistent with other theoretical work conducted in the Lee group.[196, 197] The GAUSSIAN09 program was used for all calculations, excluding re-optimisation of the radical 4-OH to the Cs symmetry point group, where GAUSSIAN 16 was used.[135, 198] These calculations were completed in collaboration with Dr. Karolina Haupa, a post doctoral researcher in the Lee group.

1.095 1.094

1.095 1.084 1.094 1.088

1.504 1.503 1.090 0.959 1.090 1.095 1.416 1.090 1.330 1.330 1.504 1.501 1.492 1.090 1.490 1.099 1.100 0.960 1.948 1.952 1.424 1.085 1.087 1.100 Br Br

precursor (i) precursor (ii) Δ E = 15.4 kJ mol -1 Δ E = 10.4 kJ mol -1

1.095 1.094 1.093 1.083 1.093 1.088

1.497 1.498

1.987 Br 1.985 Br 1.094 0.959 1.088 1.413 1.336 1.490 1.337 1.088 1.487 1.489 1.100 1.085 1.492 1.085 1.101 1.087 1.423 1.088 0.960 1.100

precursor (iii) precursor (iv) Δ E = 0.0 kJ mol -1 Δ E = 8.3 kJ mol -1

Figure 5.9: Geometric conﬁgurations and relative energies of the D1 states of the 4p rotomers.

140 Of the four configurations, precursor (iii) is the lowest in energy, by some 8.4 kJ mol−1. We can therefore expect this isomer to appear with the greatest abundance in our spectra. Following these calculations, the bromine atom was removed, and the geometry was re-optimised for each configuration. Surprisingly, each of the four configurations converged to the same geometry, described in figure 5.10. Further attempts were made to explore the energy surface of the OH torsion, however convergence (without imaginary frequencies) was only achieved for the one geometry.

1.093

1.089 1.093

1.507

1.102 119.61 119.48 124.65 1.102 1.082 1.379 108.69 1.391 1.485 120.91 108.60 1.418 0.959 1.083 1.084

Figure 5.10: Geometric details for the D0 state of the 4-OH radical

The D0 state of the 4-OH radical is assigned to Cs symmetry, with the carbon and oxygen framework lying within the plane of symmetry. By comparison with precursor (iii), after the C-Br dissociation, the terminal CH2 group rotates to line up with the plane of symmetry. Further, the methyl group rotates by ∼50◦.

◦ The terminal CH2OH group also rotates, by ∼25 , such that the OH group also lies within the plane of symmetry.

The D0 frequencies for both the precursor (iii) and the radical 4-OH are listed in table 5.1. The frequencies are labelled in the Mulliken convention. Precursor

(iii) is determined to have C1 symmetry, and as such frequencies are numbered in 0 decreasing order. For radical 4-OH, of Cs symmetry, the a modes are labelled in a decreasing order, followed by the a00 modes in a similar order. The frequencies have been scaled here by 0.9646, which is the recommended scaling factor for isoprene at the b3pw91/aug-cc-pVTZ level of theory.[199]

141 Table 5.1: List of frequencies for the D0 state of Precursor (iii) and Radical 4-OH calculated at the b3pw91/aug-cc-pVTZ level of theory. All values are in wavenumbers (cm−1). Harmonic frequencies, scaled by 0.9646, have been included for comparison with experiment. [199]

Precursor (iii) Radical 4-OH

Mode # Frequency Scaled Freq. Mode # Frequency Scaled Freq. Mode # Frequency Scaled Freq. Symm. Mode # Frequency Scaled Freq. Symm.

42 39.9 38.5 21 1215.0 1172.0 39 18.7 18.0 a00 18 1077.3 1039.2 a0 41 67.9 65.5 20 1231.0 1187.4 38 124.7 120.3 a00 17 1099.4 1060.5 a0 40 108.1 104.3 19 1233.6 1189.9 37 183.8 177.3 a00 16 1238.1 1194.3 a0 39 141.3 136.3 18 1278.3 1233.0 36 261.5 252.2 a00 15 1274.0 1228.9 a0 38 182.7 176.2 17 1363.2 1314.9 35 480.0 463.0 a00 14 1373.1 1324.5 a0 37 240.3 231.8 16 1412.9 1362.9 34 565.3 545.3 a00 13 1399.4 1349.9 a0 36 284.3 274.2 15 1441.2 1390.2 33 683.5 659.3 a00 12 1414.9 1364.8 a0 35 358.5 345.8 14 1464.3 1412.5 32 795.1 767.0 a00 11 1465.4 1413.5 a0 34 419.2 404.4 13 1476.7 1424.4 31 985.8 950.9 a00 10 1487.6 1434.9 a0 33 488.7 471.4 12 1480.5 1428.1 30 1049.2 1012.1 a00 9 1496.5 1443.5 a0 32 541.4 522.2 11 1496.8 1443.8 29 1232.8 1189.2 a00 8 1517.9 1464.2 a0 31 621.9 599.9 10 1731.8 1670.5 28 1478.9 1426.5 a00 7 2933.2 2829.4 a0 30 800.9 772.5 9 2969.7 2864.6 27 2943.5 2839.3 a00 6 3030.0 2922.7 a0 29 860.5 830.0 8 3032.8 2925.4 26 3083.3 2974.2 a00 5 3123.6 3013.0 a0 28 910.2 878.0 7 3036.8 2929.3 25 177.0 170.7 a0 4 3146.7 3035.3 a0 27 984.2 949.4 6 3085.8 2976.6 24 346.7 334.4 a0 3 3177.2 3064.7 a0 26 1026.2 989.9 5 3097.0 2987.4 23 451.6 435.6 a0 2 3241.2 3126.5 a0 25 1052.9 1015.6 4 3137.1 3026.0 22 526.9 508.2 a0 1 3863.0 3726.2 a0 24 1054.1 1016.8 3 3147.1 3035.7 21 859.5 829.1 a0 23 1094.6 1055.9 2 3172.3 3060.0 20 972.8 938.4 a0 22 1164.5 1123.3 1 3855.4 3718.9 19 1014.0 978.1 a0 The scaled harmonic frequencies and IR intensities for both the precursor (iii) and the radical 4-OH have been plotted for direct comparison with experiment, see ﬁgure 5.11. The lower limit of this FTIR is ∼550 cm−1, and so the plot has been truncated. The spectral lines have been broadened with a Gaussian function, with a FWHM value of 5 cm−1 for purposes of illustration.

(4-OH)

Figure 5.11: Scaled D0 vibrational frequencies for the precursor (iii) and the radical (4-OH)

The vibrational frequencies have been initially scaled by a factor of 0.9646, and revised scaling factors for both the low and high-frequency range are generally made following assignment.[199] The most prominent features of both spectra reside in the 1000 cm−1 - 1250 cm−1 range. Additionally, the C-Br stretching mode unique to (iii) is predicted at 599.9 cm−1.

143 5.4.3 Results and Discussion

A partial IR spectrum of precursor (iii), following 8 hours of deposition, is shown in the upper trace (blue) of figure 5.12. The most prominent impurities in the spectrum are those of hexane and ethyl acetate, confirmed by comparison with argon matrix and gas phase spectra of these radicals. [200, 201, 202] These species, especially ethyl acetate, are responsible for the most intense features in the spectrum, including the large vibrational structure near 3000 cm−1. In later experiments, efforts were made to further reduce the concentration of hexane and ethyl acetate in the sample, by vacuum distillation, however by then the precursor had all but degraded.

The middle trace (red) was taken following 3 minutes of photolysis at 248 nm. The diﬀerence between these two traces is shown in the bottom (purple) trace. After taking into account peaks belonging to the aforementioned contaminants, the remaining spectral lines which were depleted following photolysis are indicated with a faint blue band. These are seen as a negative peak in the purple trace. Pho- tolysis was stopped before the peaks disappeared completely, otherwise secondary photolysis could not be conducted on these same lines.

There are a number of instances throughout the spectrum where a sharp positive/negative paired peak are observed, and this phenomenon is attributed to a redistribution in the rotational proﬁle, rather than a photolysed/new feature. The spectral region ∼3000 cm−1 is convoluted with structure attributed to ethyl acetate, and so the two features at 2986.6 cm−1 and 2996.6 cm−1 (shoulder) are only tentatively assigned to a photoliable precursor. The peak at 699.5 cm−1 is likewise tentative. Across the spectrum, the small noise in the diﬀerence (purple) trace comes about due to changes in water vapour concentrations in the FTIR beam path. Though the volume is purged with N2, small leaks occur.

144 145 iue5.12: Figure ieidctspaswihhv eltdi nest fe raito.Genln niae ek hc aefre olwn irradiation. following formed have which peaks indicates line Green irradiation. after intensity in depleted have which peaks indicates line Absorbance (km mol -1 ) pe Trace: Upper eoiin( hours). (8 Deposition ideTrace: Middle htlss( iue,28nm). 248 minutes, (3 Photolysis Wavenumber (cm -1 ) oe Trace: Lower ieec Pooyi eoiin.Blue Deposition). - (Photolysis Diﬀerence Table 5.2: Peak positions observed to decay (precursor) and form (radical) in the matrix. All units are in wavenumbers (cm−1).

Precursor Radical

3574.6 * 934.8 * 3623.2 2996.6 817.7 * 3464.9 2986.6 * 773.1 * 3171.1 † 1663.7 * 699.5 * 1350.9 † 1016.9 1113.1 † 623.7 †

New features, produced after the photolysis, are indicated with a faint green band. Only spectral features which did not overlap with preexisting lines have been marked. These features are much less prominent across the spectral range, indicating that only a very small population of the radical has been produced. The observation of two new lines in the OH stretching region is contrary to the results of the DFT calculations, where only one conﬁguration of 4-OH is calculated to be stable.

The full list of peaks which have been observed to decay (precursor) and form (radical) are listed in table 5.2. The matrix was subjected to secondary photolysis and the behaviour of the lines was recorded. The intensity of a number of the observed peaks is barely above the level of noise in the experiment, and so response from secondary photolysis (if it even occurs) may well lie within level of noise, precluding observation.

Secondary Photolysis

Secondary photolysis was conducted with wavelengths of 520 nm, 445 nm, 405 nm, 365 nm, 308 nm, 254 nm and 214 nm. Of the lines that diminished upon primary photolysis, only those indicated with an asterisk in table 5.2 showed a response to a secondary photolysis wavelength, and only to 365 nm. The region between 1400 cm−1 - 900 cm−1 has been used to illustrate this secondary photolysis behaviour, see ﬁgure 5.13. Here the two precursor lines at 934.8 cm−1 and 1016.9 cm−1 are shown

146 to diminish further upon secondary photolysis. After the 365 nm irradiation, the peaks indicated with asterisks disappeared completely from the FTIR trace, and so no further tests could be done. ) -1 Absorbancemol (km

Wavenumber (cm -1 )

Figure 5.13: Results of the secondary photolysis at 405 nm and 365 nm. The purple trace has been adapted from the diﬀerence trace in ﬁgure 5.12. The green and blue lines indicate peaks which have increased and decreased in intensity following 405 nm and 365 nm irradiation, respectively.

The radical lines only responded to 405 nm light, and only those indicated with a dagger (†). These peaks were observed to increase in intensity following 405 nm irradiation. In figure 5.13, the lines at 1113.1 cm−1 and 1350.9 cm−1 which exhibit this behaviour are indicated with faint green bands. Neither of the spectral features at 3623.2 cm−1 or 3464.9 cm−1 were affected in any way by any of the secondary photolysis wavelengths. As their peak intensity is very weak, we cannot be certain that they were not affected by the 405 nm wavelength. However at this point we cannot pair either OH stretching frequency to the group of radical lines which grew upon 405 nm irradiation.

The peak positions and intensities recorded for the precursor do not match with those predicted by theory, see ﬁgure 5.11. The lowest frequency mode observed

147 is at 699.5 cm−1, which is ∼100 cm−1 higher in energy than the closest theoretically predicted value of 599.9 cm−1. Of the three very intense bands predicted between 1000 cm−1 - 1200 cm−1, only two bands are observed in/near to this region, at 934.8−1 and 1016.9 cm−1. Further, these bands have already been determined by secondary photolysis to relate to two diﬀerent species. There is indeed no evidence that the bands attributed to a photoliable precursor can be assigned to the 4-bromo- 3-methyl-2-buten-1-ol molecule. As a result, we cannot attribute the radical lines to the photo product of this precursor. This evidence is furthered by the secondary photolysis results for the radical, where no OH stretching frequency could be paired with any other line, preventing an assignment to the 4p radical.

5.4.4 Concluding Remarks

Further experiments were conducted with varying concentrations, sample preparation, deposition rates, annealing, temperature and photolysis sources; however neither precursor or radical spectra were ever reproduced or able to be matched to theory. Many of the features in each spectra were observed to decay or grow in intensity with time and annealing, indicating that polymerisation may have been another decay route for the 4-bromo-3-methyl-2-buten-1-ol precursor.

As the excited states of the hydroxy isoprenyl radicals have yet to be observed, likely due to short lifetimes and rapid inter-conversion followed by dissociation as exhibited by other substituted allylic radicals, matrix spectroscopy remains a viable option. Further studies should investigate alternative precursors, or pathways to form the hydroxy isoprenyl radicals. As other halogenated species of similar sizes have been investigated in the Lee laboratory, it is likely that the hydroxyl functional group was responsible for the degradation/polymerisation of the precursor. Further, more care should be taken in the sample preparation to ensure limited impurities, especially ones which absorb in the same regions as the precursor and desired radical.

148 Chapter 6

Epilogue

6.1 Conclusions of Part I

The work conducted in this part of the thesis provides valuable contributions to the fields of atmospheric chemistry and spectroscopy. Experimental methodologies were employed in the study of radical species with mass, isomer, and even rotamer specificity. This method of approach can be used to study various reactive intermediates that otherwise evade detection, or are grouped into the same mass or isomer channels. This work presents findings on atmospherically relevant radicals which play roles in urban and pristine environments, and forms a contribution to the efforts of elucidating the important hydroxy isoprenyl radicals, key to the formation of secondary organic aerosol.

The most important class of anthropogenic volatile organic compounds, responsible for 80% of the formation of urban secondary organic aerosol and photochemical smog, are the aromatic moleucles, in particular, benzene. [85, 87] The dominant loss mechanism of benzene in the atmosphere is initiated by addition of the hydroxyl radical, forming a hydroxycyclohexadienyl type radical. In chapter 3, a sample containing phenol and water was discharged, and two rotamers of the ortho-hydroxycyclohexadienyl radical were identiﬁed. The ortho radical has been

149 postulated as a reaction intermediate in the atmospheric degradation of benzene, by Mardyukov et al.[132]

The D1 ← D0 spectrum of the ortho-hydroxycyclohexadienyl radical found to be comprised of spectra pertaining to two individual rotamers, syn or anti, and so hole burning spectroscopy was employed to determine which rotamers were ac- countable for which peaks. No other isomer was found to absorb in this spectral

−1 region. The electronic origin of the D1 ← D0 transition was observed at 18200 cm for the anti isomer, and at 18218 cm−1 for the syn conformer.

Upon excitation the anti isomer was found to distort out-of-plane signiﬁ- cantly, resulting in a richer vibrational spectrum. The syn conformer was calculated to be puckered out-of-plane, lowering its symmetry from Cs to C1, however no single quantum a00 transitions were observed. Duschinsky matrices were constructed for each of the isomers, and it was found that upon excitation, the majority of anti modes contained mixed a0 and a00 character, conﬁrming the excited state to be of

C1 symmetry. For the syn rotamer however, the duschinsky matrix showed only one significant mode which contained both a0 and a00 character, which related to the formation of a local mode. Thus, by an analysis of the off diagonal elements of the matrix, and the lack of observed single quantum a00 modes, we assign the excited state to effective Cs symmetry. This is a great example of where duschinsky matrices were used as an effective tool to aid in understanding changes in symmetry upon electronic excitation. The observation of these species may aid toward the completion of the atmospheric degradation mechanism of benzene.

The most abundant biogenic volatile organic compound is isoprene, whose oxidation leads to the formation of ∼6 - 30 Tg of secondary organic aerosol each year, [26] accounting for up to 30% of the global budget for particulate organic matter.[66] The most dominant loss mechanism for isoprene is also initiated by an addition reaction with the hydroxyl radical. Despite the signiﬁcant contribution of isoprene derived secondary organic aerosol to the particulate organic matter budget, the radical species formed from this initial addition reaction had never been directly observed. The hydroxyl radical is expected to add almost exclusively to the terminal carbons of isoprene, forming resonantly stabilised radicals with allylic chromophores.

150 In order to detect these radicals, we undertook a study of the allyl and substituted allyl radicals, culminating in the observation of two new radical species 1,1-dimethyl allyl and 1,2-dimethyl allyl, radicals presumed to form from the addition of hydrogen to the 1- and 4- positions of isoprene respectively.

Isomer selective studies were undertaken to generate these two radicals individually. A premix gas containing one the appropriate halogenated precursors was discharged to break the required carbon-halogen bond, and form the radical product. The D1 ← D0 transition of the 1,1-dimethyl allyl radical was observed, with an origin at 24058 cm−1. The spectrum contained rich structure in the origin region, which was assigned to two methyl rotor progressions coupling to the electronic excitation, and to the CH2 twisting motion.

The spectrum was truncated after ∼200 cm−1, which is analogous to the truncated spectrum observed for the A˜ state of the 1-methylallyl radical by Gasser et al.[174] They attribute the limited vibrational structure to owe to the short lifetime of this state, which may also be the case for the 1,1-dimethyl allyl radical A˜ state.

The A˜ state of the 1,2-dimethyl allyl radical was sought in the same spectral region, however, as with the 2-methylallyl radical, no D1 ← D0 transition was observed. [170] For the 2-ma radical, this was attributed to the rapid internal conversion to the ground state through conical intersections, which may also be the case for the 1,2-dma radical.

The D2 ← D0 transition of the 1,1-dma radical was observed in the UV region of 38300 cm−1 - 40500 cm−1, with an origin at 38329 cm−1. In the same region, excitation from D0 to the electronic states D2 and D3 was observed for the 1,2-dma radical, with respective origins of 38501.2 cm−1 and ∼39600 cm−1. Following these studies, isoprene was discharged with water vapor, and a m/z 69 peak was observed and attributed to the hydrogen addition to isoprene. The UV region was investigated for this mass channel and a spectrum containing the signatures of the higher excited states of both the 1,1- and 1,2-dma radicals was observed. Additional features were also observed, and tentatively attributed to the inner addition products. This process and methodology was used to conﬁrm that the hydrogen atom indeed

151 added to both the terminal sites of isoprene, by comparison with the isomer specific reference spectra. The ionisation potentials of both the 1,1- and 1,2-dma radicals were measured to be 7.118 ±0.002 and 7.275 ±0.025 eV respectively. These data are key experimental findings, and can be used in time-resolved multiplexed photoionisation mass spectrometry experiments to determine effective branching ratios between isomers. [203]

Chapter 5 details our efforts towards the spectroscopic observation and identification of the hydroxy isoprenyl radicals. We began by demonstrating our efficacy at adding hydroxyl radicals to molecules with double bonds, in a series of calibration experiments on indene. Discharge products of m/z 116 and m/z 133 were observed and the carriers were confirmed as 1-indanyl and 2-hydroxy-indan-1-yl by comparison with studies conducted by Troy et al. [134, 189] These studies were extended by replacing water with heavy water, and mass peaks at m/z 117 and m/z 134 were identified, and their spectra assigned to the deuterated isotopologues. An assignment of the vibronic structure of these radicals is left as future work. Indene was replaced by isoprene and the experiments were repeated. Despite exhaustive efforts, a m/z 85 peak, corresponding to the formation of hydroxy isoprenyl radicals, were never observed.

In order to observe the hydroxy isoprenyl radicals, halogenated precursors were synthesised to yield three of the four possible addition sites, following a dissociation of the carbon-halogen bonds by the discharge. Indeed, discharge enhanced signal on the m/z 85 peak was observed for these precursors, under conditions which were used for the optimal production of the co-fragments, iodine and bromine, making this the ﬁrst direct measurement of the class of hydroxy isoprenyl radicals. Unfortunately, no spectrum could be obtained for any of the hydroxy isoprenyl radicals.

As the electronic spectra of the hydroxy isoprenyl radicals evaded observation, the precursors were sent to Taiwan, where matrix isolation spectroscopy was performed to elucidate the vibrational spectroscopy of these molecules. Un- fortunately, the samples decomposed in transit. The utilisation of MIS remains as a valid experimental technique to detect and study these radicals. An important

152 advantage of the matrix setup, is that once the hydroxy isoprenyl radical has been detected, oxygen (O2) can be seeded into the premix gas, and the peroxy addition to the hydroxy isoprenyl radical can be investigated.[194] This was the end goal of our experiments, had the initial detection of the hydroxy isoprenyl radicals yielded positive results, and is recommended as future work. Following observation, annealing of the matrix may allow for self abstraction of a hydrogen by the O2 group, forming an important QOOH intermediate.

This part of the thesis has provided contributions to the atmospheric degradation pathway of benzene, and spectroscopic insight on the use of duschinsky matrices as a valuable tool to assist in understanding changes in symmetry upon electronic excitation. The studies presented here have furthered our understanding of the role of substitution/methylation on the allyl radical and chromophore. We have devised and implemented methodologies to successfully observe isomer specific radicals, which are part of larger reaction mechanisms, such as the hydrogen addition to isoprene. Finally, we have detected, for the first time, a mass spectral identification of the class of isoprenyl radicals. We hope that the experimental and analytical techniques used within this thesis can be applied to other radicals and transient species, to shed light on a broader set of reaction mechanisms.

153 Part II

Cations in Space Chapter 7

Introduction

7.1 Composition of the Interstellar Medium

Far from empty, the vast regions occupying the space between stars contain around 10% of the total mass of our galaxy. This region is deﬁned as the interestellar medium (ISM), and consists primarily of gaseous (99%) and grain-like particles (1%). Contributions to the ISM originate from stellar events such as supernovae or the formation of planetary nebulae. [204] These events expel heavier atoms, such as carbon, nitrogen and oxygen, contributing 0.11% to the ISM, which is dominantly ﬁlled with hydrogen and helium (93.38% and 6.49% respectively).[205] The ISM is enriched by these events, whereafter collisions and agglomerations of the gaseous particles form carbon dust and organic nanoparticles. [206, 207]

Around 200 molecular species have been detected in the interstellar medium and circumstellar shells as of March 2018. [208, 209] The majority of these species were detected with radio telescopes and identiﬁed by rotational spectroscopy. Most of the detected species are limited to 1-13 carbon atoms, in part due to the experimental techniques used in their detection. Molecules with permanent dipoles may emit low energy radiation characteristic of their rotational transitions. These photons are detected in the radio to millimetre range of the spectrum, and can be iden- tiﬁed by comparison with spectra generated from laboratory experiments. Research

155 conducted at the Harvard-Smithsonian Centre for Astrophysics, led by Thaddeus and McCarthy, has resulted in the identiﬁcation of more than 50 molecules with astrophysical relevance.[210, 211, 212, 213]

Interstellar features have also been identiﬁed in the ultraviolet (UV), visible and infrared (IR) regions of the electromagnetic (EM) spectrum. Measurements within the IR region, yielding information about the vibrational energy levels and geometry of molecules, were often limited to the far IR due to contamination by absorption proﬁles of species like H2O and CO2, present in considerable concentrations in the atmosphere.[214] The introduction of orbital telescopes, such as the Spitzer Space Telescope, has allowed for a more thorough re-examination this spectral region, consequentially leading to the discovery of the fullerenes C60 and C70 in a young planetary nebula.[215]

Despite advancements in experimental techniques, telescopes and the launch- ing of additional space probes, there remain various interstellar features across the EM spectrum that have not yet been explained. These include the Red-Rectangle Bands (RRBs),[216, 217, 218] and the Diﬀuse Interstellar Bands (DIBs) which oc- cupy regions of the visible and near IR spectrum.[219] Other spectral regions containing unassigned features include the Anomalous Microwave Emissions (AME) and the Unidentiﬁed Infrared emission bands (UIRs).[220, 221] As a result, models of interstellar processes, such as the UMIST Database for Astrochemistry (UDfA), remain incomplete.[222]

7.2 The Diﬀuse Interstellar Bands

The DIBs represent a series of some ∼500 absorption lines stretching across the visible and near-IR (NIR) regions of the EM spectrum. These features reappear in spectra taken of stars which lie behind diffuse interstellar clouds, the first of which were observed in 1918, by Mary Lea Heger at the Lick Observatory, who was reporting the absorption lines of binary star systems. In 1934, Merrill returned to these diffuse lines, observing that they remained stationary with respect to the oscillating

156 features of other stars, commenting further that whilst they behaved as interstellar lines should in some regards, such as intensity, displacement and occurrence, they were widened with diﬀuse edges.[223] The observation of further bands quickly followed. [224, 225] Since this time various further studies, in particular by Herbig and Leka,([226] and references within) and Hobbs et al.,[227, 228] have extended the number of observed DIBs to ∼500.

The positive identification of a species as a carrier of a DIB requires more than just the wavelength and profile coincidence of a spectral band. DIBs are presently believed to be carbonaceous in nature, postulated to account for a portion of the cosmic carbon balance.[123] The lack of correlation between these bands argues for unique carriers, requiring origin dominated electronic transitions with no significantly extended vibronic structure.[229]

The intensity of the DIBs is another factor that must be sensibly accounted for by the oscillator strength of the proposed carrier. If we assume the carrier to be organic, modelled transitions should have an oscillator strength of at least f = 10−2. Larger oscillator strengths can be accounted for by presuming a smaller concentration of the carrier in the ISM. If a molecular transition has f ≤ 10−2, then this would require a substantially larger abundance of the carrier, which forms a contradiction with the expected abundance of carbon in diffuse clouds. [230] Absorption in the visible region indicates that the bands are related to the electronic transitions of chemical species. Absorption in the NIR can arise due to combination or overtone vibrations, however these are typically weak transitions, and would require an associated large column density of the chemical species. It is more likely that the NIR absorptions relate to low lying electronic states. The majority of the DIBs also have a FWHM of ∼1 Å, which can be attributed to rotational structure, indicating a molecular origin. [228] Finally, the intense fields of radiation present within the ISM argue that any carrier must also be sufficiently stable to photolytic degradation. [231] One such class of molecules which satisfy these selection criterion are the polycyclic aromatic hydrocarbons.[232]

157 7.2.1 Polycyclic Aromatic Hydorcarbons

The class of PAH molecules were originally postulated to account for another set of interstellar features, the UIR emission bands. These bands were observed in many regions of the ISM, including low density circumstellar gas clouds, planetary nebula and star forming regions.[233] Five key bands were identified in the 3.3 µm - 11 µm range, including 3.3 µm, 6.3 µm, 7.7 µm, 8.6 µm and 11.3 µm. The first of these is characteristic of a C-H stretching mode, the following two of a C-C stretching mode and the final two as C-H bending modes. These vibrational modes are consistent with those observed for a variety of PAH systems, [221, 234, 235] and following reviews in 1989 by Allamandola et al.[236] and Puget and Léger, [237], the PAH hypothesis has been generally accepted as explanation for the UIR features. More recently, alternative explanations have arisen for the UIRs, including the mixed aromatic/aliphatic organic nanoparticles (MAON) hypothesis. [238, 239]

It was suggested that there may be a correlation between the UIR and DIB bands, considering the ubiquity of both features across the ISM. As PAH’s were already hypothesised to account for the UIR emission bands, and considering their properties matched the selection criterion for DIBs described prior, they quickly became candidates (in both ionised and neutral forms). [232]

Prior to 2015, not a single DIB had been unambiguously and successfully assigned to a spectral carrier. Other classes of molecules have since been put forward as potential candidates, fullerenes, and fulleranes.[240, 241] With the recent obser-

+ vation C60 as the first confirmed DIB, it is very possible that further analogues of fullerenes could be responsible for other DIBs.[96, 242] Subject to intense radiation flux, the fragments of these molecules, including the class of (ionic) PAH’s, could certainly be responsible for other bands and cannot yet be discounted.

158 7.3 Protonated Species in the Interstellar Medium

+ With the recent attribution of C60 as a spectral carrier of four diﬀuse interstellar bands (DIBs) by Campbell et al.,[96] a renewed attention has focussed on the detection of other protonated species. Of key interest are molecules that could be formed from the fragments of these fullerenes, such as the positively charged polycyclic aromatic hydrocarbons (PAH’s). The class of neutral PAH radicals have been studied for a number of years, and will be explored further in chapter 8. Previous and current members of our group explored this hypothesis, by studying a range of substituted PAH’s such as triphenylene, 1- and 2-naphthyl methyl, 9-anthracenylmethyl and 1-pyrenylmethyl. [243, 244, 245] The higher electronic states of some of these species were also considered, however extremely short lifetimes of such states led to very broad absorption bands, much larger than those observed in the interstellar medium (ISM). [183] Unfortunately no correlations were made with DIBs.

Protonated PAH’s have been put forward as intermediates in the catalysis of H2 in the interstellar medium. [246] As molecular hydrogen is unlikely to form from two or three bodied collisions, due to constraints in angular momentum and energetic barriers, production has been postulated to occur on larger PAH grains. [247, 248, 249] Initially, atomic hydrogen adds to a PAH+, forming a H-PAH+,

+ addition of a second hydrogen is then able to occur, generating a H2-PAH . This + + then dissociates to produce H2 and the initial PAH . These H-PAH ’s have not only been put forward as suitable candidates for DIBs [250, 251], but also for the unidentiﬁed infrared (UIR) emission bands. [252, 253] Moreover, H-PAH+’s have long been observed as key intermediates in condensed phase electrophilic aromatic substitution reactions,[254] and have been identiﬁed in combustion experiments. [255]

Sanford et al. describe the multitude of processes for which deuterium enrichment can occur for interstellar molecules, including the class of PAH’s.[256] They infer that PAH’s can be enriched with deuterium by ion-molecule reactions within dense molecular clouds, and in fact PAH’s represent the largest reservoir for D enrichment of all ion-molecule reactions. Deuterium cations (D+) undergo

159 + + substitution reactions with H2 to generate HD + H . Protonation of HD by H3 + forms H2D . [257] PAH molecules are then prone to deuteron addition through the reaction

+ + P AH + H2D → P AH − D + H2

The PAH-D+ recovers a neutral charge by accepting an electron, which results in the expulsion of a H or D atom. Even without considerations of zero point energy, larger PAH’s have a higher chance of retaining deuterium enrichment, as there are a larger number of peripheral H atoms which could equally likely dissociate.[256]

7.4 Focus and Outlines of Part II

The second part of this thesis is concerned with the identiﬁcation and spectroscopy of radical and ionic species relevant to astrophysical processes. Following recent work conducted in our group on the proton and deuteron addition to naphthalene, we extend our studies to probe the species formed from H+ and D+ addition to the three ring system, anthracene. In order to measure the electronic spectroscopy of this adduct, we must ﬁrst measure the corresponding neutral radicals formed from the addition of H and D to anthracene. The outcomes of this experiment are twofold, as these radical adducts are also proposed candidates for DIBs within the PAH hypothesis.

As the neutral products formed from the hydrogen addition to anthracene have only been observed for higher electronic states in the UV,[258] and as the deuterated isotopologues have never been observed, the spectroscopy of each of these neutral products, and the eﬀect of deuteration, will be examined in chapter 8.

In chapter 9, the complex spectroscopy of the cations formed from the H+ and D+ addition to anthracene is explored, with reference to the recent work conducted on the H+-anthracene adduct by Alata et al. and Garkusha et al. [141, 258]. The relevance of these cations as potential DIB carriers is brieﬂy discussed. Chap- ter 10 reﬂects on the work conducted in this section of the thesis, and recommends direction for future work on this class of DIB candidates.

160 Chapter 8

The Hydrogenation and Deuteration of Anthracene

8.1 Introduction

Anthracene is a prototypical polycyclic aromatic hydrocarbon, consisting of three fused benzene rings. PAH’s have been put forward as candidates for interstellar phenomenon such as the unidentified infrared emissions and the diffuse interstellar bands.[232] Arguments for the PAH hypothesis are founded on their stability, both to electromagnetic radiation and reactions with other astrophysical species, that their relative abundance is compatible with constraints on estimated carbon quantities, and that they are likely to form in the diffuse interstellar medium. [242] Furthermore, PAH’s have been suggested to account for as much as 20 - 30% of interstellar carbon.[221]

Given the relative abundance of hydrogen within the ISM, it is appropriate to understand the interaction of hydrogen with prototypical PAH’s. The association of hydrogen to benzene breaks aromaticity and forms the cyclohexadienyl radical

(C6H7),[179] an important class of intermediate involved in many processes, such as the atmospheric oxidation of benzene, as seen in chapter 3. Addition of hydrogen to naphthalene can occur at one of two unique locations on the ring, forming the 1- and

161 2-hydronaphthyl radicals (C10H9). [259] The spectroscopy and thermochemistry of the 1- and 2-hydronaphthyl radicals has been explored by Sebree et al. to aid in the photochemical models of PAH formation in the atmosphere of Saturn’s moon, Titan. [260] The resonance stabilisation of these radicals allows their concentrations to build up in reactive and harsh environments, such as combustion chambers, ﬂames[261, 262] and the interstellar medium.[263]

The D1 ← D0 electronic transitions of resonantly stabilised radicals often lie within the visible region of the EM spectrum, making them great DIB candidates. In this work alone, we have studied the 3-membered π systems allyl, and substituted allyl radicals, all of which have electronic transitions in the 400-420 nm region, see chapter 4. The observed shift in the D1 electronic state of the allyl radical upon substitution oﬀers the possibility of ﬁne tuning a radical species for comparison with other DIB lines. The 5-membered π-system, hydroxycyclohexadienyl radical, was observed with an electronic transition in the 550 nm region.

Following naphthalene, the next-largest linear aromatic system is anthracene. Garkusha et al. made the first observation of a hydro-anthracenyl radical in a neon matrix.[258] They observed a broad UV absorption profile stretching ∼20 nm with an origin at 326.4 nm, and identified the position of three vibrational bands. The location of the additional hydrogen on the anthracenyl ring was unknown, and the authors report that the spectrum may relate to the 1H-An, the 9H-An or both isomers. As such, we refer to our recent work on the H and D attachment to naphthalene to guide the experimental direction of these studies. [264]

162 8.2 Experimental

The experimental apparatus and details are described thoroughly in chapter 2. Here we mention only those details explicit to this experiment. A sample of anthracene was heated in the range of 75-100◦C before the nozzle, which was held ∼10◦C higher to stop the sample from condensing before expansion. A gas mixture of argon and (heavy) water was passed through the sample container and collected the vapour of anthracene. The (heavy) water was heated externally to 50◦C to increase its seed ratio. As described in the previous section 2.1.2, a discharge containing (heavy) water is an eﬀective way of producing hydrogen (and deuterium) radicals, which are prone to add to aromatic species.

The mixture was introduced into the vacuum chamber by the nozzle apparatus described earlier, and a discharge of 2.1 kV ballasted by 20 kΩ resistance was struck for 130 µs across the gas. After the pulsed discharge nozzle (PDN) region, the mixture was supersonically expanded into the diﬀerentially pumped vacuum chamber and the coldest part was skimmed by a 2 mm skimmer. Within the extraction chamber, to search for the H and D addition products of anthracene, a 1+10 REMPI scheme was employed. The ﬁxed ionisation source of ∼290 nm was produced from the doubled output of a QUANTEL TDL90 laser, circulating the Pyromethane 597 dye. This wavelength was chosen as an initial ionisation source of 266 nm produced too much one-laser background signal. The excitation photons were produced from a SIRAH COBRA-STRETCH laser, using the Coumarin 540a, 503 and 480 dyes. The excitation lasers were scanned in wavelength, and the signal response for m/z 179 and 180 were monitored and recorded for the hydrogen and deuterium-addition experiments respectively.

The ionisation potential of the radicals was measured by ﬁxing the excitation laser on the origin of the radicals electronic excitation, and scanning the ionisation laser in frequency. In this way, a photo ionisation eﬃciency (PIE) curve was generated. To be able to scan the ionisation frequency, the dye in the ionising laser was changed to DCM, where the frequency doubled output was used.

163 A lifetime scan was recorded for the H-addition isomer by fixing the ionisation source, and temporally scanning the excitation laser. The lifetime is recorded as a convolution of a Gaussian and an exponential decay function. The Gaussian is itself a convolution of the two Gaussian profiles of the excitation and ionisation laser pulses, which typically have a full width at half maximum (FWHM) of 5-7 ns (depending on the laser used). Where the lifetime of the excited state is much larger than the profile of the Gaussian, deconvolution is not required and the lifetime of the excited state can be extracted from fitting an exponential decay function to the tail of the data at long time.

Hole burning experiments, previously described in chapter 3, were conducted on the anthracene + D radicals, to determine the spectral carrier of individual excitation bands. A third laser, from another SIRAH COBRA-STRETCH dye laser (using Coumarin 503) was parked on a speciﬁc transition to burn, around 100 ns prior to the 1+10 REMPI experiment described above. The hole burning laser was separately tuned to the ﬁrst two peaks observed in the Anthracene + D experiment, and the 1+10 REMPI experiment was redone for each.

164 8.3 Results and Theory

The REMPI Spectra of H-An and D-An

The results of the 1+10 REMPI experiments for the H and D addition to anthracene can be seen below in figure 8.1.The electronic spectra exhibit rich vibrational structure with multiple peaks near the origin for both species. These successive peaks might arise from origin transitions of one or more isomers, depending upon the addition site of the hydrogen atom. Unlike H-addition to phenol, forming the hydroxycyclohexadienyl radical, reported in chapter 3, there are no rotational isomers for the H-An species. There could however, be multiple configurational isomers. We explore this possibility, first for the H-An potential energy surface, with the aid of theoretical calculations on the relative energies of hydrogen addition at different sites, the predicted D1 ← D0 excitation energies of these isomers and their ionisation potentials to compare with experiment.

H + Anthracene

D + Anthracene

Wavenumbers (cm-1 )

Figure 8.1: Top: REMPI spectrum of the m/z 179 trace, the H-An radical. Bottom: REMPI spectrum of the m/z 180 trace, the D-An radical.

165 The relative intensity of hot band transitions vary with diﬀering discharge conditions, and as this was not observed to be the case, the ﬁrst peak in each spectrum is assigned as the electronic origin for each species. The radical formed from H/D addition contains 13 π electrons, as such the lowest energy transition is

D1 ← D0. This transition represents a combination of electron excitations into and out of the singly occupied π orbital. For the H-addition isomer (H-An), the D1 ← D0 electronic origin transition lies at 19115 cm−1. The D addition isomer, (D-An) lies some 3 cm−1 to the red at 19112 cm−1. This is opposite to the eﬀect of deuteration on naphthalene, whereby a ∼10 cm−1 blue shift in the origin is observed.[264] The observed vibrational bands are sharp, with a full width at half maximum (FWHM) of

−1 1.4-1.5 cm indicating a bound D1 state. The strong intensity observed in multiple vibrational bands suggests a change in geometry between the ground and excited state.

Theoretical Calculations

According to literature, H or D addition could occur at any of three unique positions around the anthracene rings, as described by the bold numbers in ﬁgure 8.2.[258] In principle, addition to the sp hybridised carbon at position 9a is possible. However, addition at this site is sterically hindered, and results in fewer resonance structures of the ensuing radical, making it energetically unfavourable. The equivalent addition site in naphthalene has likewise never been observed. [259]

1 9 8 9a 8a 2 7

3 6 4a 10a 4 10 5

Figure 8.2: Numbering convention for the Anthracene substituents, unique addition locations shown in bold.

Relative energies of formation for the 1H-An and 2H-An compared to the 9H-An isomer are displayed in table 8.1. These energies were calculated with the

166 B3LYP/6-311+G(d,p) level of theory. The 9H-An isomer the most energetically favourable to form from H addition to anthracene, by some 26.23 and 43.75 kJ mol−1 compared to addition at the 1- and 2- positions respectively. The highly energetic environment of the discharge is able to overcome large energetic barriers such as these and produce a variety of possible radicals. The relative abundance however, will depend on the relative stability and energies of formation which clearly favours production of the 9H-An isomer.

Table 8.1: Energies of formation of the 1H-An and 2H-An conformational isomers relative to the lowest energy 9H-An isomer. All energies are in kJ mol−1

1H-An 2H-An 9H-An

Relative Energy 26.10 43.50 0.0 ZPE Corrected 27.70 45.60 0.0

The adiabatic D1 ← D0 excitation energies for the 1H-An, 2H-An and 9H- An isomers were calculated at the (TD)-DFT B3LYP/6-311+G(d,p) level of theory. Zero point energies were calculated as half of the sum of the harmonic frequencies for each isomer. Calculation of the excitation energy is described by;

E = E(TD−DFT ) + ∆ESCF + ∆EZPE where we begin with the excitation energy taken from a TD-DFT output, and add to it the difference in the self-consistent field (SCF) energies of the ground and excited state, and the difference in the ground and excited state zero point energies. The results of these calculations are displayed below in table 8.2.

Table 8.2: Zero point energy corrected adiabatic excitation energies for the 1H-An, 2H-An and 9H-An conformational isomers

1H-An 2H-An 9H-An Observed

Excitation 16655 cm−1 15523 cm−1 20101 cm−1 19115 cm−1 Energy

The observed H-An electronic transition is ascribed to the ﬁrst peak in the REMPI spectrum, at 19115 cm−1. No bands were observed to the red of this

167 transition, and as such any other origin must be at higher energy. The calculated D1 −1 ← D0 transition for the 9H-An isomer, at 20101 cm , is within 5% of the observed value, with a discrepancy of 989 cm−1. The 1H-An and 2H-An isomers are calculated to absorb some 2457 cm−1 and 3589 cm−1 lower in energy, respectively. These calculations, along with the relative energies of formation, provide good evidence that the band at 19115 cm−1 should be assigned to the 9H-An isomer.

Ionisation Energies

The photo-ionisation efficiency (PIE) curves were obtained by first exciting a transition in the H-An spectrum, and then scanning the ionisation laser, the results of which are displayed in figure 8.3. The top two PIE curves result from exciting the origin and the first successive transitions in the H addition spectrum, and are labelled Ha and Hb. As we typically use the onset obtained from the origin excitation as the ionisation energy, this scan was repeated three times and averaged. We routinely assign an ionisation energy by fitting two straight lines to the noise and the ionisation onset and measuring the point of intersection. This can sometimes be challenging to determine for species where the onset is very gradual, indicating a poor Franck-Condon overlap between the excited state wavefunction and the ground vibrational state wavefunction of the cation. The point of intersection in Ha is more clearly defined than Hb, justifying its choice. The error in these measurements is often of the order of 10 meV.[109, 264] The ionisation potential for the H addition isomer is found to be 6.3392 ± 0.0010 eV. This range encompasses the value obtained by ionising through a Hb excitation scheme and can be considered consistent.

The adiabatic (and ZPE corrected) ionisation energies of the 1H-An, 2H-An and 9H-An isomers were also calculated at the B3LYP/6-311+G(d,p) level of theory, and the results are displayed below in table 8.3. All three isomers are predicted to have similar ionisation energies, and characteristic of the B3LYP level of theory, are under predicted compared to the observed value. [109, 123] As such we cannot use the ionisation energy alone as a way to discriminate between the isomers, however they are consistent the H-An species.

168 Ha 6.3392 eV (3) 110 cm -1 Hb Ha

6.3384 eV

6.3401 eV Da

6.3405 eV

Db Da D 6.3405 eV c Dc

Dd 6.3399 eV (2) De -1 57 cm Dd

6.3401 eV

Ionisation Energy (eV) Wavenumbers (cm-1 )

Figure 8.3: Photo Ionisation Efficiency curves for the for the first 2 peaks in the spectrum of the H addition species, labelled as Ha and Hb, and the first 5 peaks in the spectrum of the D addition species, labelled as Da -De. The energy represents the combined sum of the excitation and ionisation laser energies. Bracketed values indicate the number of scans averaged for each case. The central line represents the determined ionisation energy, and the grey band represents the error for this value. Additional onsets are seen for Ha and Dd, (see text). Table 8.3: Adiabatic ionisation energies calculated for the 1H-An, 2H-An and 9H-An confirmational isomers, ZPE corrected values are also shown. All units are in eV.

1H-An 2H-An 9H-An Observed

Adiabatic IE 6.2438 6.1992 6.2034 ZPE Corrected 6.3081 6.2558 6.2619 6.3392 ± 0.0010

A previous member of our group, Dr. Tyler Troy, performed a detailed theoretical evaluation of the size and type of functionals and basis sets used to calculate ionisation energies, as part of his PhD thesis.[123] After experimentally determining the ionisation energy of the trans-1-phenylallyl radical, ionisation energy calculations were performed at the B3LYP level of theory and a sequential increase basis set size from 6-31G(d) to 6-311++G(3df,3pd) was conducted. Convergence in the predicted ionisation energy was observed at the 6-311+G(d,p) level of theory. By comparisons with calculated and experimental adiabatic ionisation energies, Troy then determined the mean deviation (MD) and mean absolute deviation (MAD) of the 6-311++G(d,p) basis set to be -0.14 and 0.14 eV respectively. As ionisation energies were determined to converge by use of a smaller basis (6-311+G(d,p)), we can assume the same mean (absolute) deviation for calculations. For the purpose of Troy’s large study, the ground and excited state were assumed to have similar vibrational frequencies, such that there was no change in the ZPE upon ionisation.

By adding the average deviation (+0.14 eV) to the calculated adiabatic energies, we obtain values of 6.3838 eV, 6.3392 eV and 6.3434 eV for the 1H-An, 2H- An and 9H-An isomers respectively. The 2H-An and 9H-An radicals are the most likely candidates as they show the best match with the observed value of 6.3392 eV.

The 9H-An radical not only has the lowest energy of formation, but also the most accurately predicted electronic transition energy. In this case, the ionisation energy was not a great discriminant between the isomers, however it is consistent with a H-An species. As such we assign the REMPI spectrum of the H-An radical in ﬁgure 8.1 solely to the 9H-An isomer.

170 Zgierski et al. performed ﬂuorescence excitation spectroscopy on the 9,10-

−1 dihydroanthracene molecule.[265] They observed an S1 ← S0 origin at 37204 cm , and a long progression of a butterfly inversion mode built onto the origin. This inversion mode was observed to have a frequency of ∼27 cm−1. As there are only two peaks, other than the origin, in the first 100 cm−1 of the 9H-An spectrum, the first of these at +26.6 cm−1 is likely this same inversion coordinate, and the second at +67.7 cm−1 could either be an overtone of the butterfly mode (anharmonic in nature to account for the increasing spacing), or another low frequency mode. There are four bands in the same region for the deuterated isotopologue, D-An, as seen in

−1 ﬁgure 8.3. The ﬁrst band, Db is only +16.6 cm away from the electronic origin. If we assume for now that deuterium also adds at the 9- position, then it would only minimally contribute to the frequency of the inversion coordinate. As such, a reduction in the frequency of this coordinate, from 26.6 cm−1 to 16.6 cm−1 is not expected.

Unlike hydrogen addition to anthracene, where H-H exchange would result in formation of an identical isomer, the added deuterium could intramolecularly exchange with a hydrogen atom around the ring, forming a series of different isotopomers. These isotopomers would all have electronic transitions in similar regions, differing only by their respective ZPE’s. To first ensure that there was no other isomer contributing to this vibrational structure, PIE curves were obtained for the

D-An species, after exciting the origin and the four consecutive peaks, labelled Da and Db−e in figure 8.3. There is a larger variety in the sharpness of the onset for these PIE curves, and we have done our best to appropriately fit these onsets and noise. The intersection value ranges from 6.3400 to 6.3405 eV. Exciting through Dd and De resulted in a lower signal to noise ratio, due to their smaller excitation peak intensity, see figure 8.3. Exciting through Dc shows the clearest onset, and we use this value to record the IP, as 6.3405 ± 0.0005 eV, which encompasses the observed IE’s for all five peaks. The error is representative of the number of scans taken and the sharpness of the onsets. As all peaks lie within this small error, there is no evidence of more than one isomer contributing to the first 5 peaks of the D-An spectrum, and we can tentatively assign the spectrum to the 9D-An species, as isotopomers are still a possibility and could well have very similar ionisation energies.

171 The ∼20 meV shift in IE compared to 9H-An is consistent with a small diﬀerence in zero point energies as a result of deuteration.

In the case of Ha and Dd, we can see a second ionisation onset, which infers a change in the Franck-Condon overlap between the neutral and cation, indicating a possible further vibrational excitation of the cation. This additional onset appears some 110 cm−1 and 57 cm−1 respectively after the initial onset, and will be reviewed in chapter 9.

Hole-Burning

Whilst the ionisation energies for the ﬁrst 5 peaks in the D-An spectrum were consistent, we cannot rule out H-D exchange between the species post discharge, nor the presence of hot bands arising from a pair of low lying states. For this reason, hole-burning spectroscopy was conducted on these 5 peaks, see ﬁgure 8.4.

As mentioned in the experimental section, hole-burning is a vital spectroscopic tool in distinguishing spectral carriers. Here, the regular 1+10 REMPI experiment was conducted on the D-An electronic origin at 19112 cm−1. 100 ns prior, a hole-burning laser is scanned across the range containing the peaks Da -

De. Spectral lines that exhibit depletion come from the same isomer as the band at 19112 cm−1. This experiment was repeated, and the averaged results are displayed in figure 8.4. As the REMPI signal baseline fluctuates, the magnitude of the depletion is taken as a percentage and is at most ∼40%. The 9D-An REMPI spectrum has been scaled for purpose of illustration. We can conclude that the clear and definitive depletion of all 5 peaks, as shown in figure 8.4 relate to the same spectral carrier as the band at 19112 cm−1. This confirms that H-D exchange does not occur here, and that there is only 1 isomer/isotopomer responsible for the D-An spectrum, which we assign to the 9D-An radical.

By an evaluation of the electronic, formation and ionisation energies of the possible H addition candidates, we conﬁrm that the H-An REMPI spectrum contains only one isomer, the 9H-An radical. Upon deuteration, additional peaks

172 lsvl rvdta l ek wdt h aeseta are,asge sthe as assigned carrier, spectral same the to radical. owed 9D-An peaks con- which all conducted, that were proved experiments from clusively burning coming Hole as hypothesised isotopomers. were exchange bands H-D the isomers, additional precluded iments cm 100 the in observed were cm 19112 on at signal 180) REMPI (m/z the origin monitoring addition whilst D measured the spectrum (depletion) Hole-burning 8.4: Figure Depletion on 19112cm-1 band (%) 40 30 20 10 0 − 1 einfloigectto,ada oiainexper- ionisation as and excitation, following region Wavenumbers (cm − 1 173 . D-An HoleburningSpectrum D-An REMPISpectrum -1 ) 8.4 Discussion

As we discard the notion of multiple isomers or isotopomers, we must consider other possibilities for the difference in the origin region of the 9H-An and 9D-An spectra. That is, only three peaks were observed in the origin region of the 9H-An spectrum, compared to the five observed for 9D-An, right in figure 8.3. Spectral spacings, similar to those around the origin, are seen throughout the spectrum, indicating that this progression is likely built onto higher frequency modes. As such, once the band structure around the origin region has been understood, we can revisit and assign the remainder of the spectrum.

Origin Region of 9H-An

We turn to theory to assist our understanding of the effect of deuteration, by first beginning with hydrogenation. The ground and excited state geometries and frequencies of the 9H-An radical were calculated in GAUSSIAN 16,[135] using the (TD)-DFT B3LYP/ 6-311+G(d,p) basis set. This method and basis set were chosen as they are relatively cheap and effective, and the energies of the first 8 excited states were calculated to improve accuracy. The geometries are displayed in figure 8.5.

The ground and excited states are both of Cs symmetry, where the plane of symmetry runs through the CH2 group as indicated by the green plane in figure 8.5. When we attempted to converge the molecule to planar symmetry, (in both ground and excited states), which becomes a C2v point group, an imaginary frequency is calculated in each case, indicating a transition state structure. This mode represents a butterfly motion, symmetric about the Cs plane of symmetry. In fact, it is this motion that primarily dominates the D1 ← D0 electronic transition. This can be seen in the top of figure 8.5, where there is a more prominent angle formed in the excited state. The carbon within the CH2 group puckers out of plane, reducing the angle along the symmetry axis, formed with the opposing CH group, from 178.06◦

◦ to 176.54 . In the ground state, the two hydrogens in the CH2 group are almost

174 176.54176.54 178.06178.06

1.0851.085 1.0871.087 1.4021.402 1.4231.423 1.085 1.085 11.101.101 1.1031.103 1.417 1.435 1.421421 1.368 1.45959 105.00105.00 105.12105.12 1.384 1.5131.513 1.5071.507 1.392 11.097.097 1.381 1.0951.095 1.084 1.400 1.084 1.431 1.395 1.405 1.086 1.086

1.084 1.083

Ground State Excited State

CS CS

Figure 8.5: Left: Geometric details for the ground state of the H addition isomer, indicating bond lengths, and specific angles. The profile view emphasises the out-of-plane pucker of the sp2 hybridised carbon. Right: Geometric details for the excited state of the H addition isomer, indicating bond lengths, and specific angles. The profile view emphasises the increased out-of-plane pucker of the sp2 hybridised carbon. symmetrically spaced about the plane of the anthracene ring. In the excited state, the CH2 group rotates, to the point where one hydrogen assumes a more equitorial position and the other more axial. We can understand this geometric shift by the change in the hybridization of the 9- carbon. The preferred C-C-C angle of an sp2 carbon is 120◦. Addition of a hydrogen or deuterium changes the hybridisation of

3 ◦ this carbon to sp , which prefers a C-C-C bond angle of ∼109 . As a result the CH2 group puckers out-of-plane to be able to narrow this angle, and the four C-C bonds unique to the centre ring all contract, bringing the outer rings closer together. The calculations imply that 4 of the 6 C-C bonds within the side rings increase in bond length. The CH bond lengths remain the same upon excitation, as does the HCH angle within the CH2 group.

It is worth mentioning that the calculated geometry is very prone to the choice of basis set, size and method. In our early investigations, Drs. Yu Liu and

175 Klaas Nauta in our group calculated the structure and frequencies in the ground and excited states using a series of different sized Pople basis sets. They found that the inclusion of the diffuse functionals on the carbon atoms (‘+’) would pucker the molecule out of the plane of the rings, and that without it, the molecule would often converge to a C2v symmetry. We believe that the use of the diffuse functionals is necessary, and will further consider the energy profile of the butterfly motion through planarity.

To assign the low frequency modes of 9H-An near the origin, observed at +26.6 cm−1 and +67.7 cm−1, we calculated the radical’s excited state frequencies (presented later). Only two modes were calculated under 150 cm−1. The ﬁrst of these is an a0 butterﬂy inversion, at 37.1 cm−1, the second is an a00 out-of-plane ring

−1 twist, at 83.7 cm . As Cs symmetry is preserved upon electronic excitation, only even changes in quanta of a00 modes are allowed and the out-of-plane ring twist is only expected to be observed as an overtone ∼167.4 cm−1. Thus, the ﬁrst peak is likely a single quanta of this inversion mode, and the second, an overtone. As the spacing between successive peaks increases, the progression in this mode is clearly anharmonic. This simple assignment however, does not account for the additional peaks observed in the 9D-An spectrum.

Eﬀective Butterﬂy Potentials

To investigate this inversion coordinate further, we take a closer look at how this mode is affected by deuteration. Addition of a hydrogen atom to the 9- site of the anthracene ring at the equatorial or axial position is indistinguishable. A symmetric double well potential can describe this mode as the molecule proceeds through the butterfly inversion, see figure 8.6.

We can see that H1 moves from an equatorial position (bottom left of

ﬁgure), to an axial position (top left), and likewise the opposite for H2. Though these geometries are symmetric for 9H-An, they are not for 9D-An, as can be seen in the right side of the ﬁgure. Deuteration lifts the degeneracy of the 9H-An symmetric

176 Hʤ H axial

9H-An 9D -An Hʣ D ax

Hʤ H transition state

Hʣ D

Hʤ H

9H-An 9D eq -An equatorial Hʣ D

Figure 8.6: The inversion ‘butterﬂy’ coordinate of the anthracene ring. Left: The 9H-An isotopologue. Right: Depiction of the equatorial and axial conﬁgurations of the 9D-An isotopologue potential (in both the ground and excited state), and this perturbation introduces an asymmetry to the double wells.

Assuming that the peaks at +26.6 cm−1 and +67.7 cm−1 represent excitations of different quanta of the butterfly mode, we can fit this part of the spectrum to a symmetric double well potential. To fit the spectra, we developed an effective (empirical) potential, similar to the method of Laane and Veguilla-Berdecía. [266, 267] By beginning with known solutions for the simple harmonic oscillator, we introduced a quartic pertubation, and the eigenfunctions and eigenvalues for the double well potential were solved in a similar fashion. The double well was

4 2 approximated by the functional aq37 + bq37, (b<0), where q37 is the normal mode displacement vector for the butterfly inversion, mode ν37 (explained later). Away from equilibrium, the coefficient of the quartic term, a, determines the gradient of the walls. The quadratic coefficient, b, regulates the height of the barrier between the two wells.

As hole-burning demonstrated that no hot bands were observed in either spectrum, we assume that every transition originates in the ground, vibrationless state. For the 9H-An radical, a non-planar ground state would also have a symmetric double well potential in this coordinate. As a result, regardless of the shape of the ground state PES around the inversion plane, the wavefunction for v00 = 0 must be symmetric. In exciting from this symmetric ground state wavefunction, a non-

177 zero Franck-Condon overlap integral would only exist for a symmetric excited state. Hence, only excitations in even quanta (0, 2, 4,...) are allowed.

The 9D-An radical exhibits no such constraints upon excitation. As seen in figure 8.6, the equatorial and axial positions of the deuterium are not equivalent. To determine the extent of this effect, the zero point energy was calculated for deuterium in each position, and found to be different, (shown later in table 8.4). The difference in zero point energy lifts the degeneracy of the PES. This asymmetry means that transitions from v00 = 0 to any vibrational level in the excited state, v0, become allowed.

A similar functional form as above was used to describe the 9D-An potential energy surface by including a linear term, to account for an offset in the zero point energies of deuterium at the equatorial and axial positions. The first 5 peaks of the 9D-An spectrum were used in the fitting algorithm, as they could not be assigned to any other vibrational mode, and were assumed to be progressions of the butterfly mode. Hence, we simultaneously fit the a and b coefficients to the first 3 and 5 peaks of the 9H-An and 9D-An REMPI spectra respectively, whilst allowing c to fit independently. The calculated PES’s, as a result of these fits, are shown below in figure 8.7.

4 2 The double wells were ﬁt by the quartic functional 5.81q37 - 8.33q37. The quadratic term greatly impacts the energy level spacings early on in the spectrum.

A linear term of 13.34q37 was used to describe the oﬀset in the ZPE of the deuterium in the equatorial and axial positions, which we will refer to as 9Deq-An and 9Dax-An respectively, (ﬁgure 8.6). As the inversion proceeds through the q37 coordinate, the two stationary geometries are referred to as q37 and -q37.

As the 9H-An spectrum shows excitations in even quanta only, the band observed at +26.6 cm−1 is assigned to the ﬁrst overtone of the butterﬂy mode. The second peak, at 67.7 cm−1 is then assigned as 4 quanta of the same mode. No higher overtones are observed in the spectrum. The quartic term reproduces the increased spacing (anharmonicity) of the wells, and a submerged barrier is calculated for the

178 hr sadffrnei h P fte9D the of ZPE the in difference a is there mode. frequency lowest the of calculation frequency GAUSSIAN q the for from 3.4 of mass reduced A geometry. planar The spectra. red. REMPI in 9D-An drawn and 9H-An the to cm fit 200 potentials first well double state Excited 8.7: Figure aso . a sdfrte9-nfi.Ti a lootie rmteGAUSSIAN the 9D from obtained also q was mode This of fit. output 9D-An frequency the for used was 3.5 of mass cm eest xiain nec ieo h el h necneso ewe 9D between interconversion The well. the 9D of and side convention unbracketed each and on bracketed excitations The to mode. refers 2, butterfly (1), by the 0, of as produced quanta assigned 4 is be can and spectrum excitations (3), These the well. in the of peak sides alternate successive to Each excitation stage. this at configuration ax − 1 A emtis hseetv oeta ersnsazr on nryinduced energy point zero a represents potential effective This geometries. -An oee,w r nbet eemn hc el(q well which determine to unable are we However, . ax A,a hw nfiue86 scluae ob arels.Alre reduced larger A barrierless. be to calculated is 8.6, figure in shown as -An, rmtelna ffe sdt tte9-nsetu,i a ese that seen be can it spectrum, 9D-An the fit to used offset linear the From − 1 ftecrepnigRMIsetu r hw,wt h iuae iepositions line simulated the with shown, are spectrum REMPI corresponding the of 9H-An REMPISpectrum Simulation 9D-An REMPI Spectrum 9D-An REMPI Simulation 37 n a acltda h vrg fte9D the of average the as calculated was and , 179 eq A n 9D and -An 37 4 a sdfrte9-nfi,obtained fit, 9H-An the for used was Relative Energy (cm-1 ) Relative Energy (cm-1 ) D ax D 1 1 A ofiuain of configurations -An 37 13.34 -q , - 8.33 - q - q 37 37 q 37 - 8.33 37 eae owhich to relates ) q 37 4 2 0 0 + 5.81 q 37 4 2 + 5.81 q q q 37 37 eq 37 q A and -An 37 eq q q ∼ -An 37 37 20 split in the molecular symmetry, allowing for otherwise unallowed odd quanta of the butterfly mode to be excited.

Time-Dependent Density Functional Theory (TD-DFT) Calculations

To confirm the validity of our model, we undertook a TD-DFT study. The excited state frequencies for the isotopologues/isotopomers were calculated to confirm the relative energy of the 9Deq-An and 9Dax-An configurations, the shape of the double wells and the assignment of the origin region for both isotopologues. These calculations, displayed in table 8.4, were performed at the TD-DFT B3LYP/ 6-311+G(d,p) level of theory and frequencies have been scaled by a factor of 0.97 to fit with experiment. A full list of the calculated frequencies is provided at the end of the chapter in appendix 8.6. Frequencies were calculated for both of the 9Deq-An and 9Dax-An configurations to determine their respective ZPE’s and to assign the spectra.

The frequencies in table 8.4 have been truncated at 1200 cm−1 for comparison with experiment, the entire list is included in the Appendix. The modes of the 9H-An isotopologue have been labelled by the Mulliken Convention, where the a0 modes are labelled in decreasing order of frequency, which are then followed by

00 the a modes, in the same order. The modes of 9Deq-An and 9Dax-An have been reordered so that a mode label in 9Deq-An and 9Dax-An refers to the same molecular motion as that of 9H-An. This allows the reader to quickly see the effect on the frequency of a mode by deuteration in either configuration. Finally, the planar C2v frequencies are included, and labelled again according to the molecular motions of the 9H-An Cs geometry. These are included to give the reader more information about the character of each mode, as the C2v symmetry labels are very easily com- 0 parable to the Cs symmetry labels. All a1 and b1 modes in C2v map to a in the 00 Cs symmetry point group, similarly, all a2 and b2 modes map to a . The negative frequency in the C2v table refers to the butterfly motion of the anthracene rings. As 0 the ground and excited states are of Cs symmetry, we expect a modes to be active in single or multiple quanta, and all a00 modes to only be active in even overtones or combination bands.

180 −1 Table 8.4: Excited state frequencies (<1200 cm ) for the 9H-An radical, and the 9Deq-An and

9Dax-An conﬁgurations in the Cs symmetry point group. The C2v character representations are also shown for the 9H-An isomer, with a calculated negative ν37 frequency. Scaled frequencies (0.97) and vibrational symmetries are included for comparison with experiment. The Mulliken convention for mode labelling is applied to the 9H-An Cs conﬁguration, and others were rearranged to keep the label consistent with molecular motions. Calculations were carried out at the TD-B3LYP/6- 311+G(d,p) level of theory. All units are in wavenumbers (cm−1).

9H-An C2v 9H-An 9Deq-An 9Dax-An Mode # Freq Sc. Freq Sym Mode # Freq Sc. Freq Sym Mode # Freq Sc. Freq Sym Mode # Freq Sc. Freq Sym

0 0 0 16 1184.6 1149.1 a1 16 1186.1 1150.5 a 16 1183.3 1147.8 a 16 1185.6 1150.0 a 0 0 0 17 1163.1 1128.2 a1 17 1163.2 1128.3 a 17 1158.4 1123.6 a 17 1162.6 1127.7 a 0 0 0 18 1060.0 1028.2 a1 18 1067.2 1035.2 a 18 1067.2 1035.2 a 18 1066.9 1034.9 a 0 0 0 19 998.4 968.4 a1 19 999.2 969.2 a 19 998.9 969.0 a 19 999.1 969.1 a 0 0 0 20 960.0 931.2 b1 20 962.5 933.6 a 20 962.2 933.4 a 20 962.2 933.4 a 0 0 0 21 944.2 915.9 b1 21 946.2 917.8 a 21 933.2 905.2 a 21 931.1 903.2 a 0 0 0 22 903.5 876.4 b1 22 900.4 873.4 a 22 898.8 871.9 a 22 899.0 872.0 a 0 0 0 23 883.5 857.0 a1 23 896.8 869.9 a 23 883.9 857.4 a 23 893.8 866.9 a 0 0 0 24 875.0 848.8 b1 24 866.1 840.1 a 24 832.4 807.5 a 24 829.2 804.3 a 0 0 0 25 789.1 765.4 b1 25 789.5 765.8 a 25 785.9 762.3 a 25 786.6 763.0 a 0 0 0 26 706.7 685.5 a1 26 713.6 692.2 a 26 707.7 686.4 a 26 713.3 691.9 a 0 0 0 27 707.9 686.7 b1 27 704.2 683.0 a 27 699.9 678.9 a 27 700.9 679.9 a 0 0 0 28 641.4 622.2 b1 28 650.2 630.7 a 28 628.8 609.9 a 28 631.8 612.8 a 0 0 0 29 639.2 620.0 a1 29 635.4 616.4 a 29 638.1 618.9 a 29 643.2 623.9 a 0 0 0 30 590.9 573.1 a1 30 582.8 565.3 a 30 580.5 563.1 a 30 568.2 551.1 a 0 0 0 31 480.4 466.0 b1 31 479.9 465.5 a 31 477.7 463.4 a 31 475.6 461.4 a 0 0 0 32 404.8 392.7 a1 32 405.5 393.4 a 32 404.1 392.0 a 32 402.6 390.5 a 0 0 0 33 376.1 364.8 a1 33 372.3 361.1 a 33 371.7 360.6 a 33 370.4 359.3 a 0 0 0 34 278.5 270.2 b1 34 278.2 269.9 a 34 273.1 264.9 a 34 271.3 263.1 a 0 0 0 35 233.2 226.2 a1 35 239.3 232.1 a 35 235.9 228.9 a 35 238.4 231.2 a 0 0 0 36 150.1 145.6 b1 36 162.9 158.0 a 36 156.2 151.5 a 36 155.0 150.4 a 0 0 0 37 -37.7 -36.6 b1 37 37.1 36.0 a 37 35.9 34.8 a 37 36.0 34.9 a 00 00 00 50 1219.0 1182.5 b2 50 1224.0 1187.3 a 50 1216.5 1180.0 a 50 1208.5 1172.3 a 00 00 00 51 1194.9 1159.1 a2 51 1192.7 1156.9 a 51 985.8 956.2 a 51 973.0 943.8 a 00 00 00 52 1181.4 1146.0 b2 52 1178.9 1143.5 a 52 1179.4 1144.0 a 52 1182.3 1146.8 a 00 00 00 53 1102.7 1069.6 b2 53 1101.0 1068.0 a 53 1101.0 1068.0 a 53 1102.0 1068.9 a 00 00 00 54 963.3 934.4 b2 54 964.4 935.4 a 54 949.6 921.1 a 54 954.5 925.8 a 00 00 00 55 954.0 925.4 a2 55 958.5 929.7 a 55 958.6 929.8 a 55 960.2 931.4 a 00 00 00 56 917.9 890.3 a2 56 920.8 893.2 a 56 917.8 890.2 a 56 914.6 887.1 a 00 00 00 57 905.6 878.4 b2 57 903.0 875.9 a 57 884.1 857.6 a 57 888.2 861.5 a 00 00 00 58 808.4 784.1 b2 58 817.1 792.5 a 58 808.0 783.8 a 58 816.8 792.3 a 00 00 00 59 800.3 776.3 a2 59 800.9 776.9 a 59 800.9 776.8 a 59 800.0 776.0 a 00 00 00 60 695.1 674.2 b2 60 719.9 698.3 a 60 711.9 690.6 a 60 713.5 692.1 a 00 00 00 61 713.4 692.0 a2 61 710.9 689.6 a 61 699.9 678.9 a 61 683.7 663.2 a 00 00 00 62 662.6 642.7 a2 62 654.5 634.8 a 62 648.1 628.6 a 62 647.6 628.2 a 00 00 00 63 551.4 534.8 b2 63 567.5 550.5 a 63 567.2 550.2 a 63 566.8 549.8 a 00 00 00 64 503.6 488.5 a2 64 504.8 489.7 a 64 497.2 482.3 a 64 501.2 486.2 a 00 00 00 65 449.3 435.8 a2 65 444.8 431.4 a 65 443.7 430.4 a 65 439.1 425.9 a 00 00 00 66 402.2 390.1 a2 66 391.0 379.2 a 66 391.3 379.6 a 66 391.3 379.6 a 00 00 00 67 378.8 367.5 b2 67 375.5 364.2 a 67 373.5 362.3 a 67 375.6 364.3 a 00 00 00 68 226.5 219.7 a2 68 219.3 212.7 a 68 219.6 213.0 a 68 218.6 212.0 a 00 00 00 69 86.1 83.5 a2 69 83.7 81.2 a 69 83.8 81.3 a 69 83.7 81.2 a ZPE: 43491 ZPE: 43538 ZPE: 42806 ZPE: 42827 Within table 8.4, mode ν51 contains the most apparent deviation in frequency, which represents the in-plane wag of the CH2 group. Its frequency drops −1 −1 −1 from 1156.9 cm to 956.2 cm and 943.8 cm for 9Deq-An and 9Dax-An isomers respectively. Other modes, with higher frequency, shown in the table in the

Appendix also exhibit large deviations upon deuteration, including mode ν18, the −1 CH2 scissoring mode, which drops from 1398.2 cm in the 9H-An case to 1242.9 −1 −1 cm and 1240.9 cm in the 9Deq-An and 9Dax-An cases respectively. Upon deuteration, the CH2 symmetric and asymmetric stretching modes form local modes of a

(near independent) CH and CD stretch. The symmetric CH2 stretch, ν11, lowers −1 −1 −1 in frequency from 2944.6 cm to 2162.1 cm and 2098.1 cm , for the 9Deq-An and 9Dax-An respective CD stretching frequency. The asymmetric CH2 stretching frequency, remains similar to the CH stretch for both isomers.

As GAUSSIAN16 calculates geometries and Self Consistent Field (SCF) energies irrespective of isotopologue substitutions, there is difference in the 1-D butterfly inversion coordinate for the 9H-An and 9D-An isotopologues. The observed difference between 9H-An and 9D-An arises from the ZPE corrections. Therefore we have calculated the respective ground and excited state ZPE’s on the stationary points that describe this 1-D PES. Ground state calculations for both isotopologues were performed at the B3LYP/6-311+G(d,p) level of theory. Excited state calculations were performed at the TD-B3LYP/6-311+G(d,p) level of theory. The ZPE’s were calculated for each geometry, including the planar 9D-An (not shown) of 42707 cm−1. These energies were calculated as half of the sum of the non imaginary vibrational frequencies. The results of these calculations are displayed in figure 8.8.

For the case of 9H-An, there is a degenerate double well, with a barrier

−1 of ∼67 cm . The C2v structure is observed to be higher in energy than the Cs structure, which is reasonable, as it is calculated to be a transition state. We found that the height of this barrier is ‘extremely’ prone to subtle changes in geometry, which is affected by calculation parameters, such as different integral grid sizes, the tightness of convergence criteria, or the use of diffuse functionals. ZPE corrections alter the ground state barrier by less than 1 cm−1, however the excited state barrier

−1 −1 −1 is seen to increase by 20 cm , from 33 cm to 53 cm . The D1 ← D0 electronic

182 9H-Anthracenyl Radical

Cs C2v Cs Cs C2v Cs No ZPE Correction ZPE Corrected

9D-Anthracenyl Radical

eq.planar ax. eq.planar ax.

No ZPE Correction ZPE Corrected

Figure 8.8: Top: PES for the 1-D ν37 inversion coordinate of 9H-An, including the ZPEs. Bottom:

PES for the 1-D ν37 inversion coordinate of 9D-An, including the ZPEs. All energies were calculated relative to the lowest energy conﬁguration in the ground state. transition energy is lowered by ∼1000 cm−1 after ZPE corrections, to 20101 cm−1. The 1-D DFT calculated PES is similar to the experimentally derived PES shown in ﬁgure 8.7, however predicts a much larger barrier height between the wells.

Without ZPE corrections, the 1-D PES for the 9D-An radical remains the same as the 9H-An PES. The inclusion of ZPE is very apparent in both the ground and excited states. In the ground state, deuterium in the axial position is calculated to be 10 cm−1 higher in energy than in the equatorial position. Whilst this perturbation from the 9H-An radical is minimal, it breaks the degeneracy seen

−1 −1 in the H2 wells. The barrier here is increased by 6 cm to 73 cm . As no hot band transitions were observed, the energy diﬀerence between the v00 = 0 and v00 = 1 must be suﬃcient so that only v00 = 0 is populated in the jet cooled molecular beam. As such this barrier is likely smaller than the calculated 73 cm−1, (which would result in a larger splitting between successive energy levels).

The D1 landscape for 9D-An undergoes the largest change upon inclusion −1 of ZPE. The diﬀerence in energy between 9Deq-An and 9Dax-An extends to 31 cm , which is similar to the ‘experimentally derived’ value of ∼20 cm−1. The planar transition state lowers to the point where there is a barrierless transition between the two conﬁgurations. The D1 PES looks remarkably similar to the experimentally derived PES, and implies that the lower energy well refers to deuterium in the equatorial position. The corrected 1-D potentials for both the 9H-An and 9D-An radicals clearly demonstrate the need to consider ZPE when dealing with isotopologues.

In assigning the spectrum, we expect that the ν37 butterfly mode will form progressions built onto other modes upon excitation. As different modes will exhibit different 1-D potentials to the ν37 PES discussed above, we can expect that not all mode progressions will exactly match the spacings here. Similarly, the 1-D potential for ν37 may be perturbed by any mode it combines with.

184 Beyond the Origin Region

The assigned REMPI spectrum for the H addition isomer is shown in ﬁgure 8.9. The assignment began by ﬁrst searching for progressions of 2 and 4 quanta of the

0 ν37 inversion mode built onto other a modes. The legend describes the notation used for this spectrum, where progressions built onto a0 modes are described by vertical lines attached to the same horizontal line. Unlike the studies by Shin et al. and Zgierski et al. on the 9,10-dihydroanthracene molecule, who only observe the electronic origin and a butterﬂy inversion progression built onto it, our spectrum spans a much broader range and we can see the inversion mode is also built onto many other modes. [265, 268] Indeed, we can see that ν37 forms combination bands 0 with the a modes ν36, ν35, ν30, ν27, ν23 and ν19. In fact, only ν26 and ν21 do not 00 combine with ν37. Furthermore, only the combinations of two a modes are seen to 00 be active, ν69 and ν67. As no single quanta of any a mode is observed, there is no conﬂict in our earlier assignment of Cs symmetry.

The observed transition frequencies and assignments are given in table 8.5. Excited state calculations were computed at the TD-DFT B3LYP/6-311+G(d,p) level of theory, and scaled by 0.97 to fit with the experiment. Whilst the recommended scaling factor for this level of theory is 0.985,[136], we find 0.97 better fits the experimental observations. The first two observed transitions, assigned to 2 and

4 quanta of ν37, are the only modes where the theoretical value is taken from the ﬁt double well potential, (ﬁgure 8.7), rather than the GAUSSIAN16 output. For this reason, calculation of the absolute average deviation only accounts for the uncoupled a0 and a00 modes. The largest deviation, of only 12.3 cm−1, and average deviation of only 5.8 cm−1 is reasonable for this level of theory and size of molecule.[245]

The ‘theoretical values’ for modes combined with ν37 are calculated as the 0 −1 −1 sum of the observed a mode and 26.6 cm or 67.7 cm for 2 or 4 quanta of the ν37 mode respectively. By doing this, we can gain an understanding of the perturbations each a0 mode makes to the 1-D double well potential. Apart from the discrepancy

−1 of the outlier, ν364ν37 of 6.3 cm , all other modes only perturb the transition by less than 2 cm−1. To understand this, we examine the scaled displacement vectors

185 Table 8.5: 9-hydroanthracenyl assignments and comparison to theory, including the mean absolute deviation (MAD). The daggers (†) indicate modes not calculated by DFT, and the asterisks (*) represent tentative assignments, (see text). All units are in wavenumbers (cm−1).

9H-An

Observed Relative Assignment Theory ∆T −O

19114.8 0.0 0 0 0

19141.4 26.6 2ν37 26.6 † 0.0

19182.6 67.7 4ν37 67.7 † 0.0

19279.4 164.6 2ν69 162.4 -2.2

19283.7 168.8 ν36 158.0 -10.8

19296.6 181.7 ν692ν37 * 181.7 0.0

19310.9 196.1 ν362ν37 195.4 -0.7

19319.1 204.2 ν694ν37 * 204.2 0.0

19345.1 230.3 ν364ν37 236.6 6.3

19349.4 234.5 ν35 232.1 -2.4

19377.9 263.0 ν352ν37 261.1 -1.9

19689.5 574.7 ν30 565.3 -9.4

19716.0 601.1 ν302ν37 601.3 0.2

19755.5 640.7 ν304ν37 642.4 1.8

19807.7 692.8 ν27 683.0 -9.8

19815.1 700.2 ν26 692.2 -8.0

19834.8 720.0 ν272ν37 719.4 -0.5

19838.5 723.6 2ν67 728.4 4.8

19978.0 863.2 ν23 869.9 6.7

20005.6 890.8 ν232ν37 889.7 -1.1

20036.0 921.2 ν21 917.8 -3.4

20096.3 981.5 ν19 969.2 -12.3

20122.9 1008.1 ν192ν37 1008.1 0.0

MAD 5.8 Wavenumbers (cm -1 )

0 1 1 1 1 1 1 0 0 36 0 35 0 legend 30 0 27 0 23 0 19 0 37 2 X 1 0 4 0 37 0

2 1 2 1 69 0 1 2 26 0 67 0 21 0 X 0 37 0

1 4 X 0 37 0

Relative Wavenumbers (cm-1 )

Figure 8.9: REMPI spectrum of the 9-hydroanthracenyl radical. Assignments of the first instance 0 00 of a and a modes are made with progressions of 2 and 4 quanta of ν37 following the legend inset. Unidentified bands are marked with a star. for each of the observed modes as shown in figure 8.10. These motions are almost identical for both isotopologues, and this image can be used as a reference for both cases.

The combination band of ν364ν37 is very weak, and possibly within the limit of the noise. As this band has been identiﬁed as a potential outlier, this assignment is rather tentative. Though theory would suggest otherwise, assignment of the peaks

−1 −1 at 164.6 cm and 168.8 cm to ν36 and 2ν69 is made on the basis that we expect the ν36 motion to be more able to form a progression with the ring inversion, as it contains some of the same character.

−1 The peak observed at 235 cm is assigned to the single quantum of ν35, calculated to be 232.1 cm−1, which represents a symmetric in-plane ring twist. This

−1 motion also forms a combination band with 2ν37 some 28.5 cm higher. After 300

187 37 36 35

30 27 26

23 21 19

69 67

Figure 8.10: Scaled displacement vectors representing the active modes present in the 9- hydroanthracenyl spectrum. cm−1 of no observed peaks, the line at 574.7 cm−1 is assigned to the ring elonga-

−1 −1 tion/contraction and CH wags of ν30, predicted 9.4 cm lower at 565.3 cm . Peaks −1 −1 at 692.8 cm and 700.2 cm are assigned to single quanta of ν27 and ν26 respec- −1 tively, with the former coupling to 2ν37 at 720.0 cm . The only other A" mode, ν67, is observed in its ﬁrst overtone at 723.6 cm−1, only 4.8 cm−1 lower than predicted.

The CH wag in the plane of symmetry, ν23, is calculated at a frequency of −1 −1 869.9 cm , and is observed at 863.2 cm . This mode also combines with 2ν37, 27.6 −1 −1 −1 cm higher in frequency. Mode ν21 is observed at 921.2 cm , 3.4 cm higher than calculated. Finally, the CH wag in the plane of the anthracene rings, ν19, calculated to be 962.2 cm−1, is observed 12.3 cm−1 higher at 981.5 cm−1, pertaining the highest discrepancy from calculation, which may well be aﬀected by the double well.

−1 Following the overtone of ν69 at 164.6 cm , two peaks are identiﬁed and marked by stars. These peaks lie 17.2 cm−1 and 39.7 cm−1 higher in energy than the ν69 overtone. The ν69 motion involves the peripheral rings twisting out of plane in opposite directions. This motion would signiﬁcantly alter the potential of the

188 ν37 butterﬂy coordinate. As such, we tentatively assign these additional peaks to

2 and 4 quanta of ν37 built upon ν69. This assignment is especially tentative as no corresponding analogy is observed in the spectra of the 9D-An species. There also exist a number of small peaks ∼20,000 cm−1, however these are likely noise due to their regularity and the lack of appropriate calculated modes in this region.

The spectrum of the D addition to anthracene can be seen in ﬁgure 8.11.

All peaks owing to the 9Deq-An configuration are shown in red, and those from the 9Dax-An configuration are shown in blue. The legend similarly describes the notation for this assignment, where the 5 vertical lines represent first the transition excited from the 9Deq-An well, then the 9Dax-An well, then 2ν37 built onto to 9Deq-

An, then 9Dax-An, then 4ν37 built onto to 9Deq-An. Technically speaking, these excitations represent 1, 2, 3 and 4 quanta of the ν37 mode built onto the 9Deq-An transition.

Over the ﬁrst 300 cm−1, we can see the same structure repeated about the

0 a modes ν36 and ν35, indicating a progression of ν37 built onto these modes. There still exist a couple of unassigned peaks within the ν36 coupled progression, analogous to the spectrum of the H-addition radical. Following that, the ﬁrst overtone of mode

ν69 can be assigned based on the best comparison with theory. All assignments for the bands coming from each side of the 9Deq-An and 9Dax-An wells can be seen in table 8.6. Again, all calculations were computed at the TD-DFT B3LYP/6- 311+G(d,p) level of theory, and scaled by 0.97 to fit with the experiment. Relative origins from each well are taken by the difference in the first two peaks, of 16.6 cm−1.

The discrepancy between theoretical and observed values for ν23 and ν19 is larger for the 9D-An assignments than the 9H-An. Whilst theory calculates that these bands should appear around the same value compared to the origin, the progression of ν37 built onto these modes indicates this assignment is sensible. The largest deviation observed is for the axial excitation of the ν23 mode, observed 29.9 cm−1 above the predicted value. The larger discrepancies towards the higher energy end of the spectrum implies use of a second scaling factor for the region above 19958 cm−1 may be appropriate.

189 Table 8.6: 9-deuteroanthracenyl assignments and comparison to theory, for both the 9Deq-An and

9Dax-An conﬁgurations, including the mean absolute deviation (MAD) for each. All units are in wavenumbers (cm−1).

9Deq-An 9Dax-An

Observed Relative to 9Deq-An Assignment Theory ∆T −O Relative to 9Dax-An Assignment Theory ∆T −O

19112.5 0.0 0 0.0 0 19129.2 16.6 0.0 0 0 0

19143.8 31.3 2ν37 31.3 0 14.6

19161.7 49.2 32.6 2ν37 32.6 0

19182.8 70.3 4ν37 70.3 0 53.6

19266.8 154.2 ν36 151.5 -2.7 137.6

19280.5 168.0 2ν69 162.6 -5.4 151.3

19282.5 170.0 153.4 ν36 150.4 -3.0

19295.6 183.0 166.4 2ν69 162.4 -4.0

19298.5 185.9 ν362ν37 185.5 -0.4 169.3

19314.9 202.4 185.8 ν362ν37 186.0 0.2

19346.4 233.9 ν35 228.9 -5.0 217.2

19363.3 250.7 234.1 ν35 231.2 -2.9

19378.7 266.1 ν352ν37 265.1 -1.0 249.5

19684.0 571.5 ν30 563.1 -8.4 554.8

19692.9 580.4 563.7 ν30 551.1 -12.6

19712.9 600.4 ν302ν37 602.7 2.4 583.7

19801.3 688.8 ν27 678.9 -9.9 672.2

19804.4 691.8 ν26 686.4 -5.4 675.2

19815.7 703.1 686.5 ν27 679.9 -6.6

19826.6 714.1 697.5 ν26 691.9 -5.6

19833.9 721.4 ν272ν37 720.1 -1.3 704.8

19836.3 723.7 2ν67 724.6 0.9 707.1

19856.6 744.0 727.4 2ν67 728.6 1.2

19918.0 805.4 ν30ν35 805.3 -0.1 788.8

19927.6 815.1 798.4 ν30ν35 797.9 -0.6

19947.0 834.4 ν30ν352ν37 836.6 2.1 817.8

19958.4 845.8 ν23 857.4 11.6 829.2

19965.3 852.8 836.1 ν23 866 29.9

19987.4 874.8 ν232ν37 877.1 2.3 858.2

20029.7 917.1 ν21 905.2 -11.9 900.5

20038.8 926.2 909.6 ν21 903.2 -6.4

20098.2 985.6 ν19 969.0 -16.6 969.0

20114.6 1002.0 985.4 ν19 969.1 -16.3

20129.1 1016.6 ν192ν37 1016.9 0.3 1000.0

AVG DEV 5.6 AVG DEV 6.9 Wavenumbers (cm -1 )

1 1 35 030 0 0 0 1 1 1 1 1 1 0 36 0 35 0 legend 30 0 27 0 23 0 19 0 X 1 1 0 26 0

2 1 2 1 69 0 X 0 67 0 21 0 1 2 2 X 0 37 0 1 2 1 69 0 26 0 67 0 21 0 1 2 X 0 37 0 1 4 X 0 37 0

Relative Wavenumbers (cm-1 )

Figure 8.11: REMPI spectrum of the 9-deuteroanthracenyl radical. Assignments of the ﬁrst instance of a’ and a" modes are made with coupled progressions of 2 and 4 quanta of ν37 following the legend inset. Bands attributed to excitation of HD9Deq-An are marked in red, whilst those coming from the 9Dax-An side of the double well are marked in blue.

The mean absolute deviations, calculated again by only considering the a0

00 −1 −1 and a modes uncoupled to ν37, are 5.6 cm and 6.9 cm for the 9Deq-An and

9Dax-An radicals respectively. As these values are also considerably skewed by the outliers of ν23 and ν19, this is a reasonable assignment, conﬁrming the choice of the relatively cheap level of theory for such a large radical.

The intensities of the modes map very consistently, however there are two peaks that are observed to be much stronger upon deuteration, at positions 19826.6

−1 −1 cm and 20038.8 cm , corresponding to the 9Dax-An modes ν26 and ν21 respectively. Both of these modes contain the CDH in plane rock, and the Franck-Condon overlap could possibly be larger for motions involving the deuterium in this position.

191 Lifetime Scan of 9H-An

A scan of the lifetime for the 9H-An radical was also taken, as shown in ﬁgure 8.12.

y = 2.9 exp (-x / 280.7 ) + 1.3

Time (ns)

Figure 8.12: Lifetime scan of the D1 origin of 9-hydroanthracenyl, ﬁt to a ﬁrst order exponential decay function.

The lifetime of the first excited state was taken by fixing the excitation laser to 19115 cm−1, and scanning the ionisation laser in time. The lifetime of the state is taken from the inverse of the exponential decay constant, equal to 280.7 ± 12.5 ns. This lifetime indicates that further fluorescence experiments could be undertaken to determine the exact nature of the ground state and the size of the barrier between the wells. The lifetime is likely similar for the 9D-An radical and similar experiments would confirm the relative energies of the 9Deq-An and 9Dax-An wells in the ground state.

192 8.5 Concluding Remarks

In conclusion, we have been able to observe the first electronic transition of the radicals produced from the hydrogen and deuterium addition to anthracene. For both isotopologues, the addition site was the same, at the 9- position. This was confirmed by the relative energies of formation, and the calculated origin positions for excitation. The ionisation energies were measured, confirming the assignment and suggesting that only one isomer was present for each H and D addition. Hole burning experiments were conducted for the deuterated analogue, confirming the first 5 peaks all come from the same ground state, rather than isotopomers formed from H-D exchange or from hot bands.

0 As only one a frequency was calculated within the origin region, the ν37 butterfly inversion mode, we fit excited state double well potentials to this mode, crossing through a planar transition state. For the 9H-An radical, the two minima were identical, yielding a symmetric double well potential, resulting in only even quanta of the ν37 mode able to gain any Franck-Condon intensity upon excitation. For the 9D-An radical, the deuterium located at an equatorial or axial position yielded different zero point energies, lifting the degeneracy seen for the 9H-

An radical. This resulted in excitations to both of the 9Deq-An and 9Dax-An wells, accounting for the additional observed peaks in the REMPI spectrum.

This information helped us to assign the spectra, which both contained

0 progressions of the ν37 inversion mode built onto a large number of a modes. The −1 −1 −1 mean absolute deviations of 5.8 cm , 5.6 cm and 6.9 cm for the 9H-An, 9Deq-

An and 9Dax-An peak positions respectively, conﬁrmed the choice of the relatively cheap TD-DFT B3LYP/6-311+G(d,p) level of theory, run using GAUSSIAN16.

The lifetime of the 9H-An radical was determined to be 280.7 ±12.5 ns. As a result, the ground state potentials could be explored through ﬂuorescence spectroscopy, as the lifetime of the excited 9H-An radical is suﬃcient for these experiments.

193 8.6 Appendix to Chapter 8

A complete list of the vibrational frequencies for the D1 electronic states of the

9H-An, 9Deq-An and 9Dax-An radicals;

Table 8.7: Vibrational frequencies for the D1 electronic state of the 9H-An radical

9H-An

Mode # Frequency Symmetry Mode # Frequency Symmetry Mode # Frequency Symmetry

69 83.7 a00 46 1383.4 a00 23 896.8 a0 68 219.3 a00 45 1436.5 a00 22 900.4 a0 67 375.5 a00 44 1499.9 a00 21 946.2 a0 66 391.0 a00 43 1543.8 a00 20 962.5 a0 65 444.8 a00 42 1663.4 a00 19 999.2 a0 64 504.8 a00 41 3148.5 a00 18 1067.2 a0 63 567.5 a00 40 3157.7 a00 17 1163.2 a0 62 654.5 a00 39 3179.3 a00 16 1186.1 a0 61 710.9 a00 38 3194.2 a00 15 1273.0 a0 60 719.9 a00 37 37.1 a0 14 1312.5 a0 59 800.9 a00 36 162.9 a0 13 1363.5 a0 58 817.1 a00 35 239.3 a0 12 1389.8 a0 57 903.0 a00 34 278.2 a0 11 1441.5 a0 56 920.8 a00 33 372.3 a0 10 1469.0 a0 55 958.5 a00 32 405.5 a0 9 1516.9 a0 54 964.4 a00 31 479.9 a0 8 1616.5 a0 53 1101.0 a00 30 582.8 a0 7 2950.5 a0 52 1178.9 a00 29 635.4 a0 6 3035.7 a0 51 1192.7 a00 28 650.2 a0 5 3149.0 a0 50 1224.0 a00 27 704.2 a0 4 3149.2 a0 49 1243.4 a00 26 713.6 a0 3 3158.6 a0 48 1297.2 a00 25 789.5 a0 2 3179.6 a0 47 1355.9 a00 24 866.1 a0 1 3195.3 a0

194 Table 8.8: Vibrational frequencies for the D1 electronic state of the 9Deq-An radical

9Deq-An

Mode # Frequency Symmetry Mode # Frequency Symmetry Mode # Frequency Symmetry

69 83.8 a00 46 1374.2 a00 23 883.9 a0 68 219.6 a00 45 1430.4 a00 22 898.8 a0 67 373.5 a00 44 1497.6 a00 21 933.2 a0 66 391.3 a00 43 1543.7 a00 20 962.2 a0 65 443.7 a00 42 1663.1 a00 19 998.9 a0 64 497.2 a00 41 3148.5 a00 18 1067.2 a0 63 567.2 a00 40 3157.7 a00 17 1158.4 a0 62 648.1 a00 39 3179.3 a00 16 1183.3 a0 61 699.9 a00 38 3194.2 a00 15 1272.0 a0 60 711.9 a00 37 35.9 a0 14 1312.7 a0 59 800.9 a00 36 156.2 a0 13 1364.3 a0 58 808.0 a00 35 235.9 a0 12 1390.1 a0 57 884.1 a00 34 273.1 a0 11 1281.4 a0 56 917.8 a00 33 371.7 a0 10 1466.4 a0 55 958.6 a00 32 404.1 a0 9 1516.9 a0 54 949.6 a00 31 477.7 a0 8 1615.9 a0 53 1101.0 a00 30 580.5 a0 7 2951.3 a0 52 1179.4 a00 29 638.1 a0 6 2229.0 a0 51 985.8 a00 28 628.8 a0 5 3148.9 a0 50 1216.5 a00 27 699.9 a0 4 3149.2 a0 49 1242.9 a00 26 707.7 a0 3 3158.6 a0 48 1289.3 a00 25 785.9 a0 2 3179.6 a0 47 1343.2 a00 24 832.4 a0 1 3195.3 a0 Table 8.9: Vibrational frequencies for the D1 electronic state of the 9Dax-An radical

9Dax-An

Mode # Frequency Symmetry Mode # Frequency Symmetry Mode # Frequency Symmetry

69 83.7 a00 46 1380.8 a00 23 893.8 a0 68 218.6 a00 45 1435.3 a00 22 899.0 a0 67 375.6 a00 44 1500.1 a00 21 931.1 a0 66 391.3 a00 43 1543.6 a00 20 962.2 a0 65 439.1 a00 42 1663.4 a00 19 999.1 a0 64 501.2 a00 41 3148.5 a00 18 1066.9 a0 63 566.8 a00 40 3157.8 a00 17 1162.6 a0 62 647.6 a00 39 3179.3 a00 16 1185.6 a0 61 683.7 a00 38 3194.2 a00 15 1272.9 a0 60 713.5 a00 37 36.0 a0 14 1312.5 a0 59 800.0 a00 36 155.0 a0 13 1364.4 a0 58 816.8 a00 35 238.4 a0 12 1391.7 a0 57 888.2 a00 34 271.3 a0 11 1279.3 a0 56 914.6 a00 33 370.4 a0 10 1467.1 a0 55 960.2 a00 32 402.6 a0 9 1516.9 a0 54 954.5 a00 31 475.6 a0 8 1616.4 a0 53 1102.0 a00 30 568.2 a0 7 2163.0 a0 52 1182.3 a00 29 643.2 a0 6 3034.7 a0 51 973.0 a00 28 631.8 a0 5 3149.0 a0 50 1208.5 a00 27 700.9 a0 4 3149.2 a0 49 1242.6 a00 26 713.3 a0 3 3158.6 a0 48 1297.4 a00 25 786.6 a0 2 3179.6 a0 47 1352.6 a00 24 829.2 a0 1 3195.3 a0 Chapter 9

The Protonation and Deuteronation of Anthracene

9.1 Introduction

In recent years our group has successfully been able to measure the electronic transitions of the protonated and deuteronated naphthalene molecules, and their neutral analogues. [140, 264] Based on initial work by Sebree et al. on 1-hydronaphthyl radical [259, 260] and Alata et al. on 1-hydronaphthylium cation,[141] Krechkivska et al. performed triple-resonance spectroscopy to observe the D1 ← D0 excitation of the 1-hydronaphthylium cation with greatly improved resolution.[140] Later, the excitation spectrum of the deuteronated naphthalene was observed, and its spectroscopic features and thermochemistry were explored in relation to its isotopologue. [264]

Whilst neither species reproduced a DIB carrier, there remains a possibility that larger, protonated or deuteronated, PAH molecules are responsible for some of the DIBs. A study by Pathak and Sarre emphasised the correlation between the ﬁrst electronic transitions of H-PAH+’s, based on TD-DFT studies, and the diﬀuse interstellar bands. [269] As we tend towards larger species, the next logical choice is to investigate protonation on the three ring containing PAH, anthracene.

197 Protonated anthracene was first observed in the gas phase by Alata et al., [141] as part of their work on the detection of protonated acenes, naphthalene, anthracene and tetracene. They investigated the dependence of the electronic absorption energy as a function of the number of rings, the excited state lifetimes of these larger protonated PAH’s when compared to the very short lifetime of protonated benzene, and the charge transfer character of the excited state. Alata et al. observed the first electronic transition of protonated anthracene to have an origin at 491.43 nm. Only one spectral carrier was identified, which they assigned to be the 9-hydroanthracenylium (9H-An+) cation, based on calculations of the transition energy and energy of formation. The other possible isomers, 1H-An+ and 2H-An+,

−1 lie ∼43 and ∼59 kJ mol higher in energy respectively. The S1 ← S0 transition was observed at 2.51 eV, and the adiabatic transition energy was calculated to be 2.7 eV for the 9H-An+ isomer (2.5 eV when ZPE corrected), whilst the 1H-An+ and 2H-An+ were calculated to have transitions of 1.7 eV and 1.6 eV respectively.

In the case of protonated benzene, excitation from S0 to S1 induces a large geometric change. A very short excited state lifetime is predicted by the calculation of conical intersections. [270] As such, no vibrational structure is reported for the

+ broad UV photodissociation spectrum of C6H7 .[271]

To the contrary, each of the protonated acenes observed by Alata et al. showed resolved vibrational structure, indicating a longer excited state lifetime. [141] Indeed, for 9H-An+, the spectrum spans some 900 cm−1. In this region four vibrational bands and their progressions were assigned. In assigning this spectrum, ab initio calculations were performed with two methods, TD-DFT with a B3LYP functional and split-valence plus polarization (SVP) basis, and the resolution of the identity second order Mφller-Plesset perturbation theory (RI-MP2) functional with a resolution of the identity second-order approximate coupled-cluster (RI-CC2) basis. The former method resulted in the excited state being of planar geometry, whilst the latter gave a non-planar structure, with the carbon of the CH2 group being out of plane by ∼10◦. A peak observed ∼100 cm−1 from the origin was assigned to a single quantum of the ν69, out-of-plane butterﬂy motion of the rings, which would only be allowed if the molecule is non-planar. As further progressions of this mode are

198 observed to have much more intensity than the calculation estimates, they reason that the frame of the 9H-An+ cation must be even further out of plane than theory suggests.

Following this study, Garkusha et al. observed the electronic absorptions of protonated anthracene in a neon matrix. [258] They investigated the range from 500 to 400 nm, and within this region claim to have observed all three isomers of H-An+. The three band systems had reported origins of 493.8 nm, 453.5 nm and ∼410 nm. As such, they similarly performed TD-DFT calculations on the three isomers, but at the B3LYP/6-311G(d,p) level of theory, and determined excitation energies of 2.78, 1.92 and 1.91 eV for the 9H-An+, 1H-An+ and 2H-An+ isomers respectively, consistent with Alata’s values. However, instead of assigning 9H-An+ to the band at 2.51 eV, they reason that TD-DFT values for this level of theory in the literature are often up to 0.5 eV away from experiment, and then shift all of their values by 0.65 eV to match the experiment. As a result they reassign the origin at 2.51 eV from 9H-An+ to 1H-An+, however they then claim that this result is consistent with the assignment from Alata; a clear oversight.

In our studies on the hydrogenation (and deuteration) of anthracene, only addition at the 9- site was observed. As such, when ionising the 9-hydroanthracenyl radical, we are guaranteed to form exclusively the 9H-An+ cation. By observing the S1 ← S0 transition, we will be able to compare our ﬁndings with the previous literature and conﬁrm the band’s origin.

199 9.2 Experimental

To generate cold cations, the ionisation laser was fixed at the onset of the ionisation potential, to ensure cations were formed in the ground vibrational state. Recently, our group made use of this method to measure the electronic spectroscopy of the cold cation formed from H and D addition to naphthalene. This experiment is inherently complicated. It requires the use of three counter-propagating lasers, all which must strike the same molecule within nanoseconds of each another. It is hence extremely subject to small changes in spatial and temporal alignment. Further, upon excitation, the cation must dissociate before reaching the detector, to ensure depletion of the parent signal. As the method of detection that Alata and coworkers utilised required that the protonated acenes fragment upon excitation, we can be confident in our experimental approach. [141] Information about the excited state is obtained through the resonant dissociation of the cation, recorded as a depletion spectrum. The same SIRAH COBRASTRETCH, used for the hole burning experiments in chapter 8, was used as the depletion laser here. The dyes Coumarin 503, 480 and 460 were used to measure the range of the cations vibronic spectra, and this laser was fired 70 ns after the initial 1+10 REMPI scheme, as to not interfere with the REMPI process.

200 9.3 Results and Discussion

Threshold ionisation was employed to generate the isomer-speciﬁc, vibrationally

+ 0 cold, 9H-An cation. The 1+1 REMPI scheme proceeds through the D1 ← D0 electronic transition of 9-hydroanthracenyl at 19115 cm−1, see chapter 8. The ionisation laser is then parked at 32098 cm−1, giving a combined ionisation energy of 51213 cm−1, only 87 cm−1 above the ionisation potential. Signal was recorded as a function of the depletion laser wavelength, fired 70 ns after the 1+10 REMPI scheme. Similarly, the deuterated analogue 9D-An+ was generated by the corresponding 1+10 REMPI scheme, where excitation of the 9-deuteroanthracenyl radical at 19112 cm−1 was followed by an ionisation laser set to 32098 cm−1. This gives a combined energy of 51210 cm−1, 74 cm−1 above the observed ionisation potential of 51136 cm−1. The excitation spectra, recorded for the 9H-An+ and 9D-An+ cations, are displayed below in figure 9.1. The ionisation laser wavelength was chosen to provide maximum signal whilst not exciting the first observed transition for either isotopologue, which are observed over 200 cm−1 from the respective origins.

The observed spectra span some 1400 cm−1 from the origin, with more structure of increasing complexity expected beyond. There appear to be no intense peaks over the ﬁrst 200 cm−1, contrary to that observed by Alata et al. who assign the single quanta out of plane butterﬂy motion to a peak observed at ∼100 cm−1. To explore this contradiction, the excited state geometries and frequencies were calculated with a range of functionals, M06-2X, wB97XD, B3LYP and CAM-B3LYP, all utilising the same 6-311+G(d,p) basis set. All four methods converged to a planar excited state, with non-equivalent side rings. There was no evidence of the sp2 hybridised carbon puckering out of plane by ∼10◦.

The geometric details from the ground and excited state (TD)-DFT B3LYP calculation is shown in ﬁgure 9.2. Unlike the neutral analogue, there is no out-of- plane distortion along the butterﬂy coordinate. Instead, the molecule adopts a planar C2v symmetry in its ground state. Upon excitation, the framework distorts in the plane of the molecule, where the average outer ring size grows from 1.403 Å to 1.405 Å on the left ring and 1.407 Å on the right ring. In the left ring, the

201 9H-An+

9D-An+

20200 20400 20600 20800 21000 21200 21400 21600 21800 Wavenumbers (cm -1 )

Figure 9.1: Top: Triple resonance depletion spectrum of the m/z 179 trace, representing the 9H- An+ cation. Bottom Left: Triple resonance depletion spectrum of the m/z 180 trace, representing the 9D-An+ cation. geometric distortion in each C-C bond length is subtle. This is not the case for the ring on the right side, where large elongations and contractions are calculated for various C-C bond lengths. This non-symmetric ring expansion could arise due to a charge transfer of the electron density upon excitation, and understanding it requires further theoretical and experimental eﬀorts, outside the scope of this study. For the deuterated isomer, the structure converges similarly, though strictly the symmetry reduces from C2v and Cs, to Cs and C1, for the ground and excited states respectively. For the purposes of assignment, however, we make use of the (higher order) symmetry labels from the 9H-An+ isotopologue as the modes will more or less retain the same character.

202 1.099 1.102 1.084 1.099 1.084 1.085 1.102 1.085

1.394 1.390 1.500 1.5001.390 1.394 1.399 1.386 1.508 1.491 1.450 1.391

1.083 1.082 1.084 1.084 1.403 1.437 1.388 1.383 1.405 1.427 1.427 1.405

1.083 1.083 1.083 1.083

1.377 1.422 1.406 1.406 1.422 1.377 1.379 1.428 1.393 1.470 1.396 1.439 1.084 1.082 1.085 1.084 1.087 1.084

Ground State Excited State

C2V CS

Figure 9.2: Ground and excited state structural parameters for the 9H-An+ cation. The symmetry reduces from C2v to Cs upon excitation, as only the plane of symmetry in the plane of the anthracene rings remain.

+ + The calculated modes for the S1 ← S0 transition of 9H-An and 9D-An are shown below in table 9.1. These calculations were performed at the TD-DFT B3LYP/6-311+G(d,p) level of theory for consistency with the neutral analogues. Only modes up to 1200 cm−1 are displayed, as beyond this point the spectrum becomes too convoluted to assign peaks unambiguously. A full list of the calculated frequencies for both the 9H-An+ and 9D-An+ cations is provided at the end of the chapter in appendix 9.5.

The modes for the 9H-An+ cation are labelled according to the Mulliken

+ convention for a Cs point group, and those for 9D-An are labelled such that the same mode number represents the same molecular motion, and we retain the symmetry from the 9H-An+ calculation. The scale factors were experimentally ﬁt to be 0.991 and 0.998 for the 9H-An+ and 9D-An+ cations respectively, which indicate that the level of theory used is appropriate and also more eﬀective at calculating the cation frequencies compared to the neutral, where the chosen scale factor was 0.97.

Assignment of the 9H-An+ Cation

The assigned spectrum of the 9H-An+ cation is shown in ﬁgure 9.3, which includes the spectrum obtained by Alata et al. for comparison.[141] The mode labels assigned by Alata have been adapted to ensure that the same molecular motion is represented

203 Table 9.1: List of frequencies (under 1200cm−1) for the 9H-An+ and 9D-An+ cations, calculated with the TD-DFT B3LYP/6-311+G(d,p) level of theory. Symmetry labels are determined for the 9H-An+ cation following Mulliken convention. Mode labels were rearranged for the 9D-An+ cation to retain the same label for the same molecular motions. Frequencies were scaled by 0.991 and 0.998 for the 9H-An+ and 9D-An+ cations respectively to ﬁt with experiment. All units are in wavenumbers (cm−1).

9H-An+ 9D-An+ Mode # Freq Sc. Freq Sym Mode # Freq Sc. Freq Sym 26 1215.5 1205.1 a0 26 1214.5 1212.0 a0 27 1212.7 1202.2 a0 27 1210.8 1208.3 a0 28 1198.3 1188.0 a0 28 1197.2 1194.8 a0 29 1191.2 1180.9 a0 29 1190.6 1188.2 a0 30 1186.4 1176.2 a0 30 1187.2 1184.8 a0 31 1167.6 1157.6 a0 31 1171.2 1168.8 a0 32 1141.7 1131.8 a0 32 1142.2 1139.9 a0 33 1090.1 1080.7 a0 33 1093.1 1090.9 a0 34 1050.0 1040.9 a0 34 1051.4 1049.2 a0 35 1030.3 1021.5 a0 35 1030.2 1028.1 a0 36 904.2 896.4 a0 36 896.9 895.1 a0 37 884.3 876.7 a0 37 879.2 877.4 a0 38 800.3 793.4 a0 38 770.5 768.9 a0 39 736.1 729.8 a0 39 731.3 729.8 a0 40 662.9 657.2 a0 40 661.4 660.1 a0 41 612.6 607.3 a0 41 608.6 607.3 a0 42 598.2 593.0 a0 42 596.0 594.8 a0 43 512.4 508.0 a0 43 506.2 505.2 a0 44 379.9 376.6 a0 44 378.6 377.9 a0 45 368.1 364.9 a0 45 368.0 367.2 a0 46 231.0 229.1 a0 46 230.5 230.0 a0 48 1200.8 1190.5 a00 48 950.6 948.7 a00 49 1009.7 1001.0 a00 49 1009.7 1007.7 a00 50 993.0 984.5 a00 50 993.1 991.1 a00 51 983.7 975.2 a00 51 985.2 983.2 a00 52 973.9 965.5 a00 52 979.6 977.6 a00 53 920.9 913.0 a00 53 907.8 906.0 a00 54 894.8 887.1 a00 54 889.7 887.9 a00 55 872.5 865.0 a00 55 846.9 845.2 a00 56 790.7 783.9 a00 56 790.7 789.1 a00 57 754.1 747.6 a00 57 747.8 746.3 a00 58 716.2 710.1 a00 58 712.4 710.9 a00 59 699.0 693.0 a00 59 685.5 684.1 a00 60 549.1 544.3 a00 60 530.0 528.9 a00 61 505.3 500.9 a00 61 500.6 499.6 a00 62 441.6 437.8 a00 62 440.9 440.0 a00 63 421.8 418.1 a00 63 418.6 417.7 a00 64 359.1 356.0 a00 64 355.4 354.6 a00 65 272.7 270.4 a00 65 269.7 269.1 a00 66 224.4 222.4 a00 66 222.0 221.5 a00 67 168.7 167.3 a00 67 159.4 159.1 a00 68 104.3 103.4 a00 68 104.1 103.9 a00 69 42.4 42.1 a00 69 41.3 41.2 a00 0 1 2 3 0 0 46 0 46 0 46 0

1 2 3 1 2 3 65 0 65 0 65 0 44 0 44 0 44 0

Alata et al. 1 2 3 69 0 69 0 69 0

this work

0 0 0

1 1 1 46 0 45 0 44 0

2 2 2 1 1 1 68 0 67 0 46 0 46 0 67 0 68 0

1 1 1 1 1 1 1 1 2 2 67 0 69 0 67 0 68 0 65 0 69 0 66 0 67 0 67 0 69 0

1 2 1 1 1 46 0 68 0 46 0 67 0 69 0 0 0 0 0 0 1 0 2 0 1 0 1 0 0 0 0 0 1 0 1 1 4 0 2 1 0 0 2 0 0 2 1 1 0 2 2 2 0 1 1 3 0 0 2 2 0 1 1 0 0 2 1 0 0 0 1 2 2 42 68 40 66 46 69 67 67 69 69 67 46 69 69 67 67 69 38 69 46 68 48 69 69 44 45 69 69 47 1 0 1 0 1 0 1 0 1 0 2 0 1 0 2 0 1 0 2 0 1 0 1 0 2 0 1 0 1 0 2 0 2 0 1 0 1 0 1 0 1 0 1 0 1 0 44 62 44 67 66 67 65 66 44 68 67 44 46 67 67 67 46 67 65 44 65 67 46 1 0 1 0 1 0 2 0 1 0 1 0 1 0 1 0 1 0 2 0 1 0 1 0 1 0 45 46 46 46 67 44 65 46 46 46 45 46 65 1 0 1 0 1 0 46 46 44

Relative Wavenumbers (cm -1 )

Figure 9.3: Assignment of the excitation spectrum of the 9H-An+ cation. The top spectrum was adapted from reference [141], as are the mode labels to be consistent with our labelling scheme. by our labeling scheme, as the chosen axis of Cs symmetry is different. First and foremost, one can see that the broader peaks observed by Alata et al. are resolved into (multiple) sharper peaks, as in our case the cation’s ground state has been prepared solely in the ground vibrational state with low rotational temperature. Furthermore, laser selection from the ground state only excites a narrow range of rotational states, governed by the spectral width of the laser pulse. The origin transition is confirmed at 491.59 nm (20342.1 cm−1), as our method precludes the possibility of hot band transitions. This value is in reasonable agreement with the 491.43 nm recommended by Alata et al. Our increased resolution allows for a more accurate determination of this peak position. Confirmation of the origin position of the 9H-An+ cation simultaneously contradicts the assignment of Garkusha et al., who state that this spectral region owes to the 1H-An+ isomer.[258]

Observed peak positions and mode assignments are given in table 9.2. Both Alata et al. and Garkusha et al. respectively observe the same progression (at similar positions) of 232 cm−1 and ∼200 cm−1 built onto the origin and attribute it to the in-plane opening of the C-C-C angle around the sp hybridised carbons. In our case, this band is split into three peaks, two of which combine to give a wider ﬁrst peak. We attribute the ﬁrst peak to the same motion, a single quantum of

0 −1 −1 the a mode ν46, calculated to be 229.1 cm and observed at 226.2 cm . With no other a0 modes calculated within this region, the two other peaks are attributed to even quanta of a00 modes. The shoulder, with a peak observed at 227.4 cm−1 from the origin, is attributed to the ﬁrst overtone of the out of plane ring twist,

−1 −1 ν68, calculated to be 206.8 cm . The next lowest mode, at 237.2 cm is assigned −1 as the combination of modes ν69 and ν67, predicted at 209.3 cm . These modes are likely to couple, as the former represents the butterﬂy motion and the latter an out of plane CH2 rock coupled to a butterﬂy inversion. Though these assignments deviate from theory by 20.6 and 27.9 cm−1, the deviation per mode is only half this value. As no single quantum of the 5 low frequency a00 modes were observed over

−1 the ﬁrst 300 cm , we can be conﬁdent of our reassignment to planar Cs symmetry.

The observed band at 283.6 cm−1 is assigned as the combination of a00

−1 modes ν68 and ν67, calculated to be 270.7 cm . The following band, assigned by

206 Table 9.2: 9H-An+ assignments and comparison to theory. Unique modes are represented in blue and are used to calculate the the mean absolute deviation (MAD). All units are in wavenumbers (cm−1).

9H-An+ Assignment

Observed Relative Assignment Theory ∆T −O

20342.1 0.0 0 0 0

20568.3 226.2 ν46 229.1 2.8

20569.5 227.42 ν68 206.8 -20.6

20579.3 237.2 ν67ν69 209.3 -27.9

20625.7 283.6 ν67ν68 270.7 -12.9

20663.3 321.2 ν65ν69 312.4 -8.8

20670.7 328.52 ν67 334.5 6.0

20707.1 365.0 ν45 364.9 -0.1

20710.4 368.3 ν44 376.6 8.3

20733.7 391.6 ν66ν67 389.7 -1.8

20792.7 450.6 2ν46 452.4 1.8

20796.0 453.8 ν462ν68 453.6 -0.2

20803.7 461.6 ν46ν67ν69 463.4 1.8

20818.1 476.0 2ν672ν69 474.4 -1.6

20849.6 507.5 ν46ν67ν68 509.8 2.3

20889.4 547.3 ν46ν65ν69 547.4 0.1

20893.2 551.0 ν462ν67 554.7 3.7

20902.3 560.1 ν65ν672ν69 558.4 -1.7

20929.8 587.7 ν42 593.0 5.4

20936.3 594.1 ν44ν46 594.5 0.4

20938.2 596.1 ν442ν68 595.7 -0.4

20944.3 602.1 ν45ν67ν69 602.2 0.1

20948.6 606.5 ν44ν67ν69 605.5 -0.9

20959.4 617.3 ν46ν66ν67 617.8 0.5

20994.2 652.1 ν40 657.2 5.1

20998.1 656.0 4ν67 657.0 1.0

21002.4 660.3 ν62ν66 660.3 0.0

21016.2 674.1 3ν46 678.6 4.5

21021.7 679.6 2ν462ν68 679.8 0.2

21026.6 684.5 ν45ν65ν69 686.2 1.7

21028.2 686.1 2ν46ν67ν69 687.8 1.7

21037.9 695.7 ν442ν67 696.8 1.1

21041.9 699.8 ν462ν672ν69 702.2 2.5

21075.0 732.9 ν44ν45 733.3 0.4

21078.2 736.1 2ν44 736.6 0.5

21116.2 774.1 2ν46ν65ν69 771.8 -2.3

21119.0 776.9 2ν462ν67 779.1 2.2

21125.5 783.4 2ν662ν67 783.1 -0.2

21128.7 786.6 ν46ν65ν672ν69 784.6 -2.0

21133.6 791.4 ν462ν672ν68 793.3 1.9

21135.9 793.7 ν38 793.4 -0.4

21161.0 818.9 ν442ν46 818.9 0.0

21172.6 830.4 ν44ν46ν67ν69 831.7 1.3

21175.6 833.5 ν44ν672ν68ν69 832.9 -0.6

AVG DEV 5.0 Alata et al. as the ﬁrst overtone of the butterﬂy motion, is split into two peaks at

−1 −1 −1 321.2 cm and 328.5 cm . As 2ν69 is calculated to be only 84.2 cm , the ﬁrst of these bands are reassigned as the combination of the out of plane CH wag, ν65 with mode ν69, and the latter as the ﬁrst overtone of ν67. The band observed by Alata at 371 cm−1, is also split into two peaks at 365.0 cm−1 and 368.3 cm−1. These

0 are assigned as the a modes ν45 and ν44, corresponding to in plane ring breathing and ring rotation motions. The out-of-plane antisymmetric ring puckering, ν66 is −1 calculated to combine with ν67 at 389.7 cm matching the band observed at 391.6 cm−1.

Beyond this point, only 4 other unique modes are assigned, which have all been coloured blue in table 9.2. With the increasing number of spectral lines observed, it became apparent that we could no longer simply assign the spectrum to single quanta of a0 modes and simple combinations of two a00 modes. As such, peak positions were extracted, and the spectrum was ﬁt to linear combinations of previously assigned modes. The ∆T −O value for these modes is taken as the diﬀer- ence of the peak position and the sum of the previously observed band positions,

−1 rather than the theoretically obtained values. The small (5 cm ≤) ∆T −O values attributed to these progressions indicate harmonic character. These long progressions reﬂect the large calculated geometric change.

−1 0 The band observed at 587.7 cm is assigned to the a mode ν42, calculated at 593.0 cm−1, which represents the contraction and elongation of the peripheral rings. ν42 represents a similar motion, involving all three rings, and is assigned to −1 the band at 652.1 cm . The out of plane CH wag, ν62 is predicted to combine with −1 ν66 at 660.3 cm and is observed likewise. Finally, the mode which contains large character of the change in geometry upon electronic excitation, ν38, is calculated to have a frequency of 793.4 cm−1 and is assigned to the peak at 793.7 cm−1.

The MAD is calculated to be only 5.0 cm−1 per mode. Only assignments made with by comparison to the TD-DFT theory, coloured blue in table 9.2, were considered in calculating the MAD. The in plane a0 modes were calculated more precisely at this level of theory than the a00 modes and combinations. Krechkivska et al. conclude similarly for their assignment of the 1-hydronaphthylium cation, in

208 that the RI-CC1 method used by Alata et al. outperforms their TD-DFT B3LYP method for the out of plane modes.

Assignment of the 9D-An+ Cation

Spectral assignments for the 9D-An+ cation can be seen in ﬁgure 9.4. Modes involving more pronounced displacements of the deuterium atom are of course shifted.

The largest eﬀects of deuteration were calculated as follows; the frequency of ν48, −1 −1 representing the CH2 in plane twist, reduces from 1190.5 cm to 948.7 cm upon deuteration. The frequency of the scissoring motion, ν20 also lowers from 1390.8 cm−1 to 1243.8 cm−1. Finally, the CH symmetric and anti symmetric stretching modes become local modes when deuterated. The symmetric stretch, ν10, undergoes the largest reduction, from 2940.2 cm−1 to 2173.9 cm−1, whilst the frequency

−1 −1 of the asymmetric CH stretch, ν9 increases from 3143.0 cm to 3170.3 cm , as the deuterium atom remains roughly stationary.

Assignments and band positions for the deuterated analogue, 9D-An+, can be seen in table 9.3. With the break in strict Cs symmetry, the vibrational modes of the 9D-An+ cation need not be rigorously pure a0 and a00 motions. Upon inspection the modes are almost identical and any single quanta of a00 vibrations are likely to have minimal intensity, apart from the local mode formed about the CD stretch, outside the spectral range of our study.

In the 9D-An+ spectrum, the ﬁrst set of modes ∼230 cm−1 more distinctly show three peaks. The third of these has the most intense progressions, similar to

0 + the a mode ν46 in the 9H-An spectrum, and is assigned accordingly. Contrary + to the 9H-An assignment, the combination of ν67 and ν69 is calculated to appear before the overtone of ν68. The diﬀerence between theory and observation for these three peaks is much smaller than for the the 9H-An+, and contributes to a smaller

+ MAD value per mode. In the 9H-An spectrum, the ν67ν68 mode is calculated 12.9 cm−1 below the assigned peak, here the the same mode is calculated 17.4 cm−1 below the observed value of 280.4 cm−1, showing a consistent under prediction. A

209 0 0 0

1 1 1 2 1 1 2 67 0 69 0 46 0 67 0 45 0 44 0 46 0

2 1 1 1 1 1 1 1 1 68 0 67 0 68 0 46 0 67 0 69 0 46 0 67 0 68 0

2 2 1 2 68 0 69 0 46 0 68 0 2 0 0 0 0 0 2 0 2 0 0 1 0 1 0 2 1 0 2 3 1 1 0 1 1 0 0 2 2 0 0 2 0 1 1 0 1 0 0 1 0 1 46 68 68 46 68 67 46 68 41 69 69 40 68 69 68 65 67 68 38 46 69 1 0 1 0 1 0 2 0 1 0 1 0 2 0 1 0 1 0 1 0 1 0 2 0 2 0 1 0 1 0 2 0 1 0 45 46 45 67 65 67 46 67 45 67 40 67 46 59 67 68 40 1 0 1 0 1 0 1 0 1 0 2 0 2 0 3 0 46 46 45 40 46 46 46 67 1 0 45

Relative Wavenumbers (cm -1 )

Figure 9.4: Assignment of the excitation spectrum of the 9D-An+ cation. Table 9.3: 9D-An+ assignments and comparison to theory. Unique modes are represented in blue and are used to calculate the the mean absolute deviation (MAD). All units are in wavenumbers (cm−1).

9D-An+ assignment

Observed Relative Assignment Theory ∆T −O

20358.3 0

20588.6 215.4 ν67ν69 200.3 -15.1

20579.0 220.72 ν68 207.8 -12.8

20573.7 230.3 ν46 230.0 -0.2

20725.6 280.4 ν67ν68 263.0 -17.4

20954.3 290.92 ν682ν69 290.2 -0.7

20818.2 325.32 ν67 318.2 -7.1

20809.3 367.2 ν45 367.2 0.0

20801.6 384.7 ν44 377.9 -6.9

21050.1 443.3 ν46ν67ν69 445.6 2.3

21039.8 450.9 ν462ν68 450.9 0.0

21010.1 459.8 2ν46 460.5 0.7

21239.9 507.6 ν46ν67ν68 510.7 3.1

20683.7 545.1 2ν672ν68 546.0 0.9

20913.5 549.2 ν46ν65ν69 540.3 -8.9

21142.3 555.1 ν462ν67 555.6 0.4

20903.4 583.6 ν45ν67ν69 582.6 -1.0

20865.9 590.1 ν452ν68 587.9 -2.3

21094.4 596.0 ν45ν46 597.5 1.5

21233.5 601.1 ν41 607.3 6.2

21230.7 651.7 ν40 660.1 8.3

21184.7 681.5 2ν462ν68 680.5 -1.0

21464.3 691.7 3ν46 690.8 -1.0

21468.7 736.0 2ν46ν67ν68 737.8 1.8

21378.2 761.9 3ν672ν68ν69 761.3 -0.5

21227.2 771.9 ν38 768.9 -3.0

20743.1 774.1 ν462ν672ν68 776.2 2.1

21132.4 780.1 2ν46ν59ν65 779.5 -0.6

21130.3 783.9 2ν462ν67 785.8 1.9

21120.2 826.3 ν452ν46 827.7 1.4

20649.3 868.9 ν40ν67ν69 867.1 -1.8

20638.8 872.3 ν402ν68 872.4 0.1

20907.6 875.2 ν45ν46ν67ν68 874.8 -0.4

20941.9 881.5 ν40ν46 882.0 0.4

AVG DEV 2.3 second, wider band is also seen at 290.9 cm−1 and attributed to the combination of

−1 overtones 2ν682ν69, calculated to appear at 290.2 cm .

The next three unique bands in the 9H-An+ spectrum are also observed

0 here, the overtone of ν67, and the a modes ν45 and ν44. The intensity of the ν44 ring + rotation is much smaller in the 9D-An spectrum, and further spaced from the ν45 ring breathing mode. The following mode ν66ν67 is not assigned here, though could be one of the many smaller, unassigned peaks to the blue of the ν44 mode.

Though the ν65ν69 combination is not observed, it appears to combine with −1 −1 the ν46 mode with considerable intensity at 549.2 cm , predicted at 540.3 cm .

Furthermore, there is no observation of the ν42 mode here, rather the contraction −1 −1 and elongation of the central ring, ν41, is observed at 601.1 cm , 6.2 cm lower −1 than predicted. The ν40 mode persists, and is assigned to the band at 651.7 cm . 0 −1 The ﬁnal assigned a mode ν38, is calculated some 24.5 cm lower than the value for the 9H-An+ cation, at 768.9 cm−1, as it involves the in plane CHD wag. It is assigned to the peak at 771.9 cm−1.

The MAD for the 9D-An+ cation’s calculations is smaller than that of the 9H-An+ cation, at only 2.3 cm−1 per mode. This number is largely influenced by the better fit for the ν67ν69 and 2ν68 modes. Again, the remainder of the bands were assigned by fitting the peak positions to the combinations of previously assigned bands and do not influence the MAD. Beyond 900 cm−1, the spectrum becomes similarly complicated, and whilst assignment is possible, it would remain tenuous at best.

Planar or Puckered

In the previous study by Alata et al. two theoretical methodologies were employed to calculate the ground and excited state geometries of the 9H-An+ cation.[141] The results of these were contrasting, where (TD)-DFT methods yielded a planar structure, yet MP2 indicated a puckered structure. Our calculations utilised the (TD)- DFT method, and were consistent with a planar structure. There is no evidence

212 to discredit either the DFT or MP2 geometry calculations, and so these computational methods cannot be used alone to conclusively determine the planarity of the 9H-An+ molecule. Rather, we must rely on the evidence provided by the spectra.

1 In the paper by Alata et al., the ‘peak’ assigned to the 690 transition is within the noise level of their experiment. Indeed, to both the red and the blue of this ‘peak’, there are multiple other ‘peaks’ of similar and greater signal/noise (S/N) ratios which remain unassigned. As such, their assignment to the band at ∼100 cm−1 from the origin remains tentative at best. Alata’s assignment of the

1 690 transition is made with regard to the progression they observe in this mode, 2 3 (including 690 and 690). Our experiments have a greater resolution however, and 2 3 we have found that originally assigned 690 and 690 peaks are actually made up of 2 and 3 individual peaks respectively, none of which we assign to overtones of ν69.

The origin region of the 9D-An+ spectrum should also be considered in this argument. For for the neutral species, this out-of-plane inversion mode ν69 was observed for both the 9H-An and 9D-An species (albeit in multiple quanta). One

1 + would then expect this same 690 transition to be observed in the 9D-An experiment, which has a greater S/N ratio, however this region is absent of any peaks.

The assignment presented here, (using the vibrational frequencies generated from the TD-DFT output), is able to account for all transitions above the S/N level. Furthermore these assignments are consistent with the symmetry lowering from C2v to Cs upon excitation.

For the above reasons, we argue that the 9H-An+ and 9D-An+ cations are of a planar nature. We do however submit that the question of planarity must remain open until further experimental evidence can be provided to conﬁrm or contradict our interpretation.

213 9.4 Concluding Remarks

Beginning with well-deﬁned REMPI schemes to threshold ionise the 9-hydroanthracenyl and 9-deuteroanthracenyl radicals, we have been able to investigate the S1 ← S0 transitions of the 9H-An+ and 9D-An+ cations. Our isomer-selective techniques have been able to conﬁrm the previous work done by Alata et al., in assigning the origin transition observed at 491.59 nm (20342.1 cm−1) to the 9H-An+ cation, contrary to the assignment made by Garkusha et al.. The isotopologue, 9D-An+ was observed to have an origin at 20358.3 cm−1, at a displacement of +16.2 cm−1.

A reassignment of the 9H-An+ spectrum was made, based on TD-DFT calculations at the B3LYP/6-311+G(d,p) level of theory, which indicated a planar, Cs geometry. The choice of symmetry was justiﬁed by calculations done with the various functionals M06-2X, wB97XD and CAM-B3LYP, and the lack of single quantum

00 00 a modes observed or assigned. A single quanta of the a butterﬂy mode, ν69, assigned by Alata et al. to a band they observed at ∼100 cm−1 was not observed for either of the 9H-An+ or 9D-An+ cations in our experiments.

There were fewer differences between the 9H-An+ and 9D-An+ spectra c.f. the neutral radical analogues. The largest differences between the calculated frequencies of the two isotopologues occurred outside of the assigned range. For experiments without mass resolution, the best identifier between species is the CH and CD stretching frequencies. Unlike the neutral analogue, the intensity of the origin transitions of the cations were smaller than the vibrational structure, indicating a more pronounced change in geometry upon electronic excitation.

This technique used to examine the excited state of the cations requires that the mass changes upon excitation, and can hence only be applied to cations that photo-dissociate, unlike hole burning techniques which can also be used where there is a redistribution of electronic/vibrational levels after excitation. Furthermore, to investigate the other isomers 1H-An+ and 2H-An+, one would need to ﬁrst ﬁnd respective REMPI schemes and determine the ionisation potentials of the neutral radicals to be able to generate the cold vibrationless cations.

214 As discussed by both Alata and Garkusha, the 9H-An+ cation does not line up with any diﬀuse interstellar bands. In comparison with recent studies by Hobbs, neither of the 9H-An+ or the 9D-An+ cation electronic transitions correspond with the observed DIBs. [227, 228] The 16 cm−1 shift by deuteration brings the line closer to the peak observed at 4921 Å, and it is recommended that deuteration should be considered for any protonated species which correctly identiﬁes as a DIB spectral carrier. We suggest looking towards larger protonated and deuteronated

+ PAH’s and more stable fragments of the C60 fullerene, such as tetracene, pyrene and pentacene. Larger systems will no doubt pose experimental challenges to obtain high concentrations in molecular beams due to their decreasing vapour pressures.

215 9.5 Appendix to Chapter 9

A complete list of the vibrational frequencies for the D1 electronic state of the 9H- An+ and 9D-An+ cations:

+ Table 9.4: Vibrational frequencies for the D1 electronic state of the 9H-An

9H-An+

Mode # Frequency Symmetry Mode # Frequency Symmetry Mode # Frequency Symmetry

69 42.4 a00 46 231.0 a0 23 1324.0 a0 68 104.3 a00 45 368.1 a0 22 1343.0 a0 67 168.7 a00 44 379.9 a0 21 1377.6 a0 66 224.4 a00 43 512.4 a0 20 1402.8 a0 65 272.7 a00 42 598.2 a0 19 1419.2 a0 64 359.1 a00 41 612.6 a0 18 1448.0 a0 63 421.8 a00 40 662.9 a0 17 1458.9 a0 62 441.6 a00 39 736.1 a0 16 1504.0 a0 61 505.3 a00 38 800.3 a0 15 1534.6 a0 60 549.1 a00 37 884.3 a0 14 1551.6 a0 59 699.0 a00 36 904.2 a0 13 1566.3 a0 58 716.2 a00 35 1030.3 a0 12 1607.7 a0 57 754.1 a00 34 1050.0 a0 11 1667.4 a0 56 790.7 a00 33 1090.1 a0 10 2965.7 a0 55 872.5 a00 32 1141.7 a0 9 3170.3 a0 54 894.8 a00 31 1167.6 a0 8 3182.1 a0 53 920.9 a00 30 1186.4 a0 7 3183.2 a0 52 973.9 a00 29 1191.2 a0 6 3186.0 a0 51 983.7 a00 28 1198.3 a0 5 3195.4 a0 50 993.0 a00 27 1212.7 a0 4 3203.8 a0 49 1009.7 a00 26 1215.5 a0 3 3204.6 a0 48 1200.8 a00 25 1279.8 a0 2 3206.8 a0 47 2973.6 a00 24 1307.7 a0 1 3218.1 a0

216 + Table 9.5: Vibrational frequencies for the D1 electronic state of the 9H-An

9D-An+

Mode # Frequency Symmetry Mode # Frequency Symmetry Mode # Frequency Symmetry

69 41.3 a00 46 230.5 a0 23 1327.5 a0 68 104.1 a00 45 368.0 a0 22 1341.7 a0 67 159.4 a00 44 378.6 a0 21 1364.3 a0 66 222.0 a00 43 506.2 a0 20 1246.3 a0 65 269.7 a00 42 596.0 a0 19 1410.3 a0 64 355.4 a00 41 608.6 a0 18 1444.6 a0 63 418.6 a00 40 661.4 a0 17 1452.7 a0 62 440.9 a00 39 731.3 a0 16 1502.9 a0 61 500.6 a00 38 770.5 a0 15 1531.9 a0 60 530.0 a00 37 879.2 a0 14 1551.4 a0 59 685.5 a00 36 896.9 a0 13 1565.8 a0 58 712.4 a00 35 1030.2 a0 12 1607.5 a0 57 747.8 a00 34 1051.4 a0 11 1666.7 a0 56 790.7 a00 33 1093.1 a0 10 2178.3 a0 55 846.9 a00 32 1142.2 a0 9 3170.3 a0 54 889.7 a00 31 1171.2 a0 8 3182.1 a0 53 907.8 a00 30 1187.2 a0 7 3183.2 a0 52 979.6 a00 29 1190.6 a0 6 3185.9 a0 51 985.2 a00 28 1197.2 a0 5 3195.4 a0 50 993.1 a00 27 1210.8 a0 4 3203.8 a0 49 1009.7 a00 26 1214.5 a0 3 3204.6 a0 48 950.6 a00 25 1282.9 a0 2 3206.8 a0 47 2969.2 a00 24 1307.3 a0 1 3218.1 a0 Chapter 10

Epilogue

10.1 Conclusions of Part II

The work conducted in this part of the thesis forms a contribution to the ongoing effort toward elucidating carriers of the diffuse interstellar bands. The determination of the spectral carriers of these bands extends our fundamental knowledge of the composition of the universe and provides further insight into astronomical events. Polycyclic aromatic hydrocarbons are one class of molecular species that have been postulated to account for the DIB’s, in addition to the unidentified infrared bands, as their stability, composition and mechanisms of formation can be accounted for in the interstellar medium.[242] Indeed, they are recommended to account for as much as 20-30% of the total interstellar carbon balance.[221]

The harsh radiative environment of the ISM lends credence to the formation of resonantly stabilised radicals, and ions, from the photolytic degradation of PAHs, and by reaction with abundant hydrogen atoms and protons. RSRs and closed shell cations have been demonstrated to absorb in the visible regions of the electromagnetic spectrum, making them ideal candidates for the DIBs. Almost a century after the DIBs were ﬁrst observed by Mary Lea Hager, Campbell et al.

+ made the ﬁrst positive identiﬁcation of a DIB carrier, the cationic fullerene C60.[96] This work provided direct evidence that ionised carbonaceous polycyclic aromatics

218 were indeed present within the ISM, and as such one can postulate that molecules with a similar moiety would also be present within the ISM, likely passivated with universally abundant hydrogen.

In this work, the spectroscopy of four possible DIB candidates was examined. We report the ﬁrst observation of the D1 ← D0 electronic transitions of the 9-hydroanthracenyl and 9-deuteroanthracenyl resonantly stabilised radicals, formed from the association of hydrogen and deuterium atoms to anthracene within an electrical discharge. These radicals were ionised within our apparatus, yielding the isomer speciﬁc 9-hydroanthracyllium and 9-deuteroanthracyllium cations, analogous to the ionic products formed from the addition of a proton or deuteron to the 9- site of anthracene in the ISM. This work represents an extension to recent studies conducted within our group on the hydrogen and proton addition to naphthalene, (including the deuterated isotopologues). [140, 264]

The hydrogen and deuterium addition to anthracene was explored in section 8.3. Only one isomer was observed in each case, associated with addition at the

9- position. The origins of the D1 ← D0 electronic transition of 9H-An and 9D-An radicals were observed at 19115 cm−1 and 19112 cm−1 respectively. The ground

2 state for both species are of CS symmetry, the inversion plane containing the sp hybridised carbon and the H2 or HD substituents. The ground state geometries of both molecules are out-of-plane in a ν37 inversion coordinate, representing a but- terﬂy motion of the peripheral wings. Excitation of both isotopologues proceeds through this coordinate, resulting in a less planar D1 state.

Vibrations and overtones in this coordinate dominate the origin region for both isotopologues. Additional features were observed in this region for the 9D- An species. It was determined that the position of the deuterium atom relative to the anthracene framework shifted, (between an equatorial and axial position), as the molecule was proceeded along this coordinate. The energy of these two configurations is degenerate for the 9H-An species; however the position of the deuterium atom (eq. or ax.) influences the zero-point energy of the molecule in each configuration, splitting this degeneracy. As a result, electronic excitation was observed from both the axial and equatorial geometries. This allowed both odd and

219 even quanta of the ν37 coordinate to couple to electronic and vibronic transitions of the 9D-An species.

In all, we have measured D1 ← D0 electonic transitions of both the 9H- An and 9D-An species, and have observed a zero-point energy induced splitting of geometric degeneracy, leading to additional vibronic structure, an important spectroscopic result. This effect is expected to be observed for other chemical species and their isotopologues, where the electronic excitation proceeds through an inversion coordinate. The lifetime of the 9H-An D1 state indicates that fluorescence spectroscopy experiments may be conducted to elucidate the ground state structure and vibrational profile. The rich vibrational structure observed for these species exon- erate them as potential DIB candidates; in addition these electronic band origins do not line up with any of the DIBs. Radicals and ions which do not significantly distort upon electronic excitation would have a reduced vibrational profile, (in the limit where only the origin transition is observed), due to decreased Franck-Condon overlap with higher vibrational levels.

Following the determination of the ionisation potentials for 9H-An and 9D-An, their respective cations, 9H-An+ and 9D-An+, were created by threshold ionisation of the neutral species. Triple resonance depletion techniques were employed to uncover the vibronic structure of the respective D1 electronic states. The cations were determined to be planar in their ground and excited state. Upon excitation, the peripheral rings changed in size compared to one another, indicating the likelihood of a charge transfer state. This phenomenon should be investigated in the future, especially considering Alata et al. conclude that only molecules with even numbers of rings exhibit this charge transfer.[141] The electronic origins of the S1 + + ← S0 transitions for the 9H-An and 9D-An cations are determined to be 20342.1 cm−1 and 20358.3 cm−1 respectively.

The cation spectra contained rich vibrational structure, which became in- creasingly complicated towards higher energy. As a result, to assign combination and overtone bands in this region, a ﬁtting algorithm was employed based on the positions of assigned modes in the low energy region of the spectra.

220 The resolution aﬀorded by triple resonance spectroscopy in these experiments is far superior to the setup utilised by Alata et al. and Garkusha et al. As a direct result, additional features were observed in the spectra which assisted in a reassignment of the vibrational structure of the D1 state, and the conﬁrma- tion of a planar excited state, contrary to previous studies recommending a bent geometry.[141] Triple resonance spectroscopy requires prompt dissociation of the cation upon photo-excitation, and so is not necessarily a ubiquitous technique.

Our threshold-ionisation technique, ensuring cold preparation of the cation, allowed us to confirm the origin, as no hot bands were present in our spectra. Furthermore, our isomer selective approach of ionising a specific neutral radical (e.g. 9H-An) ensures the corresponding cation (e.g. 9H-An+) is formed, removing any isomeric ambiguity. These techniques have proven their efficacy in generating isomer specific cold cations, with improved resolution allowing for direct comparison with interstellar features.

The 9H-An+ and 9D-An+ cations cannot be attributed to DIBs, on account of their complex vibronic structure and mismatched origin position. However, their production and spectroscopic identiﬁcation adds credence to the experimental techniques used here, and should be employed in future isomer speciﬁc studies of other possible cationic candidates for the DIBs. We suggest looking towards larger protonated and deuteronated PAH’s and more stable fragments of the C60+ fullerene, such as tetracene, pyrene and pentacene.

221 References

[1] C. Parmesan and G. Yohe. A globally coherent ﬁngerprint of climate change impacts across natural systems. Nature, 421:37–42, 2003.

[2] R. J. Charlson, S. E. Schwartz, J. M. Hales, R. D. Cess, J. A. Coakley, J. E. Hansen, and D. J. Hofmann. Climate forcing by anthropogenic aerosols. Sci- ence, 255(5043):423–430, 1992.

[3] P. J. Crutzen and J. Lelieveld. Human impacts on atmospheric chemistry. Annual Review of Earth and Planetary Sciences, 29(1):17–45, 2001.

[4] M. E. Jenkin, J. C. Young, and A. R. Rickard. The mcm v3.3.1 degradation scheme for isoprene. Atmospheric Chemistry and Physics, 15(20):11433–11459, 2015. The chemical mechanistic information was taken from the Master Chemical Mechanism, MCM v3.2 (reference), via website: http://mcm.leeds.ac.uk/MCM.

[5] S. H. Chung and J. H. Seinfeld. Global distribution and climate forcing of carbonaceous aerosols. Journal of Geophysical Research: Atmospheres, 107(D19):AAC 14–1–AAC 14–33, 2002.

[6] J. Haywood and O. Boucher. Estimates of the direct and indirect radiative forcing due to tropospheric aerosols: A review. Reviews of Geophysics, 38(4):513–543, 2000.

[7] M. O. Andreae and P. J. Crutzen. Atmospheric aerosols: Biogeochemical sources and role in atmospheric chemistry. Science, 276(5315):1052–1058, 1997.

[8] M. Hori. Role of organic aerosols in the activation of cloud condensation nuclei. Earozoru Kenkyu, 18(4):257–265, 2003.

[9] C. M. Somers, B. E. McCarry, F. Malek, and J. S. Quinn. Reduction of particulate air pollution lowers the risk of heritable mutations in mice. Science, 304(5673):1008–1010, 2004.

[10] C. A. Pope III and D. W. Dockery. Health eﬀects of ﬁne particulate air pollution: Lines that connect. Journal of the Air & Waste Management Association, 56(6):709–742, 2006.

[11] C. I. Davidson, R. F. Phalen, and P. A. Solomon. Airborne particulate matter and human health: A review. Aerosol Science and Technology, 39(8):737–749, 2005.

222 [12] S. Fuzzi, M. O. Andreae, B. J. Huebert, M. Kulmala, T. C. Bond, M. Boy, S. J. Doherty, A. Guenther, M. Kanakidou, K. Kawamura, V.-M. Kerminen, U. Lohmann, L. M. Russell, and U. Pöschl. Critical assessment of the current state of scientiﬁc knowledge, terminology, and research needs concerning the role of organic aerosols in the atmosphere, climate, and global change. Atmospheric Chemistry and Physics, 6(7):2017–2038, 2006.

[13] U. Pöschl. Atmospheric aerosols: Composition, transformation, climate and health eﬀects. Angewandte Chemie International Edition, 44(46):7520–7540, 2005.

[14] H. Matsui, M. Koike, N. Takegawa, Y. Kondo, R. J. Griﬃn, Y. Miyazaki, Y. Yokouchi, and T. Ohara. Secondary organic aerosol formation in urban air: Temporal variations and possible contributions from unidentiﬁed hydrocarbons. Journal of Geophysical Research: Atmospheres, 114(D04201), 2009.

[15] B. J. Turpin, P. Saxena, and E. Andrews. Measuring and simulating particulate organics in the atmosphere: problems and prospects. Atmospheric Environment, 34(18):2983 – 3013, 2000.

[16] A. H. Goldstein and I. E. Galbally. Known and unexplored organic constituents in the earth’s atmosphere. Environmental Science & Technology, 41(5):1514– 1521, 2007.

[17] Q. Zhang, J. L. Jimenez, M. R. Canagaratna, J. D. Allan, H. Coe, I. Ul- brich, M. R. Alfarra, A. Takami, A. M. Middlebrook, Y. L. Sun, K. Dzepina, E. Dunlea, K. Docherty, P. F. DeCarlo, D. Salcedo, T. Onasch, J. T. Jayne, T. Miyoshi, A. Shimono, S. Hatakeyama, N. Takegawa, Y. Kondo, J. Schnei- der, F. Drewnick, S. Borrmann, S. Weimer, K. Demerjian, P. Williams, K. Bower, R. Bahreini, L. Cottrell, R. J. Griﬃn, J. Rautiainen, J. Y. Sun, Y. M. Zhang, and D. R. Worsnop. Ubiquity and dominance of oxygenated species in organic aerosols in anthropogenically-inﬂuenced northern hemi- sphere midlatitudes. Geophysical Research Letters, 34(13), 2007.

[18] M. Kanakidou, J. H. Seinfeld, S. N. Pandis, I. Barnes, F. J. Dentener, M. C. Facchini, R. Van Dingenen, B. Ervens, A. Nenes, C. J. Nielsen, E. Swietlicki, J. P. Putaud, Y. Balkanski, S. Fuzzi, J. Horth, G. K. Moortgat, R. Winterhal- ter, C. E. L. Myhre, K. Tsigaridis, E. Vignati, E. G. Stephanou, and J. Wilson. Organic aerosol and global climate modelling: a review. Atmospheric Chem- istry and Physics, 5(4):1053–1123, 2005.

[19] K. S. Docherty, E. A. Stone, I. M. Ulbrich, P. F. DeCarlo, D. C. Snyder, J. J. Schauer, R. E. Peltier, R. J. Weber, S. M. Murphy, J. H. Seinfeld, B. D. Grover, D. J. Eatough, and J. L. Jimenez. Apportionment of primary and secondary organic aerosols in southern california during the 2005 study of organic aerosols in riverside (soar-1). Environmental Science & Technology, 42(20):7655–7662, 2008.

[20] C. Tomasi and A. Lupi. Primary and Secondary Sources of Atmospheric Aerosol, chapter 1, pages 1–86. Wiley-Blackwell, 2016.

223 [21] J. H. Kroll, N. L. Ng, S. M. Murphy, R. C. Flagan, and J. H. Seinfeld. Sec- ondary organic aerosol formation from isoprene photooxidation. Environmen- tal Science & Technology, 40(6):1869–1877, 2006.

[22] K. Tsigaridis and M. Kanakidou. Global modelling of secondary organic aerosol in the troposphere: a sensitivity analysis. Atmospheric Chemistry and Physics, 3(5):1849–1869, 2003.

[23] M. Hallquist, J. C. Wenger, U. Baltensperger, Y. Rudich, D. Simpson, M. Claeys, J. Dommen, N. M. Donahue, C. George, A. H. Goldstein, J. F. Hamilton, H. Herrmann, T. Hoﬀmann, Y. Iinuma, M. Jang, M. E. Jenkin, J. L. Jimenez, A. Kiendler-Scharr, W. Maenhaut, G. McFiggans, Th. F. Mentel, A. Monod, A. S. H. Prévôt, J. H. Seinfeld, J. D. Surratt, R. Szmigielski, and J. Wildt. The formation, properties and impact of secondary organic aerosol: current and emerging issues. Atmospheric Chemistry and Physics, 9(14):5155– 5236, 2009.

[24] K. Tsigaridis, N. Daskalakis, M. Kanakidou, P. J. Adams, P. Artaxo, R. Ba- hadur, Y. Balkanski, S. E. Bauer, N. Bellouin, A. Benedetti, T. Bergman, T. K. Berntsen, J. P. Beukes, H. Bian, K. S. Carslaw, M. Chin, G. Curci, T. Diehl, R. C. Easter, S. J. Ghan, S. L. Gong, A. Hodzic, C. R. Hoyle, T. Iversen, S. Jathar, J. L. Jimenez, J. W. Kaiser, A. Kirkevåg, D. Koch, H. Kokkola, Y. H. Lee, G. Lin, X. Liu, G. Luo, X. Ma, G. W. Mann, N. Mi- halopoulos, J.-J. Morcrette, J.-F. Müller, G. Myhre, S. Myriokefalitakis, N. L. Ng, D. O’Donnell, J. E. Penner, L. Pozzoli, K. J. Pringle, L. M. Russell, M. Schulz, J. Sciare, Ø. Seland, D. T. Shindell, S. Sillman, R. B. Skeie, D. Spracklen, T. Stavrakou, S. D. Steenrod, T. Takemura, P. Tiitta, S. Tilmes, H. Tost, T. van Noije, P. G. van Zyl, K. von Salzen, F. Yu, Z. Wang, Z. Wang, R. A. Zaveri, H. Zhang, K. Zhang, Q. Zhang, and X. Zhang. The aerocom evaluation and intercomparison of organic aerosol in global models. Atmospheric Chemistry and Physics, 14(19):10845–10895, 2014.

[25] C. L. Heald, D. A. Ridley, S. M. Kreidenweis, and E. E. Drury. Satellite observations cap the atmospheric organic aerosol budget. Geophysical Research Letters, 37(24):L24808, 2010.

[26] A. G. Carlton, C. Wiedinmyer, and J. H. Kroll. A review of secondary organic aerosol (soa) formation from isoprene. Atmospheric Chemistry and Physics, 9(14):4987–5005, 2009.

[27] A. B. Guenther, X. Jiang, C. L. Heald, T. Sakulyanontvittaya, T. Duhl, L. K. Emmons, and X. Wang. The model of emissions of gases and aerosols from nature version 2.1 (megan2.1): an extended and updated framework for modeling biogenic emissions. Geoscientiﬁc Model Development, 5(6):1471–1492, 2012.

[28] O. Boucher, D. Randall, P. Artaxo, C. Bretherton, G. Feingold, P. Forster, V.-M. Kerminen, Y. Kondo, H. Liao, U. Lohmann, P. Rasch, S.K. Satheesh, S. Sherwood, B. Stevens, and X.Y. Zhang. Clouds and aerosols. In T.F. Stocker, D. Qin, G.-K. Plattner, M. Tignor, S.K. Allen, J. Boschung, A. Nauels, Y. Xia, V. Bex, and P.M. Midgley, editors, Climate Change 2013:

224 The Physical Science Basis. Contribution of Working Group I to the Fifth As- sessment Report of the Intergovernmental Panel on Climate Change, book section 7, page 571–658. Cambridge University Press, Cambridge, United King- dom and New York, NY, USA, 2013.

[29] P. J. Ziemann and R. Atkinson. Kinetics, products, and mechanisms of secondary organic aerosol formation. Chem. Soc. Rev., 41:6582–6605, 2012.

[30] J. G. Calvert, R. Atkinson, J. A. Kerr, S. Madronich, G. Moortgat, T. J. Wallington, and G. Yarwood. The Mechanisms of Atmospheric Oxi- dation of Alkenes. Oxford University Press, New York, 2000.

[31] R. Atkinson and J. Arey. Atmospheric degradation of volatile organic compounds. Chemical Reviews, 103(12):4605–4638, 2003.

[32] M. Lerdau, A. Guenther, and R. Monson. Plant production and emission of volatile organic compounds. BioScience, 47(6):373–383, 1997.

[33] R. Holzinger, E. Sanhueza, R. von Kuhlmann, B. Kleiss, L. Donoso, and P. J. Crutzen. Diurnal cycles and seasonal variation of isoprene and its oxidation products in the tropical savanna atmosphere. Global Biogeochemical Cycles, 16(4):1074, 2002.

[34] D. R. Gentner, R. A. Harley, A. M. Miller, and A. H. Goldstein. Diurnal and seasonal variability of gasoline-related volatile organic compound emissions in riverside, california. Environmental Science & Technology, 43(12):4247–4252, 2009.

[35] A. Borbon, J. B. Gilman, W. C. Kuster, N. Grand, S. Chevaillier, A. Colomb, C. Dolgorouky, V. Gros, M. Lopez, R. Sarda-Esteve, J. Holloway, J. Stutz, H. Petetin, S. McKeen, M. Beekmann, C. Warneke, D. D. Parrish, and J. A. Gouw. Emission ratios of anthropogenic volatile organic compounds in northern mid-latitude megacities: Observations versus emission invento- ries in los angeles and paris. Journal of Geophysical Research: Atmospheres, 118(4):2041–2057, 2013.

[36] W. B. DeMore, S. P. Sander, D. M. Golden, R. F. Hampson, M. J. Kurylo, C. J. Howard, A. R. Ravishankara, C. E. Kolb, and M. J. Molina. Chemical Kinetics and Photochemical Data for use in Stratospheric Modeling, Evaluation No. 12. NASA Panel for Data Evaluation, Jet Propulsion Laboratory Publication 97-4, Pasadena, CA, 1997.

[37] R. Atkinson. Atmospheric chemistry of vocs and nox. Atmospheric Environ- ment, 34(12):2063 – 2101, 2000.

[38] G. Lammel and J. N. Cape. Nitrous acid and nitrite in the atmosphere. Chem. Soc. Rev., 25:361–369, 1996.

[39] J. G. Calvert, R. Atkinson, K. H. Becker, R. M. Kamens, J. H. Seinfeld, T. H. Wallington, and G. Yarwood. The Mechanisms of Atmospheric Oxidation of the Aromatic Hydrocarbons. Oxford University Press, 2002.

225 [40] S. K. Hoekman. Speciated measurements and calculated reactivities of vehicle exhaust emissions from conventional and reformulated gasolines. Environmen- tal Science & Technology, 26(6):1206–1216, 1992. [41] R. Atkinson. Gas phase tropospheric chemistry of organic compounds. In R. E. Hester and R. M. Harrison, editors, Volatile Organic Compounds in the Atmosphere, volume 4, pages 65–90. The Royal Society of Chemistry, 1995. [42] R. Atkinson, J. Arey, and S. M. Aschmann. Atmospheric chemistry of alkanes: Review and recent developments. Atmospheric Environment, 42(23):5859 – 5871, 2008. Selected Papers from the First International Conference on At- mospheric Chemical Mechanisms. [43] R. Atkinson, D. L. Baulch, R. A. Cox, R. F. Hampson Jr., J. A. Kerr, M. J. Rossi, and J. Troe. Evaluated kinetic, photochemical and heterogeneous data for atmospheric chemistry: Supplement v. iupac subcommittee on gas kinetic data evaluation for atmospheric chemistry. Journal of Physical and Chemical Reference Data, 26(3):521–1011, 1997. [44] J. Eberhard, C. Muller, D. W. Stocker, and J. A. Kerr. Isomerization of alkoxy radicals under atmospheric conditions. Environmental Science & Technology, 29(1):232–241, 1995. [45] K. R. Darnall, W. P. L. Carter, A. M. Winer, A. C. Lloyd, and J. N. Pitts. Im- portance of ro2 + nitric oxide in alkyl nitrate formation from c4-c6 alkane pho- tooxidations under simulated atmospheric conditions. The Journal of Physical Chemistry, 80(17):1948–1950, 1976. [46] H. O. T. Pye and G. A. Pouliot. Modeling the role of alkanes, polycyclic aromatic hydrocarbons, and their oligomers in secondary organic aerosol formation. Environmental Science & Technology, 46(11):6041–6047, 2012. [47] B. Aumont, M. Camredon, C. Mouchel-Vallon, S. La, F. Ouzebidour, R. Val- orso, J. Lee-Taylor, and S. Madronich. Modeling the inﬂuence of alkane molecular structure on secondary organic aerosol formation. Faraday Discuss., 165:105–122, 2013. [48] D. S. Tkacik, A. A. Presto, N. M. Donahue, and A. L. Robinson. Secondary organic aerosol formation from intermediate-volatility organic compounds: Cyclic, linear, and branched alkanes. Environmental Science & Technology, 46(16):8773–8781, 2012. [49] Y. B. Lim and P. J. Ziemann. Chemistry of secondary organic aerosol formation from oh radical-initiated reactions of linear, branched, and cyclic alkanes in the presence of nox. Aerosol Science and Technology, 43(6):604–619, 2009. [50] J. F. Hunter, A. J. Carrasquillo, K. E. Daumit, and J. H. Kroll. Secondary organic aerosol formation from acyclic, monocyclic, and polycyclic alkanes. Environmental Science & Technology, 48(17):10227–10234, 2014. [51] Y. B. Lim and P. J. Ziemann. Products and mechanism of secondary organic aerosol formation from reactions of n-alkanes with oh radicals in the presence of nox. Environmental Science & Technology, 39(23):9229–9236, 2005.

226 [52] J. H. Kroll and J. H. Seinfeld. Chemistry of secondary organic aerosol: Forma- tion and evolution of low-volatility organics in the atmosphere. Atmospheric Environment, 42(16):3593 – 3624, 2008.

[53] B. J. Finlayson-Pitts and J. N. Pitts Jr. Chemistry of the Upper and Lower Atmosphere. Academic Press, San Diego, 2000.

[54] P. J. Ziemann. Eﬀects of molecular structure on the chemistry of aerosol formation from the oh-radical-initiated oxidation of alkanes and alkenes. In- ternational Reviews in Physical Chemistry, 30(2):161–195, 2011.

[55] E. C. Tuazon, S. M. Aschmann, J. Arey, and R. Atkinson. Products of the gas- phase reactions of a series of methyl-substituted ethenes with the oh radical. Environmental Science & Technology, 32(14):2106–2112, 1998.

[56] R. Koch, R. Knispel, M. Elend, M. Siese, and C. Zetzsch. Consecutive reactions of aromatic-oh adducts with no, no2 and o2: benzene, naphthalene, toluene, m- and p-xylene, hexamethylbenzene, phenol, m-cresol and aniline. Atmospheric Chemistry and Physics, 7(8):2057–2071, 2007.

[57] K. Zimmermann, R. Atkinson, and J. Arey. Eﬀect of no2 concentration on dimethylnitronaphthalene yields and isomer distribution patterns from the gas-phase oh radical-initiated reactions of selected dimethylnaphthalenes. En- vironmental Science & Technology, 46(14):7535–7542, 2012.

[58] H. J. L. Forstner, R. C. Flagan, and J. H. Seinfeld. Secondary organic aerosol from the photooxidation of aromatic hydrocarbons: molecular composition. Environmental Science & Technology, 31(5):1345–1358, 1997.

[59] R. Atkinson. Monograph no. 2. Journal of Physical and Chemical Reference Data, 2(1), 1994.

[60] J. R. Odum, T. P. W. Jungkamp, R. J. Griﬃn, R. C. Flagan, and J. H. Seinfeld. The atmospheric aerosol-forming potential of whole gasoline vapor. Science, 276(5309):96–99, 1997.

[61] R. Volkamer, P. J. Ziemann, and M. J. Molina. Secondary organic aerosol formation from acetylene (c2h2): seed eﬀect on soa yields due to organic photochemistry in the aerosol aqueous phase. Atmospheric Chemistry and Physics, 9(6):1907–1928, 2009.

[62] T. D. Sharkey and S. Yeh. Isoprene emission from plants. Annual Review of Plant Physiology and Plant Molecular Biology, 52(1):407–436, 2001.

[63] T. D. Sharkey, A. E. Wiberley, and A. R. Donohue. Isoprene emission from plants: Why and how. Annals of Botany, 101(1):5–18, 2008.

[64] M. Seemann, B. T. S. Bui, M. Wolﬀ, M. Miginiac-Maslow, and M. Rohmer. Isoprenoid biosynthesis in plant chloroplasts via the mep pathway: Direct thylakoid/ferredoxin-dependent photoreduction of gcpe/ispg. FEBS Letters, 580(6):1547 – 1552, 2006.

227 [65] M. Claeys, B. Graham, G. Vas, W. Wang, R. Vermeylen, V. Pashynska, J. Cafmeyer, P. Guyon, M. O. Andreae, P. Artaxo, and W. Maenhaut. Forma- tion of secondary organic aerosols through photooxidation of isoprene. Science, 303(5661):1173–1176, 2004.

[66] C. L. Heald, D. K. Henze, L. W. Horowitz, J. Feddema, J.-F. Lamarque, A. Guenther, P. G. Hess, F. Vitt, J. H. Seinfeld, A. H. Goldstein, and I. Fung. Predicted change in global secondary organic aerosol concentrations in response to future climate, emissions, and land use change. Journal of Geophys- ical Research: Atmospheres, 113:D05211, 2008.

[67] C. R. Hoyle, T. Berntsen, G. Myhre, and I. S. A. Isaksen. Secondary organic aerosol in the global aerosol - chemical transport model oslo ctm2. Atmospheric Chemistry and Physics, 7(21):5675–5694, 2007.

[68] D. K. Henze and J. H. Seinfeld. Global secondary organic aerosol from isoprene oxidation. Geophysical Research Letters, 33(9), 2006.

[69] P. O. Wennberg, K. H. Bates, J. D. Crounse, L. G. Dodson, R. C. McVay, L. A. Mertens, T. B. Nguyen, E. Praske, R. H. Schwantes, M. D. Smarte, J. M. St Clair, A. P. Teng, X. Zhang, and J. H. Seinfeld. Gas-phase reactions of isoprene and its major oxidation products. Chemical Reviews, Article ASAP, 2018. Article ASAP.

[70] R. Atkinson, D. L. Baulch, R. A. Cox, J. N. Crowley, R. F. Hampson, R. G. Hynes, M. E. Jenkin, M. J. Rossi, J. Troe, and IUPAC Subcommittee. Eval- uated kinetic and photochemical data for atmospheric chemistry: Volume ii - gas phase reactions of organic species. Atmospheric Chemistry and Physics, 6(11):3625–4055, 2006.

[71] J. Kesselmeier, U. Kuhn, S. Rottenberger, T. Biesenthal, A. Wolf, G. Schebeske, M. O. Andreae, P. Ciccioli, E. Brancaleoni, M. Frattoni, S. T. Oliva, M. L. Botelho, C. M. A. Silva, and T. M. Tavares. Concentrations and species composition of atmospheric volatile organic compounds (vocs) as observed during the wet and dry season in rondônia (amazonia). Journal of Geophysical Research: Atmospheres, 107(D20):LBA 20–1–LBA 20–13, 2002.

[72] F. Paulot, D. K. Henze, and P. O. Wennberg. Impact of the isoprene photochemical cascade on tropical ozone. Atmospheric Chemistry and Physics, 12(3):1307–1325, 2012.

[73] J. Jeon, J. R. Barker, and K. Song. Oh + isoprene: A direct dynamics study. Bulletin of the Korean Chemical Society, 38(6):651–660, 2017.

[74] E. E. Greenwald, S. W. North, Y. Georgievskii, and S. J. Klippenstein. A two transition state model for radical−molecule reactions: applications to isomeric branching in the oh−isoprene reaction. The Journal of Physical Chemistry A, 111(25):5582–5592, 2007.

[75] A. P. Teng, J. D. Crounse, and P. O. Wennberg. Isoprene peroxy radical dynamics. Journal of the American Chemical Society, 139(15):5367–5377, 2017.

228 [76] J. Peeters, T. L. Nguyen, and L. Vereecken. Hox radical regeneration in the oxidation of isoprene. Phys. Chem. Chem. Phys., 11:5935–5939, 2009.

[77] J. D. Crounse, F. Paulot, H. G. Kjaergaard, and P. O. Wennberg. Peroxy radical isomerization in the oxidation of isoprene. Phys. Chem. Chem. Phys., 13:13607–13613, 2011.

[78] L. Whalley, D. Stone, and D. Heard. New Insights into the Tropospheric Oxidation of Isoprene: Combining Field Measurements, Laboratory Studies, Chemical Modelling and Quantum Theory, pages 55–95. Springer Berlin Hei- delberg, Berlin, Heidelberg, 2014.

[79] J. Peeters, J.-F. Müller, T. Stavrakou, and V. S. Nguyen. Hydroxyl radical recycling in isoprene oxidation driven by hydrogen bonding and hydrogen tunneling: The upgraded lim1 mechanism. The Journal of Physical Chemistry A, 118(38):8625–8643, 2014.

[80] E. E. Greenwald, B. Ghosh, K. C. Anderson, K. S. Dooley, P. Zou, T. Selby, D. L. Osborn, G. Meloni, C. A. Taatjes, F. Goulay, and S. W. North. Isomer- selective study of the oh initiated oxidation of isoprene in the presence of o2 and no. i. the minor inner oh-addition channel. The Journal of Physical Chemistry A, 114(2):904–912, 2010.

[81] B. Ghosh, A. Bugarin, B. T. Connell, and S. W. North. Isomer-selective study of the oh-initiated oxidation of isoprene in the presence of o2 and no: 2. the major oh addition channel. The Journal of Physical Chemistry A, 114(7):2553– 2560, 2010. PMID: 20121059.

[82] I. V. Tokmakov and M. C. Lin. Kinetics and mechanism of the oh + c6h6 reaction: a detailed analysis with ﬁrst-principles calculations. The Journal of Physical Chemistry A, 106(46):11309–11326, 2002.

[83] M. Martín-Reviejo and K. Wirtz. Is benzene a precursor for secondary organic aerosol? Environmental Science & Technology, 39(4):1045–1054, 2005.

[84] N. L. Ng, J. H. Kroll, A. W. H. Chan, P. S. Chhabra, R. C. Flagan, and J. H. Seinfeld. Secondary organic aerosol formation from m-xylene, toluene, and benzene. Atmospheric Chemistry and Physics, 7(14):3909–3922, 2007.

[85] D. K. Henze, J. H. Seinfeld, N. L. Ng, J. H. Kroll, T.-M. Fu, D. J. Jacob, and C. L. Heald. Global modeling of secondary organic aerosol formation from aromatic hydrocarbons: high- vs. low-yield pathways. Atmospheric Chemistry and Physics, 8(9):2405–2420, 2008.

[86] E. Borrás and L. A. Tortajada-Genaro. Secondary organic aerosol formation from the photo-oxidation of benzene. Atmospheric Environment, 47:154 – 163, 2012.

[87] R. Volkamer, J. L. Jimenez, F. S. Martini, K. Dzepina, Q. Zhang, D. Salcedo, L. T. Molina, D. R. Worsnop, and M. J. Molina. Secondary organic aerosol formation from anthropogenic air pollution: Rapid and higher than expected. Geophysical Research Letters, 33(17):L17811, 2006.

229 [88] B. Koo, A. S. Ansari, and S. N. Pandis. Integrated approaches to modeling the organic and inorganic atmospheric aerosol components. Atmospheric Environment, 37(34):4757 – 4768, 2003.

[89] T. H. Lay, J. W. Bozzelli, and J. H. Seinfeld. Atmospheric photochemical oxidation of benzene: benzene + oh and the benzene−oh adduct (hydroxyl- 2,4-cyclohexadienyl) + o2. The Journal of Physical Chemistry, 100(16):6543– 6554, 1996.

[90] G. Ghigo and G. Tonachini. Benzene oxidation in the troposphere. theoretical investigation on the possible competition of three postulated reaction channels. Journal of the American Chemical Society, 120(27):6753–6757, 1998.

[91] C.-C. Chen, J. W. Bozzelli, and J. T. Farrell. Thermochemical properties, pathway, and kinetic analysis on the reactions of benzene with oh: an elemen- tary reaction mechanism. The Journal of Physical Chemistry A, 108(21):4632– 4652, 2004.

[92] S. Olivella, A. Solé, and J. M. Boﬁll. Theoretical mechanistic study of the oxidative degradation of benzene in the troposphere: Reaction of benzene−ho radical adduct with o2. Journal of Chemical Theory and Computation, 5(6):1607–1623, 2009. PMID: 26609853.

[93] T. A. Miller. Chemistry and chemical intermediates in supersonic free jet expansions. Science, 223(4636):545–553, 1984.

[94] A. Kovács and J. E. Rode. Modelling the matrix shift on the vibrational frequency of tho by dft-d3 calculations. The Journal of Chemical Physics, 146(12):124301, 2017.

+ [95] J. Fulara, M. Jakobi, and J. P. Maier. Electronic and infrared spectra of c60 − and c60 in neon and argon matrices. Chemical Physics Letters, 211(2):227 – 234, 1993.

[96] E. K. Campbell, M. Holz, D. Gerlich, and J. P. Maier. Laboratory conﬁrmation of c60+ as the carrier of two diﬀuse interstellar bands. Nature, 523:322, 2015.

[97] W. R. Gentry and C. F. Giese. Ten-microsecond pulsed molecular beam source and a fast ionization detector. Review of Scientiﬁc Instruments, 49(5):595–600, 1978.

[98] L. Castiglioni. Photodissociation and Excited States Dynamics of the Allyl Radical. PhD thesis, Swiss Federal Institute of Technology, ETH Zurich, 2007.

[99] D. H. Levy. Laser spectroscopy of cold gas-phase molecules. Annual Review of Physical Chemistry, 31(1):197–225, 1980.

[100] R. Schlachta, G. Lask, S. H. Tsay, and V. E. Bondybey. Pulsed discharge source of supersonically cooled transient species. Chemical Physics, 155(2):267 – 274, 1991.

230 [101] Y. Ohshima and Y. Endo. Structure of c3s studied by pulsed-discharge-nozzle fourier-transform microwave spectroscopy. Journal of Molecular Spectroscopy, 153(1):627 – 634, 1992.

[102] H. L. Snyder, B. T. Smith, T. P. Parr, and R. M. Martin. Dissociative excitation of water by metastable argon. Chemical Physics, 65(3):397 – 406, 1982.

[103] T. Mikoviny, J. D. Skalny, J. Orszagh, and N. J. Mason. The role of water and oxygen impurities on ozone production in a negative corona discharge of co2. Journal of Physics D: Applied Physics, 40(21):6646, 2007. [104] M. Zhao, Y. Cui, M. Samoc, P. N. Prasad, M. R. Unroe, and B. A. Reinhardt. Inﬂuence of two-photon absorption on third-order nonlinear optical processes as studied by degenerate four-wave mixing: The study of soluble didecyloxy substituted polyphenyls. The Journal of Chemical Physics, 95(6):3991–4001, 1991.

[105] W. C. Wiley and I. H. McLaren. Time-of-ﬂight mass spectrometer with improved resolution. Review of Scientiﬁc Instruments, 26(12):1150–1157, 1955.

[106] J. A. Syage and J. Wessel. Molecular multiphoton ionization and ion fragmentation spectroscopy. Applied Spectroscopy Reviews, 24(1-2):1–79, 1988.

[107] I. Schek and J. Jortner. Bloch equations and kinetic equations in multiphoton ionization. Chemical Physics, 97(1):1 – 11, 1985.

[108] S. H. Lin, Y. Fujimura, H. J. Neusser, and E. W. Schlag. {CHAPTER} 2 - theory of multiphoton absorption and ionization. In S. H. Lin, Y. Fujimura, H. J. Neusser, and E. W. Schlag, editors, Multiphoton Spectroscopy of Molecules, pages 7 – 69. Academic Press, 1984.

[109] O. Krechkivska, C. Wilcox, G. D. O’Connor, K. Nauta, S. H. Kable, and T. W. Schmidt. Ionization energies of three resonance-stabilized radicals: Cyclohexa- dienyl (dn, n = 0, 1, 6, 7), 1-phenylpropargyl, and methylcyclohexadienyl. The Journal of Physical Chemistry A, 118(44):10252–10258, 2014.

[110] E. Whittle, D. A. Dows, and G. C. Pimentel. Matrix isolation method for the experimental study of unstable species. The Journal of Chemical Physics, 22(11):1943–1943, 1954.

[111] I. Norman and G. Porter. Trapped atoms and radicals in a glass ‘cage’. Nature, 174:508–509, 1954.

[112] E. D. Becker and G. C. Pimentel. Spectroscopic studies of reactive molecules by the matrix isolation method. The Journal of Chemical Physics, 25(2):224– 228, 1956.

[113] T. Bally. Matrix Isolation, chapter 17, pages 795–845. Wiley-Blackwell, 2005.

[114] M. Rasanen and V. E. Bondybey. Gauche- and trans-n-butane and their ir- induced interconversion in solid neon. Chemical Physics Letters, 111(6):515 – 520, 1984.

231 [115] M. Pettersson, J. Lundell, L. Khriachtchev, and M. Räsänen. Ir spectrum of the other rotamer of formic acid, cis-hcooh. Journal of the American Chemical Society, 119(48):11715–11716, 1997.

[116] V. E. Bondybey, M. Rasanen, and A. Lammers. Chapter 10. rare-gas matrices, their photochemistry and dynamics: recent advances in selected areas. Annu. Rep. Prog. Chem., Sect. C: Phys. Chem., 95:331–372, 1999.

[117] Jan Fulara. Matrix and gas - phase spectroscopic studies of possible dib carriers. In V. Pirronello, J. Krelowski, and G. Manicò, editors, Solid State Astrochemistry, pages 175–210, Dordrecht, 2003. Springer Netherlands.

[118] M. E. Jacox. Comparison of the ground state vibrational fundamentals of diatomic molecules in the gas phase and in inert solid matrices. Journal of Molecular Spectroscopy, 113(2):286 – 301, 1985.

[119] J. V. Kranendonk. Solid Hydrogen: Theory of the Properties of Solid H2, HD and D2. Plenum Press, New York, 1983. [120] K. Sundararajan, K. Sankaran, N. Ramanathan, and R. Gopi. Production and characterization of para-hydrogen gas for matrix isolation infrared spectroscopy. Journal of Molecular Structure, 1117:181 – 191, 2016.

[121] M. E. Fajardo. Physics and chemistry at low temperatures. Pan Stanford Publishing, Singapore, 2011.

[122] B. A. Tom, S. Bhasker, Y. Miyamoto, T. Momose, and B. J. McCall. Pro- ducing and quantifying enriched para-h2. Review of Scientiﬁc Instruments, 80(1):016108, 2009.

[123] T. P. Troy. The jet-cooled spectroscopy of resonance-stabilized radicals. PhD thesis, The University of Sydney, 8 2011.

[124] M. Bahou, P. Das, Y.-F. Lee, Y.-J. Wu, and Y.-P. Lee. Infrared spectra of free radicals and protonated species produced in para-hydrogen matrices. Phys. Chem. Chem. Phys., 16:2200–2210, 2014.

[125] L. Wang, R. Wu, and C. Xu. Atmospheric oxidation mechanism of benzene. fates of alkoxy radical intermediates and revised mechanism. The Journal of Physical Chemistry A, 117(51):14163–14168, 2013.

[126] R. Volkamer, B. Klotz, I. Barnes, T. Imamura, K. Wirtz, N. Washida, K. H. Becker, and U. Platt. Oh-initiated oxidation of benzene part i. phenol formation under atmospheric conditions. Phys. Chem. Chem. Phys., 4:1598–1610, 2002.

[127] B. Klotz, R. Volkamer, M. D. Hurley, M. P. Sulbaek Andersen, O. J. Nielsen, I. Barnes, T. Imamura, K. Wirtz, K.-H. Becker, U. Platt, T. J. Wallington, and N. Washida. Oh-initiated oxidation of benzene part ii. inﬂuence of elevated no concentrations. Phys. Chem. Chem. Phys., 4:4399–4411, 2002.

232 [128] T. Berndt and O. Boge. Formation of phenol and carbonyls from the atmospheric reaction of oh radicals with benzene. Phys. Chem. Chem. Phys., 8:1205–1214, 2006.

[129] D. S. Hollman, A. C. Simmonett, and H. F. Schaefer. The benzene+oh potential energy surface: intermediates and transition states. Phys. Chem. Chem. Phys., 13:2214–2221, 2011.

[130] D. R. Glowacki, L. Wang, and M. J. Pilling. Evidence of formation of bicyclic species in the early stages of atmospheric benzene oxidation. The Journal of Physical Chemistry A, 113(18):5385–5396, 2009.

[131] T. Seta, M. Nakajima, and A. Miyoshi. High-temperature reactions of oh radicals with benzene and toluene. The Journal of Physical Chemistry A, 110(15):5081–5090, 2006.

[132] A. Mardyukov, R. Crespo-Otero, E. Sanchez-Garcia, and W. Sander. Photo- chemistry and reactivity of the phenyl radical–water system: A matrix isolation and computational study. Chemistry – A European Journal, 16(29):8679– 8689, 2010.

[133] O. Krechkivska, C. M. Wilcox, T. P. Troy, K. Nauta, B. Chan, R. Jacob, S. A. Reid, L. Radom, T. W. Schmidt, and S. H. Kable. Hydrogen-atom attack on phenol and toluene is ortho-directed. Phys. Chem. Chem. Phys., 18:8625–8636, 2016.

[134] T. P. Troy, M. Nakajima, N. Chalyavi, K. Nauta, S. H. Kable, and T. W. Schmidt. Hydroxyl addition to aromatic alkenes: Resonance-stabilized radical intermediates. The Journal of Physical Chemistry A, 116(30):7906–7915, 2012.

[135] M. J. Frisch, G. W. Trucks, H. B. Schlegel, G. E. Scuseria, M. A. Robb, J. R. Cheeseman, G. Scalmani, V. Barone, G. A. Petersson, H. Nakatsuji, X. Li, M. Caricato, A. V. Marenich, J. Bloino, B. G. Janesko, R. Gom- perts, B. Mennucci, H. P. Hratchian, J. V. Ortiz, A. F. Izmaylov, J. L. Sonnenberg, D. Williams-Young, F. Ding, F. Lipparini, F. Egidi, J. Go- ings, B. Peng, A. Petrone, T. Henderson, D. Ranasinghe, V. G. Zakrzewski, J. Gao, N. Rega, G. Zheng, W. Liang, M. Hada, M. Ehara, K. Toyota, R. Fukuda, J. Hasegawa, M. Ishida, T. Nakajima, Y. Honda, O. Kitao, H. Nakai, T. Vreven, K. Throssell, J. A. Montgomery, Jr., J. E. Peralta, F. Ogliaro, M. J. Bearpark, J. J. Heyd, E. N. Brothers, K. N. Kudin, V. N. Staroverov, T. A. Keith, R. Kobayashi, J. Normand, K. Raghavachari, A. P. Rendell, J. C. Burant, S. S. Iyengar, J. Tomasi, M. Cossi, J. M. Millam, M. Klene, C. Adamo, R. Cammi, J. W. Ochterski, R. L. Martin, K. Mo- rokuma, O. Farkas, J. B. Foresman, and D. J. Fox. Gaussian˜16 Revision A.03, 2016. Gaussian Inc., 340 Quinnipiac Street, Building 40, Wallingford CT 06492.

[136] J. P. Merrick, D. Moran, and L. Radom. An evaluation of harmonic vibrational frequency scale factors. The Journal of Physical Chemistry A, 111(45):11683– 11700, 2007.

233 [137] I. Pugliesi and K. Müller-Dethlefs. The use of multidimensional franck−condon simulations to assess model chemistries: a case study on phenol. The Journal of Physical Chemistry A, 110(14):4657–4667, 2006. a free download of the software can be found at http://www.fclab2.net.

[138] N. J. Reilly, M. Nakajima, B. A. Gibson, T. W. Schmidt, and S. H. Kable. Laser-induced ﬂuorescence and dispersed ﬂuorescence spectroscopy of jet-cooled 1-phenylpropargyl radical. The Journal of Chemical Physics, 130(14):144313, 2009.

[139] D. J. Kemp, W. D. Tuttle, F. M. S. Jones, A. M. Gardner, A. Andrejeva, J. C. A. Wakeﬁeld, and T. G. Wright. Consistent assignment of the vibrations of symmetric and asymmetric meta-disubstituted benzenes. Journal of Molecular Spectroscopy, 2018.

[140] O. Krechkivska, Y. Liu, K. L. K. Lee, K. Nauta, S. H. Kable, and T. W. Schmidt. Triple-resonance spectroscopy reveals the excitation spectrum of very cold, isomer-speciﬁc protonated naphthalene. The Journal of Physical Chemistry Letters, 4(21):3728–3732, 2013.

[141] I. Alata, C. Dedonder, M. Broquier, E. Marceca, and C. Jouvet. Role of the charge-transfer state in the electronic absorption of protonated hydrocarbon molecules. Journal of the American Chemical Society, 132(49):17483–17489, 2010.

[142] R. Atkinson. Gas-phase tropospheric chemistry of organic compounds: A review. Atmospheric Environment, 24:1–41, 1990.

[143] J. A. Miller, S. J. Klippenstein, Y. Georgievskii, L. B. Harding, W. D. Allen, and A. C. Simmonett. Reactions between resonance-stabilized radicals: Propargyl + allyl. The Journal of Physical Chemistry A, 114(14):4881–4890, 2010.

[144] H. R. Zhang, E. G. Eddings, and A. F. Saroﬁm. A journey from n-heptane to liquid transportation fuels. 1. the role of the allylic radical and its related species in aromatic precursor chemistry. Energy & Fuels, 22(2):945–953, 2008.

[145] K. M. Leung and R. P. Lindstedt. Detailed kinetic modeling of c1 — c3 alkane diﬀusion ﬂames. Combustion and Flame, 102(1):129 – 160, 1995.

[146] N. M. Marinov, M. J. Castaldi, C. F. Melius, and W. Tsang. Aromatic and polycyclic aromatic hydrocarbon formation in a premixed propane ﬂame. Combustion Science and Technology, 128(1-6):295–342, 1997.

[147] V. A. Krasnopolsky. Chemical composition of titan’s atmosphere and iono- sphere: Observations and the photochemical model. Icarus, 236:83 – 91, 2014.

[148] A. Webster. Large molecules, small radicals and the diﬀuse interstellar bands. Monthly Notices of the Royal Astronomical Society, 265(2):421–430, 1993.

[149] I. Fischer and P. Chen. Allyl-a model system for the chemical dynamics of radicals. The Journal of Physical Chemistry A, 106(17):4291–4300, 2002.

234 [150] E. Hirota, C. Yamada, and M. Okunishi. Infrared diode laser spectroscopy of the allyl radical. the ν11 band. The Journal of Chemical Physics, 97(5):2963– 2970, 1992.

2 [151] L. Castiglioni, A. Bach, and P. Chen. Spectroscopy and dynamics of a [ b1] allyl radical. Phys. Chem. Chem. Phys., 8:2591–2598, 2006. [152] C. L. Currie and D. A. Ramsay. Electronic absorption spectrum and dissociation energy of the allyl radical. The Journal of Chemical Physics, 45(2):488– 491, 1966. [153] G. Maier, H. P. Reisenauer, B. Rohde, and K. Dehnicke. Ir-, uv- und esr-spektrum des allylradikals in einer argon-matrix. Chemische Berichte, 116(2):732–740, 1983. [154] K. Tonokura and M. Koshi. Absorption spectrum and cross sections of the allyl radical measured using cavity ring-down spectroscopy: the a ← x band. The Journal of Physical Chemistry A, 104(37):8456–8461, 2000. [155] A. B. Callear and H. K. Lee. Electronic spectra of the free allyl radical and some of its simple derivatives. Trans. Faraday Soc., 64:308–316, 1968.

2 [156] J. W. Hudgens and C. S. Dulcey. Observation of the 3s a1 rydberg states of allyl and 2-methylallyl radicals with multiphoton ionization spectroscopy. The Journal of Physical Chemistry, 89(8):1505–1509, 1985. [157] J. A. Blush, D. W. Minsek, and P. Chen. Electronic spectrum of allyl and allyl- 2 2 2 2 2 2 d5 radicals: the b[1 a1] ← x[1 a2], c[2 b1] ← x[1 a2], and d[1 b2] ← x[1 a2] band systems. The Journal of Physical Chemistry, 96(25):10150–10154, 1992. [158] M. Gasser, J. Frey, J. M. Hostettler, A. Bach, and P. Chen. Vibronic structure of the 3s and 3p rydberg states of the allyl radical. The Journal of Physical Chemistry. A, 114:4704–11, 10 2010. [159] J.-C. Wu, R. Li, J.-L. Chang, and Y.-T. Chen. Rydberg states of the allyl radical observed by two-photon resonant ionization spectroscopy. The Journal of Chemical Physics, 113(17):7286–7291, 2000. [160] F A. Houle and J L. Beauchamp. Detection and investigation of allyl and ben- zyl radicals by photoelectron spectroscopy. Journal of The American Chemical Society, 100, 05 1978. [161] I. Fischer, T. Schü„ler, H.-J. Deyerl, M. Elhanine, and C. Alcaraz. Photoion- ization and dissociative photoionization of the allyl radical, c3h5. International Journal of Mass Spectrometry, 261(2):227 – 233, 2007. [162] J. M. Hostettler, L. Castiglioni, A. Bach, and P. Chen. Photochemical deactivation pathways of the a-state allyl radical. Phys. Chem. Chem. Phys., 11:8262–8265, 2009. [163] C. Chen, B. Braams, D. Y. Lee, J. M. Bowman, P. L. Houston, and D. Stranges. Evidence for vinylidene production in the photodissociation of the allyl radical. The Journal of Physical Chemistry Letters, 1(12):1875–1880, 2010.

235 [164] M. Shahu. Casscf study into the mechanism for predissociation of the allyl radical. International Journal of Quantum Chemistry, 106(2):501–506, 2006.

[165] C. Chen, B. Braams, D. Y. Lee, J. M. Bowman, P. L. Houston, and D. Stranges. The dynamics of allyl radical dissociation. The Journal of Phys- ical Chemistry A, 115(25):6797–6804, 2011.

[166] D. Stranges, P. O’Keeﬀe, G. Scotti, R. Di Santo, and P. L. Houston. Compet- ing sigmatropic shift rearrangements in excited allyl radicals. The Journal of Chemical Physics, 128(15):151101, 2008.

[167] Y. Song, M. Lucas, M. Alcaraz, J. Zhang, and C. Brazier. Ultraviolet pho- 2 2 todissociation dynamics of the allyl radical via the b a1(3s), c b2(3py), and 2 e b1(3px) electronic excited states. The Journal of Physical Chemistry A, 119(50):12318–12328, 2015.

[168] J. C. Schultz, F. A. Houle, and J. L. Beauchamp. Photoelectron spectroscopy of isomeric c4h7 radicals. implications for the thermochemistry and structures of the radicals and their corresponding carbonium ions. Journal of the Amer- ican Chemical Society, 106(24):7336–7347, 1984.

[169] C.-C. Chen, H.-C. Wu, C.-M. Tseng, Y.-H. Yang, and Y.-T. Chen. One- and two-photon excitation vibronic spectra of 2-methylallyl radical at 4.6–5.6 ev. The Journal of Chemical Physics, 119(1):241–250, 2003.

[170] M. Gasser, J. A. Frey, J. M. Hostettler, and A. Bach. Vibronic structure of the 3s rydberg state of the 2-methylallyl radical. Journal of Molecular Spectroscopy, 263(1):93 – 100, 2010.

[171]A.R o¨der, K. Issler, L. Poisson, A. Humeniuk, M. Wohlgemuth, M. Comte, F. Lepetit, I. Fischer, R. Mitric, and J. Petersen. Femtosecond dynamics of the 2-methylallyl radical: A computational and experimental study. The Journal of Chemical Physics, 147(1):013902, 2017.

[172] J. Herterich, T. Gerbich, and I. Fischer. Excited-state dynamics of the 2- methylallyl radical. ChemPhysChem, 14(17):3906–3908, 2013.

[173] M. Gasser, A. Bach, and P. Chen. Photodissociation dynamics of the 2- methylallyl radical. Phys. Chem. Chem. Phys., 10:1133–1138, 2008.

[174] M. Gasser, J. A. Frey, J. M. Hostettler, and A. Bach. Probing for non- statistical eﬀects in dissociation of the 1-methylallyl radical. Chem. Commun., 47:301–303, 2011.

[175] A. Nikitjuka, A. Nekrasova, and A. Jirgensons. Methylprenyl and prenyl protection for sulfonamides. Synlett, 26(02):183–186, 2014.

[176] J.-M. Vatèle. The prenyl group: a versatile hydroxy protecting group, remov- able chemoselectively under mild conditions. Tetrahedron, 58(28):5689 – 5698, 2002.

[177] F. L. Zhang and P. Casey. Protein prenylation: Molecular mechanisms and functional consequences. Annual review of biochemistry, 65:241–269, 02 1996.

236 [178] M. Wang and P. Casey. Protein prenylation: Unique fats make their mark on biology. Nature Reviews Molecular Cell Biology, 17:110–122, 2016.

[179] M. Nakajima, T. W. Schmidt, Y. Sumiyoshi, and Y. Endo. Rotationally- resolved excitation spectrum of the jet-cooled cyclohexadienyl radical. Chem- ical Physics Letters, 449:57–62, 11 2007.

[180] L. H. Spangler. Structural information from methyl internal rotation spectroscopy. Annual Review of Physical Chemistry, 48(1):481–510, 1997.

[181] J. R. Gascooke and W. D. Lawrance. Methyl rotor dependent vibrational interactions in toluene. The Journal of Chemical Physics, 138(13):134302, 2013.

[182] F. P. Lossing and J. C. Traeger. Free radicals by mass spectrometry xlvi. heats of formation of c5h7and c5h9 radicals and cations. International Journal of Mass Spectrometry and Ion Physics, 19(1):9 – 22, 1976.

[183] G. D. O’Connor, G. V. G. Woodhouse, T. P. Troy, and T. W. Schmidt. Double-resonance spectroscopy of radicals: higher electronic excited states of 1- and 2-naphthylmethyl, 1-phenylpropargyl and 9-anthracenylmethyl. Molec- ular Physics, 113(15-16):2138–2147, 2015.

[184] T. W. Schmidt. The electronic spectroscopy of resonance-stabilised hydrocarbon radicals. International Reviews in Physical Chemistry, 35(2):209–242, 2016.

[185] director S. E. Stein. Mass Spectra. NIST Chemistry WebBook, NIST Stan- dard Reference Database Number 69, National Institute of Standards and Technology Mass Spec Data Center, Gaithersburg MD, 20899, retrieved on- line: February 19, 2018.

[186] M. Lorenz and M. Kalesse. Synthesis of the c10−c32 core structure of spirangien a. Organic Letters, 10(19):4371–4374, 2008.

[187] W. L. Chameides, F. Fehsenfeld, M. O. Rodgers, C. Cardelino, J. Martinez, D. Parrish, W. Lonneman, D. R. Lawson, R. A. Rasmussen, P. Zimmerman, J. Greenberg, P. Mlddleton, and T. Wang. Ozone precursor relationships in the ambient atmosphere. Journal of Geophysical Research: Atmospheres, 97(D5):6037–6055, 1992.

[188] R. A. Rasmussen and M. A. K. Khalil. Isoprene over the amazon basin. Journal of Geophysical Research: Atmospheres, 93(D2):1417–1421, 1988.

[189] T. P. Troy, M. Nakajima, N. Chalyavi, R. G. C. R. Clady, K. Nauta, S. H. Kable, and T. W. Schmidt. Identiﬁcation of the jet-cooled 1-indanyl radical by electronic spectroscopy. The Journal of Physical Chemistry A, 113(38):10279– 10283, 2009.

[190] B. B. Snider and J. R. Duvall. Synthesis of the 4-methyl-1,2-oxazetidine-4- carboxylic acid moiety of the originally proposed halipeptin a and b structures. Tetrahedron Letters, 44(15):3067 – 3070, 2003.

237 [191] J. Kuk, B. S. Kim, H. Jung, S. Choi, J.-Y. Park, and S. Koo. General preparation and controlled cyclization of acyclic terpenoids. The Journal of Organic Chemistry, 73(5):1991–1994, 2008.

[192] L. Fraser-Monteiro, M. L. Fraser-Monteiro, J. J. Butler, and T. Baer. Ther- + mochemistry and dissociation dynamics of state-selected c4h8o2 ions. 3. ethyl acetate. The Journal of Physical Chemistry, 86(5):752–757, 1982.

[193] L. W. Sieck and M. Mautner. Ionization energies and entropies of cycloalkanes. kinetics of free energy controlled charge-transfer reactions. The Journal of Physical Chemistry, 86(18):3646–3650, 1982.

[194] Y.-F. Lee and Y.-P. Lee. Infrared absorption of iodomethylperoxy (syn-ich2oo) radical generated upon photolysis of ch2i2 and o2 in solid para-h2. Molecular Physics, 113(15-16):2148–2158, 2015.

[195] P.-W. Lee, P. G. Scrape, L. J. Butler, and Y.-P. Lee. Two hcl-elimination channels and two co-formation channels detected with time-resolved infrared emission upon photolysis of acryloyl chloride [ch2chc(o)cl] at 193 nm. The Journal of Physical Chemistry A, 119(28):7293–7304, 2015.

[196] C.-Y. Chou and Y.-P. Lee. Infrared absorption of 1-chloro-2-methyl-2-propyl [·c(ch3)2ch2cl] and 2-chloro-2-methylpropyl [·ch2c(ch3)2cl] radicals produced in the addition reactions of cl with isobutene (i-c4h8) in solid para-hydrogen. The Journal of Chemical Physics, 145(13):134302, 2016.

[197] J. C. Amicangelo and Y.-P. Lee. Infrared spectra of the 1-chloromethyl-1- methylallyl and 1-chloromethyl-2-methylallyl radicals isolated in solid para- hydrogen. The Journal of Physical Chemistry A, 121(46):8771–8784, 2017.

[198] M. J. Frisch, G. W. Trucks, H. B. Schlegel, G. E. Scuseria, M. A. Robb, J. R. Cheeseman, G. Scalmani, V. Barone, G. A. Petersson, H. Nakatsuji, X. Li, M. Caricato, A. Marenich, J. Bloino, B. G. Janesko, R. Gomperts, B. Mennucci, H. P. Hratchian, J. V. Ortiz, A. F. Izmaylov, J. L. Son- nenberg, D. Williams-Young, F. Ding, F. Lipparini, F. Egidi, J. Goings, B. Peng, A. Petrone, T. Henderson, D. Ranasinghe, V. G. Zakrzewski, J. Gao, N. Rega, G. Zheng, W. Liang, M. Hada, M. Ehara, K. Toyota, R. Fukuda, J. Hasegawa, M. Ishida, T. Nakajima, Y. Honda, O. Kitao, H. Nakai, T. Vreven, K. Throssell, J. A. Montgomery, Jr., J. E. Peralta, F. Ogliaro, M. Bearpark, J. J. Heyd, E. Brothers, K. N. Kudin, V. N. Staroverov, T. Keith, R. Kobayashi, J. Normand, K. Raghavachari, A. Rendell, J. C. Burant, S. S. Iyengar, J. Tomasi, M. Cossi, J. M. Millam, M. Klene, C. Adamo, R. Cammi, J. W. Ochterski, R. L. Martin, K. Morokuma, O. Farkas, J. B. Foresman, and D. J. Fox. Gaussian˜09 Revision E.01, 2016. Gaussian Inc., 340 Quinnipiac Street, Building 40, Wallingford CT 06492.

[199] R. D. Johnson III, editor. NIST Computational Chemistry Comparison and Benchmark Database, NIST Standard Reference Database Number 101. NIST, Gaithersburg, MD, 2016. http://cccbdb.nist.gov/.

238 [200] M. A. Goodman, R. L. Sweany, and R. L. Flurry. Infrared spectra of matrix- isolated, crystalline solid, and gas-phase c3-c6 n-alkanes. The Journal of Phys- ical Chemistry, 87, 05 1983.

[201] M. I. Redondo, M. V. Garcia, M. A. Raso, and W. A. P. Luck. Ir spectra of ethyl acetate and perﬂuorotertbutanol-ethyl acetate complex at diﬀerent temperatures. Journal of Molecular Structure, 218:213 – 218, 1990.

[202] T.-K. Ha, C. Pal, and P. N. Ghosh. Ethyl acetate: Gas phase infrared spectra, ab initio calculation of structure and vibrational frequencies and assignment. Spectrochimica Acta Part A: Molecular Spectroscopy, 48(8):1083 – 1090, 1992.

[203] J. D. Savee, O. Welz, C. A. Taatjes, and D. L. Osborn. New mechanistic insights to the o(3p) + propene reaction from multiplexed photoionization mass spectrometry. Phys. Chem. Chem. Phys., 14:10410–10423, 2012.

[204] R. I. Kaiser. Experimental investigation on the formation of carbon-bearing molecules in the interstellar medium via neutral−neutral reactions. Chemical Reviews, 102(5):1309–1358, 2002.

[205] T. L. Wilson and F. Matteucci. Abundances in the interstellar medium. The Astronomy and Astrophysics Review, 4(1):1–33, Mar 1992.

[206] S. Kwok, K. Volk, and B. J. Hrivnak. Chemical evolution of carbonaceous materials in the last stages of stellar evolution. Astron. Astrophys, 350:L35– L38, October 1999.

[207] S. Kwok. Organic Nanoparticles as a Component of the Interstellar Medium. In J. Cami and N. L. J. Cox, editors, The Diﬀuse Interstellar Bands, volume 297 of IAU Symposium, pages 213–215, February 2014.

[208] C. P. Endres, S. Schlemmer, P. Schilke, J. Stutzki, and H. S. P. Müller. The cologne database for molecular spectroscopy, cdms, in the virtual atomic and molecular data centre, vamdc. Journal of Molecular Spectroscopy, 327:95 – 104, 2016. New Visions of Spectroscopic Databases, Volume II.

[209] The cologne database for molecular spectroscopy - molecules in space. https: //www.astro.uni-koeln.de/cdms/molecules. Accessed: March 2018.

[210] P. Thaddeus and M. C. McCarthy. Carbon chains and rings in the laboratory and in space. Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy, 57(4):757 – 774, 2001.

[211] M. C. McCarthy, M. J. Travers, A. Kovács, W. Chen, S. E. Novick, C. A. Gottlieb, and P. Thaddeus. Detection and characterization of the cumulene carbenes h2c5 and h2c6. Science, 275(5299):518–520, 1997. [212] M. C. McCarthy, W. Chen, M. J. Travers, and P. Thaddeus. Microwave spectra of 11 polyyne carbon chains. The Astrophysical Journal Supplement Series, 129(2):611, 2000.

239 [213] M. C. McCarthy, M. J. Travers, P. Kalmus, C. A. Gottlieb, and P. Thaddeus. Laboratory detection of the c9h radical. The Astrophysical Journal Letters, 467(2):L125, 1996.

[214] T. W. Schmidt and R. G. Sharp. The optical spectroscopy of extraterrestrial molecules. Australian Journal of Chemistry, 58:69–81, 2005.

[215] J. Cami, J. Bernard-Salas, E. Peeters, and S. E. Malek. Detection of c60 and c70 in a young planetary nebula. Science, 329(5996):1180–1182, 2010. [216] M. Cohen, C. M. Anderson, A. Cowley, G. V. Coyne, W. Fawley, T. R. Gull, E. A. Harlan, G. H. Herbig, F. Holden, H. S. Hudson, R. O. Jakoubek, H. M. Johnson, K. M. Merrill, F. H. Schiﬀer, B. T. Soifer, and B. Zuckerman. The Peculiar Object HD 44179 (“The Red Rectangle”). Astrophysical Journal, 196:179–189, 1975.

[217] M. Cohen, H. Van Winckel, H. E. Bond, and T. R. Gull. Hubble space telescope imaging of hd 44179, the red rectangle. The Astronomical Journal, 127(4):2362, 2004.

[218] R. J. Glinski, P. D. Michaels, C. M. Anderson, T. W. Schmidt, R. G. Sharp, M. L. Sitko, L. S. Bernstein, and H. Van Winckel. Current assessment of the red rectangle band problem. Astrophysics and Space Science, 323(4):337–344, 2009.

[219] P. J. Sarre. The diﬀuse interstellar bands: A major problem in astronomical spectroscopy. Journal of Molecular Spectroscopy, 238(1):1 – 10, 2006.

[220] R. A. Watson, R. Rebolo, J. A. Rubiño-Martín, S. Hildebrandt, C. M. Gutiér- rez, S. Fernández-Cerezo, R. J. Hoyland, and E. S. Battistelli. Detection of anomalous microwave emission in the perseus molecular cloud with the cos- mosomas experiment. The Astrophysical Journal Letters, 624(2):L89, 2005.

[221] A. Leger and J. L. Puget. Identiﬁcation of the ‘unidentiﬁed’ IR emission features of interstellar dust. Astronomy and Astrophysics, 137:L5–L8, August 1984.

[222] D McElroy, C Walsh, A J. Markwick, M A. Cordiner, K Smith, and T J. Millar. The umist database for astrochemistry 2012. 550, 12 2012.

[223] P. W. Merrill. Unidentiﬁed interstellar lines. Publications of the Astronomical Society of the Paciﬁc, 46(272):206, 1934.

[224] P. W. Merrill. Tests of the lines λ 5780 and λ 6284 in stellar spectra. Publi- cations of the Astronomical Society of the Paciﬁc, 48:179–180, 1936.

[225] P. W. Merrill. Stationary lines in the spectrum of the binary star boss 6142. Astrophysical Journal, 83:126, 1936.

[226] G. H. Herbig. The diﬀuse interstellar bands. Annual Review of Astronomy and Astrophysics, 33(1):19–73, 1995.

240 [227] L. M. Hobbs, D. G. York, T. P. Snow, T. Oka, J. A. Thorburn, M. Bishof, S. D. Friedman, B. J. McCall, B. Rachford, P. Sonnentrucker, and D. E. Welty. A catalog of diffuse interstellar bands in the spectrum of hd 204827. The Astrophysical Journal, 680(2):1256, 2008. [228] L. M. Hobbs, D. G. York, J. A. Thorburn, T. P. Snow, M. Bishof, S. D. Fried- man, B. J. McCall, T. Oka, B. Rachford, P. Sonnentrucker, and D. E. Welty. Studies of the diffuse interstellar bands. iii. hd 183143. The Astrophysical Journal, 705(1):32, 2009. [229] B. J. McCall, M. M. Drosback, J.A. Thorburn, D. G. York, S. D. Friedman, L. M. Hobbs, B. L. Rachford, T. P. Snow, P. Sonnentrucker, and D. E. Welty. Studies of the diffuse interstellar bands. iv. the nearly perfect correlation between λ6196.0 and λ6613.6. Astrophysical Journal, 708(2):1628–1638, 2010. [230] T. P. Snow and A. N. Witt. The interstellar carbon budget and the role of carbon in dust and large molecules. Science, 270(5241):1455–1460, 1995. [231] T. P. Snow. The unidentified diffuse interstellar bands as evidence for large organic molecules in the interstellar medium. Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy, 57(4):615 – 626, 2001. [232] F. Salama, C. Joblin, and L. J. Allamandola. Absorption spectroscopy of neutral and ionized pahs. implications for the diffuse interstellar bands. In A. G. G. M. Tielens and T. P. Snow, editors, The Diffuse Interstellar Bands, volume 202 of Astrophysics and Space Science Library, page 207. Springer Netherlands, 1995. [233] R. W. Russell, B. T. Soifer, and S. P. Willner. The 4 to 8 micron spectrum of NGC 7027. Astrophysical Journal Letters, 217:L149–L153, November 1977. [234] W. W. Duley and D. A. Williams. The infrared spectrum of interstellar dust - Surface functional groups on carbon. Monthly Notices of the Royal Astro- nomical Society, 196:269–274, 1981. [235] L. J. Allamandola, A. G. G. M. Tielens, and J. R. Barker. Polycyclic aromatic hydrocarbons and the unidentified infrared emission bands - Auto exhaust along the Milky Way. Astrophysical Journal Letters, 290:L25–L28, 1985. [236] L. J. Allamandola, A. G. G. M. Tielens, and J. R. Barker. Interstellar polycyclic aromatic hydrocarbons - The infrared emission bands, the excitation/emission mechanism, and the astrophysical implications. Astrophysical Journal Supplement Series, 71:733–775, December 1989. [237] J. L. Puget and A. Léger. A new component of the interstellar matter: Small grains and large aromatic molecules. Annual Review of Astronomy and Astro- physics, 27(1):161–198, 1989. [238] S. Kwok and Y. Zhang. Mixed aromatic–aliphatic organic nanoparticles as carriers of unidentified infrared emission features. Nature, 479:80–83, 2011. [239] S. Kwok and Y. Zhang. Unidentified Infrared Emission Bands: PAHs or MAONs? The Astrophysical Journal, 771:5, July 2013.

241 [240] H. Kroto. Space, stars, c60, and soot. Science, 242(4882):1139–1145, 1988. [241] A. Webster. On the carriers of the diﬀuse interstellar bands. Monthly Notices of the Royal Astronomical Society, 263(2):385–393, 1993.

[242] T. R. Geballe. The diﬀuse interstellar bands - a brief review. Journal of Physics: Conference Series, 728(6):062005, 2016.

[243] D. L. Kokkin, N. J. Reilly, T. P. Troy, K. Nauta, and T. W. Schmidt. Gas phase spectra of all-benzenoid polycyclic aromatic hydrocarbons: Triphenylene. The Journal of Chemical Physics, 126(8):084304, 2007.

[244] N. Chalyavi, T. P. Troy, M. Nakajima, B. A. Gibson, K. Nauta, R. G. Sharp, S. H. Kable, and T. W. Schmidt. Excitation and emission spectra of jet-cooled naphthylmethyl radicals. The Journal of Physical Chemistry A, 115(27):7959– 7965, 2011.

[245] G. D. O’Connor, G. B. Bacskay, G. V. G. Woodhouse, T. P. Troy, K. Nauta, and T. W. Schmidt. Excitation spectra of large jet-cooled polycyclic aromatic hydrocarbon radicals: 9-anthracenylmethyl (c15h11) and 1-pyrenylmethyl (c17h11). The Journal of Physical Chemistry A, 117(50):13899–13907, 2013. [246] M. Hirama, T. Tokosumi, T. Ishida, and J.-I. Aihara. Possible molecular hydrogen formation mediated by the inner and outer carbon atoms of typical pah cations. Chemical Physics, 305(1):307 – 316, 2004.

[247] S. Casolo, G. F. Tantardini, and R. Martinazzo. Insights into h2 formation in space from ab initio molecular dynamics. Proceedings of the National Academy of Sciences of the United States of America, 110(17):6674–6677, 2013.

[248] L. Hornekær, A. Baurichter, V. V. Petrunin, D. Field, and A. C. Luntz. Importance of surface morphology in interstellar h2 formation. Science, 302(5652):1943–1946, 2003.

[249] V. Wakelam, E. Bron, S. Cazaux, F. Dulieu, C. Gry, P. Guillard, E. Habart, L. Hornekær, S. Morisset, G. Nyman, V. Pirronello, S. D. Price, V. Valdivia, G. Vidali, and N. Watanabe. H2 formation on interstellar dust grains: The viewpoints of theory, experiments, models and observations. Molecular Astro- physics, 9(Supplement C):1 – 36, 2017.

[250] M. Buragohain, A. Pathak, M. Hammonds, and P. J. Sarre. Theoretical quantum chemical study of protonated - deuteronated pahs: Interstellar implications. AIP Conference Proceedings, 1543(1):258–264, 2013.

[251] T. P. Snow, V. Le Page, Y. Keheyan, and V. M. Bierbaum. The interstellar chemistry of pah cations. Nature, 391:259–260, 1998.

[252] A. Ricks, G. E. Douberly, and M. A. Duncan. The infrared spectrum of protonated naphthalene and its relevance for the unidentiﬁed infrared bands. The Astrophysical Journal, 702:301–306, 2009.

242 [253] M. Bahou, Y.-J. Wu, and Y.-P. Lee. Infrared spectra of protonated coronene and its neutral counterpart in solid parahydrogen: Implications for uniden- tiﬁed interstellar infrared emission bands. Angewandte Chemie International Edition, 53(4):1021–1024, 2014.

[254] G. A. Olah, G. D. Mateescu, and Y. K. Mo. Stable carbocations. cxl. naph- thalenium ions. Journal of the American Chemical Society, 95(6):1865–1874, 1973.

[255] P. Weilmünster, A. Keller, and K.-H. Homann. Large molecules, radicals, ions, and small soot particles in fuel-rich hydrocarbon ﬂames: Part i: positive ions of polycyclic aromatic hydrocarbons(pah) in low-pressure premixed ﬂames of acetylene and oxygen. Combustion and Flame, 116(1):62 – 83, 1999.

[256] S. A. Sanford, M. P. Bernstein, and J. P. Dworkin. Assessment of the interstellar processes leading to deuterium enrichment in meteoritic organics. Meteoritics and Planetary Science, 36(8):1117–1133, 2001.

[257] T. J. Millar, A. Bennett, and E. Herbst. Deuterium fractionation in dense interstellar clouds. The Astrophysical Journal, 340:906–920, 1989.

[258] I. Garkusha, J. Fulara, A. Nagy, and J. P. Maier. Electronic Absorption Spectra of Protonated Anthracenes and Phenanthrenes, and Their Neutrals in Neon Matrices. The Astrophysical Journal, 728:131, 2011.

[259] J. A. Sebree, V. V. Kislov, A. M. Mebel, and T. S. Zwier. Spectroscopic and thermochemical consequences of site-speciﬁc h-atom addition to naphthalene. The Journal of Physical Chemistry A, 114(21):6255–6262, 2010.

[260] J. A. Sebree, V. V. Kislov, A. M. Mebel, and T. S. Zwier. Isomer speciﬁc spectroscopy of C10Hn, n = 8-12: Exploring pathways to naphthalene in Titan’s atmosphere. Faraday Discussions, 147:231, 2010.

[261] J. A. Miller and S. J. Klippenstein. The recombination of propargyl radicals: solving the master equation. The Journal of Physical Chemistry A, 105(30):7254–7266, 2001.

[262] L. B. Harding, S. J. Klippenstein, and Y. Georgievskii. On the combination reactions of hydrogen atoms with resonance-stabilized hydrocarbon radicals. The Journal of Physical Chemistry A, 111(19):3789–3801, 2007.

[263] A. M. Mebel and R. I. Kaiser. Formation of resonantly stabilised free radicals via the reactions of atomic carbon, dicarbon, and tricarbon with unsaturated hydrocarbons: theory and crossed molecular beams experiments. International Reviews in Physical Chemistry, 34(4):461–514, 2015.

[264] O. Krechkivska, C. M. Wilcox, B. Chan, R. Jacob, Y. Liu, K. Nauta, S. H. Kable, L. Radom, and T. W. Schmidt. H and d attachment to naphthalene: Spectra and thermochemistry of cold gas-phase 1-c10h9 and 1-c10h8d radicals and cations. The Journal of Physical Chemistry A, 119(13):3225–3232, 2015.

243 [265] M. Z. Zgierski, T. Chakraborty, and E. C. Lim. Inter-ring exchange interactions in the excited electronic state of bridged diaryl compounds: Exci- ton splitting in 9,10-dihydroanthracene and dibenzosuberane. The Journal of Chemical Physics, 98(12):9340–9345, 1993.

[266] J. Laane. Vibrational potential energy surfaces and conformations of molecules in ground and excited electronic states. Annual Review of Physical Chemistry, 45(1):179–211, 1994.

[267] L. A. Verguilla-Berdecia. Tunneling in a quartic, symmetric, double well potential: A simple solution using a hermite basis. Journal of Chemical Education, 70(11):928, 1993.

[268] Y.-D. Shin, H. Saigusa, M. Z. Zgierski, F. Zerbetto, and E. C. Lim. Inversion potentials in the ground and excited electronic states of 9,10- dihydroanthracene as probed by the absorption and ﬂuorescence spectra of jet-cooled molecules. The Journal of Chemical Physics, 94(5):3511–3516, 1991.

[269] A. Pathak and P. J. Sarre. Protonated pahs as carriers of diﬀuse interstellar bands. Monthly Notices of the Royal Astronomical Society: Letters, 391(1):L10–L14, 2008.

[270] M. F. Rode, A. L. Sobolewski, C. Dedonder, C. Jouvet, and O. Dopfer. Com- putational study on the photophysics of protonated benzene. The Journal of Physical Chemistry A, 113(20):5865–5873, 2009.

[271] B. S. Freiser and J. L. Beauchamp. Acid-base properties of molecules in excited electronic states utilizing ion cyclotron resonance spectroscopy. Journal of the American Chemical Society, 99(10):3214–3225, 1977.

244