Investigating the Reactions of Gas-Phase Radicals Using Distonic Ions Benjamin B

University of Wollongong Research Online

University of Wollongong Thesis Collection University of Wollongong Thesis Collections

2011 Investigating the reactions of gas-phase radicals using distonic ions Benjamin B. Kirk University of Wollongong, [email protected]

Recommended Citation Kirk, Benjamin B., Investigating the reactions of gas-phase radicals using distonic ions, Doctor of Philosophy thesis, School of Chemistry, University of Wollongong, 2011. http://ro.uow.edu.au/theses/3374

Research Online is the open access institutional repository for the University of Wollongong. For further information contact the UOW Library: [email protected]

Investigating the reactions of gas-phase radicals using distonic ions.

A thesis submitted in fulﬁlment of the requirements for the award of the degree

Doctor of Philosophy

from

The University of Wollongong

Benjamin B. Kirk BCompSci, BSci(Hons I)

School of Chemistry

2011

CERTIFICATION

I, Benjamin B. Kirk, declare that this thesis, submitted in fulﬁlment of the requirements for the award of Doctor of Philosophy, in the School of Chemistry, University of Wollongong, is wholly my own work unless otherwise referenced or acknowledged. The document has not been submitted for qualiﬁcations at any other academic institution.

Benjamin B. Kirk 03 June 2011

CONTENTS

Contents i

List of Abbreviations v

List of Figures vi

List of Schemes xi

List of Tables xiv

1 Introduction 5 1.1 Foreward ...... 5 1.2 Free radicals ...... 5 1.2.1 Generation of free radicals in the gas phase ...... 9 1.2.2 Detection and characterisation of free radicals in the gas phase . 12 1.2.3 Mass spectrometry ...... 16 1.3 Radical ions ...... 36 1.3.1 Distonic radical ions ...... 36 1.3.2 Distonic radical cations ...... 39 1.3.3 Distonic radical anions ...... 41 1.3.4 Structural characterisation of distonic ions ...... 44 1.3.5 Selective gas phase synthesis of distonic radical ions ...... 49 1.4 Distonic radical ions as models for neutral reactions ...... 54 1.5 Thesis overview ...... 57

2 Instrumentation 59 2.1 Introduction ...... 59 2.2 Photolysis ...... 60

i II CONTENTS

2.2.1 Modifications for laser photolysis ...... 64 2.2.2 Synthesis of radical ions by laser photolysis ...... 70 2.2.3 Summary ...... 84 2.3 Ion-molecule reactions ...... 85 2.3.1 Modifications for ion-molecule reactions ...... 86 2.3.2 Proof of principle ion-molecule reactions ...... 91 2.3.3 Calculating the concentration of a neutral ...... 92 2.3.4 Determining the concentration of dioxygen ...... 96 2.3.5 Measuring reaction kinetics ...... 96 2.3.6 Reaction efficiency ...... 97 2.3.7 Errors in calculated reaction rates ...... 98 2.4 Conclusion ...... 99 2.5 Materials ...... 100 2.5.1 General method for synthesising trimethylammonium iodide salts 100 2.5.2 General method for generation of iodides from alcohols ...... 101

3 Gas phase reactions of distonic phenyl radical ions 103 3.1 Introduction ...... 103 3.1.1 Degradation of the phenylperoxyl radical ...... 104 3.2 Method ...... 108 3.2.1 Mass spectrometry ...... 108 3.2.2 Computational methods ...... 109 3.2.3 Predicting relative reaction rates ...... 109 3.3 Materials ...... 111 3.3.1 Nitrogen dioxide ...... 111 3.3.2 N,N,N-trimethyl-4-nitrobenzaminium iodide ...... 111 3.4 Results ...... 112 3.4.1 Gas phase synthesis of distonic ion models for phenyl radical . . . 112 3.4.2 Comparison of distonic phenyl radical ions to the neutral analogue115 3.4.3 Structural characterisation of the distonic analogues ...... 116 3.4.4 Reaction of 4-(N,N,N-trimethylammonium)phenyl radical cation

(1) with O2 ...... 124

3.4.5 Reaction of 4-carboxylatophenyl radical anion (2) with O2 . . . . . 127 CONTENTS III

3.4.6 Charge-tag effects on the reaction of distonic phenyl radical ions

with O2 ...... 130 3.4.7 Discussion ...... 133

3.4.8 Kinetics of distonic phenyl radicals with O2 ...... 134

3.4.9 Calculation of the 4-carboxylatophenyl radical anion + O2 potential energy surface ...... 138 3.5 Conclusion ...... 142

4 Gas phase reactions of distonic phenylperoxyl radicals with NOx 147 4.1 Introduction ...... 147 4.2 Methods ...... 152 4.2.1 Materials ...... 152 4.2.2 Mass spectrometry ...... 153 4.3 Results ...... 153 4.3.1 Reaction of the phenylperoxyl radical with NO· ...... 153 4.3.2 Reaction of the alkylperoxyl radicals with NO· ...... 159 · 4.3.3 Reaction of the phenylperoxyl radical with NO2 ...... 161 4.4 Conclusion ...... 163

5 Gas phase reactions of cyclohexyl radicals 167 5.1 Introduction ...... 167 5.2 Method ...... 170 5.2.1 Mass spectrometry ...... 170 5.2.2 Computational chemistry ...... 172 5.2.3 Materials ...... 172 5.3 Results ...... 174 5.3.1 Gas phase synthesis and structural characterisation of the 4- carboxylatocyclohexyl radical anion ...... 174

5.3.2 Reaction of the 4-carboxylatocyclohexyl radical anion (E) with O2 184

5.3.3 Computation of the 4-carboxylatophenyl + O2 potential energy surface ...... 187 5.4 Conclusion ...... 194

6 Summary and conclusions 197 IV CONTENTS

Bibliography 203

A Appendix A 235

B Appendix B 239

C Appendix C: Response to Reviewers Comments 243 LISTOF ABBREVIATIONS

AC Alternating Current

CAD Collision Activated Dissociation

CI Chemical Ionisation

CID Collision Induced Dissociation

CRDS Cavity Ring-down Spectroscopy

DMDS Dimethyl disulﬁde

EI Electron Ionisation

ESI Electrospray Ionisation

ICR Ion Cyclotron Resonance

IR Infrared

LIT Linear Ion-trap

MS Mass Spectrometry

FT-ICR Fourier Transform Ion Cyclotron Resonance

FTMS Fourier Transform Mass Spectrometry

PD Photodissociation

PES Photoelectron Spectroscopy

PI Photoionisation

PIE Photoionisation Efﬁciency

QIT Quadrupole Ion-trap

SORI-CAD Sustained Off-resonance Irradiation

RF Radio Frequency

UV Ultraviolet

VUV Vacuum Ultraviolet

v LISTOF FIGURES

1.1 Synthesis of triphenylmethyl and its peroxidation...... 7 1.2 Two pathways involved in lipid peroxidation...... 8 1.3 A schematic of a cavity used in CRDS...... 13 1.4 Near-IR CRDS spectrum of the O-O stretching region of the methylperoxyl and phenylperoxyl radical...... 14 1.5 Experimental and simulated IR spectra of the phenylperoxyl radical isolated in a 10 K argon matrix...... 17 1.6 Schematic of an electron ionisation source...... 18 1.7 Measurements used for calculation of resolution (∆m) and mass resolving power (m/∆m)...... 21 1.8 A double-focusing sector mass spectrometer...... 23 1.9 A quadrupole mass analyser...... 25 1.10 The ﬂowing afterglow triple quadrupole mass spectrometer...... 27 1.11 Ion cyclotron motion ...... 30 1.12 Cubic FT-ICR cell ...... 31 1.13 The hyperbolic endcaps and ring electrode of a 3D ion-trap...... 32 1.14 The hyperbolic linear quadrupole ion-trap of the ThermFisher LTQ Mass Spectrometer...... 32 1.15 The stability region of an ion within a linear quadrupole...... 33 1.16 A stability diagram illustrating the low-mass cutoff in an ion-trap...... 34 1.17 Resonance forms of the benzoyl cation...... 37 1.18 Characterisation of distonoid radical cations...... 38

+· . + 1.19 Comparison of CH3OH and CH2OH2 by CID...... 42 1.20 CID spectrum of a) ionised ethanol and b) its β-distonic isomer...... 44

vi LIST OF FIGURES VII

1.21 Potential energy surface of the 1-propanol radical cation and its γ-distonic counterpart...... 45 1.22 (a) Comparison of reaction efﬁciency to IE-EA for a number of neutral reactions. (b) An ionic curve-crossing model...... 57

2.1 (a) Gas phase UV absorption spectrum of iodobenzene and pentaflu- oroiodobenzene. (b) UV absorption spectra of iodocyclohexane and iodoadamantane in n-heptane...... 60 2.2 Temporal variation and curvature of the m/z 172 reactant ion in the pseudo- first order decay curve obtained during reaction of tert-butyl isocyanide with the N-(3,5-dehydrophenyl)-3-fluoropyridinium ion...... 61 2.3 266 nm PD spectrum for the +10 charge state of iodo-Cytochrome c demonstrating exclusive loss of iodine atom and 266 nm PD spectrum for the +10 charge state of Cytochrome c showing no fragmentation without the presence of an iodinated residue...... 62 2.4 UV absorption spectrum of 2-thioxopyridin-1(2H)-yl isobutyrate in acetonitrile...... 63 2.5 Photographs of the modified backplate of an ThermoFisher LTQ...... 65 2.6 Schematic of the ThermoFisher LTQ back plate modifications...... 66 2.7 Photographs of the laser instrumentation and digital PCB board modified to receive a TTL trigger...... 67 2.8 Dialog windows in Xcalibur to turn on TTL trigger output ...... 68 2.9 Pulse sequences for laser pulse generation...... 69 2.10 Exploded schematic of the ion source in a ThermoFisher LTQ Mass Spec- trometer...... 70 2.11 a) Isolation and 266 nm PD of N,N,N-trimethyl-4-((2-thioxopyridin-N- yloxy)carbonyl)benzeneaminium cation. Isolation and b) 266 nm PD and c) CID of 3-iodo-N,N,N,4-tetramethylbenzeneaminium cation. Isolation and 266 nm PD of d) 4-iodomethyl-N,N,N-trimethylbenzeneaminium cation. . . 74 2.12 Isolation of the m/z 149 fragment ion generated by PD (266 nm) of 3-iodo-

N,N,N,4-tetramethylbenzeneaminium cation in the presence of O2...... 75 2.13 Isolation of the m/z 149 fragment ion generated by CID of 3-iodo-N,N,N,4-

tetramethylbenzeneaminium cation in the presence of O2...... 76 VIII LIST OF FIGURES

2.14 Isolation of the m/z 149 fragment ion generated PD (266 nm) of 4-

iodomethyl-N,N,N-trimethylbenzeneaminium cation in the presence of O2. 77 2.15 Generation of the 3-carboxylatoadamantyl radical anion from 3-(N- oxycarbonyl-2(1H)-pyridinethione)adamantane-1-carboxylate anion . . . . 80 2.16 Generation of the 3-carboxylatoadamantyl radical anion from 3- iodoadamantanecarboxylate radical anion...... 81 2.17 Comparitive spectra showing the same reactivity of m/z 178 generated by a) CID of 1,3-adamantanecarboxylate, b) PD of 3-iodoadamantanecarboxylate, c) PD and d) CID of 3-(N-oxycarbonyl-2(1H)-pyridinethione)adamantane-

1-carboxylate with O2 for 1000 ms...... 82 2.18 Plot of precursors 1,3-adamantanedicarboxylate dianion (G), 3-(N- oxycarbonyl-2(1H)-pyridinethione)adamantane-1-carboxylate anion (I) and 3-iodoadamantanecarboxylate anion (J)...... 83 2.19 Schematic of the modiﬁed ThermoFisher LTQ linear ion-trap mass spectrometer...... 87

2.20 Photographs of a) ball valve T2, b) positioning of ball valve T2 and splitter K,

and c) splitter K with connection to helium cylinder H2...... 88 2.21 a) Heating block section of the mixing system, the position of which is shown in Figure 2.19 b) The septum inlet...... 89 2.22 Reaction of a) 4-(N,N,N-trimethylammonium)benzyl radical cation and b) 2- methyl-5-(N,N,N-trimethylammonium)phenyl radical cation with dimethyl disulﬁde (DMDS)...... 92

3.1 (a) ESI-MS spectrum of N,N,N-trimethyl-4-((2-thioxopyridin-N-

yloxy)carbonyl)benzaminium cation in H2O/MeOH during positive mode + + electrospray resulting in an M ion at m/z 289. (b) Isolation of M and subsequent CID of m/z 289...... 113 3.2 ESI-MS of terephthalic acid...... 114 3.3 ESI-MS of 4-iodobenzoic acid...... 115 3.4 Calculated geometry of each distonic phenyl radical ion...... 115 3.5 Calculated spin and charge densities of each distonic phenyl radical ion. . . 116 LIST OF FIGURES IX

3.6 Isolation of the 4-(N,N,N-trimethylammonium)phenyl radical cation in the presence of 1,2-dimethyldisulﬁde. (*) represents ions present due to reaction

with O2...... 118 3.7 Isolation of the 4-carboxylatophenyl radical anion in the presence of 1,2- dimethyldisulﬁde...... 118 3.16 (a) Reaction of 4-(N,N,N-trimethylammonium)phenyl rad-

ical with O2 and (b), subsequent CID of the 4-(N,N,N- trimethylammonium)phenylperoxyl radical cation (m/z 167). (c) Reaction

18 of 4-(N,N,N-trimethylammonium)phenyl radical cation with O2 and (d), 18 subsequent CID of 4-(N,N,N-trimethylammonium)phenyl- O2-peroxyl (m/z 171)...... 126 3.18 Structure, charge and spin distributions of distonic phenyl radical ions. . . . 131 3.20 Electron binding energies of intermediates on the 4- carboxylatophenylperoxyl and 4-sulfonataphenylperoxyl radical anion degradation pathways...... 136 3.23 Unimolecular degradation of the phenylperoxyl radical...... 140 3.24 The unimolecular degradation of the 4-carboxylatophenylperoxyl radical. . . 141 3.25 Optimised transition state structures calculated at the B3LYP/6-311++G(d) level of theory. The dashed line represents changes in the bonding at the transition state...... 142 3.26 The transition states and partition function calculations of the two competing reaction pathways after formation of oxepinon-7-yl...... 143

4.1 Schematic of the two-state curve-crossing model of Ellison et al. describing the F-O stretching potential in trans-FONO...... 149 4.2 Schematic of the two-state curve-crossing model of Kosmas et al. describing the interconversion of singlet and triplet states on the −−→ CH3OONO CH3ONO2 potential energy surface...... 150 −−→ 4.3 Potential energy surfaces for CH3ONO2 trans-CH3OONO reaction (a) S0

and S1 surfaces at the CAS(14,11) level, (b) S0 surface...... 151

4.4 Reaction of 4-(N,N,N-trimethylammonium)phenyl radical with O2 and NO. . 154 4.6 Comparative CID spectra of m/z 136 ions...... 157 X LIST OF FIGURES

5.1 Optimised transition state structures calculated at the CBS-QB3 level of theory. The dashed line represents changes in the bonding at the transition state...... 177

5.2 Formation of 1-methoxycarbonylcyclohexanide by reaction of d3-methoxide with methyl cyclohexanecarboxylate...... 179 5.3 Allowing (a) 4-carboxylatocyclohexyl radical anion (E), (b) 3- carboxylatocyclohexyl radical anion (F), and (c) 1-carboxylatocyclohexyl

radical anion (H) to react in the presence of background O2...... 180 5.4 Reaction of (a) 4-carboxylatocyclohexyl radical anion (E), (b) 3- carboxylatocyclohexyl radical anion (F), and (c) 1-carboxylatocyclohexyl radical anion (H) with NO·...... 181 5.5 Subjecting (a) 3-carboxylatocyclohexylperoxyl and (b) 4- carboxylatocyclohexylperoxyl (I)) radical anions to CID at 20 (normalised collision energy)...... 182 5.6 Isolation of 4-carboxylatocyclohexyl radical anion (E) in the presence of

background O2 for increasing reaction times between 30 and 5000 ms. . . . . 185 5.7 CID spectra of a) ion of m/z 141 generated by reaction

of 4-carboxylatocyclohexyl radical anion E with O2, b) 3,4- epoxycyclohexanecarboxylate anion, and c) 4-oxocyclohexanecarboxylate anion...... 188

18 5.8 Reaction of a) 4-carboxylatocyclohexyl radical anion E with O2, and b) CID spectrum measured of the m/z 162 ion generated in this reaction. Reaction

of c) 1,4-dideutero-4-carboxylatocyclohexyl with O2, and d) CID spectrum measured of the m/z 160 ion generated in this reaction...... 189 5.9 Reaction coordinate diagram for the reaction of the 4-carboxylatocyclohexyl

−1 radical anion (E) with O2 where all energy values are presented as kcal mol . 191

18 A.1 Isolation of charge-tagged phenyl radicals in the presence of O2...... 237 LISTOF SCHEMES

1.1 Greek letter designation of distonic radical ions ...... 36 1.2 Electron transfer and intramolecular 1,6-hydrogen atom transfer in dialkyl- N-nitrosamines...... 39

1.3 Reaction of ionised cyclopropane with NH3...... 40 1.4 Structurally distinct isomers of the methanol radical cation...... 41 1.5 Some resonance structures of o-xylene demonstrating cross-conjugation and charge-spin separation...... 42 1.6 Ionspray ionisation of 1,2- 1,3- and 1,4-benzenedicarboxylatic acid ...... 43 1.7 Reaction of O–· with benzene...... 43 1.8 Reactions of radical ions with DMDS...... 46 1.9 General scheme for reaction of radical ions with t-butylisocyanide...... 46 1.10 General scheme for reaction of radical ions with allyl bromide...... 46 1.11 General scheme for reaction of radical ions with dibromodichloromethane. . 47 1.12 General scheme for reaction of radical ions with alkanes...... 47

1.13 Reaction of O2 with pyridiniumethyl isomers...... 47 1.14 Cross-conjugated resonance structures of the acetate radical anion...... 49 1.15 Diagnostic reactions of radical ions with NO·...... 49 1.16 Synthesis of distonic phenyl radical cations...... 50 1.17 CID of an enolate anion...... 51 1.18 Schematic showing photodissociation at 266 nm and sustained off- resonance irradiation collision activated dissociation (SORI-CAD) of iodinated and brominated pyridinium cations ...... 51 1.19 Reaction of atomic oxygen radical anion with benzene to form o-benzyne. . . 52 1.20 Formation of acetate radical anion by desilylation of acetyltrimethylsilane. . 53 1.21 Formation of p-benzyne ...... 53

xi XII LIST OF SCHEMES

1.22 Oxidative decarboxylation of dicarboxylates generated via ESI-MS...... 54 1.23 McLafferty rearrangement of the propyl formate ester...... 55 1.24 Model phenyl radicals undergo exclusive hydrogen atom abstraction with glycine in both (a) the gas phase and (b) solution phase...... 56 2.1 CID of 3-(N-oxycarbonyl-2(1H)-pyridinethione)adamantanecarboxylate anion...... 64 2.2 Photolysis of O-nitrososerine residue...... 69 2.3 Electrospray ionisation of (i) N,N,N-trimethyl-4-((2-thioxopyridin- N-yloxy)carbonyl)benzeneaminium iodode (1), (ii) N-(3-iodo-4- methylphenyl)trimethylammonium iodide (2) and (iii) 4-(iodomethyl)- N,N,N-trimethylbenzeneaminium iodide...... 72

2.4 Photolysis of iodotyrosine residue and its reaction with O2...... 77 2.5 Electrospray ionisation of 1,3-adamantanedicarboxylic acid ...... 78 2.6 Putative mechanisms for PD and CID of 3-(N-oxycarbonyl-2(1H)- pyridinethione)adamantane-1-carboxylate anion (I)...... 80 2.7 Putative mechanisms for PD and CID of 3-iodoadamantanecarboxylate anion (J)...... 81 2.8 Electrospray ionisation of (i) 3-iodo-N,N,N,4-tetramethylbenzaminium iodide (1) and (ii) 4-(iodomethyl)-N,N,N-trimethylbenzaminium iodide (2) and their subjection to PD results in a carbon-centred radical which is allowed to react in the presence of dimethyl disulﬁde (DMDS)...... 91 3.1 Formation of cyclopentadienone ...... 105 3.2 Electrospray ionisation of N,N,N-trimethyl-4-((2-thioxopyridin-N- yloxy)carbonyl)benzenaminium (m/z 289) and subsequent CID leads to formation of the 4-(N,N,N-trimethylammonium)phenyl radical cation (m/z 135)...... 113 3.3 (a) Electrospray ionisation of terephthalic acid leads to formation of a dianion (m/z 82). Subsequent CID of the dianion results formation of the 4- carboxylatophenyl radical anion (2) at m/z 120. (b) Electrospray ionisation of 4-iodobenzoic acid leads to formation of the 4-iodobenzoate anion (m/z 247). PD of the isolated anion results in formation of 4-carboxylatophenyl radical anion (2) at m/z 120...... 114 4.1 Reaction of alkyl or arylperoxyl radicals with NO· ...... 148 LIST OF SCHEMES XIII

4.2 Scheme describing generation of 4-(N,N,N- trimethylammonium)phenylperoxyl radical...... 154 4.5 Overall scheme for CID of the isomeric (i) N,N,N-trimethyl-4- (nitrosooxy)benzaminium cation (E) and (ii) 4-N,N,N-trimethyl-4- nitrobenzaminium cation (F)...... 158 4.6 Reaction of 4-carboxylatocyclohexylperoxyl radical anion (B) with NO·. . . . 159 4.7 Putative mechanisms resulting in loss of charge of the nascent charge-tagged 3-carboxylatocyclohexyl peroxynitrite...... 160 4.8 Reaction of 3-carboxylatoadamantylperoxyl radical anion (C) with NO·. . . . 161 4.9 Overall scheme for the reaction of 4-(N,N,N- · trimethylammonium)phenylperoxyl radical cation (B) with NO2...... 162 · 5.1 Degradation pathways of the ethylperoxyl radical (CH3CH2OO )...... 168 5.2 Possible branching pathways during the unimolecular degradation of the 4- carboxylatocyclohexylperoxyl radical anion (I)...... 171 5.3 Formation of 3- and 4-carboxylatocyclohexyl radical anions by electrospray ionisation of their parent dicarboxylic acids...... 175 5.4 Reaction energy of β-scission and hydrogen atom transfer in the 4- carboxylatocyclohexyl radical anion...... 178 5.5 Generation of the 1-carboxylatocyclohexyl radical anion by CID of 1- methoxycarbonylcyclohexanide anion...... 178 5.6 Putative mechanisms for (a) formation of m/z 60 during reaction of 1-

carboxlatocyclohexyl radical anion with O2, and (b) formation of m/z 112 during reaction of NO· with 1-carboxylatocyclohexyl radical anion...... 182 5.7 Putative isomerisation mechanism of the 4-carboxylatocyclohexylperoxyl

radical anion such that reaction of O2 is hindered...... 185 6.1 Putative mechanisms for (a) formation of m/z 60 during reaction of 1-

carboxlatocyclohexyl radical anion with O2, and (b) formation of m/z 112 during reaction of NO· with 1-carboxylatocyclohexyl radical anion...... 198 6.2 Putative isomerisation mechanism of the 4-carboxylatocyclohexylperoxyl · radical anion such that reaction of O2, NO or other neutrals is hindered. . . . 199 A.1 Mechanisms for the formation of ions measured during the reaction of the

4-carboxylato-4’-biphenyl radical anion with O2...... 236 LISTOF TABLES

1.1 Example demonstrating the low-mass cutoff...... 35 1.2 Distonic organometallate radical anions...... 48

2.1 Example of neutral concentration calculation for dimethyl disulﬁde...... 95

2.2 Comparison of second order rate constants (k2) for the ion-molecule reactions depicted in Equations 2.10-2.13 ...... 96 2.3 Example of trajectory collision rate and reaction efﬁciency calculations . . . 98

3.1 Comparison of 2nd order rate constants and efﬁciencies for the ion molecule

reactions of distonic radical ions with O2...... 135

4.1 Bond dissociation energies (DH298) for 2-methyl-2-((2-thioxopyridin-N- yloxy)carbonyl)propane, tert-butyl peroxynitrite, tert-butyl nitrate, phenyl peroxynitrite, and phenyl nitrate...... 152

B.1 Literature and calculated electron afﬁnities...... 240 B.2 Calculated bond dissociation energies...... 241

xiv ACKNOWLEDGEMENTS

Without the dedication and support of so many people this thesis would never have been written. First and foremost I would like to thank my supervisors Dr. Stephen Blanksby and Dr. Adam Trevitt. Over the last 4 years of my candidature, they have had an open door policy and created a collegial environment that made me feel more important than I probably was. This is certainly true of the entire Chemistry faculty at UOW.

To Dr. David Harman. As a research fellow when I began Honours, you guided me on my journey into research. Without your incredible knowledge and enthusiasm to teach, especially in organic synthesis, I doubt I would be where I am now. I hope I can eventually pass on what you have taught me.

I would like to thank Marc Bouillon, Andrew Stevens, Hayden Matthews and all the other organic guys that have helped when I ask what must be second nature to you. Thank you for the many useful discussions in synthesis and lab methods.

To Prof. Hilkka Kenttämaa and the Kenttämaa group, most especially Bartek and Jenn. It was most gratifying to learn so much while surrounded by so many fun, supportive and dedicated people.

To my friends, family of friends, and to the research students in Chemistry (and a select few from Biology). Thanks for all the good times. You made the last four years bearable.

To Mum, Dad, Nick, Hayden and Josh. I appreciate all the things you’ve done to support me through the years that have allowed me to get this far. If not for you, I would have had to ﬁnd a real job.

A very special mention for the mass spectrometry group past and present. I know I’m going to miss someone. Maybe you can just write your names below? Berwyck,

1 2 ACKNOWLEDGEMENTS

Celine, Claire, Dave, Jane, Jenn, Jess, Jihan, Karina, Kimberly, Leo, Linda, Marty, Michael, Michelle, Nicole, Rachel, Shane, Tom, Tony and Troy. Procrastination was a major part of our relationship, but I would not have continued past Honours if not for your continued friendship and support. No matter how I was feeling you guys were always, always there for me. Thank you for putting up with my rants. Thank you for being an unjudging shoulder. Thank you for the crazy nights out on the town and the quite nights in front of the TV. You guys are so amazing and I wish you all the very, very best in the future. ABSTRACT

Free radicals are involved as reactive intermediates in a diverse range of processes including hydrocarbon combustion, the formation of photochemical smog, lipid peroxidation, cell apoptosis, polymerisation and synthetic cyclisation reactions. The intrinsic reactivity of free radicals and the inherent difficulty involved in their controlled generation and isolation has led the experimentalist to develop ever more complex methods of determining their ultimate fate in these reactions. A systematic mass spectrometric and computational study of archetypal charge-tagged phenyl and cyclohexyl radicals was undertaken to elucidate the reactivity of these radicals towards O2. Central to this study was the modification of a commercial linear ion-trap mass spectrometer which facilitated (i) injection of neutral gases and liquid reagents into the buffer gas of the instrument and (ii) introduction of a laser pulse through the ions isolated within the ion-trap. These modifications were essential to allowing measurement of ion-molecule reactions which were used to probe the structure and reactivity of radical ions isolated within the ion-trap. In addition, photolysis experiments may now be undertaken using this instrument to probe isolated species or generate free radicals within the ion-trap by photodissociation of photoactive precursors.

Distonic phenyl and cyclohexyl radical ions with carboxylato anion and N,N,N- trimethylammonium cation motifs were synthesised using ion-trap mass spectrometry. The reactions of N,N,N-trimethylammoniumphenyl radical cation and 4- carboxylatophenyl radical anion with O2 were found to proceed through formation of a peroxyl radical intermediate. In the cationic case, the phenylperoxyl radical was generated in sufﬁcient quantity for subsequent isolation. The major degradation product in both cases was the phenoxyl radical, which differs signiﬁcantly from long-standing mechanistic and computational predictions which suggest loss of CO2 to generate cyclopentadienyl radical as the major reaction product of this reaction. Collision in-

3 4 ABSTRACT

duced dissociation of the isolated phenylperoxyl radical demonstrated that both O and CHO· may be lost directly from the phenylperoxyl radical, with a computational study · suggesting that loss of CHO may be competitive with CO2 loss at ambient temperatures.

A combined computational and experimental study suggests that the 4- · · carboxylatocyclohexylperoxyl radical anion degrades by loss of HO2 and HO to form alkenes and isomeric epoxides mirroring previous investigations of the neutral archetype. Alternative pathways leading to formation of cyclohexanone or ring-opened isomers were not observed. Subsequent isolation of the nascent cyclohexylperoxyl radical in the presence of O2 resulted in no reaction, suggesting that the radical does not isomerise to a hydroperoxycyclohexyl radical as has previously been suggested.

The reaction of the isolated 4-carboxylatocyclohexylperoxyl radical anion and · · N,N,N-trimethylammoniumphenylperoxyl radicals with NO and NO2 was also investigated. Allowing the distonic cyclohexylperoxyl radical to react with NO· resulted in no · reaction, while its reaction with NO2 was not possible as the dicarboxylate precursor – was found to be depleted by electron transfer to form NO2. In contrast, reaction of the · · isolated phenylperoxyl radical with NO and NO2 resulted in products consistent with · · ejection of NO2 and NO3, respectively, as previously reported in the literature.

While these results highlight the strengths of using a distonic radical ion approach to investigating radical reactions and their ability to model the reactivity of their neutral counterparts, limitations were also identified which may restrict their utility. This may include (i) the complicated nature of multiple neutral reagents in the ion-trap which leads to competing reactions with the isolated radicals. This may limit the investigation of step-wise reactions which require different neutral reagents. (ii) The fixed pressure of the ion-trap which may limit the collisional cooling of reactive or fragile product ions leading to their absence or low ion-count in a given spectrum. (iii) Destabilisation or degradation of reactive species such as radicals due to the isolation process which was found to impart significant energy on the isolated species, and (iv) the limited structural information provided when a product is in low abundance or collision induced dissociation fails due to the inherent structural stability of the ion resulting in ejection of the ion prior to fragmentation. ooyi laae hr n odn lcrnrmiswt ahfamn (Reaction fragment each with remains electron bonding by one generated where are cleavage, radicals homolytic free neutral (Reaction fragment, cation single a radical with remain or electrons bonding 1.1) (Reaction anion radical respectively. a 1.2), is product the then tron, poiesi.Afe aia sa tm oeueo o ihoeo oeunpaired more or one with ion or molecule of atom, another an with electrons. is paired radical are free electrons A their of spin. all opposite that such exist molecules general, In 1.2 in radicals aryl and alkyl of chemistry. role tropospheric the and into combustion insight new the providing of of with study goal radicals, overarching phenylperoxyl systematic an and a cyclohexyl is archetypal the work of following reactivity and the formation mass of ion-trap focus using disserta- primary radicals this The isolating of behind spectrometry. methods motivation powerful The simple, develop to fate. is tion ultimate complex more their ever determining develop to of experimentalist methods the they led which has with isolated difficulty and inherent generated the are and radicals organic of reactivity intrinsic The 1.1 reradicals Free Foreward 1 fteupie lcrnaie u oteacpac rls fa elec- an of loss or acceptance the to due arises electron unpaired the If 1 niehtrltclbn laae hr odbek n both and breaks bond a where cleavage, bond heterolytical Unlike 5 I NTRODUCTION

C HAPTER 1 6 INTRODUCTION

1.3).1 Furthermore, neutral free radicals may also be generated from an anion by photolysis or electrolysis resulting in removal of an electron (Reaction 1.4).

− − A + e −−→ A · (1.1)

+ − A −−→ A · + e (1.2)

AB −−→ A· + B· (1.3)

− − A −−→ A· + e (1.4)

One example of a neutral free radical is atomic chlorine, which may be formed by photolysis of molecular chlorine resulting in the generation of two chlorine atoms each with an unpaired electron (Reaction 1.5). Atomic chlorine readily abstracts hydrogen from aliphatic and aromatic2 hydrocarbons to form carbon-centred radicals and HCl (Reaction 1.6).3,4,5,6,7

−−→hv · Cl2 2Cl (1.5)

RH + Cl· −−→ R· + HCl (1.6)

Radicals are, typically, highly reactive species, in many cases seeking an electron with which to reach a stable valence octet. Exceptions are usually radicals stabilised by conjugated π-orbital systems, and/or radicals which are sterically hindered.8 The inherent reactivity of radicals leads to a situation where isolating radicals free from the perturbative effects of secondary reactions due to self-reaction and/or reaction of the radical with solvents or precursors, is quite difﬁcult. The mechanisms involved in processes of such reactive and transient species are typically inferred by end product analyses, and consequently, often tainted by ambiguity. There exists therefore, wide ranging interest in isolating radicals and directly characterising their reactivity without the complication of such interferences.

Organic chemists make use of free radical chemistry in many reactions including radical cyclisations, radical coupling reactions, and radical mediated polymerisation. The ﬁrst successful isolation of an organic free radical was, like many monumental scientiﬁc discoveries, achieved serendipitously. In 1900, Moses Gomberg, while attempting to synthesise hexaphenylethane by a Wurtz-like coupling of chlorotriphenyl- FREE RADICALS 7

methane with silver, isolated instead the stable triphenylmethyl radical (Figure 1.1a).9 Triphenylmethyl is one example of a radical stabilised by steric hindrance and conjugation. In this case the three bulky phenyl rings which are bonded to the methylene group slow or prevent access to the radical site by large molecules and the unpaired electron may delocalise over the three phenyl rings. In solution, the triphenylmethyl radical is not the sole molecule; the radical is in equilibrium with a dimer. On exposure to air Gomberg found the radical to undergo a facile oxidation that was shown to result in a peroxide (Figure 1.1b). Furthermore, the radical was found to react with iodine to form iodotriphenylmethane. This work famously concluded with Gomberg asserting, "This work will be continued and I wish to reserve the ﬁeld for myself."9

Ag O2 Cl O O

AgCl (a) (b)

Figure 1.1: Synthesis of triphenylmethyl and its peroxidation.9

Free radicals are ubiquitous in the body and are involved in many biochemical processes including neural signalling, viral defence and cell membrane damage.10 Chronic inﬂammation is one example, which can lead to increased lipid peroxidation.11,12,13,14 Lipid peroxidation occurs when an oxidant abstracts a hydrogen from an unsaturated fatty acid, resulting in formation of a carbon-centred radical. As in the case of the triphenylmethyl radical, these radicals may react with dioxygen to form peroxyl radicals, which may subsequently abstract hydrogen from another lipid or hydrocarbon (Figure 1.2i) or undergo rearrangements to form a new molecule.

A peroxyl radical isomerisation pathway is shown in Figure 1.2(ii). Cyclisation of the peroxyl radical results in a bridgehead dioxolane which can rearrange to eject malondialdehyde (MDA). The radical intermediates formed during lipid peroxidation have been shown to degrade into a range of saturated and α,β-unsaturated aldehyde and ketones. Many of these byproducts (MDA, for example) can modify DNA bases or react with proteins, deregulating cellular homeostatis and can ultimately result in carcinogenesis.11,12,13,14 In contrast to the triphenylmethyl radical, these transient 8 INTRODUCTION

H H H [ox] H

H O 5-exo ii) O OO OO

RH i)

H H H + O O

MDA HOO

Figure 1.2: Two pathways involved in lipid peroxidation.11 radical intermediates are not isolable. They react quickly within their environment, and are thus short-lived. Instead, and typical of many radical reactions, the mechanisms involved have been proposed based on the characterisation of the stable end products.

While the former two examples have described radical reactions which occur in solution, radicals are also pervasive in the gas phase. Free radicals play an important role in the formation and removal of organic materials and other anthropogenic pollutants from the troposphere. An oxidant, primarily HO· in the troposphere, initiates these processes by abstracting a hydrogen from a hydrocarbon, in this example an alkane, to produce an alkyl radical (Reaction 1.7).15,5 The fate of alkyl radicals (R·) in air is their reaction with O2 which results in the generation of alkylperoxyl radicals (Reaction 1.8).15,5 Alkylperoxyl radicals (ROO·) may react with another molecule or decompose unimolecularly to form smaller molecules, which ultimately results in their conversion to CO2 and H2O. These processes, however, may result in formation of atmospheric pollutants, for example, alkylperoxyl radicals may react with NO· (Reaction 1.9) to

· 15,5 generate NO2, an important component of photochemical smog (Reaction 1.10). FREE RADICALS 9

+ · −−→ · + RH HO R H2O (1.7) · + −−→ · R O2 RO2 (1.8) · + . −−→ ∗ RO2 NO ROONO (1.9) ∗ −−→ · + . ROONO RO NO2 (1.10)

Radicals generated in all of these processes are only present in very low concentrations, as even in the gas phase radicals will react quickly upon encountering a reaction partner. The difﬁculty of isolating the key reactive intermediates means evidence for many purported mechanisms are supported primarily by measurement of reaction products. This thesis aims to introduce and demonstrate a systematic approach to synthesising radicals and directly probing reactive intermediates and reaction products.

1.2.1 Generation of free radicals in the gas phase

To probe short-lived free radicals one must develop methods with which to generate and isolate them with speciﬁcity and free from interfering reactants. Additionally, the radicals must be produced in quantities within the limit of detection of the measuring instrumentation. Removing interferences will simplify the analysis of experimental data. While many important radical reactions occur in the solution phase, and the aim of many investigations is to study radical reactions which occur in solution, analysing radical reactions in the gas phase offers many advantages. Radicals are known to react with many solvents. For example, many radicals will abstract hydrogen from alkanes typically used as solvents. The fundamental advantage of investigating radical reactions in the gas phase is the removal of solvent and counter-ion interactions, which may complicate analysis of the reaction. In addition, gas phase reactions may be undertaken at varying pressures, with radical precursor and reactant molecules diluted in an inert gas such as argon or helium. Lowering the concentration of radicals may seem counter- intuitive, however, this decreases the frequency of reactive collisions and therefore increases the lifetime of the radicals and the time in which they may be detected.

Before identifying or measuring the concentration of radicals, however, one must have a means with which to generate them. The next three sections describe methods which are commonly used to generate free radicals in the gas phase, speciﬁcally: 10 INTRODUCTION

radiolysis, pyrolysis and photolysis.

1.2.1.1 Radiolysis

Radiolysis encompasses any method which uses radiation from nuclear decay to dissociate a chemical bond, but may also include electromagnetic radiation at microwave and radio-frequency wavelengths as well as X-ray and γ-radiation from a linear acceler- ator. Radioactive decay sources such as 60Co produce γ-radiation16 while 210Po and 242Cm produce α-radiation.17 While there have been examples of gas phase experiments utilising radioactive decay sources,16 their use is typically limited to solution phase experiments. In contrast, microwave sources have been particularly useful in gas phase studies performed in ﬂow systems.18,19,20,21,22,23

Microwave discharge reactions typically generate a reactive species such as atomic

24 hydrogen from H2 or atomic nitrogen from N2 by direct excitation. The reactive species is allowed to react with a secondary reagent to produce the desired radical. A common gas phase method of generating the atmospherically relevant HO· is the reaction of atomic ﬂuorine with water vapour in a microwave discharge ﬂow tube as outlined in Reactions 1.11 and 1.12.23,20

−−→ · F2 2F (1.11) · + −−→ · + F H2O HO HF (1.12)

· Similarly, HO may be produced by reaction of atomic hydrogen with N2O as outlined in Reaction 1.13.21

· + −−→ · + H N2O HO N2 (1.13)

Rate constants for the reaction of HO· with many hydrocarbons have been measured using HO· generated in this manner.20

1.2.1.2 Pyrolysis and combustion

Pyrolysis and combustion generate radicals among other products, by rapidly heating a molecule or fuel to promote dissociation. In contrast to combustion, pyrolysis FREE RADICALS 11

is undertaken in the absence of oxygen. In both cases, premixed gases containing small concentrations of a radical precursor may be used to measure the reaction of a radical with reactant molecules. Common instrumentation used for both pyrolysis and combustion include shock tubes21,25,26,27 and high temperature reactors.28,29,30 These methods are typically used to model combustion reactions at high temperature and/or pressure.

A shock tube contains a low and high pressure region separated by a diaphragm. In the high pressure region, a low molecular weight molecule such as helium or hydrogen is held, while in the low pressure region, the radical precursor and reactant gases are introduced into an inert gas such as argon or xenon, collectively known as the test gas. When the diaphragm is ruptured, either by pressure or mechanically, a shock wave is produced which propagates through the low pressure region increasing the temperature and pressure of the test gas resulting in pyrolysis of the fuel.21,25,26,27 A large volume dump reservoir may be connected to the shock tube to quench the reaction, which stops reﬂecting shock waves which may reheat the test gas. The product gas may be analysed directly, or sampled over the duration of the experiment for ofﬂine analysis.

In heated flow reactors, reactant molecules are introduced into the flow of an inert gas which passes through a heated block or flow tube held at high temperature.31,32 Un- like a shock tube, flow reactors are operated continuously and products are always being generated. In a flow tube, the reaction products and intermediates can additionally be analysed by sampling at various positions along the flow tube.

A hyperthermal nozzle can be used to produce a clean beam of radicals by super- sonically pulsing a radical precursor seeded into an inert gas through a heated nozzle held at temperatures as high as 2100 K.28,29,30 The resulting radical beam is cooled by supersonic expansion through the nozzle exit. The near-sonic ﬂow of the reactant gas minimises recombination and other bimolecular reactions of the reactive species prior to its expansion into vacuum for analysis or sampling. For example, Nandi et 29 − al. used this technique to synthesise methyl radicals. Azomethane (CH3N NCH3) or iodomethane (CH3I) were seeded into argon and pulsed through a hyperthermal nozzle held at 1150 K. The resulting molecular beam was ionised and analysed by

+ photoionisation mass spectrometry where CH3 was the only ion detected, conﬁrming the only product was the methyl radical. 12 INTRODUCTION

1.2.1.3 Photolysis

Photolysis is any chemical reaction which occurs due to the exposure of an atom or molecule to light (typically electromagnetic radical lower than the infrared (IR); < 740 nm).1 The flash photolysis method uses a short, intense flash of light to produce free radicals.1 Early flash photolysis experiments by Norrish and Porter33,34,35,36 in the late 1940s made use of high intensity flash tubes to achieve photolysis, while work in the 1960s by Lindqvist introduced the use of monochromatic light produced by lasers.37 In both cases a low concentration of the radical precursor is seeded into a flow of an inert gas which is passed through a reaction tube. The laser or flash lamp is directed down, or through the reaction tube resulting in photolysis of the precursor and generation of radicals.38,33,34,35,37

Unlike most pyrolysis experiments, laser photolysis can be used to selectively generate radicals without affecting the buffer and reactant gases. In this way, radicals may be produced from photoactive precursors and rapidly thermalised by collisions with the buffer gas, while the reactant and buffer gasses are essentially transparent to the laser pulse. In addition, because radicals are produced by photolysis and not thermolysis the reaction tube may be heated or cooled to any temperature allowing the generation of radicals over a wide temperature range, not just those higher than the decomposition temperature of the precursor.

Phenyl radicals have been generated from various photolabile precursors including nitrosobenzene,39 acetophenone39,40 and iodobenzene.41 Yu and Lin, for example, used an excimer laser at 248 nm and 193 nm to photolyse nitrosobenzene and acetophenone, respectively, to generate phenyl radicals. In these experiments, the phenyl radicals were successfully synthesised by laser photolysis and allowed to react with

·42 42 NO and O2. Cavity ring-down spectroscopy was then used to measure the kinetics of these association reactions.

1.2.2 Detection and characterisation of free radicals in the gas phase

The primary goal of detecting free radicals is to obtain information about their structure and spectroscopic properties and to elucidate their reactivity. The term reactivity is rather ambiguous. In general, reactivity can encompass a range of properties which deﬁne a radical reaction all of which are fundamentally important. These include, a) FREE RADICALS 13

what molecules a radical will react with, b) the rate at which the radical will undergo this reaction, c) the mechanisms by which the reaction will proceed, and d) the intermediates and products which are ultimately formed.

While reaction rates can be measured using many experimental methods, some as simple as a titration, investigation of radical reactions often requires more sophisticated techniques to overcome the issues of low concentration and lifetime of the radicals being studied. The methods introduced in the previous section have been used to overcome issues of radical synthesis, and in general, have been shown to generate radicals in concentrations within the limits of detection of spectroscopic instrumentation. The following methods: cavity ring-down spectroscopy, matrix-isolation and mass spectrometry, are three examples of techniques routinely used for the spectroscopic analysis of radicals.

1.2.2.1 Cavity ring-down spectroscopy

Cavity ring-down spectroscopy (CRDS) is a sensitive direct laser absorption method. It is characterised by its long effective pathlength (up to many kilometres) over other direct absorption methods, which increases the limit of detection over single pass methods. Unlike conventional absorption spectroscopy which measure the magnitude of absorption through a sample, CRDS methods measure the rate of absorption of a laser pulse conﬁned in an optical cavity.43 A simpliﬁed schematic is shown in Figure 1.3.

High reflectivity mirrors

Light Source Detector

Optical Cavity

Figure 1.3: A schematic of a cavity used in CRDS.

In a typical experiment a laser pulse is ﬁrst injected in an optical cavity. The light 14 INTRODUCTION

Figure 1.4: The Near-IR O-O stretching region of the methylperoxyl and phenylperoxyl radical spectrum 40 1 1 measured by Just et al. Vibrations of the methylperoxyl radical and two vibrations (120 and 14O) of the phenylperoxyl radical are indicated. is reﬂected many times between two highly reﬂective (R > 0.999) mirrors and over time decays. If a sample is introduced which absorbs at the same wavelength as the laser pulse, the rate of decay will increase. An absorption spectrum can be obtained by varying the wavelength of the incident laser pulse. Although CRDS is a highly sensitive absorption technique, only samples which absorb in the wavelengths output by the laser can be analysed. Furthermore, complex gas mixtures with overlapping absorption bands can complicate analysis. CRDS is also subject to background and cavity mode effects which can decrease the resolution of the spectrum obtained.43

CRDS has been employed by many groups to measure electronic spectra and the kinetics of gas phase radical reactions. Lin and co-workers have measured rates for

39 the reaction of the phenyl radical with a number of molecules including O2 and NO·,42 both important reactions in the troposphere, using CRDS. In these experiments, CRDS was undertaken at a ﬁxed wavelength corresponding to electronic transitions in phenylperoxyl and phenyl radicals. The Miller group have used CRDS in a number of recent studies, systematically modelling and experimentally measuring near-IR electronic spectra of peroxyl radicals including ethylperoxyl,44 propylperoxyl,45 butylperoxyl,46 FREE RADICALS 15

pentylperoxyl,47 cyclohexylperoxyl48 and phenylperoxyl radicals40 to gain insight into their geometric, electronic and vibrational structure, which will allow identification of these radicals and their isomers in future kinetic studies. For example, in their study of the phenylperoxyl radical, Miller and co-workers observed the Ã2A′-X˜ 2A′′ origin band and a number of excited state vibrational frequencies. The spectrum shown in Figure 1.4 shows the O-O stretching region of both the methylperoxyl and phenylperoxyl radicals measured during these experiments.40 Traditionally, the B˜′′-XA˜ ′′ transition in the UV is used to measure alkylperoxyl radicals. This region, however, is usually broad and structureless making identification of overlapping alkylperoxyl radicals difficult. The identification of sharp vibrational bands should allow discrimination of different alkylperoxyl radicals in a sample (as shown in Figure 1.4).

1.2.2.2 Matrix isolation

When a sample contains a low concentration of radicals, CRDS may suffer from sensitivity issues. The matrix isolation technique was developed by Whittle, Dows and Pimentel in 1954 to overcome the inherent difﬁculty in isolating sufﬁcient quantities of radicals for analysis.49 This method provides a means to isolate a reactive intermediate in solid matrices of rare gases such as argon or xenon or relatively inert gases such as nitrogen at cryogenic temperatures, thus preventing diffusion and inhibiting any reaction of the intermediate; effectively trapping the target radical in an inert cage.50 Importantly, matrix isolation concentrates these radical intermediates so that spectroscopic experiments, such as IR or UV/visible spectroscopy, may be undertaken to characterise the electronic structure of the trapped species.

In a typical experiment a gaseous active species and matrix is sprayed via a pulsed nozzle onto a plate which has been cryogenically cooled to a temperature suitable to ensure immediate condensation of the material. The active species can be produced by a number of means including chemical reaction, pyrolysis, glow discharge, or photolysis; the only limitation being the time required to produce and deposit the species onto the plate so as to minimise any reaction between synthesis and deposition. The use of photolysis allows formation of the active material within the matrix after deposition of a photoactive precursor. Infrared or UV/visible spectroscopy may then be used to characterise the target. While this is essentially a solid-phase technique, when suitable matrices are employed (e.g., argon, which is only weakly interacting) infrared spectra 16 INTRODUCTION

can be comparable to those measured in the gas phase.50 Complementary high-level computational modelling is often required to aid in the assignment of electronic spectra if they have not previously been measured (see below examples).

An excellent demonstration of matrix isolation is the work of Nandi et al. in which the methylperoxyl radical was generated by alternating deposition of (i) the methyl

. radical CH3 generated by pulsing azomethane (CH3NNCH3) or methyl iodide CH3I through a hyperthermal supersonic nozzle and (ii) O2 doped in argon onto a CsI plate cyrogenically cooled to 20 K.29 This enabled the measurement of infrared and linear dichroism spectra which were then used in combination with computational models to assign vibrational frequencies. Recently, in a similar experiment from the same group, − − − · Jochnowitz et al. isolated the propargylperoxyl radical (CH−C CH2 OO ) to show that the reaction of propargyl radical with O2 leads only to formation of the propargylper- − − − · 51 oxyl radical, and not the isomeric allenylperoxyl radical CH2 C CH OO . More recently, Mardyukov and Sander52 used matrix isolation and IR spectroscopy to isolate and characterise the phenylperoxyl radical in an argon matrix at 10 K. A testament to the difﬁculty in isolating the phenylperoxyl radical is that this 2009 article was the ﬁrst reported isolation of the phenylperoxyl radical, previously predicted as an important intermediate during the oxidative degradation of benzene and aromatic aggregates.53,54,55,56,57,58

1.2.2.3 Synchrotron photoionisation mass spectrometry

Many of the radical sources introduced above have been coupled to a mass spectrometer for the analysis of the intermediates and products of radical reactions. For example, Osborn et al.38 developed an instrument where a ﬂash photolysis ﬂow tube was coupled to a photoionisation source with the resulting ions mass analysed by a small magnetic sector. This instrument allowed the selective synthesis of radicals and analysis of reaction products by photoionisation mass spectrometry.

1.2.3 Mass spectrometry

Mass spectrometers can be broken down into four components: volatilisation, ionisation, mass analysis and detection. Mass spectrometry involves ﬁrst volatilising and ionising an analyte, separating the resulting ions by their mass to charge ratio (m/z), and ﬁnally, detecting the mass analysed ions. There are many methods of ionisation, mass FREE RADICALS 17

Figure 1.5: Experimental and simulated IR spectra of the phenylperoxyl radical isolated in a 10 K argon matrix, with original ﬁgure caption, as reported by Mardyukov and Sander.52 analysis and detection. The following sections will introduce mass spectrometry as relevant to this introduction and the following chapters. The next section will describe ionisation techniques used to produce the charged particles fundamental to mass spectrometry including, electron ionisation, chemical ionisation and photoionisation. Following this section will be a description of four methods of mass analysis and detection including, the magnetic and electric sector, quadrupole mass ﬁlter, Fourier- transfer ion cyclotron resonance and the quadrupole ion-trap.59,60

1.2.3.1 Ionisation

1.2.3.1.1 Electron ionisation In a typical electron ionisation source61,59,60 a heated filament generates a beam of electrons which are accelerated by a potential difference between the filament and a trap anode (shown in Figure 1.6). To ionise organic molecules, electrons with typically 70 eV of energy are produced by a the potential difference between the heated filament and trap electrode. The energy imparted on an organic molecule when an electron impacts the sample is significant and in addition to ionisation yielding a molecular ion (Reaction 1.14), the radical cation produced may undergo significant fragmentation. 18 INTRODUCTION

The fragmentation patterns of organic molecules are quite reproducible and can therefore be used as a ﬁngerprint to identify different compounds.

− + − MX + e −−→ MX · + 2e (Ionisation) (1.14)

+ + MX · −−→ M · + X (Dissociative Rearrangement) (1.15)

+ + MX · −−→ M + X· (Dissociative Ionisation) (1.16)

Heated filament

Neutral Electron beam

+ + + Pusher + To mass analyser + + + + + + + +

Trap anode

Figure 1.6: Schematic of an electron ionisation source. An electron beam is produced by a heated ﬁlament which passes through the volatilised sample towards a trap anode. A pusher electrode guides any ionised sample towards the mass analyser.

Electron ionisation sources can also be used to generate negatively charged radical ions. This process is termed electron capture (EC) ionisation. During positive electron ionisation high energy (70 eV) electrons are used to overcome the ionisation energy of the molecule. If the energy of the electron is lower than the ionisation energy of a molecule, no ionisation will take place. In contrast, EC employs close to thermal energy electrons. Depending on the electron afﬁnity of the molecule and the energy of the electrons there are three mechanisms which can occur (for more information see Gross60 and Dillard62):

− − MX + e −−→ MX · (Resonance capture) (1.17)

− − MX + e −−→ M + X· (Dissociative resonance capture) (1.18)

− − + − MX + e −−→ M + X + e (Ion-pair formation) (1.19) FREE RADICALS 19

Resonance electron capture results in a molecular ion (a radical anion) while both dissociative resonance capture and ion-pair formation result in fragment ions. While EC has been used for a range of applications, EC is sensitive to ion source conditions such as the source temperature, operating pressure, buffer gas and source contamination, which inhibits its use in identifying molecules by database comparison.60,62

1.2.3.1.2 Chemical ionisation Chemical ionisation (CI) is used to denote bimolecular processes which result in ionisation of the target analyte. As previously noted, EI results in signiﬁcant fragmentation due to the high energy imparted by the incoming electron, in some cases this results in a very low yield of the molecular ion itself. Determination of the molecular mass of an analyte is essential to its identiﬁcation, so although the fragmentation observed during EI aids in structural elucidation of the analyte, it is not always desired. CI is a soft ionisation method complementary to EI which can overcome this shortfall.

+ − −−→ +· CH4 e CH4 (Reagent ionisation) (1.20) +· + −−→ + + · CH4 CH4 CH5 CH3 (Reagent auto-ionisation) (1.21) + + −−→ + + M CH5 MH CH4 (Proton transfer) (1.22)

In CI, a reagent gas is ﬁrst ionised by high energy electrons, followed by a subsequent ion-molecule reaction with the neutral analyte resulting in either a) proton transfer, b) anion transfer, c) electrophilic addition or d) charge transfer.60,59 The resulting ion has a potential energy proportional to the reaction exothermicity of the transfer process, which is typically much lower than the energy imparted during EI. This results in less fragmentation, and a higher yield of the quasi-molecular ion.

+ + MA + X −−→ M + XA (Anion transfer) (1.23)

+ + M + X −−→ [MX] (Electrophilic addition) (1.24)

+ + M + X · −−→ M · + X (Charge exchange) (1.25) 20 INTRODUCTION

1.2.3.1.3 Photoionisation An alternative method of overcoming the extensive fragmentation commonly associated with electron ionisation is photoionisation (PI).63,64,38,65,66 PI makes use of either ﬁxed frequency or a tunable light source to provide discrete wavelengths of light in the 7 - 11 eV range which, on irradiation of an amenable analyte, results in ionisation closer to its ionisation threshold resulting in little excess energy and therefore little, or no fragmentation.67 In addition, the ionising wavelength can be selected so that interfering species, such as the carrier gas from a GC instrument, or solvent and nebulising gases used in atmospheric pressure ionisation methods like electrospray, are not ionised. PI can also be used to generate ions for chemical ionisation, where the typical reactions, such as charge transfer, proton transfer and anion transfer, may subsequently ionise the analyte.67

Recent advances in the use of photoionisation have coupled highly tunable (7 - − − 25 eV), high photon flux (1021 photons cm 2 sr 1) medium resolution (E/∆E ≈ 1000 meV) vacuum ultraviolet (VUV) radiation produced at a synchrotron beamline.38,65,66 The fine-tunability of the VUV source in this case, not only allows ionisation of a wide range of samples, but allows accurate measurement of photoionisation efficiency (PIE) spectra which show the number of ions produced as a function of the photon energy. PIE spectra can be used to identify and characterise the analytes, for example, structural isomers, which typically result in distinct PIE spectra.38

1.2.3.2 Mass analysers

The ability of a mass analyser to discriminate between two ions of different mass is known by two terms, resolving power and resolution. In the past there has been contention between use of the two terms as they are often used interchangeably. In 2006, the IUPAC Analytical Chemistry Division released a provisional recommendation of mass spectrometry terms68 which defined resolution as the "smallest mass difference ∆m between two equal magnitude peaks so that the valley between them is specified fraction of the peak height", and resolving power as "the observed mass divided by the difference between two masses that can be separated: m/∆m." It was noted that resolving power requires definition of the method by which ∆m was measured (i.e., percentage of maximum peak height) and the mass at which the measurement was made. There are two methods generally used to calculate mass resolving power R. The FREE RADICALS 21

ﬁrst involves calculation of the resolution ∆m between two peaks which are separated by a 10% valley and is typically used to calculate the mass resolving power of sector instruments. In this case the mass resolving power is therefore given by Equation 1.26.

m R = (1.26) ∆m

The second considers that the peak width it proportional to the resolving power, and is simply the quotient of the measured mass and width of the measured peak at half of its full height (full width at half maximum, FWHM) as shown in Equation 1.27. This method has the advantage of requiring only one peak for the calculation of mass resolving power.

m R = (1.27) FWHM

Measurement of these values is illustrated in the Figure 1.7. In this diagram, m is 1000, while ∆m is 0.2 giving a mass resolving power of 1000/0.2 = 5000. If the second method is employed, the mass resolving power equates to 1000/0.4 = 2500. As shown, the method used to calculate mass resolving power must be speciﬁed in order to compare measured values.

m m + Δm Δm 100 Full width at half maximum 80 (FWHM)

60 % 10% 40 Valley

0 1000.0 1000.2 m/z

Figure 1.7: Measurements used for calculation of resolution (∆m) and mass resolving power (m/∆m), including the 10% valley method, and full width at half maximum (FWHM), based on the diagram of Henderson and McIndoe.59 22 INTRODUCTION

It can be seen that a mass spectrometer with a constant resolving power will have a variable resolution over its entire mass range. In contrast, instruments based on ion- traps and quadrupoles commonly achieve a unit mass resolution which implies that the instrument achieves a constant resolution, which may resolve peaks 0.1 - 1 Da apart depending on the instrument, over the entire mass range.

1.2.3.2.1 Electric and magnetic sector Sector instruments refer to electric and magnetic sector mass analysers which use an electric or magnetic ﬁeld to discriminate ions based on their m/z ratio. Common instrumentation incorporate both an electric and magnetic sector in tandem, which are known as double-focusing mass analysers. Instruments with additional sectors have also been constructed for tandem experiments, experiments in which an ion is mass selected and fragmented by various activation methods to gain additional structural information. The trajectory of an ion travelling through a sector analyser is governed by the Lorentz force equation given by

FL = q(E + v × B), (1.28) where FL is the Lorentz force, E is electric field strength, v is the velocity of the ion, B is the magnetic field strength, × is the cross product operator, and q is the charge on the ion. Thus in an electric field the force on the ion in the direction of the electric field simplifies to

FL = qE, (1.29) while in an homogeneous magnetic ﬁeld, the force on the ion is perpendicular to the direction of the ion motion and is given by

FL = qvB. (1.30)

In a magnetic sector instrument a mass spectrum is produced either by scanning the accelerating voltage of the ions while keeping the magnetic field constant, or by scanning the magnetic field strength and keeping the accelerating voltage constant. This results in ions of a specific m/z being focussed through a slit and onto a detector. FREE RADICALS 23

The resolution of the instrument may be increased by decreasing the slit width, but this will also decrease the sensitivity of the instrument. Alternatively, an array detector may be used, in this case there is no slit and all ions are measured at the same time.

Figure 1.8: A double-focusing sector mass spectrometer. Diagram taken from Ligon.69

While the overall purpose of a sector instrument is to discriminate ions based on their m/z ratio, more speciﬁcally, an electric sector focuses ions of the same kinetic energy, while a magnetic sector separates ions by momentum to charge (mv/z). The incoming velocity dependence of the magnetic sector means that the deﬂection of ions of the same mass with differing entrance velocities will vary, decreasing the resolution and sensitivity of the instrument. As shown in Figure 1.8, electric sectors may be used to focus ions of the same kinetic energy and are therefore used to reduce the distribution of ion kinetic energies entering the magnetic sector, resulting an overall increase in resolution.60,59,69

Structural elucidation of ions may be accomplished by identifying fragment ions. Ions traversing tandem sector instruments may be fragmented in the ﬁeld-free drift regions before the ﬁrst sector, or between subsequent sectors of the mass spectrometer by accelerating the ions at high voltage (ion energy of 3 - 10 keV) to collide with an inert gas, typically helium or argon. When a collision occurs, the kinetic energy of the ion is converted into internal energy, resulting in excitation of the ion to high energy electronic excited states. The maximum converted energy can be calculated using

Equation 1.31, where ELAB is the energy in the laboratory frame (the energy of the ion due to the accelerating voltage), Mi is the mass of the ion, Mn is the mass of the neutral and ECOM is the centre of mass corrected collision energy. 24 INTRODUCTION

Mn ECOM = ELAB , (1.31) Mn + Mi

As can be deduced from this equation, increasing the mass of the neutral results in a larger conversion of kinetic to internal energy upon an individual collision. This excitation energy is rapidly converted to vibrational energy which is redistributed among the covalent bonds of the ion. If the internal energy exceeds the dissociation energy of a particular bond fragmentation may occur.70

Doubling-focusing sectors may be operated at resolving powers as high as 100,000, but are typically used at around 25,000. While a higher resolution is obtained using tandem double-focusing instruments (for example, a four sector instrument), a single double-focusing mass spectrometer may be used to perform tandem analysis.71 In a reverse geometry instrument, where the magnetic sector is located before the electric sector, a precursor ion may be selected using the magnetic sector, and collisional dissociated in the subsequent ﬁeld free drift region. The kinetic energy of the resulting fragments is proportional to their mass, thus scanning the electric sector results in a product ion spectrum. This type of experiment is known as mass-analysed ion kinetic energy spectrometry (MIKES), with the width of the product ion peaks characteristic of the kinetic energy distribution resulting from the dissociation process. If this information is not required, the large peak widths is simply a loss of resolution.71 This may be improved by increasing the ion kinetic energy before or after collisions. The sectors may also be used in a linked conﬁguration but this may result in poor resolution of the parent ion resulting in contamination of the product spectrum with fragments from neighbouring ions.71

Sector instruments are typically used for high resolution mass spectrometry and accurate mass measurement. While their use is declining over cheaper and smaller instruments such as time of ﬂight mass analysers, sector instruments are still in common use.

1.2.3.2.2 Quadrupole mass analyser A quadrupole mass analyser consists of four identical parallel rods. While the ideal quadrupole ﬁeld is produced by hyperbolic rods, cylindrical rods are more common. Mass analysis is accomplished by applying a ﬁxed DC voltage of opposite polarity and a FREE RADICALS 25

superimposed radio frequency (RF) voltage to each opposing pair of electrically coupled rods (as shown in Figure 1.9). The trajectory of ions through the quadrupole can be simpliﬁed by considering the effect of each coupled pair of rods individually on two positively charged ions: one light ion, and one heavy ion.59,60

Figure 1.9: A quadrupole mass analyser. Ions of a specific mass to charge (m/z) have a stable trajectory through the electric field produced by the four rods. All other ions are deflected and neutralised on the rods. Diagram based on Henderson and McIndoe.59 U is the DC voltage while V cosωt is the RF voltage.

First, consider the two positively charged rods. With no RF voltage, the two ions are repelled by both rods and would continue through the mass analyser. When the RF voltage is applied and the overall potential is negative, the two positively charged ions will be attracted to the rods, while when positive, the ions will tend towards the centre of the rods. Due to inertia, it is harder to deﬂect the ion of high mass, while the low mass ion will be accelerated at a faster rate towards the rods. The low mass ion will eventually hit the rod, while the heavy ion will pass through. In this manner, the positive rods act as a high mass ﬁlter; only ions greater than a certain mass will pass through the mass analyser.72

Now consider the two negatively charged rods. With no RF voltage, the two positively charged ions are attracted to the rods. When the RF voltage is applied and the overall potential becomes positive, the two ions are repelled and are pushed towards the centre of the rods. This time, the lighter ion is deflected more when the overall potential is positive, while the heavy ion continues on its collision course with the rods. In this manner, the negatively charged rods act as a low mass filter; only ions less than a certain mass will pass through the mass analyser. Ions are thereby filtered by the combination of a high and low mass filter which allows only ions of a single m/z ratio to pass through 26 INTRODUCTION

the quadrupole.72

Quadrupole mass analysers can achieve a modest resolution typically in the range of 1000 to 3000.59,60 They can also be run at unit mass resolution which varies the resolution depending on the m/z range so that ions which are one mass unit apart may be discriminated while maintaining a fast scan speed.59

CID in an quadrupole is accomplished by coupling multiple quadrupoles together, and is similar to CID as employed in a sector instrument.59,60 A common geometry used for CID experiments is the triple quadrupole. The first quadrupole is used to mass select a specific ion, the second quadrupole is used as an ion guide and collision cell, while the final quadrupole mass analyses the products of CID. The collision cell is filled with a heavy inert gas, typically nitrogen or argon, called the collision gas, while the mass selected ions are accelerated at a high potential between the first quadruple and the collision cell. As described previously with reference to the sector analyser, the ion collides the collision gas resulting in conversion of kinetic energy to internal energy (Equation 1.31), which results in fragmentation if the ion gains internal energy over its dissociation threshold. CID energy can be set by varying the accelerating potential between the first and second quadrupole. In general, CID in a triple quadruple is lower in energy (commonly < 100 eV) than that of a sector instrument and the use of low accelerating voltages may result in spectra similar to those measured in ion-trap mass analysers.73

1.2.3.2.3 Flowing afterglow The flowing afterglow experiment was first developed by Ferguson and co-workers in the early 1960s to investigate the reaction rates of a number of atmospherically important gases with helium ions at thermal temperatures.18,74 The instrument consists of an ion source, reaction flow tube, ion sampling system and quadrupole mass analyser (Figure 1.10). Ions can be generated by a variety of methods including microwave discharge, electron impact, chemical ionisation, and laser ablation. Typically, the ion precursor is seeded into a stream of high purity helium which passes through the ionisation source to generate the ions. The resulting gas mixture passes through a de Laval nozzle which cools and accelerates the ion/gas mixture into a reaction flow tube. Depending on the required ion temperature, the ions can alternatively be cooled by FREE RADICALS 27

Figure 1.10: The flowing afterglow triple quadrupole mass spectrometer. Taken from Squires.75 collisions with the buffer gas. The drift tube and buffer gas may be heated so that experiments can be undertaken over a wide temperature range. Ports along the length of the flow tube allow neutral molecules to be injected into the gas stream at set intervals to react with the generated ions. Alternatively, a moveable injection port can be used allowing the introduction of neutrals at distance (or kinetic reaction time) along the flow tube. The product ions are then sampled through an orifice or skimmer cone and analysed using a quadrupole mass filter. The ion abundance and reaction time or gas pressure can then be used to measure ion-molecule reaction rates.75

Over the past 40 years, flowing afterglow instrumentation has significantly advanced. For example, in 1976 Adams and Smith developed the Selected Ion Flow Tube (SIFT) which couples a quadrupole mass filter to the ion source to allow the selection of a single ion for introduction into the flow tube, dramatically increasing the sensitivity of the instrument.76 Indeed, the versatility provided by a SIFT-MS instrument is demonstrated by its recent commercialisation in the form of a volatile organic compound quantification device.77 In 1983, Squires and Cooks coupled a triple quadrupole mass analyser to the ion detection region (depicted in Figure 1.10), allowing CID and secondary ion-molecule reactions for structural analysis of ions sampled from the flow tube, and later, Squires and co-workers constructed an electrospray ionisation (ESI) source to allow solution borne ions, including multiply charged ions, to be injected into their flow tube.78

Flowing afterglow has been fundamental to investigations of gas phase radical reactions. Villalta et al. used a neutral ﬂow tube coupled to a ﬂowing afterglow mass 28 INTRODUCTION

spectrometer79 to study an archetypical example of photochemical smog formation. In · this experiment, CH3O2 radicals are synthesised and injected into the neutral ﬂow tube, and reacted with NO· which is added at different locations along the same ﬂow tube. The resulting kinetically resolved reaction is measured by injecting the neutral product

+ stream into the ﬂowing afterglow of O2 ions generated by EI of O2 in a stream of helium, resulting in chemical ionisation of the neutral products and allowing measurement of the reaction products by mass spectrometry.

1.2.3.2.4 Fourier transform ion cyclotron resonance Fourier transform ion cyclotron resonance (FT-ICR) or Fourier transform mass spectrometry (FTMS)* was developed by Comisarow and Marshall80 and is the evolution of the ion cyclotron resonance (ICR) method introduced by Lawrence in 1930.81 The development of FTMS arose with the growing maturity of FT-NMR.82 The increased sensitivity and resolving power obtained by applying Fourier methods to NMR were highly desired in the mass spectrometry ﬁeld and the similarities evident between NMR and ICR, most notably their dependance on frequency, sparked interest in applying these techniques to ICR.82 In addition, Fourier techniques reduce, by several orders of magnitude, the time required to detect a wide frequency window (or mass range) which was historically undertaken by sweeping the magnetic ﬁeld strength. This was not only slow, but limited the resolving power of these instruments.83

The ICR method employs the Lorentz force equations as introduced for magnetic sector mass analysers in Equation 1.30. An ion in a circular orbit will experience a centripedal force Fc equal to

Fc = ma, (1.32) where m is the mass of the ion and a is acceleration. Centripedal acceleration a can be calculated using the equation

v2 a = , (1.33) r *While an FTMS was not used in the experiments reported in this research thesis, the author was graciously hosted by the Kenttämaa laboratory (Purdue University, IN) where experiments similar to those reported in Chapter 3 were undertaken on FTMS instrumentation. FREE RADICALS 29

where r is the radius of the orbit. Replacing the acceleration term a in Equation 1.32 gives

mv2 Fc = , (1.34) r which can be substituted into Equation 1.30 to give

mv2 qV B = , (1.35) r which simpliﬁes to

mv qB = . (1.36) r

The frequency of the circular orbit f can be calculated by

v f = , (1.37) 2πr which on substitution into Equation 1.36 gives the fundamental equation of ion cyclotron resonance which allows calculation of the cyclotron frequency fc of an ion of a given charge q and mass m in a magnetic ﬁeld B

qB fc = . (1.38) 2πm

Ions are detected in an FTMS by exciting the trapped ions into a large radius near to the detection plates. An RF voltage applied at the cyclotron frequency of a single m/z ion results in the resonant ions being excited into a large cyclotron radius. These ions are excited coherently, that is, the ions are packed as closely together before and after the excitation and travel in a coherent packet around the cell.84 After excitation, the ions retain their large cyclotron radius and as the coherent packet of ions pass alternate detection plates (see Figure 1.12) a signal is produced which may be acquired over a number of seconds. If a broadband excitation is used, all m/z ions may be excited at once and the resulting signal (called a transient) is a superposition of frequencies and amplitudes proportional to the cyclotron frequency and concentration of each m/z ion.84 Applying a Fourier transform to this transient results in a spectrum of frequencies 30 INTRODUCTION

which may be converted to masses by calibration using the cyclotron motion equations (Equation 1.38).84

In contrast to magnetic sector and electric sector instruments, the kinetic energy of the ions does not have a large effect on the resolution of the resulting spectrum, as demonstrated in Equation 1.38.84 The resolution obtained during FTMS is proportional to the length (acquisition time) of the transient. Ion packets lose coherency due to collisions with neutral molecules and coalescence with ions of similar cyclotron frequency. Long transients and thus higher resolution is obtained by providing ultra-high vacuum and minimising space-charge effects.84 FTMS can routinely obtain resolutions higher than any other mass analyser in excess of 100,000, and in some cases higher than 1,000,000.84

- v qv x B X B

Figure 1.11: Ion cyclotron motion. The orbital motion of an ion moving in a magnetic ﬁeld directed into the page.

There are four major components to an FTMS system, a strong magnetic field, the analyser cell, an ultra-high vacuum system and the data system (which includes all of the electronic devices used to control the system). The higher the strength of the magnetic field, the higher the possible resolving power.85 Typical magnetic field strengths on commercial instruments are between 1 and 9 Tesla, but there are a number of instruments with higher field magnets. The magnetic field drives the ions into cyclotron motion which traps the ions radially.

Analyser cells are available in many different geometries, one of which is the cubic cell (Figure 1.12). The cubic cell is made of six plates on which electronic potentials can be applied to trap, excite and detect ions isolated within. The cell is aligned with the magnetic field and a small trapping voltage (typically 1-2 V) is applied to the two plates perpendicular to the magnetic field to produce a dipolar field which traps ions axially. The addition four plates parallel to the field are used for excitation and detection.84

Collisions with background gas during detection can affect the resolution and mass FREE RADICALS 31

Excitation Plates

- +

+ -

Trapping Plates Detection Plates

Figure 1.12: Cubic FT-ICR cell including its alignment with a magnetic ﬁeld, and electronic circuitry. accuracy of the instrument. Therefore a vacuum system which can provide a working pressure on the order of 10−9 to 10−10 Torr is required for effective performance.84

FTMS can be coupled with many ionisation techniques and is extensible in a way which allows creative experiments without modifying hardware. Ionisation can be accomplished within the analyser cell, or ions can be generated externally and funnelled into the analyser cell by ion optics. Unlike sectors and quadrupole where tandem experiments are undertaken by coupling subsequent mass analysers (tandem- in-space), tandem experiments in an FTMS are achieved by sequential excitation and ejection events within the same cell (tandem-in-time).84 The primary method of ion fragmentation used in FTMS is sustained off-resonance collision activated dissociation (SORI-CAD). During SORI-CAD, a low energy (~1 - 20 eV) supplementary AC voltage is applied with a frequency that is ~75 - 4000 Hz off-resonance to the cyclotron frequency of the parent ion. This causes the ion to undergo multiple acceleration and deceleration events as the ion cyclotron frequency and the applied frequency go in and out of phase. The kinetic energy of the ion is converted into internal energy by multiple low-energy collisions with an inert buffer gas, typically pulsed into the instrument to a reach a pressure of 10−6 - 10−5 Torr. This gradual excitation process leads to fragmentation at ion internal energies close to the dissociation thresholds of the ion.84,86,87,88,89 An alternative method is on-resonance (or direct) CAD, where the supplementary AC voltage is applied at the cyclotron frequency of the parent ion. While this method also results in fragmentation of the parent ion, fragment ions are produced away from the centre of the analyser cell which is detrimental to both instrument sensitivity and resolution (for an in-depth discussion see cited references).84,85 32 INTRODUCTION

1.2.3.2.5 Quadrupole ion-traps Quadrupole ion-trap (QIT) are manufactured in 2D (or linear) and 3D geometries. A 3D ion-trap contains two hyperbolic end-caps and a ring electrode as shown in Figure 1.13. The overlap of DC and RF potentials applied to the endcaps and ring electrode produce a quadrupole field which traps the ions within the trap. In contrast, a linear ion-trap has a similar geometry to a conventional quadrupole mass analyser (as described previously in Section 1.2.3.2.2). The main advantage of a linear quadrupole is the increased ion storage capacity which significantly increases the sensitivity of the instrument.90,91 An example of a commercial linear ion-trap, the ThermoFisher LTQ linear ion-trap is shown in Figure 1.14. Ions within a linear ion-trap are isolated axially (z-direction) by a DC potential applied to the end caps of the ion-trap, while they are isolated radially (x- and y-direction) by a quadrupolar field produced by the overlap of DC and AC potentials applied to the four quadrupole rods. While ion trapping is possible with a standard single quadruple, the LTQ contains a quadrupole which is divided into three sections where the first and third sections are the end caps. This design minimises fringe field distortions in the trapping and resonance excitation fields which minimises unwanted ion excitation.90

Figure 1.13: The hyperbolic endcaps and ring Figure 1.14: The hyperbolic linear quadrupole ion- electrode of a 3D ion-trap.91, 92 trap of the ThermFisher LTQ Mass Spectrometer.90

The instrument used in subsequent chapters is a linear ion-trap so, while 3D ion- traps operate under similar conditions, the rest of this section will deal speciﬁcally with linear ion-traps.

The secular frequency (ω0) of an ion in a linear ion-trap, that is, the frequency at which the ion oscillates within the trapping ﬁeld, is proportional to the RF drive frequency (Ω) and the ion’s q-parameter (q). The secular frequency of an ion can be calculated using the following equations:91 FREE RADICALS 33

√ q2 b0 = (1.39) 2 b0Ω ω0 = (1.40) 2

The stability of an ion with a quadrupole ﬁeld is dictated by the Mathieu equation. The general solution to this equation for a linear ion-trap is given by:91

4eV0−p q = (1.41) 2Ω2 mr0

In this equation q is the stability factor or q-parameter, m is the mass of the ion, r0 is the distance from the centre of the trap to the quadrupole rod, Ω is the main RF frequency, V0−p is the peak-to-peak amplitude of the RF potential and e is elementary charge. Figure 1.15 illustrates the stability of an ion versus its q-parameter value. The y- axis ax term is proportional to the DC voltage applied to the quadrupole, which is 0 V in a linear ion-trap.91 Ions which have a q-parameter between 0 and 0.908 are considered stable, and should maintain a stable trajectory within the ion trap.

Figure 1.15: A diagram representing the stability of an ion within a linear quadrupole, taken from Douglas et al.93

CID in a quadrupole ion-trap is a slow energy accumulation similar to SORI-CAD and on-resonance CAD used in FTMS. A supplementary AC ’tickle’ voltage of ∼1 V is applied at the secular frequency of the isolated ion resulting in the ion gaining kinetic 34 INTRODUCTION

energy. The secular frequency (the frequency at which the ion oscillates while conﬁned) is constant at a given q-parameter and RF drive frequency, so the increase in kinetic energy must also lead to an increase in the radial displacement of the ion. The ion continues to collide with the helium buffer gas and, as many collisions occur, the excess kinetic energy is slowly converted into internal energy. As the internal energy overcomes the dissociation threshold of the ion, fragmentation occurs. The gradual heating of the ion results in the lowest energy fragmentation processes occurring, unlike the high energy single collision fragmentation which is usually observed by tandem-in-space mass spectrometry (a triple quadrupole for example).91,94

0.0

0.908

0.000 0.500 1.000 1.500

q-parameter

Figure 1.16: A stability diagram illustrating the low-mass cutoff in an ion-trap. The size of the bubble is proportional to the mass of an ion. The amplitude of the RF potential V is set so that the red ion is isolated at a q-parameter value of 0.250. Any ion whose q-parameter is increased past a value of 0.908 due to variation of V is destabilised and neutralised on a rod, or ejected from the ion-trap.

Ions are isolated within the ion trap by selective destabilisation. The m/z being

90 isolated is placed at a q of 0.830 by varying V0−p . This establishes a low-mass cutoff as trapped ions reestablish their trajectories based on the new RF amplitude. Any ions with a q-parameter greater than 0.908 will be destabilised and ejected from the ion- trap. For example, in an ion trap with Ω = 1.2 Mhz, r = 4 cm and isolated ion of m/z

250 held at a q of 0.830 requires a V0−p = 489.1 V. It follows that any ions with a m/z of less than m/z 228 will have a q-parameter value greater than 0.908 and will no longer maintain stable trajectories within the ion-trap (demonstrated in Table 1.1).91 A multi- frequency waveform containing a superposition of sine components spaced every 500 FREE RADICALS 35

Hz at all resonant frequencies between 5 - 500 Khz with a notch removed at the m/z being isolated is applied to eject all other unwanted ions.90,95

Table 1.1: Example parameters for Equation 1.41 demonstrating the low-mass cutoff.

RF Drive Frequency (Ω) 1.2 MHz Radius (r ) 4 cm

RF Amplitude (V0−p ) 489.1 V Mass (m) 500 Da 250 Da 150 Da 100 Da Q-parameter (q) 0.415 0.830 1.383 2.075

Detection of ions in an linear ion-trap is accomplished by sweeping V0−p such that ions are sequentially brought to a q-parameter of 0.880. The secular frequency ω of ions at a q-parameter of 0.880 can be calculated based on Equation 1.40. A supplementary RF potential is applied at this frequency so that only those ions isolated at a q-parameter of 0.880 gain kinetic energy. This supplementary potential with an amplitude optimised to maintain resolution (usually 20 mV/m/z with an intercept of 3 V) results in the selective destabilisation of these ions in the x-direction and ejection through exit slits in the hyperbolic rods (shown in Figure 1.14) onto detection dynodes.90,95

Non-destructive image signal detection with Fourier transform analysis, analogous to detection in FTMS, has also been applied to quadrupole ion-traps,96,97 however, higher operating pressures which decrease the time a transient may be acquired and AC coupling between the RF drive frequency and the detector electrodes which causes significant noise, has meant operation of these modified instruments has been confined to research laboratories. A recent report by Xu et al. applied this method to a miniaturised linear ion-trap with promising results.98 Image signal detection requires no dynode and electron multiplier which significantly simplifies ion-trap design and allows higher working pressures. In these experiments, narrow band detection of secondary ion motion components induced by a constant AC voltage were measured instead of the secular frequency of the trapped ions. Similarly to FTMS, the secular frequency component of the radial amplitude quickly decays at higher pressures due to ion-neutral collisions which limits the transient acquisition time. This technique is currently limited, however, to targeted analyses such as single-reaction monitoring (SRM) or multi-reaction monitoring (MRM).98 36 INTRODUCTION

1.3 Radical ions

Radicals ions are synonymous with mass spectrometry, as both electron ionisation and chemical ionisation techniques yield radical ion products. Radical ions are atoms or molecules which have gained or lost an electron resulting in a open shelled ion. In addition to ionisation, radical ions are often encountered during fragmentation of even- electron ions during tandem mass spectrometry experiments, including more modern fragmentation techniques such as electron capture dissociation99 (ECD) and electron transfer dissociation100 (ETD). These techniques are essentially post-source analogues of negative ion electron ionisation and chemical ionisation and are used primarily to fragment large biomolecules to elucidate their structure. Radical ions also play key roles

–· 10 as reactive intermediates in cell defence mechanisms (superoxide O2 for example), and also many organic syntheses such as electrocyclisations, photochemical rearrangements and electron transfer reactions.101,102,103

1.3.1 Distonic radical ions

In contrast to conventional radical ions containing a charge and unpaired electron localised on a single atom, distonic radical ions, the etymology stemming from the Greek diestos and Latin distans meaning separate, are a subclass of radical ions in which the radical and charge are located on separate atoms. While the existence of charge-spin separated radical ions had been postulated decades prior, interest in distonic radical ions quickly developed when it emerged that they may be more stable energetically than their conventional radical ion isomers.104

Distonic radical ions may be classiﬁed by the location of the radical relative to the charge by assignment of a Greek letter. For example, the distonic methyleneoxonium + . radical cation CH2OH2 containing the radical and charge on adjacent atoms is clas- + α . β siﬁed as -distonic, while CH2CH2OH2 is -distonic, and so on (examples shown in Scheme 1.1).

CH2OH2 CH2CH2OH2 CH2CH2CH2OH2 α−distonic ion β−distonic ion γ−distonic ion

Scheme 1.1: Greek letter designation of distonic radical ions, radical ions in which the radical and charged moieties are located on separate atoms.105,104, 106, 107, 1 RADICAL IONS 37

Historically, the formalised definitions by IUPAC and the initial coining of the term make reference only to radical cations. The IUPAC Gold Book, to date, contains no mention of distonic radical anions and distonic radical cations are defined as per the initial coining by Yates, Bouma, and Radom as "radical cations in which the charge and radical site are separated".106 This definition was refined in a later communication by the authors to suggest that use of the term be confined to radical cations generated by the ionisation of zwitterions or ylides107 and biradicals108 such that strict localisation of the charge and radical site in all resonance forms of the radical ion is maintained. The official IUPAC definition has not been updated to reflect this change, and indeed, distonic tends to be understood by the more liberal IUPAC definition in the literature.

The strict definition by Yates, Bouma and Radom motivated Eberlin and co- workers109 to introduce the term distonoid (distonic like) to encompass radical ions which display distonic character without formally falling under the distonic definition. This may include radical ions which display varying degrees of charge-spin separation, but with conventional radical ion resonance forms (e.g., benzoyl cation as shown in Scheme 1.17), thus failing a predicate of the distonic classification.

O O C C

Figure 1.17: Resonance forms of the benzoyl cation.109

The authors undertook extensive computation of the Mulliken spin and charge distributions of many putative distonoid ions (shown in Figure 1.18) which uncovered a range of radical ions which displayed signiﬁcant charge-spin separation while not formally classiﬁed as distonic. Furthermore, ion-molecule reactions of putative distonoid ions with neutral molecules usually employed to characterise distonic radical ions (introduced in Section 1.3.4) demonstrated that many distonoid ions display the same dual radical/ion reactivity of distonic radical ions.

While this work was significant, providing terminology which would allow the research community to remain true to the formal definition of distonic, the distonoid classification has not gained traction. As reported by Munsch et al.,110 the term distonic is now generally used to denote ions with separated charge and radical centres whether 38 INTRODUCTION

Figure 1.18: Calculations undertaken by Tomazela et al. at the B3LYP/6-311+G(d,p) level of theory set to characterise distonoid radical cations.109† The signed numbers are the calculated atomic charge, while the numbers labelled with an e represent the atomic spin. or not they fall into the formal deﬁnition of ionised zwitterions, ylides or biradicals. Indeed, there are now examples in the literature of the use of distonic to describe more exotic charge-spin separated species such as biradicals,111,112,113,114,115 triradicals,116 nitrenes117,118 and carbenes.117,119,120

The term distonic, as used in this thesis, will denote radical anions or cations which display characteristic charge-spin separation as determined by diagnostic molecular orbital calculations or ion-molecule reactions. Ion-molecule reactions which are characteristic of distonic radical ions will be introduced later in this chapter. These diagnostic tools aim to distinguish whether a putative distonic radical ion displays independent radical reactivity, a property which is lacking during the reaction of conventional radical ions.

The charge present on a distonic radical ion make them amenable to isolation using ion-trap mass spectrometry, thus a suitably manufactured charged moiety may act as a handle, or charge-tag with which to allow manipulation of a model radical in the gas phase. The following sections provide an account of the discovery of distonic radical RADICAL IONS 39

ions, their synthesis, and the evolution of their use as neutral radical models.

1.3.2 Distonic radical cations

The existence of stable charge-spin separated radical cations was ﬁrst proposed by Billets, Jaffé and Kaplan to explain the selective nature of an ion-molecule reaction resulting in a proton transfer reaction between two dialkyl-N-nitrosamines (Reaction 1.42).121,122

′ ′ +· + −−→ + + [R2NNO] R2NNO [R2NNOH] products (1.42)

Billets and co-workers observed that proton transfer only took place when the neu-

’ tral species in the reaction, R2NNO, had an equal or greater basicity than R2NNO. This speciﬁcity could not be explained by a reaction involving direct hydrogen abstraction

’ from the R2 alkyl chain (as shown in Scheme 1.2a) as hydrogen atom abstraction should not be affected by the basicity of the ions, only the availability of protons on the neutral

+· molecule. Instead it was suggested that [R2NNO] may be present as a charge-spin separated radical cation from which the reaction can proceed via direct proton transfer (Scheme 1.2b). Indeed, previous mass spectrometric studies had hinted at charge-spin separated species as intermediates in McLafferty rearrangements during collisional activation of radical cations.123,124,125

H H2 H2 H CH CH3 R' C R C 3 R CH R C N CH3 N CH3 N CH3 H2C O N CH3 H2C a) + + + N N N N N N N N O O R' O R' O O H

H H H 2 2 H2 CH R C R' C R C 3 R CH b) N CH2 N CH N CH N CH2 3 + 2 H2C + N H N N N N N O O O R' O O H H

Scheme 1.2: a) Electron transfer between dialkyl-N-nitrosamines and subsequent hydrogen abstraction. b) Intramolecular 1,6-hydrogen atom transfer to an intermediate charge-separated radical cation and subsequent proton transfer suggested by Billets et al.121

The selective nature of ion-molecule reactions was also the subject of an investigation published a year later by Gross and McLafferty in which the speciﬁcity of

+· the products observed during the reaction of NH3 with C3H5 isomers cyclopentane 40 INTRODUCTION

and propene was questioned. Loss of ethyl radical and ethylene was only observed during reaction of NH3 with ionised cyclopentane. It was suggested that addition of

NH3 to the isomers resulted in distinct charge-separated collision complexes. The reaction of NH3 with ionised cyclopropane was proposed to result in formation of the

3-ammoniumpropyl radical cation (A), whereas addition of NH3 to ionised propene was to yield 2-ammoniumpropyl radical cation (B). As illustrated in Scheme 1.3(a) and (b), respectively, intermediate A could conceivably lose both ethylene or ethyl radical, while B could not. There was no structural characterisation of the collision complexes undertaken in this study, indeed both ethylene and ethyl radical could form from a conventional radical cation isomer of A.

NH3

- C2H4 NH3 NH2 a) + NH 3 CH2 CH3

A - H

- CH CH 3 2 H2C NH2

NH3 b) No reaction + NH3

Scheme 1.3: Scheme proposed by Gross and McLafferty126 for the reaction of ionised cyclopropane with NH3.

It was not until 10 years later that Crow, Gross and Bursey127 undertook the ﬁrst mass spectrometric experiment demonstrating formation of a stable charge-spin separated radical cation. Collisional activation spectra were measured to structurally charac-

+· terise four [C3H8O] isomers synthesised by ionisation of 1-propanol, 2-propanol, +· methyl ethyl ether and of a m/z 60 fragment of 1,2-dimethoxyethane. The [C3H8O] isomer resulting from ionisation of 1,2-dimethyoxyethane was shown to exhibit a · collisional activation spectrum including characteristic loss of CH2 and an abundant +· β [CH3OH] ion which was rationalised as arising from formation of the -distonic + · isomer CH3OHCH2CH2.

The term distonic was not coined until a computational study speciﬁcally investigating low-energy charge-separated radical cations was undertaken by Bouma, Nobes and Radom.104 The motivation behind this study was to explain ESR experiments RADICAL IONS 41

. + which had identiﬁed the methyleneoxonium radical cation ( CH2OH2) in the solution- phase, achieved by γ-irradiation of protonated methanol at cryogenic temperatures.128

+· Calculation of the [CH4O] potential energy surface demonstrated that the methyle- . + neoxonium radical cation CH2OH2 was more stable energetically than the conventional +· −1 radical cation CH4O by 45 kJ mol . The results of the computational study suggested . + 105 that it may be possible to form CH2OH2 in the gas phase. Bouma et al. and Holmes et al.129 both set out to synthesise the two isomeric radical cations and probe their structure by mass spectrometry using collisional activation (Scheme 1.4).

EI CID a) CH3 OH CH3 OH

O OH H EI CID b) H2CO H2C OH2 O OH H

Scheme 1.4: The two experiments undertaken by Bouma et al. which demonstrated structurally distinct isomers of the methanol radical cation.105

It was hypothesised that EI of ethylene glycol would generate the putative dis-

. + tonic isomer CH2OH2, while EI of methanol would yield the methanol radical cation +· CH3OH . As shown in Figure 1.19, subjecting the two ionised isomers to collisional activation resulted in distinct mass spectra conﬁrming the earlier work of theory.

1.3.3 Distonic radical anions

Initial motivation for synthesising distonic radical anions stemmed from their im- portance in investigating the bond energies and heats of formation of more unusual chemical species, for example, carbenes, alkynes and biradicals.130 Distonic radical anions were synthesised ﬁrst in the gas phase by Harrison and Jennings in 1975 during reaction of O–· with acetone (shown in Reaction 1.43).131 However, Harrison and

+· Jennings suggested that H2 was abstracted from the same methyl group. Charge and radical separation were not considered or proven until deuterium labelling experiments of 1,1,1-trideuteroacetone undertaken by Dawson et al. demonstrated that HDO was the dominant loss during its reaction with O–·, resulting in the charge separated radical anion shown in Reaction 1.43.132 42 INTRODUCTION

+· . + 105 Figure 1.19: a) Table of CID data for CH3OH and CH2OH2 measured by Bouma et al.. b) CID spectrum +· . + 129 of CH3OH and CH2OH2 measured by Holmes et al.

−· + −−→ . − + O CH3COCD3 CH2COCD2 HDO (1.43)

This method was later utilised by Bruins et al. who reported, and conﬁrmed by deuterium labelling, that reaction of m-xylene with O–· resulted in formation of an [M - 2H]−. ion in which the two hydrogens were removed from separate methyl groups.133 Quite similarly, prior predictions were that abstraction occurred from the same methyl group.133

These examples were reported prior to coining of the term distonic. While there is signiﬁcant delocalisation of charge and spin (demonstrated in Scheme 1.5), these examples still fail the more restrictive distonic deﬁnition as both radical anions are cross-conjugated.

H2C CH2 H2C CH2 H2C CH2 H2C CH2

Scheme 1.5: Some resonance structures of o-xylene demonstrating cross-conjugation and charge-spin separation.

It was not until a theoretical investigation by Pius and Chandrasekhar in 1988 RADICAL IONS 43

that distonic was used to classify a radical anion.134 This investigation focussed on very small organometallic radical anions which compared distonic and conventional . – –· = isomers of the form CH2XH and CH3X (where X Li, BeH, CH2, Na, MgH or AlH2), respectively. In contrast to the distonic radical cation system previously studied, these distonic radical anions were found to be less stable than their conventional radical ion counterparts. It was suggested that there was a competitive preference between the electropositive X atom adopting a structure with a reduced negative charge and the unpaired electron bearing carbon maintaining a stable octet. In most cases the

. – latter dominates; an exception was CH2BH3 which was calculated as isoenergetic to –· its conventional isomer CH3BH2 (shown in Reaction 1.44).

−· −−→ . − ∆ ◦ = −1 CH3BH2 CH2BH3 H f 1.1kcalmol (1.44)

A year later, Siu et al. published a report wherein they synthesized 2-, 3- and 4- carboxylatophenyl radicals while investigating multiply charged ions during ionspray ionisation (shown in Scheme 1.6).135

CO H CO CO 2 Ionspray 2 CID 2

-2H+ CO2H CO2

Scheme 1.6: Ionspray ionisation of 1,2- 1,3- and 1,4-benzenedicarboxylatic acid results in formation of a dianion which upon subjection to CID undergoes oxidative decarboxylation ejecting CO2 and an electron.

Their distonic nature and their potential as radical models was not proposed, however, until Guo and Grabowski generated 2-carboxylatophenyl by reaction of o-benzyne radical anion with CO2 during ﬂowing afterglow experiments (shown in Scheme 1.7), where they concluded by suggesting that the inert nature of the benzoate anion may prove useful in investigating the reactivity of radicals using ion-molecule techniques.136

CO2 + O + CO2

- H2O

Scheme 1.7: Reaction of O–· with benzene results in 1,2-H+· abstraction and formation of o-benzyne. When o-benzyne is allowed to react in the presence of CO2, 2-carboxylatophenyl forms. Adapted from Guo and Grabowski.136 44 INTRODUCTION

1.3.4 Structural characterisation of distonic ions

To quote Squires, any "synthetic method without structural veriﬁcation or some means for chemically characterising the product is of dubious worth",75 and without a doubt, a central element of studying distonic radical ions is to conﬁrm that a distonic radical ion was in fact isolated and not its conventional counterpart. The following section introduces mass spectrometric methods, including collisional activation and ion-molecule reactions, typically used to differentiate and structurally characterise distonic radicals.

1.3.4.1 Collisional activation / collision-induced dissociation

In the past, CID was successfully employed using sector and triple quadrupole mass spectrometers to characterise distonic radical ions as distonic and conventional ions often fragmented yielding characteristic spectra. Distonic radical cations are commonly found to lose even-electron fragments to yield old-electron products, while CID of conventional radical cations tends to result in loss of hydrogen and alkyl radicals resulting in primarily even-electron fragment ions. An example of this method is shown

+· in Figure 1.20 where CID spectra of the conventional ([CH3CH2OH] ) and distonic + . 73 ( CH2CH2OH2) isomers of ethanol are compared. In these spectra the conventional · · radical cation (Figure 1.20a) displays a loss of H and CH3 while the distonic isomer

(Figure 1.20b) loses H2O, demonstrating thir structural dissimilarity, and these characteristic loses.

EI CID CH3CH2OH CH3CH2OH m/z 46

EI CID HOCH2CH2CH2OH CH2O + CH2CH2OH2 m/z 46

Figure 1.20: CID spectrum of a) ionised ethanol and b) its β-distonic isomer, taken from Wysocki and Kenttämaa.73

The success of the CID method is, however, predicated on undertaking an exhaus- tive comparison to the CID spectra of all possible distonic and conventional isomers of the radical ion.137 Characterisation by CID can be further complicated when the RADICAL IONS 45

conventional and distonic ion isomerise to a common intermediate prior to fragmentation, a possibility if the barrier to isomerisation between the distonic and conventional radical ion is lower than barriers to dissociation.138,139,140,141 For example, Wysocki and Kenttämaa reported that CID of both propanol and its γ-distonic isomer in a quadrupole ion-trap resulted in identical spectra. Calculation of a potential energy surface suggested that isomerisation to the distonic radical ion and subsequent H2O loss was lower in energy than direct fragmentation pathways (PES shown in Figure 1.21).73

Figure 1.21: Potential energy surface of the 1-propanol radical cation and its γ-distonic counterpart, taken from Wysocki and Kenttämaa.73

1.3.4.2 Ion-molecule reactions

Since the invention of chemical ionisation, the reaction of radical ions with neutral molecules has been an active area of research in mass spectrometry. Ion-molecule reactions retain a key role in the synthesis and characterisation of distonic radical ions. Neutral radicals typically react by radical-radical combination, bimolecular homolytic substitution (SH2 reactions such as hydrogen abstraction), or addition to unsaturated bonds. Conventional radical ions undergo reactions typical of chemical ionisation: primarily charge-transfer. Distonic radical ions tend to react differently to their conventional radical ion analogues, with charge and radical moieties reacting independently. Their reactivity is therefore characteristic of the functionalisation of the charged and radical moieties which may result in reactions typical of ions (charge transfer), and/or an independent radical (radical-radical combination, SH2 reaction, and double or triple bond addition). Radical-like reactivity is typically the distinguishing characteristic sep- 46 INTRODUCTION

arating a distonic radical ion from its conventional counterpart, thus radical-trapping molecules are routinely employed to identify and characterise distonic radical ions.

O O a) + CH3SSCH3 + CH3SSCH3

O O b) + CH3SSCH3 + CH3S S

Scheme 1.8: The reactions of a) ionised oxetane and b) the distonic methyleneoxoniumethyl radical cation with DMDS result in distinct products, as demonstrated in Stirk et al.142

Kenttämaa and co-workers have contributed seminal works on the structural elucidation of distonic radical cations through publication of several reviews of the state of distonic radical cation research.143,137,144 The group demonstrated that distonic radical cations react distinctly from their conventional radical cations when allowed to react with dimethyl disulﬁde142 or dimethyl diselenide.145,146 The distonic species

. . were found to react by SCH3 or SeCH3 abstraction, respectively, while conventional radical cations were found to undergo charge exchange (Scheme 1.8). Abstraction of

. SCH3 from dimethyl disulﬁde has since become the prototypical validation of a distonic radical cation.

There are a suite of other reagents routinely employed to characterise distonic radical ions. These include t-butylisocyanide, which transfers .CN:147

C X R + N X R C N X R NC +

Scheme 1.9: General reaction of t-butylisocyanide with a radical cation.

Allyl bromide, which may transfer bromine atom and/or allyl:109,148

Br X R Br + X R +

Br + X R + Br X R

Br + X R + Br X R Scheme 1.10: General reaction of allyl bromide with a radical cation.

CBr4, CBr2Cl2 and CBrCl3 which transfer bromine atom and CHBr3 which may transfer both hydrogen and bromine atoms:148 RADICAL IONS 47

Br Cl Cl X R + C X R Br + C Br Cl Br Cl

H Br Br X R + C X R H + C Br Br Br Br Scheme 1.11: General reaction of dibromodichloromethane with a radical cation.

Cyclohexane, tetrahydrofuran, acetone, toluene and 1,4-dioxane have been used as hydrogen atom sources and reaction of distonic radical ions in their presence results in hydrogen atom transfer.148,109

X R + R'H X RH + R' , Scheme 1.12: General reaction of alkanes with a radical cation.

Conventional radical ions are typically unreactive, or undergo charge-transfer analogous to Scheme 1.8(a) in the presence of these reagents.

149,109,118,150 O2 has also been shown to react selectively with distonic radicals. Yu et 150 149 β al. and later Sorrilha et al. found O2 to add to the -distonic 2-ethylpyridinium radical cation forming 2-ethylperoxylpyridinium, whereas no reaction was observed for the conventional and α-distonic isomers (shown in Scheme 1.13). Yu et al. have suggested that 1-pyridiniumethyl radical does not react with O2 due to delocalisation of the unpaired electron, whereas in the case of 2-pyridiniumethyl, the unpaired electron

150 is isolated from the ring. The reaction of O2 with distonic radicals is a central theme of this research thesis.

+ C2H4 + O a) 2 N N N O O

CH 3 + C5H5N + O b) 2 No reaction - C H 2 4 N

Scheme 1.13: Allowing a) 2-pyridiniumethyl radical cation and b) 1-pyridiniumethyl radical cation to react in 150 the presence of O2.

Squires and co-workers undertook a review of characteristic reactions of distonic radical anions while investigating distonic organometallate radical anions.151 These reactions were based on those reagents previously utilised to characterise distonic · · radical cations including NO , NO2,O2, CH3SSCH3, and CS2. These reactions were used 48 INTRODUCTION

to probe a collection of aluminate (AlMe3) and borate (BF3 and BH3) functionalised radical anions (shown in Table 1.2) to identify distonic radical anions which displayed chemical reactivity characteristic of a free radical.

Table 1.2: Table of distonic organometallate radical anions‡ investigated by Hill et al..151

The Brønsted basicity of the distonic radical anion was found to be a major factor in the overall reactivity of the radical ions investigated. Reactions of distonic radical anions with a high gas phase basicity were observed to undergo competitive proton- transfer reactions with some neutral reagents, whereas, those with a low gas phase basicity typically reacted via radical type reactions. The reaction of dimethyl disulﬁde with these ions, for example, was shown to result in proton transfer and formation

–152 · of CH3SCH2S which was highly competitive with the prototypical CH3S radical transfer characteristic of distonic radical cations.152,153 RADICAL IONS 49

· The high electron afﬁnity of NO2 was found in many cases to result in electron transfer instead of the expected characteristic radical-radical combination. Both NO· and O2 appear to be convenient alternatives for characterising distonic radical anions as they are aprotic, with a relatively low electron afﬁnity, and therefore reactivity towards anions. Analogous to the distonic radical cation example introduced above, · Harman and Blanksby found that both O2 and NO to react with the distonic 3- carboxylatoadamantyl radical anion to form the 3-carboxylatoadamantylperoxyl radical anion and 3-nitrosoadamantanecarboxylate anion, respectively.154,155

O O O O O O O O O O

H H H H H H H H H H

Scheme 1.14: Cross-conjugated resonance structures of the acetate radical anion.

Cross conjugation occurs when there are common molecular orbitals shared between distinct resonance structures, an example of which is shown in Scheme 1.14.

As shown in Scheme 1.15, Wenthold and Squires observed an E1cb like elimination reaction that occurs after addition of NO· to both the 2-methylpropan-2-id-1-yl radical anion and acetate radical anion resulting in formation of an α-distonic alkylcarboxylate radical anion followed by subsequent ejection of a nitroso carbanion.156,153 This reactivity appears characteristic of cross-conjugated radical anions.

1.3.5 Selective gas phase synthesis of distonic radical ions

The observation of distonic radical ions during ionisation and fragmentation by mass spectrometry and the suggestion that distonic radical ions may be used to model their neutral counterparts led to an interest in the deliberate, selective synthesis of these

H2C H2C NO + NO E1cb a) CH2 CH2 H2CNO + H2C C CH2

H2C H2C

O O NO + NO E1cb b) CH2 CH2 H2C NO + CO2 O O

Scheme 1.15: Reaction of a) 2-methylpropan-2-id-1-yl radical anion and b) acetate radical anion with NO· 156, 153 results in an E1cb like reaction, as reported by Wenthold and Squires. 50 INTRODUCTION

species. In the following sections, a number of methods are introduced which have been used to selectively synthesise both charge-separated and distonic radical ions.

1.3.5.1 Fragmentation of even-electron precursors

Beginning in 1990, work by the Kenttämaa group centred around the structural characterisation of radical cations,73 and the use of distonic phenyl radical cations as models of phenyl radical reactivity.157,158 In 1995, Smith and Kenttämaa introduced a general method of synthesising phenyl radicals with inert charge carriers in an FT-ICR (Scheme 1.16) This involved synthesising multiply halogenated benzene derivatives which were ionised by EI and allowed to react with a nucleophile such as P(OCH3)3, OP(OCH3)3 157,159,158 SP(OCH3)3 S(CH3)3 OS(CH3)2, pyridine or isomers of phenylpyridine. The charge-tagged precursor is then subjected to CID, ejecting a halogen atom to yield a distonic radical cation.

I I N N N

EI CID

- I

I I I

Scheme 1.16: Synthesis of distonic phenyl radical cations.

More recently, the group have synthesised bi- and tri-radicals of benzene using this method in order to structurally characterise and investigate their reactivity.113,114,160,161

While not a common method, fragmentation of negatively charged, even-electron ions by high energy CID has been shown to yield radical ions (Scheme 1.17). Bowie162 demonstrated that CID of an enolate anion C (see Scheme 1.17) resulted in many even-electron fragments including alkyl radical loss from the ester (D), which may presumably rearrange to the α-carboxylate radical anion (E). Using this method, Maas and Nibbering used monocarboxylate anions of diethyl glutarate and dimethyl 3,3-

- · 163 dimethylglutarate as precursors to the acetate radical anion CO2CH2.

1.3.5.2 Photodissociation of halides

Coupling a ﬂashlamp or laser to an ion-trap mass spectrometer is a common method of investigating photodetachment, and fragmenting peptides by photodissociation RADICAL IONS 51

O O O O CID R CH3 R R R C O - CH C O C O C O H 3 H H H CD E

Scheme 1.17: CID of an enolate anion results in alkyl loss from the ester and formation of an α-carboxylate radical anion.

(PD).164,165,166,167,168,169,170,171,172 While previous work made use of the extensive fragmentation afforded by short wavelength lasers, Kenttämaa and co-workers took a more discriminatory approach, coupling a Nd:YAG laser to an FT-ICR mass spectrometer to selectively photodissociate bromine and/or iodine substituted phenyl radicals. Using this method the group has synthesised distonic radical and biradical cations by PD using 266 and 355 nm radiation (Scheme 1.18).114,173,174 Although synthesis of distonic radical ions was not a focus of their research, Julian and co-workers used this technique to selectively synthesise tyrosinyl and histidinyl radicals on proteins by 266 nm PD of iodinated tyrosine or histidine residues, after which subjection of the nascent radicals to CID resulted in peptide fragmentation.175,176,177

F F

N SORI-CAD N

or PD

Br Br F I EI N

F F I I

N PD N

Scheme 1.18: Photodissociation (266 nm) and sustained off-resonance irradiation collision activated dissociation (SORI-CAD) of iodinated and brominated pyridinium cations affords distonic radical cations, as demonstrated by Thoen et al.114

1.3.5.3 Reactions with atomic oxygen radical anion

Atomic oxygen radical anion O–· has been used in many experiments to synthesise radical anions. In general, it is generated by EI of N2O. The available literature has been extensively reviewed by Lee and Grabowski.178 The most signiﬁcant reaction of 52 INTRODUCTION

+· this ion in the context of charge-spin separated radical anions is H2 abstraction. For –· +· example, the reaction of O with benzene results in 1,2-H2 abstraction and formation 136 of the o-benzyne radical anion and H2O (shown in Scheme 1.19). Nibbering and co- workers have also made use of O–· to synthesise a number of radical anions including − . . – 179,180,181 CH2SCHCN and CH2CO2.

+ O 66% - H2O

O + O 34% - H

Scheme 1.19: Reaction of atomic oxygen radical anion with benzene to form o-benzyne.

A disadvantage of this method is that many side-reactions may occur during reaction of O–·, including addition of O–· and loss of H· or formation of isobaric ions

+· –· due to competing H2 abstractions. The reaction of 1,1,1-trideuteroacetone with O described in Section 1.3.3 for example, was shown to result in a mixed ion population,

+· +· with 44% of products formed from 1,1-H2 abstraction and 56% from 1,3-H hydrogen abstraction. In the unlabelled case, separation of the two isomers would not be possible.

+· Charge and radical separation only occurs when H2 abstractions occurs from separated functional groups (i.e., not during 1,1-H+· abstraction). It is also pertinent to note that many of these species do not fall under the more restrictive deﬁnition of distonic due to delocalisation of either radical or charge, or cross-conjugation. If required, this may be overcome by introducing a functional group like BF3 using ion-molecule reactions (as described in Section 1.3.4.2) to completely localise radical and charge on separate atoms.

1.3.5.4 Fluorine induced desilylation

The first reported regiospecific synthesis of carbanions in the gas phase was demonstrated by DePuy et al.182 This technique, known as fluorine induced desilylation, results in the nucleophilic displacement of a carbonanion from an alkylsilane by fluoride anion F–). Around the same time, the Bowie group found hydroxide (HO–) and alkoxides (RO–) to react similarly with alkylsilanes.183 While charge-separated radical anions were never the focus of DePuy’s research, an investigation into formation of the acetyl anion RADICAL IONS 53

by ﬂuorine induced desilylation resulted in the formation of the acetate radical anion

(shown in Scheme 1.20). The acetyl anion was shown to react with O2 by O atom transfer – · to form CH3CO2 and, in addition, by a concerted O atom transfer and loss of HO to yield . – 184 the acetate radical anion CH2CO2.

O F + O2 C OC CH3 O2C CH3 + O (H3C)3Si CH3 - (H3C)3SiF + O2 O2C CH2 + HO

Scheme 1.20: Desilylation of acetyltrimethylsilane resulting in formation of the acetyl anion and its subsequent - · 184 reaction with O2 generating acetate radical anion ( CO2CH2).

The double desilylation method as introduced by Wenthold, Hu and Squires (shown in Scheme 1.21), is a variation of the desilylation method in which a bis(trimethylsilyl) functionalised molecule is first subjected to desilylation by fluoride ion followed by subsequent reaction of the nascent trimethylsilylcarbanion with molecular fluorine. This reaction proceeds by electron transfer to the fluorine molecule via an ion-molecule collision complex followed by a second desilylation ultimately resulting in formation of p-benzyne, a distonic radical anion.156

SiMe3 SiMe3 SiMe3 - F Me3SiF F2 F2 F2 - F

- Me3SiF

SiMe3

Scheme 1.21: Fluoride induced desilylation of 1,4-bis(trimethylsilyl)benzene resulting in formation of p- benzyne.156

Wenthold, Hu and Squires have made extensive use of the double desilylation method to synthesise a number of distonic radical anions including m- and p-benzyne (for example, p-benzyne, Scheme 1.21) and acetate radical anion.156,153 In these exper- · iments they characterised the ions by their reaction with CS2, NO and CO2, the latter demonstrating the formation of distonic m- and p-carboxylatophenyl radical anions by ion-molecule reaction. These secondary distonic radical anions were allowed to react

· 156,153 further with NO2 to form m- and p-nitrobenzoate. In later work, the Squires group extended this method to synthesise distonic biradical anions.112 54 INTRODUCTION

1.3.5.5 Oxidative decarboxylation of dianions

Kass and co-workers have developed a method for the synthesis of radical anions by way of oxidative decarboxylation of dicarboxylate dianions using ESI-FTMS (shown in Scheme 1.22).185,130,186,187,188 The method was generalised after several investigations in the literature, wherein CID of dicarboxylate dianions led not only to the expected decarboxylation, but concomitant electron loss, resulting in formation of a radical anion. This was proposed to occur due to a lowering of the electron afﬁnity of the nascent anion formed after CO2 loss. This resulted in an increased Coulombic repulsion as the distance between the two charge centres decreased, ultimately leading to electron ejection and formation of a radical anion.130,135,189

CO H CO CO 2 ESI 2 CID 2 CID

-2H+ CO2H CO2

Scheme 1.22: Oxidative decarboxylation of dicarboxylates generated via ESI-MS.

Blanksby and co-workers have utilised this method to synthesise the distonic bridgehead radical, 3-carboxylatoadamantyl radical anion, from 1,3-adamantane dicarboxylic acid (as described later, in Chapter 2, Scheme 2.5). This method allowed them to investigate the reactivity of a strained carbon-centred radical.154,155

Prior to this general method, the gas phase synthesis of radical anions was limited to reactions of atomic oxygen anion O–· and the double desilylation method, both of which involve signiﬁcant preparation, and result in a number of interfering byproducts.130 In contrast, dicarboxylic acids in many cases are readily obtained or synthesised. ESI is able to generate dianions without the use or generation of small reactive and interfering species such as O–· and F–. Finally, one is not reliant or limited by the volatility of the precursor.

1.4 Distonic radical ions as models for neutral reactions

While distonic radical ions were once a hypothesis and curiosity to describe unusual products observed during ion-molecule reactions undertaken using mass spectrometry, they have now further evolved into a useful tool to model neutral radical reactivity. In addition, their intermediacy has also been demonstrated in both solution and gas phase DISTONIC RADICAL IONS AS MODELS FOR NEUTRAL REACTIONS 55

reactions. One example is the fragmentation of alkylformate esters, which have been shown to undergo stepwise McLafferty rearrangements (shown in Scheme 1.23) to form a distonic radical cation intermediate137,138,123 in both the gas phase (demonstrated by mass spectrometry)190 and solution phase (demonstrated by matrix isolation and ESR).191

H H H H H CH3 O CH O CH O 3 3 C3H6 + HCOOH H O H O H O

Scheme 1.23: McLafferty rearrangement of the propyl formate ester resulting in formation of a distonic radical cation intermediate before fragmentation. Taken from Zha et al.190

Many research groups have contributed to ongoing research of distonic radical ions; of significance certainly is the seminal work of Guo and Grabowski who suggested distonic radical ions may be used as analogues of their neutral counterparts.136 The Kenttämaa research group have and continue to provide many significant contributions to this area of research. "Charge-Site Effects on the Radical Reactivity of Distonic Ions",148 was a timely investigation into the effects of the charge-tag on the reaction of the radical moiety of a distonic phenyl radical ion. There were two major findings of this study: i) the charge-tag has a significant effect on the second order rate of reaction due to perturbation of the electrophilicity or nucleophilicity of the radical moiety, and ii) the charge-tag did not affect how a radical would react with a neutral molecule, for example, whether it would undergo hydrogen atom abstraction, therefore, distonic radical ions make a good qualitative model of radical reactivity.

With this understanding, the Kenttämaa group have continued to use distonic radical cations as models of neutral reactivity. It was shown that ion-molecule reactions occurred independently of the charge-tag. For example, a model radical would abstract hydrogen from a given hydrocarbon whether a positively or negatively charged moiety was employed (shown above in Figure 1.22a).148 By tailoring the charge-tag and substituents one can study the effect of differing the partial charge at the radical moiety. The group have also investigated the reactivity of mono-, bi- and triradicals towards peptides and nucleotides.115,116,192,193,111 They found the gas phase reactivity of their model 3-pyridiniumphenyl radical cation to agree favourably with the results obtained in the solution phase, where reaction of 4-carboxyphenyl radicals with glycine resulted 56 INTRODUCTION

in exclusive hydrogen atom abstraction (illustrated in Scheme 1.24).193,194

O H O N H2N OH H2N O N N N N H (a) SORI-CAD OH O + + O - I H N I H H2N OH O

CO2H O CO2H H2N O OH H2N (b) + OH hv NaOH, H2O

I H

Scheme 1.24: Experiments by (a) Li et al.193 and (b) Braslau and Anderson194 which demonstrate that model phenyl radicals undergo exclusive hydrogen atom abstraction with glycine in both (a) the gas phase and (b) solution phase.

During their earlier work investigating ion-molecule reactions of distonic phenyl radical cations, Kenttmaa¨ and co-workers158 reported that increasing the electron deficiency at the radical site increased the reaction efficiency. This was not particularly surprising, as the electrophilicity of the radical moiety as well as the stability of the products are well known to influence the rate of radical reactions.8 Recent publications by the group include quantification of this effect on the efficiency of H-abstraction reactions, and indeed, they have shown that it is possible to predict reaction efficiency trends between a positive and negative charge-tagged phenyl radical ion reacting with various neutral reagents (displayed in Figure 1.22a) based on ionic curve-crossing diagrams which model polar contributions to the transition state (illustrated in Figure 1.22b).148,192 These findings demonstrated conclusively that the charged moiety has an influence on reaction efficiency.

While care must be taken in choosing appropriately located charge-tags which structurally and electronically separate the charged and radical moieties, distonic radical ions appear to be a convenient method of synthesising highly reactive radical intermediates free of interferences. This is certainly apparent when comparing this method to, for example, reaction ﬂow tube experiments where radical-radical recombination and radical-precursor reactions may occur resulting in many interfering side- products and complicating overall analysis. The charge on the functional group of each THESIS OVERVIEW 57

(a) (b) Figure 1.22: (a) Comparison of reaction efficiency to IE-EA for a number of neutral reactions.148 (1) represents the 3-pyridiniumphenyl radical cation, while (2) represents the 3-carboxylatophenyl radical anion. (b) Ionic curve-crossing model employed by Petzold et al. to rationalise the difference in reaction efficiencies displayed in (a).148 model radical stops radical-radical recombination, due to Coulombic repulsion, which simplifies the study of radical reactions. The charge on these radical ions, however, alters the electrophilicity or nucleophilicity of the radical site, significantly perturbing the kinetics of these reactions. This means that comparison of reaction rates between charge-tagged systems must be considered in context, and in general, rates measured for distonic radical ion systems are not amenable to comparison with neutral radicals. As discussed above, however, the distonic nature of these radical ions results in a charge- tagged carbon-centred radical whose reactivity towards various neutral reactions has been shown to model its neutral counterpart. As such, distonic radical ions should be extremely useful in elucidating the mechanisms of radical reactions by stepwise analysis of radical-neutral reactions.

1.5 Thesis overview

The following four chapters will introduce and detail the synthesis and utilisation of distonic radical ions to explore the reactivity of phenyl and cyclohexyl radicals.

• Chapter 2 is split into two sections which describe the modiﬁcations made to a commercial linear ion-trap to allow a) addition of neutral reagents into the buffer 58 INTRODUCTION

gas of the mass spectrometer for observation of ion-molecule reactions, and b) to allow laser irradiation of trapped ions. In addition, the chapter details two proof of concept experiments which introduce the utility of these modiﬁcations to the synthesis and study of distonic radical ions.

• Chapter 3 introduces aromatic combustion, and includes an ion-molecule and computational study of the oxidation of the phenyl radical by dioxygen using a distonic radical ion model.

• Chapter 4 takes model phenylperoxyl radicals synthesised in the previous chapter and provides an investigation into the reaction of these radicals with nitric oxide and nitrogen dioxide, important reactions during the production of tropospheric pollution.

• Chapter 5 introduces the use of distonic radical anions to study the aliphatic cyclohexyl radical, and similarly to Chapter 3, describes an ion-molecule and computational study of the reaction of a model cyclohexyl with dioxygen. nsbeun hpeso hsthesis. described this experiments of chapters the subsequent and in instrument modified this of that application experiments concept the of introduce proof details chapter this flow addition, In gas reactions. neutral molecule a of delivery the ( allow to inlet gas ( buffer the primary of the precursors; modification photolabile with other (ii) and ions halide from mass-selected ions radical of distonic irradiation synthesising plate of laser posterior goal allow the to to window instrument quartz a the of of installation the ion- (i) linear LTQ spectrometer: ThermoFisher mass a trap to modifications substantive two details chapter This 2.1 I e.g. e.g. NSTRUMENTATION ae,dmty iufieadallidd)t aiiaeteivsiaino ion- of investigation the facilitate to iodide) reagents allyl liquid and point disulfide boiling dimethyl low water, and , dioxide) nitrogen and oxide nitric oxygen, , Introduction OMRCLLNA ION LINEAR COMMERICAL M : 59 DFCTOST A TO ODIFICATIONS SPECTROMETER - RPMASS TRAP

C HAPTER 2 . 60 INSTRUMENTATION

Figure 2.1: (a) Gas phase UV absorption spectrum of iodobenzene and pentaﬂuoroiodobenzene. (b) UV absorption spectra of iodocyclohexane and iodoadamantane in n-heptane. Spectra taken from Kavita et al.196

2.2 Photolysis

The photolysis of iodinated alkanes and arenes is a convenient method of producing neutral carbon-centred radicals in the gas and solution phases.195,196 UV light is often sufﬁcient to promote an electron to a pre-dissociative electronic state resulting in formation of an alkyl or aryl radical and atomic iodine.197 As shown in Figure 2.1, iodoadamantane, iodocyclohexane and iodobenzene all show absorption features in the 200 to 300 nm wavelength range.196,198,199

There are numerous research groups coupling lasers to ion-trap mass spectrometers as such instruments focus ions in a fixed region of space, permitting good overlap between ions and laser radiation. Free electron, excimer and CO2 lasers have been coupled to FT-ICR mass spectrometers to dissociate peptides.169,170,171,200 Commer- cial 3D quadrupole ion-traps have also been modified to allow the introduction of UV and IR wavelength lasers for similar applications, for example, infrared multiple photon dissociation (IRMPD)201,202,203,204 and femtosecond laser induced dissociation (fs-LID).205,206 In these examples, relatively non-selective fragmentation is realised by multiphoton absorption and strong-field ionisation.

The Kenttämaa group extended the laser/ion-trap methodology in a more selective manner to synthesise distonic radical ions from halide precursors.114 The fourth harmonic (266 nm) of a Nd:YAG laser was sufﬁcient to photolyse chloro-, bromo- and PHOTOLYSIS 61

iodobenzene species to yield phenyl radicals. The relative efficiency of photolysis was − − found to follow the bond dissociation energy of the C6H5 X bond, (DH298(C I) = 66.7 −1 207,208 −1 −1 208 kcal mol ; DH298(C−Br) ≈ 81 kcal mol ; DH298(C−Cl) ≈ 96 kcal mol ), with the weakest bond dissociating in the highest yields, in some cases with an efficiency as high as 66%. As shown in Figure 2.2, this report demonstrated that photolysis and SORI- CAD of N-(3,5-dibromophenyl)-3-fluoropyridinium ion resulted in different ion populations. The population of m/z 172 ions generated by PD and SORI-CAD were found to react quite differently towards tert-butylisocyanide. Allowing the m/z 172 ion generated by photolysis to react with tert-butylisocyanide resulted in the expected linear pseudo- first order decay. The linear decay suggested generation of a single isomer, assigned as N-(3,5-dehydrophenyl)-3-fluoropyridinium ion. In contrast, when generated by SORI- CAD the m/z 172 ion reacted with a lower efficiency, and no product was formed at m/z 198. Furthermore, a non-linear psuedo-first order decay profile measured in this case suggested a mixture of isomeric ions, probably due to rearrangement via intramolecular hydrogen atom migration.

The laser used in this experiment could not be centred through the ion cloud. This meant multiple laser pulses were required to generate a signiﬁcant ion population and raises the possibility of multi-photon processes resulting in product ions being lost or further fragmented by subsequent laser pulses. This suggested that the photolysis efﬁciency could be further improved by utilisation of an instrument with a clear line of sight through the ion storage region.

Figure 2.2: Temporal variation and curvature of the m/z 172 reactant ion in the pseudo-ﬁrst order decay curve obtained during reaction of tert-butyl isocyanide with the N-(3,5-dehydrophenyl)-3-ﬂuoropyridinium ion (m/z 172) produced by from the brominated precursor using (a) SORI-CAD and (b) 266 nm PD demonstrating that the m/z 172 ion population contained contributions from less reactive isomeric species. Taken from Thoen et al..114

Recently, Ly and Julian coupled the 266 nm fourth harmonic generated from the 62 INSTRUMENTATION

output of a Nd:YAG laser to a ThermoFisher LTQ linear ion-trap (LIT) mass spectrometer. Proteins and peptides with iodine modified histidine and tyrosine residues were subjected to irradiation, resulting in exclusive I· loss.175,176 Subsequent CID of the resulting radical ion resulted in residue specific radical directed fragmentation. As demonstrated in Figure 2.3, photodissociation was selective for the modified protein (in this case cyctochrome c); when the unmodified protein was subjected to laser irradiation there were no products. This suggests that either the C-I bond is directly excited at 266 nm or the phenyl ring of tyrosine absorbs the incident energy after which internal vibrational redistribution results in dissociation of the C-I bond. In these experiments efficiencies of up to 75% were achieved from a single laser pulse, whereas CID of the same precursor ion resulted in exclusive even-electron peptide fragmentation.

Figure 2.3: (a) 266 nm PD spectrum for the +10 charge state of iodo-Cytochrome c demonstrating exclusive loss of iodine atom.175 (b) 266 nm PD spectrum for the +10 charge state of Cytochrome c showing no fragmentation without the presence of an iodinated residue. Taken from Ly and Julian.175

This high efficiency plausibly results from strong absorption cross-sections and good overlap between the laser pulse and the LTQ ion cloud. Blake et al. estimated the diameter of the ion cloud in the ThermoFisher LTQ LIT during protein soft landing experiments. In these experiments, peptides and proteins were purified by mass- selection and subsequently ejected through the 2 mm hole in the back lens of the ion- trap assembly onto a surface. The authors estimated the ion cloud isolated within the trap to be up to 4 mm in diameter by observation of the residue present on the inside of the back lens (a photograph of the equivalent assembly from our own instrumentation is shown later in Figure 2.5c).172,209 The orifice in the back lens through which the laser was introduced during these experiments restricted the laser to a beam diameter of 2 mm. A worst case overlap between the ion cloud and laser pulse in which the ions are evenly distributed in a 4 mm diameter cylinder is therefore 25%. The high efficiencies PHOTOLYSIS 63

of >75%175,176 in initial reports and recently reported efﬁciencies of >97%210 by Ly and Julian suggest that the ion cloud is signiﬁcantly more radially constrained.

These examples clearly demonstrate the use of laser photolysis as an efﬁcient method of selectively synthesising carbon-centred radicals from iodine-substituted organic ions within an ion-trap. Many halogen derived molecules are available from commercial manufacturers. Furthermore, iodine derivatives, highly susceptible to photolysis at 266 nm, are readily synthesised by the halo-dehydroxylation of alcohols by hydrogen iodide (HI)155,211 or N-iodosuccinimide and triphenylphosphine,212 or the halogen substitution reaction of Finkelstein using sodium or potassium iodide.213 As such, laser photolysis may be used to expand the range of distonic radical ion precursors.

There are many photolabile precursors which may be used in addition to halogen derivatives. Barton esters are commonly employed to synthesise carbon-centred radicals in the solution phase.214,8 The N-O ester linkage is relatively weak, and indeed, has been shown to dissociate under heat or UV radiation in solution.214,8 A typical solution phase UV spectrum of a model Barton Ester is shown in Figure 2.4, with strong absorption bands below 400 nm accounting for the susceptibility to photolysis at short visible and UV wavelengths.215,216,217

Figure 2.4: UV absorption spectrum of 2-thioxopyridin-1(2H)-yl isobutyrate in acetonitrile. Taken from Dietlin et al.216

Barton esters have also been coupled with CID in the gas phase154,155 to produce carbon-centred phenyl and bridgehead radicals (Scheme 2.1). Similarly, nitrate esters of · the amino acid serine and serine residues on peptides have been shown to lose NO2 and 218 CH2O on irradiation in the solution phase to generate regiospeciﬁc peptide radicals. By analogy, in the gas phase, ionised and mass-selected serine nitrate esters have been subjected to CID, producing a similar suite of peptide radicals (Scheme 2.2).219 These derivatives establish a wide range of putative precursors to regioselectively synthesise 64 INSTRUMENTATION

carbon-centred radical ions in the gas phase using photolysis.

S O O O O N CID CID S - CO2 - N CO2 CO2 CO2

Scheme 2.1: CID of 3-(N-oxycarbonyl-2(1H)-pyridinethione)adamantanecarboxylate anion. Adapted from Harman and Blanksby.154

2.2.1 Modiﬁcations for laser photolysis

In this section, the modifications to a ThermoFisher LTQ LIT are described. These modifications are similar to those undertaken by Ly and Julian175,176 to the same model mass spectrometer. This instrument is well suited to this type of modification as it has an aluminium plate located directly behind the ion-trap which is easily removed for coupling of a secondary mass analyser such as an FT-ICR analyser or Orbitrap.95,220,221 Removal of this plate exposes the ion-trap assembly with a 2 mm orifice positioned centrally in the back lens (see Figure 2.5c).

For the purpose of experiments reported in this thesis, the aluminium back plate (a spare generously provided by ThermoFisher Scientific) has been modified to facilitate the introduction of a laser pulse through the orifice in the back lens (a schematic of the modifications is shown in Figure 2.6). As shown in the photographs displayed in Figure 2.5, this was accomplished by milling a 10 mm hole in the back plate aligned with the back lens orifice, installing a 2.75 in. quartz viewport with a CF flange (Part No. 9722005, MDC Vacuum Products, LLC, Hayward, CA) and sealing with a Viton® gasket (Part No. 9912220, MDC Vacuum Products, LLC, Hayward, CA) The quartz window is specified for wavelengths from 200 nm to 2.0 µm with 90% transmission efficiency,222 and is therefore suitable for the 266, 355, 532 and 1064 nm wavelengths produced by the Nd:YAG laser (Continuum Minilite™ II) used in these experiments.

The diameter of the laser beam used in these experiments is larger than the 2 mm back plate oriﬁce, therefore the laser pulse passing through this oriﬁce is restricted to a 2 mm beam diameter. Assuming that the ion cloud is evenly distributed suggests that at a minimum ~25% of ions are affected by each pulse of the laser. As detailed PHOTOLYSIS 65

Figure 2.5: a) Outside of modiﬁed back plate. b) Inside of modiﬁed back plate. c) Back plate removed from ThermoFisher LTQ exposing the rear of the back lens and ion-trap assembly. 66 INSTRUMENTATION

7.880

7.085 0.275 Ø 0.450

Ø 0.250

2.175 120° R 1.156 Ø 1.902

Ø 1.654

4.925 Ø 0.250

Ø 0.394

0.853 0.500

0.590 5.500 4.125

6.174 0.375

Figure 2.6: Schematic of the ThermoFisher LTQ back plate modiﬁcations. Units are inches. Adapted from a design supplied courtesy of Prof. Ryan Julian (University of California, Riverside, USA)

. PHOTOLYSIS 67

Figure 2.7: a) Side view of laser setup showing laser (L) and prisms (P1 and P2) in relation to the modiﬁed back plate (M). b) Top view of laser setup showing laser and prisms. c) Digital PCB board and trigger output pin location. 68 INSTRUMENTATION

Figure 2.8: a) Trigger window and b) MS/MS dialog in the LTQTune software (within the Xcalibur suite). PHOTOLYSIS 69

NO2 O O hv H - NO H - CH O H N 2 N 2 N O O O

. Scheme 2.2: Photolysis of O-nitrososerine residue, adapted from Easton et al.218 and Wee et al.219

Activation time Signal (V)

TTL (low) Gate (low)

Gated single pulse mode Time (ms) Signal (V)

TTL (low) Gate (low)

Gated continuous pulse mode Time (ms)

Figure 2.9: Gated output from instrument, and TTL pulses generated by the signal generator which triggers the Q-switch of the laser. above (see Section 2.2), efficiencies of up to 75% have been realised, so the overlap between laser and ion cloud may in fact be much greater. For example, an efficiency of 75% (assuming a C-I dissociation quantum efficiency of 1.0 for the target) for example equates to an ion cloud diameter closer to 3.5 mm. Blake et al. found that the diameter of the ion cloud was affected by the q-parameter of the ion-trap,209 but experiments in our laboratory (data not shown) found that changing the q-parameter resulted in no significant increase or decrease in PD efficiency suggesting that any change in the overlap of the laser with the ion cloud is minimal.

The laser is aligned through the LTQ by adjusting the two prisms P1 and P2 (shown in Figures 2.7a and b). A schematic of the ion source optics in the LTQ is shown in Figure 2.10. The front ion optics are removed and alignment is complete when the laser pulse passes unhindered through the instrument and observed on a card. When the front optics section is in place, the beam hits the skimmer; there is no beam dump, and over time this leads to observable discolouration on the skimmer. 70 INSTRUMENTATION

Figure 2.10: Exploded schematic of the ion source in a ThermoFisher LTQ Mass Spectrometer.95

In order to trigger the laser, the trigger output from the mass spectrometer must be coupled to the flashlamp trigger input of the laser controller via a signal delay generator which amplifies the trigger pulse sent from the instrument. The trigger output pin on the Digital PCB of the mass spectrometer (shown in Figure 2.7c) is connected to a female BNC socket which is installed on the side of the mass spectrometer housing. The trigger output pin provides a ±3.2 V (signal high) gate signal to a signal delay generator (Quantum Composers Model 9514+) that generates the 5 V TTL pulse to trigger the flashlamp of the laser. The Q-switch of the laser is internally triggered with the Q- switch delay optimised for maximum pulse energy. The signal generator has been programmed for two modes of operation, (i) single trigger per activation and (ii) trigger every 100 ms of the activation (multishot; Figure 2.9). The trigger is set in the Xcalibur software (Figure 2.8a) to fire at the start of the activation sequence of a scan and the gate signal remains high for the duration of that activation.

2.2.2 Synthesis of radical ions by laser photolysis

The primary goal of coupling a pulsed laser to the LTQ was to provide an additional means of generating distonic radical ions within the ion trap. In the next two sections, negative and positive ions with iodo and Barton ester moieties are surveyed for their PHOTOLYSIS 71

suitability as precursors for regioselective generation of distonic radical ions by laser photolysis. Furthermore, a comparison is made between subjecting the precursors to photolysis and collision induced dissociation.

2.2.2.1 Distonic radical cations

In this section, photolysis yield (ΦM ) is defined in Equation 2.1 where [M] represents the ion abundance of a m/z M photolysis product and [TIC]0 is the total ion abundance of the precursor prior to irradiation. Conversion efficiency (ε) is defined as shown in

Equation 2.2 where [TIC]0 represents the total ion abundance of the precursor prior to dissociation (either by irradiation or CID) and [TIC]1 represents the total ion abundance of the product ion spectrum after dissociation. ΦM , therefore, represents the yield of a product (of m/z M) compared with the initial abundance of the precursor and ε describes ion retention. A larger ε is more desirable as it signiﬁes a greater number of ions (products and non-fragmented precursor ions) are retained in the ion trap after dissociation. It should be noted, however, that the measured ion abundances are absolute counts and not measured relative to an internal standard so any ﬂuctuations in the concentration of the isolated precursor may lead to errors in the measured ion abundance.

[M] ΦM = × 100% (2.1) [TIC]0

[TIC]1 ε = × 100% (2.2) [TIC]0

As described in Scheme 2.3(i), infusing a solution of the Barton ester N,N,N- trimethyl-4-((2-thioxopyridin-N-yloxy)carbonyl)benzeneaminium iodide (1), in + MeOH/H2O via positive electrospray resulted in an M cation at m/z 276 assigned as the Barton ester N,N,N-trimethyl-4-((2-thioxopyridin-N-yloxy)carbonyl)benzeneaminium cation (Scheme 2.3i, A). When A is isolated and subjected to a single laser pulse at 266 nm (2.4 mJ/pulse), the ion undergoes N-O bond homolysis to lose 2-thiopyridin-N-yl radical, after which the nascent carbonyloxyl radical ejects CO2 to generate an [M - CO2 +. -C5H4NS] radical cation at m/z 135 assigned as the 4-trimethylammoniumphenyl radical cation B (Figure 2.11a). The Barton ester derivative in this case is susceptible to PD with photolysis proceeding with Φ135 = 21% and ε = 44%. Interestingly, CID of 72 INSTRUMENTATION

NMe3I NMe3 NMe3

ESI m/z 289 PD (i) - -I - CO2

- C5H3SN N N B O O O O m/z 135 1 S A S

NMe3 I NMe3 NMe3

ESI PD (ii) -I- -I I I C D 2 m/z 276 m/z 149

NMe3 I NMe3 NMe3

ESI PD (iii) -I- -I

E F 3 I m/z 276 I m/z 149

Scheme 2.3: Electrospray ionisation of (i) N,N,N-trimethyl-4-((2-thioxopyridin-N- yloxy)carbonyl)benzeneaminium iodode (1), (ii) N-(3-iodo-4-methylphenyl)trimethylammonium iodide (2) and (iii) 4-(iodomethyl)-N,N,N-trimethylbenzeneaminium iodide results in cations A, C and E which are subjected to photolysis at 266 nm resulting in formation of a carbon-centred radical.

A also results in formation of B, but with a lower efficiency of ε = 15% due either to formation of ions lower in mass than the low mass cutoff, or a low trapping efficiency of the product ions which could conceivably occur if the parent or fragment ions gain sufficient energy during CID to be ejected from the ion trap.

Infusing a solution of N-(3-iodo-4-methylphenyl)trimethylammonium iodide + (Scheme 2.3ii, 2) in MeOH via positive electrospray results in an M ion at m/z 276 assigned as N-(3-iodo-4-methylphenyl)trimethylammonium cation (C). When C is isolated and irradiated by a 266 nm laser pulse (2.4 mJ/pulse) the [M - I]+. radical cation (D) is generated (Figure 2.11b, Scheme 2.3ii) with Φ149 = 15%, Φ134 = 1% and ε = 96%. Sim- ilarly, irradiation of 4-(iodomethyl)-N,N,N-trimethylbenzaminium cation (E) results in + loss of I· to form a [M - I] . ion assigned as the 4-N,N,N-trimethylammoniumbenzyl radical cation (F) (Figure 2.11d, Scheme 2.3iii) that proceeds with Φ149 = 7%, Φ134 = 6% and ε = 85%. While PD of the isomeric ions C and E results in a similar product ion abundance (Φ149 + Φ134), PD of the benzyl iodide (E) resulted in a higher abundance of · the m/z 134 side-product which is presumably generated by CH3 loss from the nascent PHOTOLYSIS 73

+. benzyl radical at m/z 149. Subjecting C to CID resulted in a major ion at [M - CH3] at +. + m/z 261 and minor [M - I] and [M - I - CH3] ions at m/z 149 and m/z 134, respectively (Figure 2.11c).

The different relative abundances of the m/z 134 to m/z 149 products generated during PD of C and E may be rationalised by comparing relative C−I bond dissociation energies of benzyl and phenyl analogues. A 266 nm photon delivers 107.6 −1 ∼ −1 − 223 kcal mol of energy to the precursor, of which 45 kcal mol (C6H5CH2 I) is used to overcome the C−I bond energy. Therefore, ∼62 kcal mol−1 of excess energy is distributed between the departing iodine atom and the benzyl radical. This +− is sufﬁcient energy to subsequently overcome the (CH3)2N CH3 bond dissociation +− ≈ −1 energy (DH298((CH3)2(C6H4CH2)N CH3) 57 kcal mol , see Appendix B.2). In − 208 ∼ −1 ∼ contrast, the bond dissociation energy of C6H5 I is 65.4 kcal mol leaving only 42 kcal mol−1 distributed between the phenyl radical analogue and the departing iodine · atom. This leaves the trimethylammonium cation with insufﬁcient energy to eject CH3 +− ≈ −1 (DH298((CH3)2(C6H4)N CH3) 47 kcal mol , see Appendix B.2) unless the ion has gained internal energy by other means such as energetic collisions. This may therefore

+. explain the lower yield of the [M - I - CH3] in the case of phenyl iodide analogues such as C.

In order to characterise the structure of the [M - I]+. ions generated by the two different modes of activation, their reactivity towards O2 was investigated. When the * +. PD product D is isolated in the presence of adventitious O2, , product ions [M + O2] + (+32 Da) and [M + O2 - OH] (+15 Da) ions at m/z 181 and m/z 164, respectively, are generated (Figure 2.12). The reaction of D with O2 is analogous to the reaction of the tyrosinyl radical investigated by Moore et al..224 As shown in Scheme 2.4, PD of 3-iodotyrosine results in formation of a σ-radical located adjacent to the hydroxyl moiety of the tyrosine. Isolation of this radical in the presence of O2 results in O2 addition and elimination of HO· facilitated by the hydroxyl group. By analogy, the assignment of the m/z 164 from reaction of D with O2 is N,N,N-trimethyl-4-methylenyl- 3-oxybenzaminium cation, as depicted in Figure 2.12.

In contrast, isolation of the [M - I]+. ion generated by CID of D for 10 seconds under + the same conditions resulted in only a minor [M + O2 - OH] (+15 Da) ion at m/z 164 9 −3 155 *Oxygen present in the ion-trap. Estimated at [O2] = 3 x 10 molecules cm . Due primarly to transmission from the ionisation source, which is under atmospheric pressure. 74 INSTRUMENTATION

Figure 2.11: a) Isolation and 266 nm PD of N,N,N-trimethyl-4-((2-thioxopyridin-N- yloxy)carbonyl)benzeneaminium cation. Isolation and b) 266 nm PD and c) CID of 3-iodo- N,N,N,4-tetramethylbenzeneaminium cation. Isolation and 266 nm PD of d) 4-iodomethyl-N,N,N- trimethylbenzeneaminium cation. PHOTOLYSIS 75

NMe3 NMe3 NMe3 NMe3 NMe3

PD +O2 -I - HO I O O O C O OH m/z 276 [M - I] [M + O2] H [M + O2 - HO] m/z 149 m/z 181 m/z 164 D

Figure 2.12: Isolation of the m/z 149 fragment ion generated by PD (266 nm) of 3-iodo-N,N,N,4- tetramethylbenzeneaminium cation in the presence of O2.

(Figure 2.13), suggesting that while the two products are isobaric and generated from the same precursor they contain structurally distinct ion populations. While it could appear that a minor proportion of the [M - I]+. ion population generated by CID of C is D, it seems logical that a significant contribution may arise by isomerisation to the resonance stabilised benzyl radical (F) as it is similarly unreactive towards O2 (Figure 2.14). However, F is not the only possible structure of this ion. Three putative mechanisms (a), (b), and (c) are shown in Figure 2.13. The first mechanism (iiia) involves direct 1,4-hydrogen atom transfer, the second (iiib) an assisted hydrogen transfer by a nascent iodine atom, with the final (iiic) iodine atom loss resulting in formation of a · vibrationally excited complex in which CH3 migrates from the amine to the phenyl ring.

PD and CID of C were found to result in signiﬁcantly different decomposition products. PD of C resulted primarily in loss of iodine (see Figure 2.11b) while CID of · C resulted in a major loss of CH3 (see Figure 2.11(c)). 266 nm irradiation provides 107.5 kcal mol−1 of energy while CID in an ion trap is believed to occur due to a collision- mediated gradual heating of the isolated ion.92 While the energy imparted by photolysis is greater, this result can be explained by considering that PD involves an electronic excitation while CID is a gradual vibrational excitation process. PD may result in direct 76 INSTRUMENTATION

NMe3

iiia) - I [M - I] +O m/z 149 2 CID H

NMe3 NMe3 NMe3 [M - I] iiib) m/z 149 F CID I C I H m/z 276 H I iiic)

CID NMe3 NMe2 - I [M - I] m/z 149

Figure 2.13: Isolation of the m/z 149 fragment ion generated by CID of 3-iodo-N,N,N,4- tetramethylbenzeneaminium cation in the presence of O2. excitation of the ion to a pre-dissociative electronic state, in which case dissociation of the C-I bond occurs before extensive energy redistribution. Alternatively, PD may involve absorption of incident energy by the phenyl ring followed by subsequent energy redistribution into vibrational degrees of freedom, including the C-I bond.175,176,210 Considering PD of C in this context, the energy absorbed by the aromatic ring and redistributed into vibrational degrees of freedom is sufﬁcient to dissociate the C-I bond, whereas during CID, the vibrational excitation process is comparatively slow and energy is spread evenly amongst the vibrational degrees of freedom. Therefore dissociation − +− ≈ −1 of the isoenergetic N CH3 bond (DH298((CH3)2(C6H4X)N CH3) 47 kcal mol , see Appendix B.2) occurs prior to the ion gaining sufﬁcient energy for C-I bond dissociation.

While a deﬁnitive structure of the [M - I]+. CID product ion was not determined in this study, this example illustrates the ability of photolysis to generate different radical PHOTOLYSIS 77

NMe3 NMe3

PD +O2 -I

E [M - I] m/z 149 m/z 276 I F

Figure 2.14: Isolation of the m/z 149 fragment ion generated PD (266 nm) of 4-iodomethyl-N,N,N- trimethylbenzeneaminium cation in the presence of O2. ions to those generated by CID, as previously reported in the literature.114 In this case, photolysis was the only method of synthesising the target radical cation D. CID resulted in generation of m/z 149 ions, the majority of which failed to react to a signiﬁcant extent in the presence of O2 (compare Figure 2.12 and Figure 2.13 where the reaction times were the same). This assignment is supported by ion-molecule reactions with O2 which compared well with an analogous PD experiment involving 3-iodotyrosine. This was a clear demonstration of employing photolysis to selectively cleave C-I bonds to generate carbon-centred radicals.

O H I H H O O O O O O

PD + O2 - I - HO OH OH OH OH H3N H3N H3N H3N O O O O

224 Scheme 2.4: Photolysis of iodotyrosine residue and its reaction with O2, Moore et al. 78 INSTRUMENTATION

2.2.2.2 Distonic radical anions

The 3-carboxylatoadamantyl radical anion (H) is a prototypical example of a distonic carbon-centred radical anion. Harman and Blanksby have previously demonstrated that H could be readily synthesised in the gas phase.154,155 A methanolic solution of 1,3-adamantanedicarboxylic acid (4) can be infused by negative electrospray ionisation to afford an [M - 2H]2− ion G. When subjected to CID, G undergoes oxidative decarboxylation to yield H (shown in Scheme 2.5).154,155 Furthermore, the reactivity of G towards

O2 has previously been characterised. As discussed above, iodinated aromatic and aliphatic compounds and Barton ester derivatives are expected to absorb in the short wavelength visible and UV. 3-(N-oxycarbonyl-2(1H)-pyridinethione)adamantane-1- carboxylic acid (5) and 3-iodoadamantanecarboxylic acid (6) were therefore logical targets for these experiments as both molecules are expected to yield H when subject to PD (shown in Figure 2.15 and Figure 2.16). In the following experiments, a comparison of the reactivity of two radical anions formed by PD of 5 and 6 towards O2 is made to those previously generated by CID of the dicarboxylic acid 4 to determine whether all the three precursors form the same radical anion product.

O CO2H CO2 O

ESI CID + O2 + -2H - CO2 - e- CO2H CO2 CO2 CO2

2- 4 [M - 2H] [M - 2H - CO2] [M + O2] m/z 111 m/z 178 m/z 210 G H

Scheme 2.5: Electrospray ionisation of 1,3-adamantanedicarboxylic acid (4) results in an [M - 2H]2− dianion which undergoes oxidative decarboxylation when subjected to CID to yield G, the 3-carboxylatoadamantyl radical anion.

− 5 was dissolved in MeOH/H2O and infused via electrospray to yield an [M - H] ion − (I) at m/z 332 (Figure 2.15). The [M - H] ion was isolated and subjected to PD at 266 nm leading to loss of the 2-thiopyridin-N-yl moiety (-110 Da) and an abundant [M - 2H

.− - CO2] ion at m/z 178. A minor product at m/z 110 was assigned as the pyridine-2(1H)- thione anion (shown in Scheme 2.6;† Figure 2.15a). CID of I led similarly to formation

† – Direct elimination of C5H4SN is unlikely given the relevant transition state would be a 7-membered ring and furthermore, the adamantane cage would hinder interaction of the carboxylate with the O- – N bond. Ejection of C5H4SN and electron transfer from either H or K given the electron afﬁnity of – C5H4SN of 2.8 eV is lower than that of most comparable alkylcarboxylate anions (see Table B.1 for more information). PHOTOLYSIS 79

.− of an [M - H - C5H4SN - CO2] ion at m/z 178 but with a greater abundance of the side product at m/z 110 (Figure 2.15b)).

− 6 was dissolved in MeOH and infused via negative electrospray to yield an [M - H] ion (J) at m/z 305 (Figure 2.16). On irradiation of the isolated ion J with a 266 nm laser − pulse, two products, an [M - H - I] . ion at m/z 178 and an ion at m/z 127 assigned as I– were formed (Figure 2.16, Figure 2.16a). CID of J resulted only in formation of I– at m/z 127 (Figure 2.16b), presumably following one of the mechanisms shown in Scheme 2.7.‡

In order to characterise the m/z 178 ions produced by each method, their reactivity with O2 was compared. The ions were isolated in the presence of O2 for the same period of time and the difference in abundance of the M− and [M + 32]− ion was measured. The m/z 178 ions formed from oxidative decarboxylation of the dianion

G, PD of the iodide J, PD and CID of the Barton ester I were allowed to react with O2 under identical conditions. As evident in Figure 2.17, the resulting spectra were found to be identical demonstrating that the m/z 178 ion generated in each case is the 3- carboxylatoadamantyl radical anion H.

While each precursor generated H by either PD, CID or both, they were found to react with varying efﬁciencies. This data is summarised in Figure 2.18. Subjecting the Barton ester derivative I to PD with a laser energy of 2.4 mJ per pulse at 266 nm resulted in H with Φ178 = 15% and ε = 58%. Yield of an m/z 110 side product was measured 3 with Φ110 = 4%. The overall ion abundance of H was measured at (1 - 5) ×10 . CID of I 4 proceeded with Φ178 = 30% and ε = 45% with an overall ion yield of (5 - 10) ×10 ions. At the same wavelength and energy, PD of the iodo precursor J proceeded with Φ178 = 5.0% – and ε = 72% for formation of H with a side-product yield of I measured at Φ127 = 2.7%. An ion abundance for H generated from this precursor was typically (1 - 10) ×105. PD of the dicarboxylate G results in a yield of Φ178 = 2% and conversion efﬁciency of ε = 11% for an ion abundance of 7.5 ×106 while CID of G typically results in a ε = 14%, for an ion abundance of H around 1.0 ×10(6−8).

The low apparent photolysis yield (Φ) afforded by photolysis of carboxylate anions at 266 nm may arise due to electron detachment in addition to a low photolysis cross- section at 266 nm.225,226,227 Wang et al., for example, determined the electron afﬁnity

‡Direct elimination of I– is not considered due to the ring-strain of the resulting adamantene cage. Ejection of I· and electron transfer from either H or K given the electron afﬁnity of I· of 3.059 eV is lower than that of most comparable alkylcarboxylate anions (see Table B.1 for more information). 80 INSTRUMENTATION

CO2 CO2 CO2

S S ESI CID/PD O O N N - CO2 - C5H3SN O O 5 [M - H] [M - H - CO2 - C5H4SN] m/z 332 m/z 178 I H

Figure 2.15: Comparitive a) PD and b) CID of 3-(N-oxycarbonyl-2(1H)-pyridinethione)adamantane-1- carboxylate anion to generate the 3-carboxylatoadamantyl radical (H, m/z 178).

CO2 CO2 CO2 CO2 CO2

S S ESI CID/PD CID/PD - O (i) O S O - CO2, -e N N - CO2 - N 5 O [M - H] O O [M - H - CO2 - C5H4NS] K m/z 332 m/z 178 I H

CO2 CO2H S

H S CID/PD N (ii) + CO + O 2 N O m/z 110 [M - H] m/z 332

O OH CO2 S

H S CID/PD N (iii) + CO + O 2 N m/z 110 [M - H] O m/z 332

Scheme 2.6: Putative mechanisms for PD and CID of 3-(N-oxycarbonyl-2(1H)-pyridinethione)adamantane-1- carboxylate anion (I). PHOTOLYSIS 81

I I

ESI PD

-2H+ -I

CO2H CO2 CO2 - 6 [M - H] [M - H - I] m/z 305 m/z 178 J H

Figure 2.16: Comparitive spectra of the a) PD and b) CID of 3-iodoadamantanecarboxylate. PD results in formation of 3-carboxylatoadamantyl radical anion (H, m/z 178) while CID results in exclusive formation of iodide.

CO2H CO2 CO2 CO2 (i) PD + I - CO , -e- I I 2

6 [M - H] [M - H - I] K m/z 305 m/z 178 J H

CO2 CO2H

H (ii) CID + I m/z 127 I [M - H] m/z 305

CO2 CO2H

H (iii) CID + I m/z 127 I [M - H] m/z 305

Scheme 2.7: Putative mechanisms for PD and CID of 3-iodoadamantanecarboxylate anion (J). 82 INSTRUMENTATION

Figure 2.17: Comparitive spectra showing the same reactivity of m/z 178 generated by a) CID of 1,3- adamantanecarboxylate, b) PD of 3-iodoadamantanecarboxylate, c) PD and d) CID of 3-(N-oxycarbonyl- 2(1H)-pyridinethione)adamantane-1-carboxylate with O2 for 1000 ms. PHOTOLYSIS 83

Figure 2.18: Plot of precursors 1,3-adamantanedicarboxylate dianion (G), 3-(N-oxycarbonyl-2(1H)- pyridinethione)adamantane-1-carboxylate anion (I) and 3-iodoadamantanecarboxylate anion (J) and dissociation technique against total ion abundance, radical yield (C) (see Equation 2.1) and conversion efﬁciency (p) (see Equation 2.2).

· of the acetyloxyl radical (CH3COO ) to be 3.250 eV which suggests that a 266 nm photon (4.667 eV) is sufficient to photodetach an electron from a carboxylate anion. Detachment of an electron from a carboxylate anion will result in a loss of charge and thus no product ion will be observed during mass spectrometry. This will, therefore, lower the conversion efficiency (ε). The product branching of any monocarboxylate halide system will therefore depend on the relative cross-sections of (i) photodetachment resulting in loss of charge and (ii) photodissociation resulting in generation of the target radical anion at a given wavelength. Dicarboxylate dianions have also been shown to lose both charges on subjection to PD.228 Near 0 eV electrons were observed in photoelectron imaging experiments of straight chain alkyl dicarboxylates that was rationalised by spontaneous secondary decarboxylation and autodetachment from the nascent carboxylatoalkyl radical anion. This certainly explains the low conversion efficiency of most dicarboxylate dianions which is far lower than typically realised during ion-trap mass spectrometry.229

Analysing the conversion efﬁciency (ε) is one method of quantifying electron detachment, however, this value will contain errors due to ions ejected before detection. For example, PD of J may result in photolytic cleavage of the C-I bond but as this photon

−1 energy (266 nm = 107.5 kcal mol ) is greater than the C-I bond energy (DH298(C4H9 - I) = 50 kcal mol−1),207,208 any excess energy will be partitioned between the alkyl 84 INSTRUMENTATION

radical and the departing iodine atom. Therefore, in addition to photodetachment, it is conceivable that ion losses could occur due to the alkyl radical gaining sufﬁcient translational energy to be ejected from the ion-trap.

Overall, CID and PD of G proceeded with conversion efficiencies (ε) of 11-14%, signifying high ion losses (as demonstrated in Figure 2.18). In contrast, CID and PD of I and PD of J resulted in a much higher conversion efficiency of 40%. While PD and CID of I resulted in relatively high photolysis yield and conversion efficiency, the highest yield of isolable ions was realised by the oxidative decarboxylation of G by CID. A high ESI and trapping efficiency afforded by use of the dicarboxylic acid 4 resulted in a significantly higher precursor ion abundance. This compensated for the low conversion efficiency usually afforded by CID of dicarboxylates and ultimately resulted in a greater abundance of the target radical anion H.

2.2.3 Summary

In this section, the modiﬁcations made to an ThermoFisher LTQ LIT to allow coupling of a Nd:YAG laser to the instrument have been described. Photolysis experiments of examples of photolabile precursors were undertaken in order to assess whether distonic radical ions could successfully be synthesised by employing 266 nm photons generated by the coupled laser. Indeed, ions containing iodide and Barton ester functional groups were found to be effective precursors to carbon-centred radicals when subjected to PD.

Relative to CID, PD results in higher energy deposited into the target ion. The absorbing moiety is both the carbon-halogen bond and any chromophores (in these cases, a phenyl ring or pyridinethione). Absorption by the C-X bond results in pro- motion of an electron into a dissociative or pre-dissociative excited state resulting in dissociation of the C-X bond and generation of the radical prior to energy redistribution within the precursor ion. Vibrational coupling between the aromatic ring and C-I bond may also contribute signiﬁcantly to the yield of radicals. While both PD and CID of C resulted in loss of I·, PD was shown to generate the target 2-methyl-5-(N,N,N- trimethylammonium)phenyl radical cation (C), while CID appears to have resulted in generation of an isobaric product with lower reactivity towards O2 (as demonstrated in Figure 2.12 and Figure 2.13).

While CID of dicarboxylic acids is a convenient method of generating distonic ION-MOLECULE REACTIONS 85

radical anions, this method is not always suitable. For example, dicarboxylic acid moieties separated by less than three carbons may not form dianions in yields high enough for isolation and generation of radical ions. This situation arises as the Coulom- bic repulsion between the two negative charges increases with decreasing separation between the two functional groups. Furthermore, dianions generally have a high proton affinity, so the presence of neutral reagents with a moderate gas phase acidity may result in proton transfer and depletion of the precursor dianion (e.g. dimethyl disulfide, which will be discussed in Chapter 5). Similarly, electron transfer may proceed in the presence of reagents with a modest electron affinity (> 1 eV). In both cases, generation of the target radical anion is hindered. The use of photolabile monocarboxylate precursors may therefore provide an alternative method of synthesising and investigating radical ions under these conditions.

2.3 Ion-molecule reactions

The use of mass spectrometry to observe the reactivity of ions with neutral molecules has a rich history with beginnings in the chemical ionisation method which, by design, involves ion-molecule reactions.230 According to Gronert in a 2001 review,231 in the previous decade over 1000 papers had been published that directly related to the study of ion-molecule reactions by mass spectrometry. Mass spectrometry has been used to study the reactions of organic anions and cations in bimolecular addition, substitution and elimination reactions among many others.231 As described in the introduction to this thesis (Chapter 1), ion-molecule reactions are intimately involved in the synthesis and characterisation of radical ions. Many custom built mass spectrometers have been constructed to undertake ion-molecule reactions. Modifying a commercial quadrupole ion-trap for use in ion-molecule studies is an alternative. In the early 1990s, McLuckey, Glish and Van Berkel94 installed a leak valve into their Fisher ITD (the ion-trap mass analyser from the ThermoFinngan MAT GC-MS) to introduce neutral reagents into the ion storage region. In this experiment charge-state determination of peptide fragments generated by CID was realised using 1,6-diaminohexane as a proton transfer reagent to remove a single charge from the mass selected fragment.94 Similar alterations have been made by Brodbelt et al. on the same model mass spectrometer to investigate the reaction of alcohols and diols with diethyl ether ions.232 86 INSTRUMENTATION

Adaptions made to the more recent ThermoFisher LCQ 3D ion-trap by the Gronert and O’Hair groups include the addition of a gas mixing manifold that allowed the introduction of reactant molecules into the helium buffer gas without directly modifying the vacuum manifold in which the ion-trap resides.233,231,234,235 Recent undertakings in our laboratory155 and the Kenttämaa laboratory236 have resulted in the modification of a ThermoFisher LTQ linear quadrupole ion-traps (LIT) for ion-molecule reactions. A LIT allows up to a 15-fold increase in the ion storage ability of a 3D ion-trap, thus ion-molecule reactions can be undertaken with greater precision and accuracy.90 The modifications described below were adapted from the modifications made by the Gronert235 and O’Hair233 groups which allow introduction of neutral reagents into the helium buffer gas of this commercial LIT.

2.3.1 Modiﬁcations for ion-molecule reactions

The addition of neutral reagents to the ion-trap mass spectrometer allows the observation of diagnostic reactions between ions (including radical ions) synthesised and isolated within the ion storage region. These reactions aid in structural elucidation, for example, the addition of O2 can be used to determine whether an isolated radical ion is distonic,150,154,155 and may also be used to probe reaction kinetics.

Minor modifications were made to a ThermoFisher LTQ mass spectrometer to facilitate addition of neutral reagents into the helium buffer gas utilised by the instrument. This involved installation of a ball-valve which controls the input of helium between normal and ion-molecule operating conditions. In normal operating mode the buffer gas passes through the native flow splitter which vents helium to maintain the required helium flow. The schematic shown in Figure 2.19 provides an overview of the modifications made to the instrument and the construction of the gas-mixing manifold. In contrast, during experiments undertaken in ion-molecule mode the helium buffer gas is supplied from a separate cylinder (H1) and passes through a gas-mixing manifold through which neutral reagents may be added. The gas mixture passes directly into the ion-trap and the required flow must be maintained by the gas mixing manifold.

All tubing used in the modification is 1/8 in. stainless steel (SS). The liquid and gas reagents are introduced into a flow of Ultra High Purity (UHP) helium provided by a regulated cylinder H1 (3-5 psi) via a septum (B, Part No. 041890, Eduro Blue Shimadzu ION-MOLECULE REACTIONS 87 Schematic of the modified ThermoFisher LTQ linear ion-trap mass spectrometer. Figure 2.19: 88 INSTRUMENTATION

Figure 2.20: Photographs of a) ball valve T2, b) positioning of ball valve T2 and splitter K, and c) splitter K with connection to helium cylinder H2. ION-MOLECULE REACTIONS 89

Septa, SGE Analytical, Australia) positioned in a union tee (Part No. SS-200-3, Swagelok, Australia). The septum section of the tubing is coiled around a solid copper block (A) machined with a groove to allow good thermal contact with the tubing. The temperature of the copper block is set and maintained by a BTC-9090 Fuzzy Logic controller (C, Brainchild Electronic Co., Ltd, Taiwan) that regulates a cartridge heater (Part No. 837- 622, RS Components, Australia) with feedback from a thermocouple (Cat No. QM1283, Wire K Type Thermocouple, Jaycar, Australia) installed in opposite ends of the copper block (an enlarged schematic shown in Figure 2.21). The tubing and copper block are wound in ﬁbreglass cloth (Part No. 09097108, Blackwards, Australia) to retain heat within the heated section. The copper block can be heated between room temperature and 250 °C and is used to raise the vapour pressure of low boiling point liquids. Heating tape is also wound around the variable leak valves and sections of tubing to further aid in the volatilisation and of reagents and to reduce condensation. A union tee (D, Part No. SS-200-3, Swagelok, Australia) splits the ﬂow of the mixture to a pressure gauge (E,

Part No. PGI-50M-PG30-LAQX, Swagelok, Australia) and a variable leak valve (V1, Model 203, Granville-Phillips, USA) metering the split-flow (the split-flow is measured by flow meter F, Model). After the heated section, a length of tubing is coiled to add path-length to ensure adequate mixing of the reagent vapour with helium. This length of tubing is connected to a second variable leak value (V2, Model 203, Granville-Phillips, USA) which meters the flow to the instrument.

Tee Septum

1/8 in. SS Tubing Tee with septum B

Copper block Cartridge heater

Figure 2.21: a) Heating block section of the mixing system, the position of which is shown in Figure 2.19 b) The septum inlet. 90 INSTRUMENTATION

The operating modes of the mass spectrometer are selected by a three-way ball valve

(T2, Part No. SS-41GXS2-1466, Swagelok, Australia) with output from the valve directed to the helium buffer gas inlet line to the ion-trap. In normal operating mode, helium is introduced from a separate helium cylinder (40 ±10 psi, H2) which is regulated by a ﬂow splitter (K) that vents helium to maintain a pressure of ~2.5 mTorr within the ion trap. In ion-molecule mode, the ﬂow is set and maintained by the aforementioned variable leak valve (V2). There are separate helium tanks and SS tubing for each operating mode to minimise any cross contamination between normal and ion-molecule modes (photographs are shown in Figure 2.20).

The addition of reagent is controlled using a Harvard Pump 11 Pico Plus (Harvard Apparatus, USA) syringe pump which has an operating range of 1.3 pL min−1 to 0.8788 mL min−1 depending on the size of the syringe used. The total flow of helium/reagent mixture introduced in ion-molecule operating mode is controlled by two variable leak valves V1 and V2. Leak valve V2 controls the flow of gas mixture to the instrument and is set to provide pressure equal to that obtained in normal operating mode, representing a flow of approximately 2.2 mL min−1, and corresponding to a typical working pressure inside the trap of ~2.5 mTorr. Leak valve V1 is a split-flow, which has the effect of increasing the total flow of helium, while maintaining the same flow of reagent, resulting in reagent gas dilution (vide infra).

After the leak valve V2 a union tee (G) is connected to three-way ball valve T1 (Part

No. SS-41GXS2-1466, Swagelok, Australia). Connected to separate outputs of T1 are a rotary vacuum pump and a restrictive capillary (J, 1/16" OD × 25 µM ID × 100 mm L PEEKsil™ tubing). Selection of the restrictive capillary line allows the addition of higher boiling point liquids, high vapour pressure solids, or corrosive species to the helium buffer gas without coating the length of tubing usually travelled by the reagents when injected via the septum, or damaging components inside the mixing system. When reagents are added via the septum, T1 is usually set to the off position, however, the restrictive capillary may be used in tandem with the septum inlet (B) to add secondary reagents to the helium buffer gas.

By setting T1 to the vacuum line, setting the instrument mode to normal operating mode, turning off the split-ﬂow and pressure gauge valves (S2 and S3, Part No. SS-41GS2,

Swagelok, Australia), and ﬁnally, turning off the helium at shut-off valve S1 (Part No. SS- ION-MOLECULE REACTIONS 91

41GS2, Swagelok, Australia) or the cylinder, the mixing system may be placed under high vacuum to assist in removing residual reagents from the manifold.

2.3.2 Proof of principle ion-molecule reactions

In this section, the reaction of dimethyl disulﬁde (DMDS) with putative distonic radical ions is described as a proof of principle and to introduce the reader to typical ion- molecule experiments detailed in the following chapters. As described in Chapter 1, the reaction of DMDS with a distonic radical cation results in a +47 Da adduct,142 while reaction with a non-distonic radical cation typically results in charge-transfer or no reaction.

In this experiment, DMDS was used to characterise and differentiate two isomers, 2-methyl-5-(N,N,N-trimethylammonium)phenyl radical cation (D) and 4-(N,N,N- trimethylammonium)benzyl radical cation (F) by comparison of their respective reactions. Aqueous solutions of 3-iodo-N,N,N,4-tetramethylbenzaminium iodide (2) and 4-(iodomethyl)-N,N,N-trimethylbenzaminium iodide (3) in MeOH were infused via positive electrospray, in both cases yielding an ion at m/z 276. As depicted in Scheme 2.8, the respective m/z 276 ions were isolated and subjected to photolysis (single pulse at 266 nm) resulting in ions at m/z 149 (D and F).

NMe3 I NMe3 NMe3 NMe3

ESI PD + CH3SSCH3 (i) + CH3S -I -I I I S C D 2 m/z 276 m/z 149 m/z 196

NMe3 I NMe3 NMe3

ESI PD + CH3SSCH3 (ii) No reaction -I -I

E F I m/z 276 I m/z 149 3

Scheme 2.8: Electrospray ionisation of (i) 3-iodo-N,N,N,4-tetramethylbenzaminium iodide (1) and (ii) 4- (iodomethyl)-N,N,N-trimethylbenzaminium iodide (2) and their subjection to PD results in a carbon-centred radical which is allowed to react in the presence of dimethyl disulﬁde (DMDS).

DMDS was seeded in the helium buffer gas at a ﬂow rate of 0.004 µL min−1 with split ﬂow of 50.1 µL min−1 for a total neutral density of ~7.9 × 109 molecules cm−3 92 INSTRUMENTATION

within the ion-trap. As shown in Figure 2.22, the methylphenyl analogue D reacts to form a signiﬁcant abundance of an [M + 47]+ ion consistent with addition of thiomethyl · (CH3S ) to the radical while the benzyl analogue F reacts negligibly with DMDS over · a 10000 ms isolation time. Formation of a CH3S adduct during reaction with DMDS is characteristic of a distonic radical cation,142 which supports the assignment of D as distonic. Indeed the unpaired electron in F is expected to reside in a π-orbital and is therefore stabilised by delocalisation over the benzene ring, while D is a σ-radical and thus localised to a single carbon atom.

Figure 2.22: Reaction of a) 4-(N,N,N-trimethylammonium)benzyl radical cation and b) 2-methyl-5-(N,N,N- trimethylammonium)phenyl radical cation with dimethyl disulﬁde (DMDS).

2.3.3 Calculating the concentration of a neutral

In the following experiments, rate constants are measured for bimolecular reactions. In order to calculate rate constants for bimolecular reactions the concentration of at least ION-MOLECULE REACTIONS 93

one reactant or product must be known. Under these experimental conditions, it is not possible to determine the concentration of the reactant radical anion, or products, therefore, the reaction rates are measured under psuedo-ﬁrst order conditions, where the neutral molecule added to the helium buffer gas is in excess. To then determine the real second-order rate constant the concentration of the neutral molecule in excess must be known.

The concentration of neutral reagent can be calculated from the ratio of neutral flow rate and total flow of helium. The total flow of helium is made up of flow to the instrument and the split flow. The flow of helium to the instrument was measured using the following procedure.

The pressure in the mixing manifold is set to a helium pressure of 5 psi (259 Torr).

Leak valve V2 is adjusted so that the pressure inside the ion-trap is 2.5 mTorr. For explanation, the reading on the ion gauge is uncalibrated, and it has been found that the addition of some gases (such as ozone) may cause readings from the ion gauge to drift. It is therefore essential during each use of the mixing manifold to adjust V2 to obtain an ion gauge pressure reading equal to that obtained in normal mode, which is read directly prior to using the instrument in ion-molecule mode. The ﬂow of helium in normal mode is supplied independently of any feedback from the ion gauge, thus the presumption is made that under these conditions the current ion gauge pressure reading is representative of an ion-trap pressure of 2.5 mTorr.

One must then consider the difference in pressure before and after leak valve V2.

Suppose the pressure on the mixing manifold side of leak valve V2 is Pm and the pressure on the instrument side of the valve is Pi . Pressure Pi is therefore equal to the ion- trap pressure of 2.5 mTorr, while Pm is set manually to 5 psi (259 Torr) at the helium 5 regulator. The difference between the two values is ~10 , thus Pi is essentially zero and the difference in pressure (∆P) is thus 5 psi (259 Torr). At this pressure difference the ﬂow of helium from the mixing manifold results in a pressure within the ion trap of 2.5 mTorr.

If the connection to the instrument is removed, Pi is equal to atmospheric pressure

(14.7 psi, 760 Torr). To maintain a constant ∆P, Pm is increased by 14.7 psi (760 Torr) to 19.7 psi (1019 Torr). Measurement of the ﬂow rate from the mixing manifold qinst is now equal to that obtained when the mixing manifold is being operated under normal 94 INSTRUMENTATION

vacuum conditions. Changing the split ﬂow can alter ∆P, therefore it is important to set the split ﬂow while the mixing manifold is open to the instrument so that

−1 V2 may be set accordingly. An Finst value of 2.2 ± 0.04 mL min was obtained by multiple measurements at different split ﬂow rates. This value will be used in all future calculations for the ﬂow rate of helium from the mixing manifold to the instrument.

The total flow of helium (QHe ) is therefore equal to the combined split flow (qspli t ) ′ ′ and instrument flow (qinst ). The molar flow rate of helium (QHe ) and neutral (QN ) can be obtained by accounting for the molecular weight (W ) and density (ρ) of the neutral and helium.

QHe = qinst + qspli t (2.3)

′ Q × ρ Q = (2.4) W

The partial pressure of the neutral within the ion-trap PN may be calculated from the total ion-trap pressure of 2.5 mTorr (PT ) and ratio of neutral to helium concentration.

The ﬁnal term in the following equation accounts for effusion (WN and WHe are the molecular weight of the neutral and helium, respectively) , as lighter molecules will be removed faster from the ion-trap into the surrounding vacuum than heavier molecules.

√ ′ QN WN PN = PT × ′ × (2.5) QHe WHe

The ﬁnal concentration of the neutral (molecules cm−3) is subsequently calculated using the idea gas law equation with the temperature T equal to the measured buffer gas temperature of 307 K154 and the gas constant R.

n PV = nRT, M = (2.6) V

nRT P = (2.7) V P = MRT (2.8) P M = (2.9) RT ION-MOLECULE REACTIONS 95

An example of this calculation is demonstrated in Table 2.1.

Table 2.1: Example of neutral concentration calculation for dimethyl disulﬁde.

−6 −1 Flowrate Neutral (Fn) 4.0 x 10 mL min −1 Density of Neutral (ρn) 1.06 g mL −1 Molecular Weight Neutral (MWn) 94.199 g mol −1 Split Flow (qspli t ) 50.1 mL min −1 Instrument Flow (qinst ) 2.2 mL min −1 Total Helium Flow (QHe ) 52.3 mL min Temperature (T ) 307 K −1 Molecular Weight Helium (MWh) 4.003 g mol −4 −1 Density of Helium (ρh) 1.656 x 10 g mL −3 Trap Pressure (PT ) 2.5 x 10 Torr ′ −3 −1 Flowrate Helium (QHe ) 2.1636 x 10 mol min ′ −8 −1 Flowrate Neutral (QN ) 4.5011 x 10 mol min ′ ′ −5 Fn / QHe 2.0804 x 10 −7 Pressure Neutral in Trap (PN ) 2.5230 x 10 Torr −5 Pressure Neutral in Trap (PN ) 3.3631 x 10 Pa Concentration Neutral 7.94 x 109 molecules cm−3

To validate this approach, Harman and Blanksby compared second order rate constants for four reactions (described in Equations 2.10-2.13) measured using this instrumentation (shown in Table 2.2) to those reported in the literature.155 In all cases, the measured values were within the experimental error estimated for each method. This demonstrated that the current instrumentation is able to provide gas phase rate measurements of reasonable accuracy (the precision of this method will be discussed in Section 2.3.7), and validated the method described for determining neutral reagent concentration within the ion trap.

35 − + − −−→ 35 − + − Cl CH3 I Cl CH3 I (2.10) 79 − + − −−→ 79 − + − Br CH3 I Br CH3 I (2.11) − − + − −−→ − − + − (CF3)2CH O CH3CH2 Br (CF3)2CH O CH2CH3 Br (2.12a) − − + − −−→ − + − + − (CF3)2CH O CH3CH2 Br (CF3)2CH OH CH2−CH2 Br (2.12b) - - CO2 CO2

+ − − − −−→ SCH3 + − · CH3 S S CH3 CH3 S (2.13) 96 INSTRUMENTATION

Table 2.2: Comparison of second order rate constants (k2) for the ion-molecule reactions depicted in Equa- tions 2.10-2.13, determined using the current method and by FA-SIFT or FT-ICR techniques. Reproduced from Harman and Blanksby.155

Reaction k2 (Current) k2 (Literature) Literature − − − − Equation (cm3molecule 1s 1)(cm3molecule 1s 1) Method

− − (2.10) 1.2 ± 0.3 x 10 10 1.66 ± 0.03 x 10 10 FA-SIFT237 − − (2.11) 3.0 ± 0.4 x 10 11 2.89 ± 0.09 x 10 11 FA-SIFT237 − − (2.12a)&(2.12b) 5.6 ± 0.6 x 10 12 7.83 ± 0.33 x 10 12 FA-SIFT237 − − (2.13) 3.75 ± 0.1 x 10 12 8 ± 4 x 10 12 FT-ICR148

2.3.4 Determining the concentration of dioxygen

Harman and Blanksby measured the second order rate constant of the addition of

18 O-labelled dioxygen to the 3-carboxylatoadamantyl radical anion at k2 = 8.5 ± 0.4 ×10−11 cm3 molecule−1 s−1.155 It is therefore possible to determine the concentration of dioxygen within the ion-trap by measuring the pseudo-ﬁrst order rate constant k1 of the addition of dioxygen to the 3-carboxylatoadamantyl radical anion and dividing it by the known second order rate constant for the reaction.

k1 k2 = (2.14) [O2] −1 k1(s ) [O ] = (2.15) 2 3 −1 −1 k2(cm molecules s )

2.3.5 Measuring reaction kinetics

The mass spectrometer used in these experiments has an automatic gain control (AGC) feature which attempts to regulate the concentration of ions stored in the ion-trap. Under normal operating conditions, AGC is set to allow 104 ions into the trap, while the concentration of neutral reagents seeded into the helium buffer gas is typically in the order of 109 − 1012 molecules cm−3. The ratio of neutral reagent to reactant ion is therefore in excess of 105, clearly establishing the reactant ions as the limiting reagent. Therefore, the reaction proceeds under pseudo-ﬁrst order conditions with the neutral concentration assumed to remain constant. Pseudo-ﬁrst order rates k1 are calculated by plotting the natural logarithm of the decrease in ion abundance of the reactant radical ION-MOLECULE REACTIONS 97

[A] against reaction time t, deﬁned as the interval between the isolation of the selected ion and ejection of all ions from the trap for analysis. The ion abundance is normalised against the total ion abundance to account for ﬂuctuations in the number of isolated ions. This yields a linear relationship with the slope equal to −k1 as shown in equation 2.16.

ln[A] = −k1t + ln[A]0 (2.16)

The second-order rate constant is then derived from the concentration of the neutral reagent [B] using equation 2.17.

−1 3 −1 −1 k1(s ) k2(cm molecules s ) = (2.17) [B]

2.3.6 Reaction efﬁciency

Reaction efﬁciency Φ is deﬁned as the ratio of second-order rate constant k2 to the theoretical collison rate kcol . Theoretical collision rates were calculated using the parametrised trajectory collision theory of Su and Chesnavich238 as implemented by Dr. Shuji Kato in a SigmaPlot routine (University of Colorado, Boulder) which was converted to an excel spreadsheet. The trajectory collision rate kcol is calculated as a correction to the Langevin collision rate kL shown in Equations 2.18 to 2.21.

µD m1m2 x = µ = (2.18) 2µkB T m1 + m2

p 2πq α k = p (2.19) L µ

 ( )  + 2 (x 0.509) + 0.9754 x ≤ 2 = × 10.526 kcol kL  (2.20) (0.4767x + 0.6200) x > 2

k2 Φ = × 100% (2.21) kcol In these equations, α is the polarizability of the neutral, µ is the reduced mass of the ion and neutral (using molecular weights m1 and m2 for the ion and neutral respectively), µD is the dipole moment of the neutral, kB is the Boltzmann constant, q 98 INSTRUMENTATION

is the elementary charge, and T is temperature. An example of this calculation is shown in Table 2.3.

ϕ Table 2.3: Example of trajectory collision rate calculation (kcol ) and reaction efﬁciency ( ) for reaction of 4-(N,N,N-trimethylammonium)phenyl radical cation (B) with dimethyl disulﬁde. † Literature values from Grabowski and Zhang.152

Neutral dimethyl disulfide ·– Ion C9H13N −1 Molecular Weight Neutral (m1) 94 g mol −1 Molecular Weight Ion (m2) 135 g mol Polarizability Neutral (α) 10.5† Å3 † Dipole Moment Neutral (µD ) 1.9 debye Temperature (T ) 307 K Reduced Mass (µ) 55.41485 −1 g mol−1 23 −1 Avagadro Constant (NA) 6.0221415 x 10 mol −16 −1 Boltzmann Constant (kB ) 1.38066 x 10 erg K Polarizability (α) 1.05 x 10−23 cm3 −18 Dipole Moment (µD ) 1.9 x 10 statC cm Reduced Mass (µ) 9.20 x 10−23 g − Unit Charge (q) 4.80320427 x 10 10 statC −9 −1 3 −1 Langevin Collision Rate (kL) 1.019 x 10 molecules cm s Trajectory Correction (x) 2.01 −9 −1 3 −1 Trajectory Collision Rate (kcol ) 1.61075 x 10 molecules cm s −1 −1 Pseudo-first order Rate (k1) 7.986 x 10 s −10 −1 3 −1 Second order Rate (k2) 1.006 x 10 molecules cm s Reaction Efficiency (Φ) 6.1 %

2.3.7 Errors in calculated reaction rates

Random errors for reaction rate calculations were typically 2σ ≤ 10%, where σ is the average of the standard deviations obtained from the least-squares fits to the pseudo-first order decay profiles. These errors were attributed to random fluctuations in the neutral reagent concentration and random fluctuations in the total ion population. Systematic errors include the concentration of neutral (which is based on an uncalibrated readout from an ion gauge), product ions which are formed but not detected (such as products which lose their charge), product ions of very low abundance ignored in calculation of the total ion signal, and product ions with a m/z of less than 50 Da. Further error may be introduced by mass discrimination effects.93 Mass discrimination in these experiments is likely to arise due to space-charge effects and reaction of ions with residual solvent inside the ion-trap. Both effects cause curvature in pseudo-first order decay plots. CONCLUSION 99

Reaction of ions with the background gas will likely present as adduct ions, for example, an [M + 18] for addition of water. In general, this adduct formation is significantly slower than the reaction being measured. The use of MS3 and higher MSn experiments for kinetic measurements largely precludes space-charge effects, as subsequent MS experiments result in isolation of only part of the initial ion packet such that over- filling of the ion-trap is minimised by the time kinetic measurements are made. Rate measurements typically contained linear pseudo-first order decay curves and therefore mass discrimination effects are considered minimal. Rates have, therefore, not been corrected and any minimal effects are assumed to contribute to the overall error.

The concentration of O2 was derived from the second-order rate constant of the reaction of the 3-carboxylatoadamantyl radical with O2 measured by Harman and Blanksby.154,155 A pseudo-first order rate constant of this reaction was obtained prior to a given experiment to measure and confirm the stability of O2 concentration. The error in the second-order rate constant of this reaction was given as ±5%. Overall, an accuracy of ±50% is ascribed to the measured rate coefficients to take into account both random and systematic errors, while precision is ±10%.

2.4 Conclusion

This chapter has detailed the modifications made to a ThermoFisher LTQ linear ion- trap mass spectrometer and examples have shown that these modifications have been successful in accommodating both ion-molecule reactions and photodissociation of isolated ions. Photodissociation of halogen and Barton ester derived ions at 266 nm was demonstrated and resulted in the successful synthesis of distonic radical ions with a positive or negative charge, while the addition of dimethyl disulfide to the helium buffer gas via the gas mixing manifold was shown to distinguish between putative distonic radical ions.

The following chapters will detail extensive studies of aliphatic and aromatic radicals which make use of these modiﬁcations for both synthesising and characterising these reactive intermediates. 100 INSTRUMENTATION

2.5 Materials

1,2-Dimethyldisulﬁde, 1,3-adamantanedicarboxylic acid and methyl 4- bromomethylbenzoate was purchased from Sigma-Aldrich (St. Louis, MO). 3-Iodoadamantanecarboxylic acid, N,N,N-trimethyl-4-((2-thioxopyridin- N-yloxy)carbonyl)benzeneaminium iodode and 3-(N-oxycarbonyl-2(1H)- pyridinethione)adamantane-1-carboxylic acid were synthesised by Dr. David G. Harman (University of Wollongong).154 AR grade solvents were used for synthesis and HPLC grade solvents were used for mass spectrometry experiments which were purchased from AJAX Finechem (Taren Point, Australia). 3-Iodo-4-methylaniline was purchased from Alfa Aesar (Ward Hill, MA). 4-(Dimethylamino)benzyl alcohol was purchased from TCI Chemicals. 4-Iodobenzoic acid was purchased from BDH laboratory Supplies (now part of VWR International, West Chester, PA).

2.5.1 General method for synthesising trimethylammonium iodide

salts

The general method used for generating the quaternary trimethylammonium iodide salts from primary or tertiary amines was to dissolve or suspend 100 mg of the parent amine and 1.1 mol. eq. of K2CO3 in dry MeOH (15 mL). 10 mol. eq. of MeI was then added, and the solution heated at 60 °C until the starting material and intermediate methyl and dimethyl amines disappeared (as followed by TLC 1:3 EtOAc:Petroleum Spirit). If the reaction was incomplete after 6 hours and there was no evidence of its continuation, further MeI was added. After the reaction was complete, the solvent was removed in vacuo and the crude product washed with acetone, water and diethyl ether. If further puriﬁcation was required, the crude compound was recrystallised from

CH3CN/Et2O.

Conversion of dimethyl amines to their trimethylammonium salt was achieved by the following method. To the starting dimethylamine (100 mg) dissolved in acetone was added 5 mol. eq. of MeI, and the reaction heated to 50 °C. The product precipitated as the trimethylammonium iodide salt formed. When there was no starting material remaining (as followed by TLC) the solvent was removed in vacuo to yield the crude product. The crude product was then washed with cold acetone and dried to yield the MATERIALS 101

product in sufﬁcient purity for further reaction.

2.5.2 General method for generation of iodides from alcohols

100 mg of an alcohol was suspended or dissolved in 2 mL of 55% HI in a capped reaction vial and allowed to stir at 60 °C overnight. If the product precipitated, the reaction mixture was cooled to 0 °C, ﬁltered and washed with cold water (cold diethyl ether or acetonitrile for quaternary ammonium iodide salts) to remove all colour, then dried in vacuo to yield the product. If the product remained in solution, Na2S2O3 powder was added with agitation of the reaction mixture until the solution was clear or the product precipitated. If the product continued to remain in solution it is extracted with copious

Et2O, dichloromethane or EtOAc. The extracts were then combined and washed with aqueous Na2S2O3, then dried over MgSO4. The solvent was then removed in vacuo to yield the product.

2.5.2.1 N-(4-iodomethylphenyl)-N,N,N-trimethylammonium iodide (2)

Synthesised from 4-(dimethylamino)benzyl alcohol by N-methylation followed by iodo-dehydroxylation using HI. Recrystallised from CH3CN to yield colourless crystals. 1 H NMR (500 MHz, DMSO-d6) δ7.90 (d, J = 8.6 Hz, 2H), δ7.67 (d, J = 8.8 Hz, 2H), δ4.67 13 (s, 2H), δ3.59 (s, 9H). C NMR (125 Mhz, DMSO-d6) δ = 146.1, 142.1, 130.2, 120.8, 56.4, 4.5. ESI-MS (+ve) M+ 276.

2.5.2.2 N-(3-iodo-4-methylphenyl)-N,N,N-trimethylammonium iodide (3)

Synthesised from 3-iodo-4-methylaniline by N-methylation. Recrystallised from 1 δ CH3CN/Et2O to yield a colourless powder. H NMR (500 MHz, DMSO-d6) 8.35 (s, H), δ7.90 (d, J = 7.3 Hz, H), δ7.56 (d, J = 8.8 Hz, H), δ3.57 (s, 9H), δ2.43 (s, 3H). 13C NMR (125 + Mhz, DMSO-d6) δ = 145.4, 143.5, 130.4, 130.0, 120.3, 101.8, 56.5, 26.8. ESI-MS (+ve) M 276.

oeue uha oun n yee r rsn nlreqatte nAsrla fuels. Australian in quantities large in present are xylenes and toluene as such molecules eto 1o h ulQaiySadrsAt20 pcfial iisteqatt of quantity the v/v. 45% as limits high as specifically be can 2000 aromatics total Act but v/v, Standards 1% at Quality benzene Fuel the aromatic of by 21 replaced section be to removal its to led eventually compounds. toxicity its but fuel, in agent a nce olmttecnetaino edi ul o005gLb aur 2002. January 1 by g/L 0.005 to fuels in lead hydrocarbons of concentration aromatic the limit to polycyclic enacted was as such aggregates and compounds aromatic of production the in soot. result of fuels levels in high compounds aromatic of of levels high downside, that major is A however, ignition efficiency. engine premature combustion of decrease likelihood may which the knock) reducing (engine octane fuel increase that additives are experimental. theo- both and studies retical many of subject the been has light, understanding this to in and, central chemistry is combustion molecules aromatic substituted and benzene of Oxidation 3.1 h rsneo rmtccmonsi ul ed otepouto n release and production the to leads fuels in compounds aromatic of presence The * h hs u flae erlbgnudrteFe ult tnad c 00weeregulation where 2000 Act Standards Quality Fuel the under began petrol leaded of out phase The Introduction G * nAsrla h ulSadr Pto)Dtriain20 aeunder made 2001 Determination (Petrol) Standard Fuel the Australia, In SPAERATOSO DISTONIC OF REACTIONS PHASE AS 248 239,15,240,241,242,39,243,244,245,246,40,247,66 nteps,ttaty edwsue sa anti-knocking an as used was lead tetraethyl past, the In HNLRDCLIONS RADICAL PHENYL 103 249 rmtccompounds Aromatic sarsl,aromatic result, a As 249

C HAPTER 3 104 GAS PHASE REACTIONS OF DISTONIC PHENYL RADICAL IONS

(PAH) into the troposphere. This can occur due to fuel leakage from petrol stations or incomplete combustion and exhaust from engines.250,251,252,15,56,253,254,255,256 The presence of aromatic molecules in the troposphere is of interest due to their toxicity257 and role in smog formation.15,258,252 Furthermore, while the presence of benzene in modern fuels is declining, it remains a fundamental model with which to investigate larger aromatic molecules.

The removal of aromatic molecules, for example benzene, from the troposphere is initiated by reactive species, primarily HO·, which may either add to benzene to yield a · 2-hydroxycyclohexadienyl radical (HOC6H6), or abstract a hydrogen atom to yield the · 259 phenyl radical (C6H5 ) and H2O. In the latter reaction, the resulting phenyl radical may react with O2, readily available in the troposphere, to generate a phenylperoxyl · radical (C6H5OO ) as outlined in the scheme below.

+ · → · + C6H6 HO C6H5 H2O + · → · C6H6 HO HOC6H6 · + → · C6H5 O2 C6H5OO

The phenylperoxyl radical is central to the oxidative degradation of benzene and the formation of byproducts. Due to its elusive nature, the majority of mechanistic studies have largely been conﬁned to quantum chemical modelling. It was not until recently that Mardyukov and Sander isolated the phenylperoxyl radical using low temperature matrix isolation methods. This study utilised infrared spectroscopy to investigate the spectroscopic properties of the radical.52 While both low and high temperature reaction schemes have been constructed based on end-product analysis, there has been limited experimental work and associated computational modelling aimed at probing these reactions at ambient temperatures.39,244

3.1.1 Degradation of the phenylperoxyl radical

Experimental and computational studies predict all reaction pathways associated with the oxidative degradation of the phenyl radical to be initiated by the barrierless addition of O2. This exothermic reaction results in formation of a phenylperoxyl radical ∆ · + −−→ · = − −1 39,55 ( rxnH298[C6H5 O2 C6H5OO ] 46.3 kcal mol ). Without collisional cool- INTRODUCTION 105

ing this excited collision complex may simply decompose back to reactants.39

There is much debate in the literature as to the reaction mechanisms involved in the isomerisation and degradation of the phenylperoxyl radical. Tokmakov et al.55 undertook an in-depth theoretical study of the degradation of the phenylperoxyl radical, building on the framework of Carpenter et al. and Fadden et al.53,54,57,58 and proposed a number of reaction pathways to the major products previously predicted or observed from high-temperature experimental studies. Two major pathways typically invoked when investigating phenyl radical oxidation include (i) loss of CO2 to form cyclopentadienyl radical and (ii) loss of O atom resulting in phenoxyl radical. In the context of this study we include a third channel, (iii) loss of .CHO radical to form cyclopentadienone (C5H4O). In this section bold letters identify structures depicted in Scheme 3.1 which summarises the lowest energy pathways (as reported by Tokmakov et al.55) associated with formation of these three products. Computational studies · + suggest that the degradation of the phenylperoxyl radical may yield both (i) C5H5 CO2 · + + . and (ii) C6H5O O, while it has been suggested that (iii) C5H4O CHO would only be · formed at higher temperatures. Thermal degradation of C6H5O results in formation of · 55 · C5H5 and CO, and thus as a product of two major putative reaction pathways, C5H5 is considered the major reaction product of phenylperoxyl radical degradation. In the subsequent discussion all three reaction pathways will be introduced in the context of the current literature.

O O O O (3P) O CO O2

(ii) AB CD E CO O OH O O O O OCO CO2 O2 H

O O O (i) I J F K L O OH O CO H O O G C H O H O O H H H C C (iiia) M N O P Q (iiib) (iv) O O CHO O CO H C O C O O C C O O M R U V W

Scheme 3.1: Formation of cyclopentadienone proposed by Carpenter (Blue),53, 54 and extended by Fadden and Haddad (Red) and Sebbar et al. (Black).57,58, 260, 261 106 GAS PHASE REACTIONS OF DISTONIC PHENYL RADICAL IONS

Yu and Lin39 investigated the low temperature reaction kinetics of the phenyl radical and O2. The growth of the phenylperoxyl radical was measured by cavity ring-down spectroscopy over a wide temperature range. Furthermore, Yu and Lin demonstrated that phenoxyl radical was not formed below 473 K. This experiment, however, assumes that the phenoxyl radical is unreactive on the experimental timescale which may not be the case. Phenoxyl radical may, for example, react by hydrogen abstraction to form phenol, which could not have been detected by this technique.

The ejection of O atom to form phenoxyl radical (ii) is a seemingly simple degradation pathway (Scheme 3.1 B¨C). Theory has shown this reaction, relative to reactants, to have a transition state barrier of −10.2 kcal mol−1 almost equal to the reaction − energy at −10.4 kcal mol 1 (G2M level calculated by Tokmokov et al.).55 In contrast, all other energetically competitive reaction pathways involve multiple isomerisation · + steps before end-product generation. Phenoxyl radical was the subject of C6H5 O2 crossed molecular beam experiments conducted by both Gu and Kaiser, and Albert and Davis.243,262 The mechanisms of O atom ejection proposed by the two groups were, however, signiﬁcantly different. Gu and Kaiser concluded from their experiments that the phenylperoxyl radical has a lifetime shorter than 0.01 ps, suggesting that the formation of phenoxyl radical occurs directly by a stripping-like mechanism. Albert and Davis summarily disputed these ﬁndings, demonstrating that under their experimental conditions the phenylperoxyl radical is long-lived on a rotational timescale, and that an O atom is therefore ejected subsequent to formation of a relatively long-lived phenylperoxyl radical intermediate. Phenoxyl radical was shown to be competitive with both isomerisation prior to degradation of the phenylperoxyl radical and decomposition back to reactants. The temperature of these reactions are based on the collision energy of the reactants. Albert and Davis argued that the differences observed in the respective experiments lies in the way the reactants are formed. An estimate of the excess vibrational energy of the phenylperoxyl radical prior to degradation was given at 93 kcal mol−1 (Albert and Davis) and 107 kcal mol−1 (Gu and Kaiser) demonstrating that the reactions were measured at a higher than ambient temperature.262 There is still debate over the presence of phenoxyl radical in low temperature phenyl radical oxidation.

Reaction pathways (i) and (iii) involve the intermediary oxepinon-7-yl radical pro- INTRODUCTION 107

posed by Carpenter.53 This intermediate can be formed by the isomerisation of the phenylperoxyl by ipso or ortho attack to form a 1,1- or 1,2-diepoxy intermediate followed by ring expansion to form the oxepinon-7-yl radical (Scheme 3.1 G¨H; ∆rxnH298 = −96.4 kcal mol−1). The high exothermicity of this reaction can fuel subsequent rearrangement over transition state barriers that are high relative to H but are still well · + 53,57,58,263,261,260,55 below the C6H5 O2 entrance channel. Additional products identi- . ﬁed by high temperature combustion-related experiments include CO, CHO and CO2. Ejection of CO from cyclopenta-2,4-dienylcarbonyl is one source of cyclopentadienyl ¨ ¨ with an overall loss of CO2 from the phenylperoxyl radical (B C F). Alternatively, the oxepinon-7-yl radical may ring contract to form cyclopenta-2,4-dienyloxycarbonyl ¨ ¨ which may subsequently release CO2 to yield cyclopentadienyl radical (H I F). This reaction has been calculated to be highly exothermic at -114 kcal mol−1. The lowest energy pathway resulting in loss of CO is shown in Scheme 3.1(iv) and leads to formation of 3-pyranyl radical after ejection of CO. This product was predicted to reside in a deep potential well, -90.7 kcal mol−1 below reactants. However, the reaction pathway resulting in formation of 3-pyranyl radical is still higher in energy than formation of cyclopentadienyl radical.

Cyclopentadienone (Scheme 3.1 M) was first identified by Rotzoll et al. and later Chai et al. in high temperature experiments probing the oxidation of benzene.32,264 Fadden and Haddad proposed two mechanisms to account for its generation.57,58 Both mechanisms proceed subsequent to formation of the oxepinon-7-yl radical, the first resulting in ring opening of the oxepinon-7-yl radical (H¨N) followed by loss of CO and subsequent 1,5-hydrogen atom transfer and ring closure to the cyclopentenone radical (N¨Q). Elimination of atomic hydrogen follows to yield cyclopentadienone (Q¨M).

The second proposed mechanism is initiated by addition of O2 to the cyclopentadienyl radical (F¨K) followed by 1,3-hydrogen atom transfer and subsequent loss of the hydroxyl radical (K¨M) to yield cyclopentadienone. These reaction pathways both contain signiﬁcant barriers and a number of isomerisation steps, and it was therefore concluded that formation of cyclopentadienone would only occur during high temperature oxidation. Recently however, a low energy mechanism was proposed in an investigation be Sebbar et al.261 in which a more direct mechanism was suggested. This pathway proceeds by ring contraction of the oxepinon-7-yl radical which occurs by β- 108 GAS PHASE REACTIONS OF DISTONIC PHENYL RADICAL IONS

scission and subsequent ring closure of the nascent 2,4-hexadiene-1,6-dial-1-yl radical to form the 2-formylcyclopent-4-en-3-yl radical, followed by β-scission and ejection of CHO· to form cyclopentadienone (N¨M). It is clear more experimental studies are required to unravel the products of phenyl radical oxidation.

This chapter details a study of the formation and isolation of distonic radical ion models of the phenylperoxyl radical and their subsequent degradation using a modiﬁed commercial linear quadrupole ion-trap mass spectrometer. A range of models were investigated including 4-carboxylatophenyl, 3-carboxylatophenyl, 4- carboxylatobiphenyl, 4-sulfonatophenyl, and 4-sulfonatobiphenyl radical anions and 4-trimethylammoniumphenyl radical cation, in order to explore the effect of charge, distance between the charge and radical, as well as any direct substituent effects on the reactivity of the phenyl radical. The reaction of the phenyl and phenylperoxyl · · radicals towards NO and NO2 was investigated as a means of probing the distonic character of the phenyl radical analogues as detailed later in Chapter 4, to observe the tropospherically relevant bimolecular reaction between a phenylperoxyl radical and · · NO and NO2. Furthermore, potential energy surfaces describing the unimolecular degradation of both neutral and charge-tagged phenylperoxyl radical were constructed using computational methods, and a comparison of the reaction pathways made to ion- molecule reactions measured using ion-trap mass spectrometry.

3.2 Method

3.2.1 Mass spectrometry

Experiments were performed on a modified ThermoFisher LTQ (San Jose, CA) linear quadrupole ion-trap mass spectrometer90 fitted with a conventional IonMax electrospray ionisation source and operating Xcalibur 2.0 SUR1 software. Ions were generated by infusion at 3 - 5 µLmin−1 of a methanolic solution (10 - 100 µM) into the electrospray ion source. In some instances aqueous ammonia was added to raise the pH and facilitate the formation of dianions. Typical instrumental settings in negative ion mode were: spray voltage (-3.5 kV), capillary temperature (200 - 250 °C), sheath gas flow between 10 - 30 (arbitrary units), sweep and auxillary gas flow between 0 - 10 (arbitrary units). In positive ion mode typical conditions were: spray voltage (3.5 - 4.0 kV), METHOD 109

capillary temperature (150 °C), sheath gas ﬂow (10 - 25 arbitrary units), sweep and auxillary gas ﬂow set between 0 - 10 (arbitrary units). For collision induced dissociation experiments, ions were mass-selected with a window of 1 - 4 Th, using a q-parameter of 0.250, and the fragmentation energy applied was typically 10 - 45 (arbitrary units) with an excitation time of 30 ms (unless otherwise noted).

Modifications to the mass spectrometer for the introduction of neutral gases into the ion trap region of the instrument were described in Chapter 2. Briefly, neutral liquids and gases are introduced into a flow of Ultra High Purity (UHP) helium (3 - 5 psi) via a heated septum inlet (25 - 250 °C). The neutral flow is controlled using a syringe pump, while helium is supplied via a variable leak valve to provide an estimated ion-trap pressure of 2.5 mTorr.155 The pressure within the ion-trap has previously been confirmed by Harman and Blanksby, using rates from a known reaction.155 The temperature of the vacuum manifold surrounding the ion trap was previously measured at 307 ±1 K,155 which is taken as being the effective temperature for ion-molecule reactions observed herein.235,265 Reaction times of 30 - 10000 ms were set using the excitation time parameter within the control software using a fragmentation energy of 0 (arbitrary units). All spectra presented are an average of at least 50 scans.

3.2.2 Computational methods

All calculations were undertaken using the hybrid density functional theory B3LYP method266,267 and the 6-311++G(d,p) basis set within the GAUSSIAN03 suite of pro- grams,268 with selected higher accuracy energies calculated using the composite G3SX theory.269,270 All stationary points on the potential energy surface were characterised as either minima (no imaginary frequencies) or transition states (one imaginary frequency) by calculation of the frequencies using analytical gradient procedures. All reported energies include zero-point energy corrections. Minima between transition states were conﬁrmed by calculation of the intrinsic reaction coordinate (IRC).

3.2.3 Predicting relative reaction rates

The reaction rate modelled by transition state theory is shown in Equation 3.1.271 The pre-exponential factor in this equation may be used to estimate the entropy of competing reaction pathways.271 In this equation, Q and Q‡ represent the molecular 110 GAS PHASE REACTIONS OF DISTONIC PHENYL RADICAL IONS

partition function of the starting minimum and transition state structures, respectively, kB is the Boltzmann constant, T is temperature, h is Planck’s constant, ∆H0 is the reaction enthalpy (with zero-point correction) and R is the gas constant.

‡ ‡ ∆H kB T Q 0 k = e RT (3.1) h Q

Q = QV QRQT QE (3.2)

As shown in Equation 3.2, the partition function Q is made up of contributions from the vibrational (QV ), translational (QT ) and rotational (QR ) partition functions in addition to an electronic contribution (QE ). Because these species are anions and cations with no degenerate ground states, QE = 1. The transition states of interest in this chapter are unimolecular rearrangements with a common starting minimum, therefore, the rotational and translational partition functions should be essentially equal.271 Q

271,272,273 thus simpliﬁes to the vibrational partition function (QV ). Ions isolated within a linear ion-trap are expected to be thermalised rapidly by the bath gas but may not follow a Boltzmann distribution of internal energies due to the presence of an isolation potential (DC trapping potential applied to trap ions in the x-direction of the ion-trap, see Section 1.2.3.2.5 for more information) and RF drive (quadrupole AC ﬁeld applied to trap ions radially).235,89 As such, results obtained using this method may only be used qualitatively. The effective temperature of the ions within the ThermoFinnigan LTQ has previously been measured at 307 K and this value is used for T .155

As shown in Equation 3.3 the vibrational partition function (QV ) is the product of vibrational partition functions for each normal mode (after removal of the imaginary normal mode of the transition state) of a molecule where the vibrational partition function of each normal mode is deﬁned in Equation 3.4,274

QV = q(v˜1)q(v˜2)...q(v˜n) (3.3)

1 q(v˜) = (3.4) 1 − e−βhcν˜ where β = 1/kB T , c the speed of light and ν˜ is the calculated frequency of the normal 274 mode. The result of this calculation, the vibrational partition function QV , measures MATERIALS 111

the density of states available for the given temperature at the transition state. As kBT /h is constant for a given potential energy surface, when comparison is made between two competing transition states which connect to a common starting minimum a ‡ qualitative estimate can be made by direct comparison of QV values for the respective transition states.

3.3 Materials

Terephthalic acid was purchased from BDH Lab Supplies (now part of VWR Inter- national, West Chester, PA). Biphenyl-4,4’-dicarboxylic (Fluka), 1,2-dimethyldisulﬁde, 2,3,5,6-tetradeutero-1,4-benzene dicarboxylic acid, 4-iodobenzoic acid and nitric oxide (98.5%) was purchased from Sigma-Aldrich (St. Louis, MO). N,N,N-Trimethyl-4-((2- thioxopyridin-N-yloxy)carbonyl)benzeneaminium iodode, 4-sulfobenzoic acid and 4’- sulfobiphenyl-4-carboxylic acid were synthesised by Dr. David G. Harman (University of Wollongong, Australia). Benzene-1,4-(13C)-dicarboxylic acid (92% 13C) was obtained form CDN Isotopotential energy surface (Pointe Claire Quebec, Canada). Isotopically

18 labelled oxygen O2 (95%) was obtained from Cambridge Isotope Laboratories (An- dover, MA). 4-Nitrobenzoic acid, methanol (HPLC grade) and ammonia solution (28%, AR grade) were obtained from Ajax Finechem (Sydney, Australia). All commercial compounds were used without further puriﬁcation.

3.3.1 Nitrogen dioxide

· 275 NO2 was synthesised by the established method of Mattson et al. on a 50 mL scale. In this method, a 60 mL syringe is ﬁlled with 50 mL of NO·. A second syringe is ﬁlled with

25 mL O2. Both syringes are connected to a PEEK union via 22 gauge LC needles and · · the O2 is injected into the NO to produce 50 mL of NO2.

3.3.2 N,N,N-trimethyl-4-nitrobenzaminium iodide

While there is a literature method for the synthesis of puriﬁed N,N,N-trimethyl-4- nitrobenzaminium iodide, these procedures use dimethylsulfate, picric acid and benzene, all three of which are carcinogenic, toxic or unstable.276 While the following methylation of 4-nitroaniline did not produce an isolable product, there was sufﬁcient product generated in situ for characterisation by tandem mass spectrometry. The 112 GAS PHASE REACTIONS OF DISTONIC PHENYL RADICAL IONS

methylation was carried out as follows. To a solution of 4-nitroaniline (200 mg, 1.45 mmol) in dry MeOH (20 mL) was added potassium carbonate (200 mg, 1.45 mmol) and iodomethane (900 µL, 14.5 mmol) which was stirred under N2 at reﬂux for 24 hours. 50

µL of the solution was diluted in 50:50 MeOH:H2O to µM concentration and infused via positive ion electrospray ionisation to yield a complicated spectrum including signals assigned as the protonated and sodiated adducts of 4-nitrobenzaminium, N-methyl-4- nitrobenzaminium, N,N-dimethyl-4-nitrobenzaminium cations and a minor signal at m/z 181 assigned as the target N,N,N-trimethyl-4-nitrobenzaminium cation.

3.4 Results

3.4.1 Gas phase synthesis of distonic ion models for phenyl radical

As introduced in Chapter 1, the Kenttämaa group has previously reported generation of positive charge-tagged phenyl radicals using chemical ionisation and FT- ICR,148,158,114,157,159,277 however, a mass spectrometric synthesis for positive charge- tagged distonic phenyl radical using electrospray ionisation has not previously been reported. One method is to introduce an N-acyloxypyridine-2-thione functional group, often referred to as a Barton ester,214 to a molecule capable of ionisation during positive ion electrospray. Harman and Blanksby have shown previously that molecular ions, including the Barton ester moiety, CID undergo N-O bond homolysis when subjected to CID, with subsequent decarboxylation resulting in formation of a carbon-centred radical.154 In these experiments (as previously demonstrated in Section 2.2.2.1), N,N,N- trimethyl-4-((2-thioxopyridin-N-yloxy)carbonyl)benzeneaminium cation is used as a precursor for 4-(N,N,N-trimethylammonium)phenyl radical cation (1) as described in Scheme 3.2. The amine functionality has been methylated in order to impart a ﬁxed charge and to minimise interaction between the charge tag and radical both sterically and electronically.

Experimentally, positive ion electrospray of an H2O/MeOH solution of N,N,N- trimethyl-4-((2-thioxopyridin-N-yloxy)carbonyl)benzaminium iodide resulted in an + + M ion at m/z 289 (Figure 3.1a). As shown in Figure 3.1(b), isolation of the M ion and + subsequent CID resulted in formation of an [M - 154] ion at m/z 135, consistent with loss of the Barton ester functionality to form the 4-(N,N,N-trimethylammonium)phenyl RESULTS 113

NMe3 I NMe3 NMe3 NMe3

ESI CID

- I - C5H4NS - CO2

N N m/z 135 O O O O O O 1 S m/z 289 S Scheme 3.2: Electrospray ionisation of N,N,N-trimethyl-4-((2-thioxopyridin-N-yloxy)carbonyl)benzenaminium (m/z 289) and subsequent CID leads to formation of the 4-(N,N,N-trimethylammonium)phenyl radical cation (m/z 135).

Figure 3.1: (a) ESI-MS spectrum of N,N,N-trimethyl-4-((2-thioxopyridin-N-yloxy)carbonyl)benzaminium cation + + in H2O/MeOH during positive mode electrospray resulting in an M ion at m/z 289. (b) Isolation of M and subsequent CID of m/z 289. radical cation (1).

The 4-carboxylatophenyl radical anion (2) has been generated previously by the oxidative decarboxylation of 1,4-benzenedicarboxylate dianion.135 1,4- benzenedicarboxylate dianion can be generated by electrospray ionisation of terephthalic acid or electron ionisation, chemical ionisation or laser desorption ionisation of dibenzyl terephthalate followed by subsequent collision induced dissociation (CID) to yield the distonic 4-carboxylatophenyl radical anion.130,135 The inert carboxylate charge tag and the inherent structural stability of the phenyl ring establish this molecule as a clear choice for investigating the reactivity of the phenyl radical in negative mode electrospray ionisation mass spectrometry.

Infusion of a MeOH/NH4OH solution of terephthalic acid into the mass spectrome- − − ter via the electrospray ionisation source resulted in [M - H] and [M - 2H]2 parent ions at m/z 165 and m/z 82 respectively (Figure 3.2a). CID of the dianion resulted in formation 114 GAS PHASE REACTIONS OF DISTONIC PHENYL RADICAL IONS

CO2H CO2

ESI (i) CID -2H+ -CO2 -e- CO2H CO2 m/z 82

CO2H CO2 PD CO2 ESI m/z 120 (ii) -I -H+ 2

I I m/z 247 Scheme 3.3: (a) Electrospray ionisation of terephthalic acid leads to formation of a dianion (m/z 82). Subsequent CID of the dianion results formation of the 4-carboxylatophenyl radical anion (2) at m/z 120. (b) Electrospray ionisation of 4-iodobenzoic acid leads to formation of the 4-iodobenzoate anion (m/z 247). PD of the isolated anion results in formation of 4-carboxylatophenyl radical anion (2) at m/z 120.

−. of an [M - 2H - CO2] radical anion at m/z 120 (Figure 3.2b). The ion at m/z 120 is consistent with previously measured CID spectra of 1,4-benzenedicarboxylate dianion, and thus assigned as the distonic 4-carboxylatophenyl radical anion (1).135,185,154,155

Figure 3.2: (a) ESI-MS of terephthalic acid in MeOH:NH4OH during negative mode electrospray resulting in an [M - H]− ion at m/z 165 and [M - 2H]2− ion at m/z 82. (b) Isolation of [M - 2H]2− and subsequent CID of m/z 82.

Similarly, this ion was also synthesised by photodissociation of 4-iodobenzoate anion. In this case, infusion of a MeOH/NH4OH solution of 4-iodobenzoic acid via − electrospray into the mass spectrometer resulted in an [M - H] ion at m/z 247 (Figure 3.3ii). This ion was subsequently isolated and subjected to a single 266 nm laser pulse − resulting in formation of an [M - I] . ion at m/z 120 consistent with formation of 2 (Figure 3.3b). RESULTS 115

Figure 3.3: (a) ESI-MS of 4-iodobenzoic acid in MeOH:NH4OH during negative mode electrospray resulting in an [M - H]− ion at m/z 247. (b) Isolation of the [M - H]1− ion and subsequent laser pulse at 266 nm results in formation of 4-carboxylatophenyl radical anion (2) at m/z 120.

Figure 3.4: Calculated geometry of each distonic phenyl radical ion.

3.4.2 Comparison of distonic phenyl radical ions to the neutral

analogue

A comparison of the neutral phenyl radical to both positive and negative charge-tagged phenyl radical ions synthesised in these experiments was undertaken by electronic structure calculations. These calculations, summarised in Figure 3.4, demonstrate that both neutral and distonic models have similar geometries with (a) the bond angle around the carbon radical centre within 1°, (b) bond lengths of within 0.05 Å.

Mulliken spin and charge distributions are often inaccurate due to a large theory and basis set dependence,278 however, they may give an idea of the localisation of spin and charge on a ion. Thoen et al., for example, compared Mulliken spin and charge distributions between neutral and distonic radical ions to support the distonic structure and electronic similarities between the distonic models and neutral phenyl radical.158 Mulliken spin and atomic charge distributions were calculated and are 116 GAS PHASE REACTIONS OF DISTONIC PHENYL RADICAL IONS

Charge: 0.915 Charge: -1.004

NMe3 CO2

(a)(b) (c)

Spin: 0.817 Spin: 0.793 Spin: 0.773 Charge: 0.453 Charge: 0.641 Charge: 0.262 1 2 Figure 3.5: Calculated spin and charge densities of each distonic phenyl radical ion. summarised in Figure 3.5. The spin distributions of the neutral and charge-tagged analogues were found to reside primarily on the dehydrogenated carbon suggesting a highly localised σ radical. The charge around the radical centre was found to be slightly more positive than the neutral phenyl radical for 1, while in the case of 2 the charge was slightly more negative. This suggests that in the positive charge-tagged case the radical centre is more electron deficient leading to a more electrophilic radical than the neutral phenyl radical, while conversely, the negative charge-tagged phenyl radical is expected to be less electrophilic. Increased electrophilicity of the radical centre has previously been associated with an increased reaction efficiency,148 suggesting that in the positive charge-tagged case the model phenyl radical may react more efficiently with nucleophilic neutral reagents.

These calculations suggest that the reaction rate of the distonic phenyl radical ion will differ to that of the neutral analogue, however, they have very similar structure and spin density which suggests that the distonic radical ions may model the reactions of neutral phenyl radical, albeit at different reaction rates. An experimental investigation using different charge-tags and variable distances between charge and radical was required to conﬁrm this hypothesis, which is discussed in detail later.

3.4.3 Structural characterisation of the distonic analogues

As introduced in Chapter 1, it is important to conﬁrm the structure of a radical ion before concluding anything about its behaviour. The reactions of DMDS and NO· were thus used as structural probes to conﬁrm the assignment of our distonic radical ions.

. The abstraction of thiomethyl ( SCH3) by distonic radical cations from 1,2- dimethyldisulﬁde (DMDS) has previously been used as an indicator of distonic char- RESULTS 117

O2C N

+ NO - CO2 O NO CO2 CO2 N

+ NO + NO

- CO2 m/z 120 m/z 106 m/z 136 NO NO NO Scheme 3.4: Overall scheme for reaction of 4-carboxylatophenyl radical anion with NO·.

+ acter.142 1 was allowed to react with DMDS and resulted in an [M + 47] ion at m/z 182 (Figure 3.6). Similarly, allowing 2 (formed by PD) to react in the presence of DMDS − (Figure 3.7) resulted in formation of an [M + 47] ion at m/z 167. In both cases this [M +

. 47] ion was indicative of SCH3 abstraction from DMDS.

NO· is expected to react by radical-radical combination with distonic radical ions. + Allowing 1 to react with NO· results in formation of an intense [M + 30] peak at m/z 165 − (Figure 3.8), while reaction of the 2 resulted in a minor [M + 30] peak at m/z 150 and an abundant [M + 16]− signal at 136 (Figure 3.9). Both [M + 30] ions are consistent with formation of an NO· addition product.

− Isolating of the abundant [M + 16] ion at m/z 136 formed during reaction of 2 with − NO· and subjecting it to CID resulted in a single [M - 30] ion at m/z 106 assigned as loss of NO· (Figure 3.10). Comparison of the reaction of unlabelled 4-carboxylatophenyl and 4-(13C)-carboxylatophenyl radical anion with NO· (Figure 3.11a and Figure 3.11b, respectively) reveals that the m/z 136 ion is not mass shifted when the carboxylate charge-tag is 13C labelled. This suggests that the [M + 16]− ion occurs after loss of

13 CO2. Both spectra contain an ion at m/z 106 demonstrating that, in the C case, this 13 – ion similarly contains no CO2 charge-tag. Isolation of the minor m/z 106 ion for 30 − ms resulted in the facile addition of NO· to form an [M + 30] ion at m/z 136 (data not shown). These data suggest one of two putative mechanisms may occur after initial · · addition of NO to 2, (i) direct displacement of CO2 by NO or (ii) loss of CO2 followed by subsequent addition of NO·.

Similarly, the two charge-tagged radical ions were allowed to react in the presence · of NO2. Figure 3.12 depicts the spectrum of 1 allowed to react in the presence of · NO2. Product ions present in this spectrum at m/z 136, 151, 165 and 167 are due to 118 GAS PHASE REACTIONS OF DISTONIC PHENYL RADICAL IONS

Figure 3.6: Isolation of the 4-(N,N,N-trimethylammonium)phenyl radical cation in the presence of 1,2- dimethyldisulﬁde. (*) represents ions present due to reaction with O2.

Figure 3.7: Isolation of the 4-carboxylatophenyl radical anion in the presence of 1,2-dimethyldisulﬁde.

Figure 3.8: Isolation of 4-(N,N,N-trimethylammonium)phenyl radical cation (m/z 135) in the presence of NO·. (*) represents ions present due to reaction with O2. RESULTS 119

Figure 3.9: Reaction of 4-carboxylatophenyl radical anion (m/z 120) with NO·. (*) represents ions present due to reaction with O2

Figure 3.10: CID of the m/z 136 ion formed during the reaction of 4-carboxylatophenyl radical anion in the presence of NO·.

Figure 3.11: Isolation of (a). 4-carboxylatophenyl, and (b). 4-13C-carboxylatophenyl radical anion in the · 13 presence of NO . Note the presence of the m/z 136 ion in (b) which indicates the CO2 has been lost. (*) represents ions present due to reaction with NO2. 120 GAS PHASE REACTIONS OF DISTONIC PHENYL RADICAL IONS

· the presence of NO and O2 in the buffer gas and are assigned as 4-oxocyclohexa- 2,5-dien-N,N-dimethyliminium cation, N,N,N-trimethylammoniumphenoxyl radical cation, and N,N,N-trimethyl-4-nitrosobenzaminium cation, N,N,N- trimethylammoniumphenylperoxyl radical cation, respectively. The characterisation and assignment of these ions are discussed in detail in the following section and + Chapter 4. The [M + 46] ion at m/z 181 is consistent with radical-radical addition of · NO2 to 1 and is assigned as N,N,N-trimethyl-4-nitrobenzaminium cation. · Allowing 2 to react in the presence of NO2 results in the spectrum shown in Figure 3.13. The major reaction product is an ion at m/z 138. The previous reaction between · · 2 and NO led to the hypothesis that this ion is similarly formed due to addition of NO2 followed by loss of CO2. This mechanism is summarised in Scheme 3.5 and as shown, · would result in an ion at m/z 122. This ion could presumably react with a second NO2 molecule, the energy of which could drive a loss of NO· to form 4-nitrophenoxide with m/z 138. The assignment of this ion follows comparative CID spectra between this m/z · 138 ion and the isobaric ion formed by reaction of NO2 with authentic 4-nitrobenzenide anions derived from 4-nitrobenzoic acid. The overall experiment is described in Scheme 3.6. 2 was synthesised by photodissociation of 4-iodobenzoate anion at 266 nm, as a high ion count was required for subsequent CID characterisation. For explanation, · NO2 depletes the 1,4-benzenedicarboxylate dianion by electron transfer, so while it is possible to form 2 by the usual oxidative decarboxlyation method, an optimal ion count was obtained by photodissociation. As shown in Figure 3.14, subjecting both m/z 138 ions to gradually increasing CID energy results in a major fragmentation onset to generate m/z 108 at the same CID energy. This suggests that the structures of both m/z 138 ions are identical. Furthermore, this suggests that the m/z 122 ion produced during · reaction of 2 with NO2 is indeed 4-nitrobenzenide anion.

As described in Scheme 3.5, four putative mechanisms could be invoked to describe · this overall process: (i) NO2 could directly displace CO2 after the preliminary formation of 4-nitrobenzoate followed by subsequent ipso-attack of the nitro oxygen to form a transient three-membered ring intermediate. This intermediate could presumably undergo β-scission resulting in formation of a nitrosophenoxide. The unpaired electron in the ring may then direct β-scission, ejecting NO· to reintroduce aromaticity. The process may alternatively proceed after 4-nitrobenzoate anion loses CO2, due to excess RESULTS 121

Figure 3.12: Isolation of 4-(N,N,N-trimethylammonium)phenyl radical cation (m/z 135) in the presence of · NO2.

· Figure 3.13: Isolation of 4-carboxylatophenyl radical anion (m/z 120) in the presence of NO2.

· · energy produced when NO2 adds to 2. (ii) NO2 may add to 4-nitrobenzoate anion via the nitrogen whereby the mechanism would proceed as just described. (iii) The electron

· · 279 afﬁnity of NO2 (Ea (NO2) = 2.273 eV) is higher than that calculated for 4-nitrophenyl · † (Ea (NO2C6H5) = 1.97 eV, see Appendix B Table B.1). The reaction pathway begins by · – electron transfer from 4-nitrobenzenide to NO2, generating NO2. The charge is centred – on the oxygen in this case and NO2 adds as a nucleophile after which the unpaired · β · electron on the ring displaces NO by -scission. (iv) The initial addition of NO2 drives · · loss of both CO2 and NO resulting in 4-oxylylbenzenide which may add NO2 to form 4-nitrophenoxide.

Overall, the formation of radical-radical combination products during reaction of

† · Calculation of the adiabatic electron afﬁnity of C6H5 at the B3LYP/6-311++G(d,p) level resulted in · a value of 1.087 eV, consistent with the NIST Chemistry Webbook which places the Ea of C6H5 at 1.0960 · ±0.0060 eV. This benchmark provides conﬁdence in the value calculated for NO2C6H5 radical at 1.97 eV. 122 GAS PHASE REACTIONS OF DISTONIC PHENYL RADICAL IONS

O O O O O O O2C NO2 N N N N O O

- CO2 - NO

NO2 NO2 NO2 NO2 NO2

(i) + NO (ii) + NO 2 2 O

O O N CO CO2 O O 2 N O + NO + NO2 - CO2 2 (iii) - NO m/z 138 m/z 120 NO NO NO 2 NO2 NO2 NO2 2 2 m/z 166 m/z 122

(iv) - NO

+ NO2 m/z 92

O O

· Scheme 3.5: Overall scheme of the reaction of 4-carboxylatophenyl radical anion (m/z 120) with NO2.

4-iodobenzoic acid

4-nitrobenzoic acid

10 [138][108]/

20 30 40

Normalised Collision Energy (arb)

Figure 3.14: Graph depicting [138]/[108] ion abundance ratio generated by CID of the m/z 138 ion synthesised · from 4-nitrobenzoic acid and 4-iodobenzoic acid in the presence of NO2. The dissociation onset is the same in both cases suggesting the m/z 138 generated in both cases is the same ion.

· · · NO and NO2 with 1 and 2 in addition to abstraction of (CH3S ) by both ions support the distonic characterisation of the 4-(N,N,N-trimethylammonium)phenyl radical cation and 4-carboxylatophenyl radical anion. In the case of the anion 2, there appears to be · · sufﬁcient energy produced during reaction of NO and NO2 to drive decarboxylation and as such this process may be expected during reaction of 2 with other neutral reagents including O2.

3.4.3.1 Reaction of the 4-(N,N,N-trimethylammonium)phenyl radical cation (1) · with NO2

· An interesting side experiment was undertaken to investigate whether NO2 adds via · ˜ 2 nitrogen or oxygen. The ground state of NO2 (X A1) formally places the unpaired · electron on the nitrogen. To add through the oxygen, NO2 would have to be excited 2 to its ﬁrst excited state (Ã B2) where the unpaired electron is located on oxygen, but RESULTS 123

CO2H CO2 CO2 CO2 O

ESI PD + NO2 + NO2 CID (i) + - H - CO2 - NO

m/z 120 I I NO2 NO2 NO2 m/z 166 m/z 122 m/z 138

CO2H CO2 O

ESI CID + NO2 CID (ii) + - H - CO2 - NO

NO2 NO2 NO2 NO2 m/z 166 m/z 122 m/z 138 Scheme 3.6: Experimental scheme for indirect comparison of the m/z 122 ions synthesised from (i) 4- nitrobenzoic acid and (ii) 4-iodobenzoic acid.

−1 280 · this requires signiﬁcant energy at 27.8 kcal mol . The presence of O2 and NO in the buffer gas presents a unique opportunity to determine how this reaction proceeds.

NMe3 NMe3 NMe2

+ NO2 CID (i) - H3C

m/z 135 N N O O O O m/z 181 m/z 166

NMe3 NMe3 NMe3 NMe3

+ O2 + NO CID (ii) - NO

m/z 135 O O O O N m/z 151 m/z 151 m/z 181 Scheme 3.7: Experimental scheme for comparison of the m/z 181 ion synthesised from (i) reaction · · of NO2 with 4-(N,N,N-trimethylammonium)phenyl radical cation and (ii) reaction of NO with 4-(N,N,N- trimethylammonium)phenyloxyl radical cation.

As summarised in Scheme 3.7, the nitrite (N,N,N-4-(nitrosooxy)benzaminium cation, m/z 181) may be synthesised by allowing N,N,N-trimethylammoniumphenoxyl radical cation (m/z 151) to react in the presence of NO·. Comparative CID spec- · tra of the m/z 181 ion synthesised by reaction of (a) 1 with NO2 and (b) N,N,N- trimethylammoniumphenoxyl radical cation with NO· are displayed in Figure 3.15(a) and (b), respectively. In the former case, the major fragment ions are [M - 15]+ and [M · · + - 46] assigned, respectively, as loss of CH3 and NO2. In the latter, [M - 30] , assigned as loss of NO·, is the sole product. These data demonstrate that the m/z 181 ion in each case is unique. An exclusive loss of NO· during CID of the nitrite is not surprising given the N-O bond is relatively weak and formation of a stable phenoxyl radical would drive 124 GAS PHASE REACTIONS OF DISTONIC PHENYL RADICAL IONS

Figure 3.15: Comparative CID spectra of m/z 181 ion synthesised from (a) authentic N,N,N-trimethyl-4- · nitrobenzaminium iodide, (b) reaction of 4-(N,N,N-trimethylammonium)phenyl radical cation with NO2 and (c) reaction of 4-(N,N,N-trimethylammonium)phenyloxyl radical cation with NO·.

· this reaction. The major loss of CH3 from the N-nitro derivative is telling, as it suggests − − + that the C NO2 bond is stronger than the CH3 N (CH3)2 bond. While a minor loss of · · NO is present, loss of NO2 appears more favourable which suggests either (i) there is little nitrite present in the m/z 181 ion population or (ii) that isomerisation to the nitrite derivative prior to fragmentation does not occur to any major extent.

3.4.4 Reaction of 4-(N,N,N-trimethylammonium)phenyl radical

cation (1) with O2

Isolation of 4-(N,N,N-trimethylammonium)phenyl radical cation (1) and trapping over periods of 30 - 7000 ms in the presence of adventitious O2 resulted in an abundant [M + + 32] ion at m/z 167 consistent with addition of O2 (Figure 3.16a). When the m/z 135 ion 18 + is isolated in the presence of O2 an [M + 36] ion at m/z 171 increases in abundance 18 + which is consistent with the addition of O2 (Figure 3.16c). The [M + 32] ion in Figure 3.16(a) was therefore assigned as the 4-(N,N,N-trimethylammonium)phenylperoxyl radical cation.

+ Minor products present in Figure 3.16a include an [M + O2 - 16] ion at m/z 151, [M + + + O2 - 29] ion at m/z 138, and [M + O2 - 43] ion at m/z 124. Repeating this reaction in the 18 18 + presence of O2 (Figure 3.16c) resulted in corresponding ions at [M + O2 - 18] ion at 18 + 18 m/z 153, [M + O2 - 31] ion at m/z 140 and [M + O2 - 45] ion at m/z 126, respectively. RESULTS 125

. . These ions were therefore assigned as loss of O atom, loss of CHO, and loss of CH2CHO, respectively.

Isolation of the 4-(N,N,N-trimethylammonium)phenylperoxyl radical cation at m/z + 167 and subsequent CID resulted in an abundant [M - 32] ion signal at m/z 135 consistent with loss of O2 and assigned as the parent 1 (Figure 3.16b). There was a similar array of product ions to those in the reaction spectrum (Figure 3.16a): [M + O2 + + + - 16] at m/z 151, [M + O2 - 29] at m/z 138 and [M + O2 - 43] at m/z 124, which were 18 similarly assigned after comparison to the CID spectrum of the O2 labelled 4-(N,N,N- 18 trimethylammonium)phenyl- O2-peroxyl radical cation (Figure 3.16d). There was no + [M + O2 - 44] ion (m/z 123) in the reaction (Figure 3.16a) nor in the CID spectrum of m/z 167 (Figure 3.16b) suggesting under the conditions of this experiment that direct

CO2 loss does not occur.

O O O O O C C O O C O O O (ii) (iii) (iv)

H H

Me3N Me3N Me3N Me3N Me3N

m/z 167 - C H (i) 2 2 - CO - C2H3O - HO

O O O C C C O O

N NMe3 NMe3 Me3N

m/z 124 m/z 124

- CH3 - CO

O O - CH3

Me2N m/z 124 N

. Scheme 3.8: Putative mechanisms for loss of CH2CHO from the 4-(N,N,N- trimethylammonium)phenylperoxyl radical cation. All mechanisms begin at the oxepinon-7-yl radical intermediate.

The [M + O2 - 43] loss observed during reaction of 1 with O2 was not observed during reaction with any other charge-tag (vide infra). As such, it seems likely that the N,N,N-trimethylammonium charge-tag plays in role in the mechanism. Four putative mechanisms for this loss are displayed in Scheme 3.8, two which involve the charge-tag (Scheme 3.8i and 3.8iv) and two which do not (Scheme 3.8ii and 3.8iii). All mechanisms follow formation of the oxypinon-7-yl radical intermediate. In the ﬁrst mechanism, (i) CO is ejected following β-scission after which the open chain undergoes ring closure 126 GAS PHASE REACTIONS OF DISTONIC PHENYL RADICAL IONS

Figure 3.16: (a) Reaction of 4-(N,N,N-trimethylammonium)phenyl radical with O2 and (b), subsequent CID of the 4-(N,N,N-trimethylammonium)phenylperoxyl radical cation (m/z 167). (c) Reaction 18 of 4-(N,N,N-trimethylammonium)phenyl radical cation with O2 and (d), subsequent CID of 4-(N,N,N- 18 trimethylammonium)phenyl- O2-peroxyl (m/z 171). RESULTS 127

. to form 4H-pyran. Ejection of CH3 results in formation of N-methyl-N-(4H-pyran- 4-ylidene)methanaminium cation. Both (ii) and (iii) result in formation of N,N,N- trimethyl-4-oxobut-3-en-1-yn-1-aminium cation. During (ii) the radical migrates such that it is alpha to the aldehyde. Concomitant 1,3-hydrogen atom transfer and β-scission results in ejection of 2-oxyethanyl radical to form the cation. In mechanism (iii) the radical migrates to the aldehyde oxygen followed by 1,5-hydrogen atom transfer with consecutive β-scissions resulting in ejection of acetylene and HO·. Finally, (iv) is similar to a mechanism proposed by Tokmakov et al. whereby the radical migrates to the oxygen followed by subsequent ring closure to form the 2H-pyran-2-carbonyl radical.55

. Ejection of CO and subsequent loss of CH3 yields the 4H-pyran as during mechanism (i). Mechanisms (i) and (iv) would be more likely to occur as they involve relatively loose transition states which would be kinetically favoured and lower in energy. While not undertaken during these experiments, evidence to support one of these two mechanisms could be obtained by synthesising the N,N,N-13C-trimethylammonium charge-tagged

. precursor. If a CH3 group was lost from the charge-tag as shown in mechanisms (i) and 13 (iv), reaction of this C-labelled ion with O2 would result in an ion containing only two 13C-labelled methyl groups resulting in a +2 Da shift relative to the unlabelled ion and generating an ion of m/z 126. If mechanisms (ii) or (iii) occurred the resulting ion would contain all three 13C-labelled methyl groups and the ion would arise at m/z 127.

3.4.5 Reaction of 4-carboxylatophenyl radical anion (2) with O2

Figures 3.17(a-d) illustrate comparative spectra resulting from isolation of (a) un-

13 labelled, (b) C-carboxylato labelled, and (c) D4-labelled 4-carboxylatophenyl in the presence of adventitious O2, while (d) depicts the reaction of unlabelled 4- 18 carboxylatophenyl with O2.

Isolation of unlabelled 4-carboxylatophenyl (Figure 3.17a) resulted in an abundant − signal at [M + 16] consistent with addition of O2 and loss of O atom. This assignment 18 was conﬁrmed in the spectrum measured during reaction with O2 (Figure 3.17d) − 18 which features an abundant [M + 18] ion consistent with addition of O2 and loss of an 18O atom. A set of peaks around m/z 152 is present in Figure 3.17(a) and is consistent with addition of O2, suggesting formation of 4-carboxylatophenylperoxyl radical anion. These peaks are shifted by around 4 Da during reaction of the 4-carboxylatophenyl 128 GAS PHASE REACTIONS OF DISTONIC PHENYL RADICAL IONS

18 radical anion with O2 (see Figure 3.17d) which provides further evidence to support the origin of these ions as an O2 adduct. The signal at m/z 151 may represent ejection of H· to form, for example, a 1,2- or 1,3-dioxetane after ortho- or meta-attack by the peroxyl radical (as depicted in Scheme 3.9). In both cases, however, these ions were not isolable, either due to their low abundance or instability, even when the isolation width was increased from the usual 1-2 Da mass window to a width of 10-15 Da. Therefore, further structural characterisation of these ions was not possible.

CO2 CO2 CO2

+O2 - H H O O O O

CO2 CO2 H O - H O

O O Scheme 3.9: Attack by the nascent phenylperoxyl radical at the ortho or meta position results in H-atom ejection.

This group of signals may alternatively occur if the ions were fragile leading to fragmentation of the ions after ejection from the ion trap but prior to detection. If fragmentation occurred during ejection, the kinetic energy of the ion would be redistributed between the fragments which may result in the ions reaching the conversion dynodes in a incoherent packet causing phantom peaks and peak broadening.

− The [M - 12] ion at m/z 108 present in Figure 3.17(a) represents a loss of CO2 from 4- carboxylatophenylperoxyl radical anion. This signal may occur due either to ejection of

CO2 from the carboxylate charge-tag or direct ejection of CO2 after isomerisation of the ¨ 18 phenylperoxyl radical (Scheme 3.1 B F). Reaction of 2 with O2 would be expected to − yield an [M - 12] ion if this signal occurred due to direct loss of CO2, however, no such ion was present. Conversely, there was an [M - 8]− ion indicative of an unlabelled loss

13 – of CO or CO2 from phenoxyl or phenylperoxyl radicals, respectively. The CO2 charge- tagged phenyl radical measured in the spectrum shown in Figure 3.17(b) features no − − [M - 12] ion but does contain an [M - 13] ion pointing to ejection of CO2 from the 13 C-labelled charge-tag. There is therefore, no evidence to suggest that CO2 is lost due to isomerisation of 4-carboxylatophenylperoxyl radical (c.f. Scheme 3.1 H ¨ F) under these experimental conditions. RESULTS 129

Figure 3.17: Reaction of (a) 3-carboxylatophenyl, (b) 4-(13C)-carboxylatophenyl and (c) 4-carboxylato- 18 2,3,5,6-tetradeuterophenyl with O2, and reaction of (d) 4-carboxylatophenyl with O2. 130 GAS PHASE REACTIONS OF DISTONIC PHENYL RADICAL IONS

The [M + 3]− ion featured in Figure 3.17(b) appears to occur due to the direct

. 13 ejection of CHO from the nascent phenylperoxyl radical. The CO2-labelled spectrum illustrated in Figure 3.17(a) features an [M + 3]− ion suggesting the ejected fragment contains no relation to the carboxylate charge-tag. An [M + 2]− ion was present in the

D4-labelled ring (Figure 3.17c) indicating one deuterium atom had been lost. Finally, Figure 3.17(d) contains an [M + 5]− ion, demonstrating the incorporation of extraneous

18 18 − O2 followed by a loss incorporating O. This conﬁrms that the [M+3] ion arises after . addition of O2 with subsequent loss of CHO.

3.4.6 Charge-tag effects on the reaction of distonic phenyl radical

ions with O2

To investigate the effect of the charge-tag on the reaction of the model phenyl radical, isophthalic acid, 4-sulfobenzoic acid, 4’-sulfobiphenyl-4-carboxylic acid and 4,4’- biphenyldicarboxylic acid were used as precursors to formation of 3-carboxylatophenyl (3), 4-sulfonatophenyl (4), 4’-sulfonatobiphen-4-yl (5) and 4’-carboxylatobiphen-4-yl radical anion (6) and allowed to react in the presence of O2 (Figure 3.19b-f). Impor- tantly, these ions represent models with different (a) charge-tags, (b) electronic coupling between the charge and radical and (c) distance between the charge and radical.

Electronic structure calculations including Mulliken charge and spin distributions were undertaken and are summarised in Figure 3.18. In all cases the unpaired electron was found to be highly localised at the radical centre suggesting in each case a σ-radical. The four additional charge-tagged models (3-6) were found to have a Mulliken charge distribution at the radical centre more negative than that of the neutral phenyl radical as was calculated for the 4-carboxylatophenyl radical anion (2). The 3-carboxylatophenyl radical anion was found to have very similar spin and charge distributions to the neutral phenyl radical.

Allowing the four additional charge-tagged phenyl radical ions (3-6) to react in the presence of O2 resulted in the spectra shown in Figure 3.19(a)-(d). All spectra contained − − · [M + 16] and [M + 3] ions consistent with addition of O2 and loss of O and CHO , 18 respectively. O2-labelled experiments all display a +2 Da shift which is consistent with these assignments (labelled spectra are displayed in Appendix A, Figure A.1). Reaction − of 4-6 with O2 resulted in very minor [M + 32] ions consistent with addition of O2 which RESULTS 131

Charge: -0.826 Charge: -0.445

CO2 SO3

Charge: -1.004 Charge: -1.036 Charge: -0.774

CO2 CO2 SO3

(a) (b) (c) (d) (e) (f)

Spin: 0.817 Spin: 0.773 Spin: 0.800 Spin: 0.788 Spin: 0.780 Spin: 0.778 Charge: 0.453 Charge: 0.262 Charge: 0.408 Charge: 0.168 Charge: 0.171 Charge: 0.357 2 3 4 5 6 Figure 3.18: Structure, charge and spin distributions of (a) phenyl radical and (b) 4-carboxylatophenyl, (c) 3- carboxylatophenyl, (d) 4’-carboxylatobiphen-4-yl, (e) 4-sulfonatophenyl, and (f) 4’-sulfonatobiphen-4-yl radical anions. Mulliken Charge and spin distributions calculated at the B3LYP/6-311++G(d,p) level of theory.

18 are shifted by +4 Da during O2 labelling experiments conﬁrming the incorporation of

O2. Similar to reaction of the 4-carboxylatophenyl radical anion (2) with O2 (shown previously in Figure 3.17a), [M + 31]− ions were also observed suggesting either ejection of H atom from the [M + 32]− adduct, or a pseudo-stable ion.

The major peak observed during reaction of 3 was the [M + 16]− ion, consistent with addition of O2 and loss of O. In contrast, the major ion during reaction of 5 and 6 was − . the [M + 3] ion, suggesting addition of O2 and loss of CHO. In all four cases there − 18 was an [M - 12] ion consistent with addition of O2 and loss of CO2. However, O2- labelling experiments displayed +4 Da shifts during reaction of 3 and 4, demonstrating that the ions contained extraneous oxygen and therefore could not arise by this pathway (Appendix A, Figure A.1). While the [M + 12]− ion observed during 5 and 6 is consistent with ejection of CO2 from the charge-tagged phenylperoxyl radical, it is a minor ion over more than two orders of magnitude lower in abundance than the dominant [M + 3]− ion.

As shown in Figure 3.19(d), the reaction of 4 with O2 resulted in a number of paired ions separated by 3 Da including m/z 156 and 159, m/z 184 and 187, m/z 196 and 199,

18 and ﬁnally m/z 212 and 215. O2-labelling experiments show that all of these ions arise due to incorporation of oxygen. Furthermore, the ions at m/z 187, 199 and 215 appear to contain more than two labelled oxygens suggesting that these ions arise from multiple additions of O2 (Appendix A, Figure A.1d). These data were in direct contrast to all other charge-tagged phenyl radical ions investigated, where there was no evidence of more than a single O2 addition. Discussion of the mechanism and structure of these tri- 132 GAS PHASE REACTIONS OF DISTONIC PHENYL RADICAL IONS

Figure 3.19: Isolation of the (a) 3-carboxylatophenyl (3), (b) 4-sulfonatophenyl (6), (c) 4-sulfonatobiphenyl (5), and (d) 4-carboxylatobiphenyl (4) radical anions in the presence of O2. RESULTS 133

and tetra-oxygenated ions is beyond the scope of this investigation, however, putative mechanisms for each of these ions is given in Appendix A, Scheme A.1.

Importantly, all charge-tagged models 1-6 display common product ions consistent

. with addition of O2 followed by loss of either O and CHO. While reaction of sulfonato charge-tagged models (5 and 6) resulted in ions consistent with loss of CO2 following addition of O2 these ions were orders of magnitude lower in abundance than the − dominant [M + O2 - CHO] ion. This suggests that (a) direct CO2 loss is only a minor reaction pathway during this reaction and (b) the major reaction pathway is either loss of O or loss of .CHO from the nascent peroxyl radical.

3.4.7 Discussion

The reactions of negative and positive charge-tagged phenyl radicals with O2 were found to follow two common branching pathways. The initial reaction of the radical cation 1 with O2 was found to result in formation of a stable and isolable phenylperoxyl ¨ radical (A B), while product ions present after reaction of 2 with O2 suggested formation of a phenylperoxyl radical intermediate. Independent of the charge-tag was formation of the phenoxyl radical following ejection of O (B ¨ C). The phenoxyl radical was the major degradation product observed for both 1 and 2. The second pathway observed was loss of .CHO presumably resulting in formation of cyclopentadienone (M).

There was no evidence to suggest that the formation of cyclopentadienone (M) followed the two mechanisms proposed by Fadden and Hadden (shown in Scheme 3.1),57,58 as both mechanisms require formation of intermediates which were not observed. Using the reaction of 2 with O2 as an example, the ﬁrst channel would proceed through the intermediates F¨K¨L which would require formation of ions at ¨ [M + O2-CO2] and [M + O2-CO2+O2]. The second mechanism proceeds through O P or Q, suggesting an [M + O2-CO] ion should be present should the reaction follow this mechanism. ¨ The CO2 loss channel (B F) was unexpectedly absent from these reactions. Reac- tion of the 4-(N,N,N-trimethylammonium)phenyl radical cation with O2 (Figure 3.16) + showed no [M - 12] ion to indicate an addition of O2 and loss of CO2. Analysing the product ions generated during the reaction of the negative charge-tagged radicals 134 GAS PHASE REACTIONS OF DISTONIC PHENYL RADICAL IONS

18 with O2 revealed that in all cases, except that of the 4-sulfonatobiphenyl radical, the [M - 12]− ion measured did not form by this reaction channel. In the case of the − 4-sulfonatobiphenyl radical, an ion at [M + O2 - CO2] is present in the unlabelled 18 18 − spectrum, and as would be expected, there is a corresponding [M + O2 -C O2] ion 18 measured during reaction with O2. However, this ion is lower in abundance than the 18 18 − [M + O2 - CH O] signal by two orders of magnitude.

3.4.8 Kinetics of distonic phenyl radicals with O2

Reaction kinetics were measured for the reactions of distonic ions 1 - 6. The second- order rate constants derived from the pseudo-ﬁrst order rates measured during these experiments is reported in Table 3.1. The rates recorded for 2 and 3 were calculated using the natural logarithm of the parent ion decay and were not normalised to the ion count. Yu and Lin measured a second order rate for the reaction of phenyl radical = − × −11 −1 3 −1 with O2 at kO2 (1.13 2.50) 10 molecules cm s (310 K). When considering an absolute error in these calculations of 50 % (errors discussed in Section 2.3.7), the rates measured for all negative charged-tagged phenyl radicals except 2 and 3 were = × −11 in good agreement with this value. The values measured for 2 of kO2 8.0 10 −1 3 −1 = × −11 −1 3 −1 molecules cm s and for 3 of kO2 6.0 10 molecules cm s were around two times larger. Comparatively, Sommeling et al. measured both solution and = × −12 ≥ −13 gas phase rates for the reaction of 2 with O2 at kO2 6.3 10 and kO2 10 molecules−1cm3s−1, respectively.281 These rates are signiﬁcantly lower than the rate measured during these experiments.

Efficiencies for the reaction of radical ions 1 and 4 - 6 with O2 were similar at between 3.9 and 5.3%, while those for 2 and 3 were 13 and 10%, respectively. This suggests that a collision of O2 and 2 or 3 is 5 - 8% more likely to result in a successful reaction. A reaction efficiency of 6% was estimated for the neutral phenyl radical using the second order rate constant reported by Yu and Lin39 and an estimated hard sphere collision rate of 1.81 x 10−10 molecules−1 cm3 s−1.‡ While there are obvious differences between the neutral and charged phenyl radical systems, the efficiencies calculated for 1 and 4 - 6 were surprisingly consistent with this value. All efficiencies were lower than that of the bridgehead aliphatic model of 3-carboxylatoadamantyl radical anion (previously

‡ = πσ2 8kB T 1/2 σ2 = + 2 Hard sphere collision rate estimated using khs AB ( πµ ) , where AB (RA RB ) . Molecular radii for benzene (R = 1.75 Å) and O (R = 1.55 Å) were obtained from O’Hanlon.282 C6H6 2 O2 RESULTS 135

Radical Rate Constant k2 Collision Rate kcol Efﬁciency Φ − − − − (molecules 1cm3s 1)(molecules 1cm3s 1) (%)

− − phenyl§ (1.13-2.50) x 10 11 1.81 x 10 10 6 − 4-carboxylatophenyl# ≥ 10 13 − − 3-carboxylatoadamantyl† 8.5 x 10 11 5.66 x 10 10 15 − − (1) 4-(N,N,N-trimethylammonium)phenyl 5.3 x 10 11 5.79 x 10 10 5.3 − − (2) 4-carboxylatophenyl‡ 8.0 x 10 11 5.86 x 10 10 13 − − (3) 3-carboxylatophenyl‡ 6.0 x 10 11 5.86 x 10 10 10 − − (4) 4-carboxylatobiphenyl 2.8 x 10 11 5.62 x 10 10 5.0 − − (5) 4-sulfonatobiphenyl 2.8 x 10 11 5.55 x 10 10 5.0 − − (6) 4-sulfonatophenyl 2.2 x 10 11 5.72 x 10 10 3.9

Table 3.1: Comparison of 2nd order rate constants and efﬁciencies for the ion molecule reactions of distonic § 39 # 281 † radical ions with O2. Value from Yu and Lin. Value from Sommeling et al. Value is the reaction with 18 155 ‡ O2, refer to Harman and Blanksby. The ion abundance was not normalised to the total ion population, see Section 2.3.5. measured at 15%) which suggests that relative to its collision rate the aliphatic model is more reactive towards O2.

As shown in Figure 3.21(i) and (ii), the pseudo-ﬁrst order decay plot of 2 and 3 reacting with O2 was found to possess signiﬁcant curvature when normalising the ion abundance of the parent radical to total ion abundance. In stark contrast, reaction of

4-sulfonatophenyl radical anion with O2 resulted in the linear pseudo-first order decay plot shown in Figure 3.21(iii). There appeared to be two distinct components to the plots of 2 and 3 suggesting that two parallel pathways may contribute to the reaction rate. Plotting the ion count of each ion over time resulted in the graph shown in Figure 3.22. Notably, the sum of the product ions does not account for the decay of the parent ion, which suggests that there are products formed which are not isolated within the ion trap. To account for low mass ions, the instrument was set to low mass mode, which decreases the low-mass cut off of m/z 30 while isolating an ion of m/z 120, but requires separate tuning parameters, tends to lower the ion count, and reduces the maximum measurable ion to a m/z of 200 Da or lower. There were no ions, however, of significantly high abundance to account for the curvature in the decay plot. These data, and the lack of curvature in the unnormalised pseudo-first order decay plot suggest any ion losses are occurring after addition of O2.

If the electron binding energy of a minimum was lower than the reaction enthalpy of its formation then it follows that the intermediate may undergo electron 136 GAS PHASE REACTIONS OF DISTONIC PHENYL RADICAL IONS

O O S O

120 O O C

100 O O S O 80 O O O O O O 60 O C 70.0 O O O +SO C O S O O 3 O 40 46.3 O O O O 37.7 +O C +O 20 2 O O 2 O S O O 20.7 +CO2 O +SO3 (kcal/mol) O O

0 0.0 0.0 -4.5 O -5.7 -20 O Ac O -13.1 O O C O O S Energy O O -40 +CO O O 2 -45.2 -60 -56.5 -50.6 Bc O O -80 O O

-100 -98.3 -97.2 -120 Hc

Figure 3.20: Electron binding energy for the 4-carboxylatophenylperoxyl, 5-carboxylatooxepinon-7- yl and oxepinon-5-id-7-yl radical anions relative to 4-carboxylatophenyl radical anion + O2 and 4- sulfonatophenylperoxyl, 5-sulfonatooxepinon-7-yl and oxepinon-5-id-7-yl radical anions relative to 4- sulfonatophenyl + O2. Calculations were undertaken using G3SX theory. detachment, losing its charge, without producing any products measurable by mass spectrometry. The electron binding energy of the 4-carboxylatophenylperoxyl and 5-carboxylatooxepinon-7-yl radical anions, intermediates of the 4-carboxylatophenyl radical anion + O2 reaction where this anomaly occurs, and 4-sulfonatophenylperoxyl and 5-sulfonatooxepinon-7-yl radical anions, where it did not, were calculated at the composite G3SX level of theory to investigate this hypothesis. The 4- carboxylatophenylperoxyl radical anion resides -45.2 kcal mol−1 below reactants, with an electron binding energy of 91.5 kcal mol−1 (see Figure 3.20), while the rearranged 5-carboxylatooxepinon-7-yl radical anion, which resides in a deeper potential well - 98.3 kcal mol−1 below reactants, has an accessible electron binding energy of 93.8 kcal mol−1. In contrast, the electron binding energy of the 4-sulfonatophenylperoxyl and 5- sulfadionatooxepinon-7-yl radical anions were predicted at 70 kcal mol−1 and 20.7 kcal mol−1 higher than the energy of the reactants.

It is also possible that electron attachment occurs subsequent to decarboxylation due to formation of a loosely bound anion, for example, intermediates Hc and Nc . Decarboxylation results in products -56.5 kcal mol−1 and -58.2 kcal mol−1 below barrier, respectively, while the binding energies were calculated at 43.4 kcal mol−1 and 57.2 kcal mol−1, respectively (shown in Figure 3.20). This suggests that the reaction energy is sufﬁcient to drive electron detachment from these intermediates. Ejection of the sulfonato group from 5-sulfonatooxepinon-7-yl radical anion results in a product -5.7 kcal mol−1 below reactants, well below the energy required for electron detachment (see RESULTS 137

-1

(a)

-2 ln([120]/TIC)

-3

0 100 200 300 400 500

Reaction Time (ms)

-1

-2

(b) ln([120]/TIC)

-3

-4

0 100 200 300 400 500 600

Reaction Time (ms)

0.0

-0.2

-0.6

0 50 100 150

Reaction Time (ms)

Figure 3.21: Kinetic plot of a) the reaction of 4-carboxylatophenyl, b) 3-carboxylatophenyl and c) 4- sulfonatophenyl radical anions with O2

40000

35000

30000

120

25000

123

20000 136 Ion Count 15000

10000

5000

0 100 200 300 400 500

Reaction Time (ms)

Figure 3.22: Plot of the ion count of the four major ions observed during reaction of 4-carboxylatophenyl radical anion (2) with O2. 138 GAS PHASE REACTIONS OF DISTONIC PHENYL RADICAL IONS

Figure 3.20). These predictions account not only for the curvature of the normalised pseudo-ﬁrst order decay plot, but may also explain the low yield of the cyclopentadienone Nc (compared with sulfonato charged-tagged phenyl radicals 5 and 6), as decarboxylation and electron detachment of Nc would result in a low apparent yield of this product. These calculations support the hypothesis that electron detachment occurs after rearrangement of the 3- and 4-carboxylatophenylperoxyl radical anion, which ultimately accounts for the curvature observed in the normalised pseudo-ﬁrst order decay.

3.4.9 Calculation of the 4-carboxylatophenyl radical anion + O2 potential energy surface

The following calculations were undertaken in an attempt to account for the absence

. of a direct CO2 loss channel and the presence of both O atom and CHO loss channels during mass spectrometric measurements of the reaction of phenyl and O2. Both loss of O atom and .CHO have been proposed as high temperature processes, and have not been observed in reactions undertaken at temperatures below 473 K. The following calculations suggest, however, that both loss of O and CHO· may be competitive with the long-standing mechanistic and computationally determined pathways which predict loss of CO2 as the dominant channel. The potential energy surfaces of the phenyl +

O2 and 4-carboxylatophenyl + O2 reactions (Figure 3.23 and Figure 3.24, respectively) were calculated at the B3LYP/6-311++G(d,p) level of theory which included unscaled zero-point energies. The geometries, frequencies and energies of the maxima and

55 minima on the phenyl + O2 potential energy surface calculated by Tokmokov et al. were augmented with the cyclopentadienone + .CHO channel and recalculated at a consistent level of theory. In the following section, bold letters X denote structures on the neutral potential energy surface while a letter with a subscript Xc denotes the charge-tagged potential energy surface.

There were minor disparities found between the neutral and carboxylato charge- tagged surface. The neutral surface was found to contain a minimum S before formation of T, while on the carboxylato surface Tc was formed directly from the 4- carboxylatophenylperoxyl radical (Bc ). On the charge-tagged surface, the formation of H from G was found to include an additional minimum (Uc ), while generation of RESULTS 139

cyclopentadienyl (F) was found to proceed directly from I. These additional minima all occur in ﬂat regions of their respective surfaces and lie in shallow potential wells. The presence of these additional minima may be theory and/or basis set dependant and are unlikely to signiﬁcantly perturb the product distribution.

On neutral and charge-tagged potential energy surfaces, loss of atomic O is shown

−1 −1 to have barriers of 33.7 kcal mol and 32.5 kcal mol for B and BC , respectively, and in both cases were found to lie below the energy of the reactants. Given the high exothermicity of the other reaction channels the abundance of phenoxyl radical observed during reaction of 2 with O2 may be surprising, however, two points may be raised. Firstly, the barrier height of the reaction (32.5 kcal mol−1) is lower than

. −1 those encountered during loss of CO2 or CHO (41.7 and 47.8 kcal mol , respectively). Secondly, as explained by Albert and Davis,262 all other reaction pathways require multiple isomerisation steps through tight transition states, whereas formation of phenoxyl, where the energy of the products is much higher, passes through a single loose transition state, and is therefore kinetically favoured.

. Reaction pathways resulting in loss of CO2 and CHO must pass through the oxepinon-7-yl radical which is calculated to reside in a deep potential well of -90.3

−1 (H) and -100.0 kcal mol (Hc ) on the neutral and charge-tagged surfaces, respectively.

The highest barrier encountered during loss of CO2 and formation of cyclopentadienyl (F) was the tight transition state associated with ring closure of oxepinon-7-yl radical (H¨I, shown in Figure 3.25a) and 5-carboxylatooxepin-2-on-7-yl radical anion

−1 −1 (Hc ¨Ic ), which require 39.1 kcal mol (-51.2 kcal mol below reactants) and 41.7 kcal mol−1 (-58.3 kcal mol−1 below reactants) respectively.

The highest barrier encountered on the cyclopentadienone (M) formation pathway is the loose transition state associated with ejection of .CHO requiring 36.5 kcal mol−1 (Q¨M, shown in Figure 3.25b; -51.6 kcal mol−1 below reactants) and 37.8 kcal mol−1

−1 (Qc ¨Mc ; -63.2 kcal mol below reactants) on the neutral and charge-tagged surfaces respectively. On both potential energy surfaces lower barriers are associated with loss

. . of CHO compared to CO2 loss. This suggests that loss of CHO is thermodynamically favoured. Furthermore, by comparison of the geometry of the largest transition state barriers found on each reaction pathway, loss of CO2 encounters a tight, ring-closing transition state (Figure 3.25a), while ejection of .CHO passes through a loose, β-scission 140 GAS PHASE REACTIONS OF DISTONIC PHENYL RADICAL IONS 1+Gdp ihZEcorrection. ZPE with 311++G(d,p) al; et. Tokmakov from taken 3.23: Figure

Energy (kcal/mol) -120 -110 -100 -150 -140 -90 -60 -30 -80 -70 -50 -20 -10 -40 10 0 h nmlclrdgaaino h hnleoy aia.Rltv nrisicueZEcreto n acltda h 3Y/-1+Gdp ee.(Black, level. B3LYP/6-311++G(d,p) the at calculated and correction ZPE include energies Relative radical. phenylperoxyl the of degradation unimolecular The 0.0 A +O 2 -42.4 B O O 55 le adnadHda mechanism, Haddad and Fadden Blue, [-15.3] [-8.7] [3.3] (-5.0) 261 ) O -19.9 (-8.3) -8.8 2.1 G O S O O H [-18.9] [1.9] (-6.7) (G2M) -60.1 -6.4 O T O H O C [-58.8] + O ( 3 P) -90.3 H O 58 O eacltda 3Y/-1+Gdp ihZEcreto;Rd ebre l,rcluae tB3LYP/6- at recalculated al., et Sebbar Red, correction; ZPE with B3LYP/6-311++G(d,p) at recalculated [-65.7] [-51.2] O -74.8 C -80.6 O I O H N O [-65.4] [-57.0] [-38.0] J -61.7* -41.3 + CO -88.1 O R C O H OCO O O H [-51.6] [-46.5] [-39.4] -107.5 -74.3 -52.4 M C F P O + CHO + CO + CO O 2 [-63.1] H -91.3 H Q C O + CO [-31.7] -32.3 M C O + H + CO RESULTS 141 2 +CO + CHO O O c c O F M O -59.7 -109.5 O O O O [-71.0] H O [-63.2] H O O O c R O 101.0 [-77.9] [-82.2] O O O O H c c I C N O O O O -85.7 -82.5 [-75.8] [-58.3] O c O H O -100.0 O [-36.8] [-69.8] O H O O c c O T U O -70.4 O -37.6 OO [-22.7] P) c O 3 O c G C O + O ( O -27.7 O O O -16.2 [-2.7] [-25.8] O O c O B -48.7 O 2 +O O c A 0.0 O The unimolecular degradation of the 4-carboxylatophenylperoxyl radical. Relative energies include uncorrected ZPE correction and calculated at the B3LYP/6- 0

-40 -10 -20 -50 -70 -80 -30 -60 -90

-140 -150 -100 -110 -120 ) kcal/mol ( Energy Figure 3.24: 311++G(d,p) level. 142 GAS PHASE REACTIONS OF DISTONIC PHENYL RADICAL IONS

(a) Ring contraction of oxepinon-7- (b) .CHO leaving to form cyclopenta- yl. (H¨I) dienone. (R¨M)

Figure 3.25: Optimised transition state structures calculated at the B3LYP/6-311++G(d) level of theory. The dashed line represents changes in the bonding at the transition state. transition state (Figure 3.25b), suggesting the latter may also be kinetically favoured.

The entropy of the competing transition states following the intermediate oxepinon- ‡ 7-yl radical (H) were examined, and the vibrational partition function QV calculated for both channels (shown in Figure 3.26; these calculations are described in Section 3.2.3). The vibrational partition functions calculated with and without the N,N,N- trimethylammonium and carboxylato charge-tags were found to be higher during the β-scission transition state (Figure 3.26b). This suggests that the β-scission transition state has a higher number of low energy vibrational states with which to accommodate transmission through this transition state. This provides evidence to suggest that formation of cyclopentadienone, which depends on passage through the (H¨N) transition state, will be kinetically favoured.

Considering its large reaction exothermicity, direct ejection of CO2 from the nascent phenylperoxyl radical may present an attractive means of explaining the presence of

42 CO2 in previous studies, however, current experiments and calculations suggest that loss of CO2 may not proceed by this mechanism at lower temperatures.

3.5 Conclusion

The reaction of the phenyl radical with O2 was investigated using distonic radical analogues to elucidate the initial steps in the oxidative degradation of benzene. A positive charge-tagged analogue of a key intermediate in this reaction, the phenylperoxyl radi- CONCLUSION 143

Figure 3.26: The transition states and partition function calculations of the two competing reaction pathways after formation of oxepinon-7-yl.

cal, was successfully isolated. Second-order rate constants for the addition of O2 to the = − × −11 −1 3 −1 charge-tagged analogues were measured at kO2 (2.2 8.0) 10 molecules cm s , the lower values are in good agreement with previously reported rates for the reaction = − × −11 −1 3 −1 of phenyl radical with O2 of kO2 (1.13 2.50) 10 molecules cm s at 310 K.

At low temperatures, reaction of the phenyl radical with O2 is believed to be dominated by the addition-stabilisation reaction resulting in formation of the phenylperoxyl radical.39 If the phenylperoxyl radical is stabilised in the troposphere, degradation will most probably be dominated by photolytic pathways to yield phenoxyl radical and atomic oxygen.283,284 Under these experimental conditions only reaction of the positive charge-tagged (N,N,N-trimethylammonium)phenyl radical cation resulted in formation of a stabilised phenylperoxyl radical. CID of this phenylperoxyl radical

. . resulted in loss of O, CHO, O2 and CH2CHO. During the negative charge-tagged − experiments an [M + O2] ion was observed but was very low in abundance and was not isolable. However, an array of ions include losses of O and .CHO arising from degradation of a vibrationally excited phenylperoxyl radical were observed.

While phenoxyl radical has not been observed in previous low temperature studies, the major degradation product identiﬁed during degradation of N,N,N- trimethylammonium, 3-carboxylato and 4-carboxylatophenyl radical ions was loss of O 144 GAS PHASE REACTIONS OF DISTONIC PHENYL RADICAL IONS

to form an ion assigned as their respective charge-tagged phenoxyl radical ion. The absence of phenoxyl radical in previous experiments39 may be due to secondary reaction of the phenoxyl radical with neutral hydrocarbons present during these experiments.

While H2O and other solvents used during electrospray ionisation (such as MeOH) are present in the ion trap, as demonstrated by the proton transfer and H2O complexes typically present during isolation of 1,4-benzenedicarboxylate dianion, it would appear that their reactivity towards the phenoxyl radical is minimal. Biomolecular reactions with other analytes is not possible as any isolated species will be charged and thus undergo Coulombic repulsion prior to reaction. As such, under the conditions of this experiment the phenoxyl radical is detected prior to any secondary reaction. Previous experiments were also undertaken at a higher pressure (20 - 80 Torr39 as opposed to 2.5 mTorr during this experiment), which may aid in stabilisation of the nascent phenylperoxyl radical, in which case no degradation products would be observed.

A minor product observed in most cases, and the major ion during reaction of

4-sulfonatophenyl and 4-sulfonato-4’-biphenyl was the addition of O2 and loss of .CHO. This product had only been observed previously in high temperature combustion studies. A mechanism consistent with its formation at ambient temperatures was recently suggested by Sebbar et al.261 These experiments were the ﬁrst direct observation of the loss of .CHO during the degradation of the phenylperoxyl radical. A comparison of this reaction pathway with that of the formation of cyclopentadienyl radical was therefore undertaken by calculation of the potential energy surface of the neutral and negative charge-tagged analogues of the phenyl radical with O2. These calculations predicted ejection of .CHO and formation of cyclopentadienone from the phenylperoxyl radical and 4-carboxylatophenylperoxyl radical anion to be competitive both thermodynamically and entropically with the widely favoured pathway involving direct ejection of CO2 to form cyclopentadienyl radical. Addition of the charge-tag appeared, thermodynamically, to promote the CO2 ejection pathway suggesting that if a direct loss of CO2 was to occur during these experiments that it would in fact be more energetically favourable than during neutral experiments. These data suggest that direct CO2 loss from the phenylperoxyl radical may not be signiﬁcant to the production of cyclopentadienyl at low temperatures. This would suggest that cyclopentadienyl is generated primarily from thermal decomposition of the phenoxyl radical. Furthermore, CONCLUSION 145

· formation of CO2 may instead form by an indirect mechanism involving CHO or CO . + · −−→ + · 285,286,287 + · −−→ + · 288,289,290 (e.g. CHO HO CO HO2, CO HO CO2 H ).

Overall, these ﬁndings suggest that at low temperatures phenylperoxyl radical in most cases degrades back to reactants. While this process requires more energy than other degradation channels, it is entropically favourable.§ However, if the phenylperoxyl radical degrades to products the major unimolecular degradation product is phenoxyl radical. There was little evidence to support direct loss of CO2, generally held to be the most favourable reaction pathway during this process, which suggests that this pathway may not be as signiﬁcant as previously predicted. A computational investigation demonstrated that loss of .CHO is energetically favourable and competitive with loss

. of CO2. In addition, experiments provided evidence to support CHO loss directly from the nascent phenylperoxyl radical.

§ ∆ = −1 −1 ∆ = Calculation at B3LYP/6-311++G(d,p) level of theory shows S AC 131.1 cal mol K and SCC − − 127.9 cal mol 1 K 1.

otesgaueboncluainascae ihpoohmclso.Furthermore, smog. photochemical with toxiﬁca- associated and colouration brown injury signature the lung to to leads ppm) 500 - (100 doses tion. high in and ppm) 0.5 (< tions NO doses long. carbons four than atmosphere larger the in is signiﬁcant group be R to the believed when only is but troposphere, the in formation RONO of formation The (ROO radicals Peroxyl 4.1 ie n ipydcmoe akt reactants. to back decomposes simply and lived (RONO nitrate stable the to isomerisation RONO form can NO generate to decompose can ROONO excited vibrationally the of mation ROO to NO of addition 4.1, Scheme NO 297,298 2 · Introduction HNLEOY AIASWITH RADICALS PHENYLPEROXYL ocnrtosi h rpshr r motn o ayraos nlow In reasons. many for important are troposphere the in concentrations 2 · NO stxcaddrcl soitdwt h xcraino eprtr condi- respiratory of exacerbation the with associated directly and toxic is G 2 · SPAERATOSO DISTONIC OF REACTIONS PHASE AS soecmoeto rpshrcarpluinaddrcl contributes directly and pollution air tropospheric of component one is 2 lentvl,teprxntiemyudrodrc unimolecular direct undergo may peroxynitrite the Alternatively, . · r eitr fN xdto ntetoopee ssonin shown As troposphere. the in oxidation NO of mediators are ) 2 csa ikfrNO for sink a as acts 2 · n ihyratv RO reactive highly and · aiasi o eeal eonsdt euti for- in result to recognised generally now is radicals 147 2 * ). oeue hsprxntieintermediate peroxynitrite This molecule. 15,241,291,242,240,245,246,239,292,293,294,295,296 · n a hrfr opt ihNO with compete therefore may and 5 hnRi ml,RONO small, is R When · aias RO radicals; · recombination NO 2 C HAPTER sshort is 4 X 2 · 148 GAS PHASE REACTIONS OF DISTONIC PHENYLPEROXYL RADICALS WITH NOX

· the concentration of NO2 in the troposphere is proportional to the concentration of · tropospheric O3 as photoactivation of NO2 at wavelengths below 420 nm leads to 3 formation of O( P) which reacts with O2 to form O3.

Changes in the concentration of O3 in the troposphere also has wide ranging effects. > 1 O3 is a greenhouse gas which absorbs in the near UV ( 336 nm) to generate O( D) 1 + 1 −−→ · · and O2. O( D) reacts with water vapour (H2O O( D) 2 HO ) to form HO which · subsequently oxidise hydrocarbons. In this manner, O3 acts as a source of HO which aid 5 in cleansing of hydrocarbons from the atmosphere. O3 is toxic to animals and humans * and as a consequence high concentrations of O3 in the troposphere is harmful to living organisms.297,300,15

· + · → ∗ RO2 NO ROONO (4.1) ∗ → · + · ROONO RO NO2 (4.2) ∗ + → ROONO M RONO2 (4.3) · + · → RO NO2 RONO2 (4.4) → + RCH2ONO2 RCHO HONO (4.5)

Scheme 4.1: Reaction of alkyl or arylperoxyl radicals with NO· proposed by Stimac and Barker where an asterisk denotes a vibrationally excited molecule and M is a non-reactive collision partner.291

The reaction of ROO· radicals with NO· has been widely debated in the literature. The uncertainty lies in the formation and subsequent degradation or isomerisation of the nascent peroxynitrite (ROONO). For example, statistical rate theory models based on two competing reaction pathways (Reactions 4.2 and 4.3) have been unable · to satisfactorily explain the relative concentrations of RONO2 and NO2 measured in the troposphere.301 Despite this fact, there are three generally held theories as to how the reaction of peroxyl radicals with NO· proceeds: (i) Reactions 4.2 and 4.3 proceed through ˜ 2 a common tight transition state involving an avoided crossing of the ground (X A1) and 2 · · ﬁrst excited states (Ã B2) of NO2 resulting in either dissociation of ROONO to RO and · 294 NO2 or isomerisation to the nitrate RONO2 (the Ellison model, shown in Figure 4.1);

(ii) RONO2 forms by a two-step mechanism whereby the reaction ﬁrst proceeds by −−→ · + . dissociation (ROONO RO NO2) through the aforementioned transition state and · + . −−→ concludes by subsequent recombination of the two products (RO NO2 RONO2; *Ozone concentrations of 80 ppb over 6.6 hours have been shown to result in coughing and respiratory depression in humans299 INTRODUCTION 149

the Arenas model, shown in Figure 4.3);292,301,291,295 (iii) RO· is formed by direct abstract of atomic oxygen from the ROO· radical by NO·.302

In general, (iii) is not considered to be plausible, with the majority of computational work ignoring this pathway. In the Ellison model the conical intersection (overlap of a ground and excited state potential energy surface) was predicted to result in an avoided crossing which leads to direct isomerisation of ROONO to RONO2. Kosmas −−→ et al. suggested a similar model in the case of CH3OONO CH3ONO2 (shown in Figure 4.2). In direct contrast, the Arenas model predicts the conical intersection to distort the ROONO potential energy surface such that there is no direct isomerisation pathway on the ground state singlet potential energy surface and instead suggested that two barrierless reactions occur as just described, ﬁrst dissociation of ROONO to · · RO and NO2, then recombination to RONO2. Kosmas and co-workers point out that this reaction would only proceed at low pressures after formation of a weakly bound complex of the alkoxy radical and NO·. At higher pressures in the atmosphere, RO· would either decompose or react rapidly with the abundant O2 to form a carbonyl and · 296,303 HO2.

Figure 4.1: Schematic of the two-state curve-crossing model of Ellison et al.294 describing the F-O stretching potential in trans-FONO.

Experiments aimed at measuring the rate of the association reaction between HO· · and NO2 have identiﬁed the most simple case of HOONO using cavity ring-down infrared spectroscopy.304,305 Furthermore, the cis,cis-HOONO isomer has been generated and probed by infrared spectroscopy in a cryogenic matrix,306,307 but the presence of ROONO has not been directly measured in hydrocarbon systems.

There are no reports in the literature investigating the direct reaction of phenylper- 150 GAS PHASE REACTIONS OF DISTONIC PHENYLPEROXYL RADICALS WITH NOX

Figure 4.2: Schematic of the two-state curve-crossing model of Kosmas et al.296 describing the interconver- −−−→ sion of singlet and triplet states on the CH3OONO CH3ONO2 potential energy surface calculated at the B3LYP/6-311++G(d,p) level of theory.

· · oxyl radicals + NO . The direct probing of gas phase phenylperoxyl radical + NO2 has only recently been reported by Jagiella and Zebel.308 It is generally accepted that the · · reaction between ROO radicals and NO2 results in formation of a peroxynitrate RO2NO2 which subsequently degrades back to reactants.5 In contrast, Jagiella and Zebel found · · · the reaction of NO2 with C6H5OO to ultimately yield phenoxyl radical C6H5O and · 308 NO3, with no RO2NO2 detected. This was proposed to occur due to weakening − · of the O O bond after addition of NO2, most probably driven by formation of the resonance stabilised phenoxyl radical. While it was concluded that this reaction would · not signiﬁcantly increase the formation of NO3 in the atmosphere, this aromatic system is interesting in its divergence from alkane systems where the intermediate degrades back to reactants.

Calculations undertaken herein and summarised in Table 4.1 demonstrate that peroxynitrite and nitrate bearing alkyl radicals have a signiﬁcant O-O and O-N bond energies. As per the Ellison model, the barrier to bond dissociation of the alkylperoxyni- ˜ 2 2 trite (2) will reside between the energy of decomposition to the X A1 and Ã B2 states of · −1 NO2 (14.2 to 41.9 kcalmol ) while the nitrate decomposes without barrier to the ground state (41.4 kcalmol−1). These values are comparable to that of Barton esters which have been shown as stable upon isolation in an ion trap,155 thus one would optimistically expect that if formed and stabilised by collisional cooling alkyl peroxynitrites and INTRODUCTION 151

295 −−−→ Figure 4.3: Figure from Arenas et al. depicting potential energy surfaces for CH3ONO2 trans- CH3OONO reaction (a) S0 and S1 surfaces at the CAS(14,11) level, (b) S0 surface shown with molecular structures for the four corners of the surfaces. 152 GAS PHASE REACTIONS OF DISTONIC PHENYLPEROXYL RADICALS WITH NOX nitrates should be isolable. The bond dissociation energy for the phenyl peroxynitrite (4) was calculated to reside between -8.4 and 19.4 kcalmol−1 for dissociation to the ˜ 2 2 · X A1 and Ã B2 states of NO2, respectively. While higher level multiconﬁgurational calculations are required to predict the actual barrier to this dissociation channel, it is expected to reside between the energy of these two values which are relatively low and suggest that the phenyl peroxynitrite may be metastable.

Bond DH298 − (kcalmol 1)

′ 1 tBuC(O)O−NR 28.8 2 tBuO−ONO 14.2 (41.9) t − 3 BuO NO2 41.4 4 PhO−ONO -8.4 (19.4) − 5 PhO NO2 18.9

Table 4.1: Bond dissociation energies (DH298) calculated at the G3SX(MP3) level for (1) 2-methyl-2-((2- thioxopyridin-N-yloxy)carbonyl)propane, (2) tert-butyl peroxynitrite, (3) tert-butyl nitrate, (4) phenyl peroxyni- 2 · 280 trite, and (5) phenyl nitrate. The numbers in brackets are dissociation to the Ã B2 excited state of NO2.

This chapter presents an investigation of alkyl and arylperoxyl radical reactivity · · towards NO and NO2 in an effort to synthesise and isolate ROONO, RONO2 and

RO2NO2. Distonic radical ion models of the phenyl radical, cyclohexyl radical and adamantyl radical, namely 4-(N,N,N-trimethylammonium)phenyl radical cation, 4- carboxylatocyclohexyl radical anion and 4-carboxylatoadamantyl radical anion, were

ﬁrst allowed to react with O2 to form their respective peroxyl radicals and subsequently · · allowed to react in the presence of NO and NO2.

4.2 Methods

4.2.1 Materials

Nitric oxide (98.5%), 1,4-cyclohexane dicarboxylic acid (99%) and 1,3- adamantanedicarboxylic acid (98%) was purchased from Sigma-Aldrich (St. Louis, MO).

O2 was purchased from BOC Gases Australia Ltd. (Sydney, Australia). N,N,N-trimethyl- 4-((2-thioxopyridin-N-yloxy)carbonyl)benzeneaminium iodide was synthesised by

13 Dr. David G. Harman (University of Wollongong, Australia). Benzene-1,4-( CO2)- 13 dicarboxylic acid (92% CO2) was obtained form CDN Isotopotential energy surface RESULTS 153

18 (Pointe Claire, Quebec, Canada). Isotopically labelled oxygen O2 (95%) was obtained from Cambridge Isotope Laboratories (Andover, MA). 4-nitrobenzoic acid, methanol (HPLC grade) and ammonia solution (28%, AR grade) were obtained from Ajax Finechem (Sydney, Australia). All commercial compounds were used without further · puriﬁcation. The synthesis of NO2 is described in Section 3.3.1.

4.2.2 Mass spectrometry

The modiﬁcations to the ThermoFisher LTQ linear ion-trap mass spectrometer used to introduce neutral reagents into the helium buffer gas and the general mass spectrometric method have been described previously in Sections 2.3.1 and 3.2.1. Speciﬁc to this chapter is the presence of multiple neutral reagents in the mass spectrometer.

Under normal operating conditions this instrument contains approximately [O2] = 109 molecules cm−3 presumably due to introduction of air through the atmospheric pressure ionisation source. In the following experiments, low concentrations of NO·

10 11 −3 · 10 11 (between 5 x 10 and 2 x 10 molecules cm ) and NO2 (between 8 x 10 and 3 x 10 − molecules cm 3) were seeded into the helium buffer gas via the gas mixing manifold · · described in Section 2.3.1. The presence of both O2 and NO or NO2 in the helium buffer gas resulted in many competing reactions. For example, all three molecules undergo facile radical-radical combination reactions with alkyl and aryl radicals (as reported in Chapters 2 and 3), but because the concentration of O2 is fixed synthesising sufficient alkyl and arylperoxyl radicals in the presence of varying concentrations of · · nitrogen oxides was found to be difficult. As such, the concentration of NO or NO2 was varied during the experiment in order to optimise the ion count of a desired ion and measurement of the exact neutral reagent concentration and reaction rate was not undertaken.

4.3 Results

4.3.1 Reaction of the phenylperoxyl radical with NO·

A solution of N,N,N-trimethyl-4-((2-thioxopyridin-N-yloxy)carbonyl)benzaminium iodide (1) in MeOH:H2O was infused via positive electrospray into the mass spectrometer + + to yield an M ion at m/z 289 (overall method described in Scheme 4.2). The M ion 154 GAS PHASE REACTIONS OF DISTONIC PHENYLPEROXYL RADICALS WITH NOX

+ was isolated and subjected to CID resulting in formation of an [M - C5H4NS] ion at m/z 135 consistent with formation of 4-(N,N,N-trimethylammonium)phenyl radical cation

(A) as described in Chapter 3. This ion was allowed to react in the presence of O2 and · · NO resulting in product ions associated with (i) addition of O2 to A, (ii) addition of NO · + to A and (iii) addition of NO to the [M + O2] adduct (spectrum shown in Figure 4.4). · The ions associated with addition of O2 and NO to A have been described in Chapter + 3 and will not be further discussed. The [M + O2] adduct ion at m/z 167 previously identiﬁed as the 4-(N,N,N-trimethylammonium)phenylperoxyl radical cation (B) was · isolated and allowed to react in the presence of O2 and NO , the spectrum of which is shown in Figure 4.5.

Figure 4.4: Isolation of 4-(N,N,N-trimethylammonium)phenyl radical cation (m/z 135) in the presence of O2 · and NO . Ions with an asterisk are products of reactions involving O2.

NMe3 I NMe3 NMe3 NMe3

ESI CID + O2 + NO

- I - C5H4NS

N N m/z 135 O m/z 167 O O O O A O B 1 S m/z 289 S Scheme 4.2: Electrospray ionisation of N,N,N-trimethyl-4-((2-thioxopyridin-N-yloxy)carbonyl)benzenaminium iodide and subsequent CID of the ion formed (m/z 289) leads to formation of the 4-(N,N,N- trimethylammonium)phenyl radical cation at m/z 135. Allowing this ion to react in the presence of O2 results in formation of the 4-(N,N,N-trimethylammonium)phenylperoxyl radical (B).

The dominant feature in this spectrum is an ion at m/z 151 which is consistent with · · · · + addition of NO to the peroxyl radical B followed by loss of NO2 [M + NO - NO2] to form the 4-(N,N,N-trimethylammonium)phenoxyl radical cation (D in Scheme 4.3). In an effort to conﬁrm the assignment of this ion, CID spectra were measured to compare RESULTS 155

Figure 4.5: Ion-molecule reaction of 4-(N,N,N-trimethylammonium)phenylperoxyl radical cation (m/z 167) with NO·. between the m/z 151 ion synthesised in this experiment to that of the m/z 151 ion synthesised as described in Chapter 3.

NMe3 NMe3 NMe3 NMe3

+ NO + NO + NO2

O O N O O O O O N O m/z 167 m/z 197 m/z 151 m/z 181 B C DE Scheme 4.3: Overall scheme of the reaction of 4-(N,N,N-trimethylammonium)phenyl radical cation (m/z 135) with NO·.

Isolation of the m/z 151 ions resulted in a low ion count presumably due to low ion- trapping efﬁciencies across subsequent ion isolation and ion-molecule reaction steps. The m/z 151 ions generated for comparison by CID experiments were instead generated and isolated in a single isolation ion isolation step by reaction of A in the presence of · O2 and NO . As shown in Figure 3.16(a) in Chapter 3 formation of an m/z 151 ion by the reaction of A with O2 is very minor, thus the majority of the ion population at m/z 151 occurs due to reaction of the phenylperoxyl radical B with NO·. In this manner, the m/z 151 product ion is generated in the same sequence of reactions (Scheme 4.4(i)) without the ion losses incurred by the subsequent isolation and ion-molecule reaction of B.

The m/z 151 ions synthesised as described in Scheme 4.4(i) and (ii) were subject to CID. In both cases this resulted in a single product ion at m/z 136. In order to compare these m/z 136 ions they were subsequently isolated and subjected to CID. This resulted in the two groups of comparative CID spectra displayed in Figure 4.6(a) 156 GAS PHASE REACTIONS OF DISTONIC PHENYLPEROXYL RADICALS WITH NOX and (b). The onset of major fragmentation was at 30% normalised collision energy and was found to be a loss of 28 Da, most likely loss of CO, to generate an ion at m/z 108. At all three collision energies, the ratio of the m/z 108 to m/z 136 ion count was the same. Magniﬁcation of the spectra by 50x revealed similar product ions at m/z 67, 79, 81 and 121 between the two groups of spectra. There were slight discrepancies of ions and abundances at m/z 55, 77 and 91 and 93 but these were most likely due to isobaric impurities or simply the low ion count of the parent m/z 136 ion which may result in some product ions not being detected. Overall, these suggest that the major constituent of the ion population produced by reaction of B and NO· was 4-(N,N,N- trimethylammonium)phenoxyl radical cation (D).

The m/z 181 present in Figure 4.5 is assigned as the secondary addition of NO· to the phenoxyl radical (D) at m/z 151. This assignment was conﬁrmed by isolating D in the · presence of O2 and NO . The spectrum obtained (shown in Figure 4.7) reveals an [M + + 30] ion at m/z 181 consistent with addition of NO·. This reaction is predicted to result in formation of N,N,N-trimethyl-4-(nitrosooxy)benzaminium cation (E). To conﬁrm this assignment a comparative CID study between this ion and independently synthesised N,N,N-trimethyl-4-nitrobenzaminium cation (F) was undertaken. An overall description of the CID pathways observed is shown in Scheme 4.5. As shown in Figure 4.8, isolation of the putatively assigned E and subsequent CID resulted in an [M - 30]+ ion consistent with loss of NO· to form D at m/z 151. In contrast, the major losses observed when subjecting F to CID (spectrum displayed in Figure 3.15a) were [M - 15]+ and [M

+ . · - 46] ions assigned as loss of CH3 and NO2, respectively. While it was not possible to obtain an authentic precursor for E, a loss of NO· is likely the major loss of a nitrite species as the N-O bond is relatively weak and formation of the stable phenoxyl radical would drive the process. These data thus support the assigned nitrite structure E for the m/z 181 ion formed by reaction between the phenoxyl radical (D) and NO·.

Examination of the spectrum in Figure 4.5 at 1000 times magniﬁcation reveals an + [M + 30] ion at m/z 197 consistent with addition of NO· to B. This ion is tentatively assigned as the target ROONO or RONO2 molecules, but due to the low abundance of this ion it was not isolable and as such was not amenable to further characterisation. · Furthermore, while every action was taken to minimise the presence of NO2 from the · · instrument, the ion at m/z 197 could also be formed by addition of NO2 to the RO radical RESULTS 157

NMe3 NMe3 NMe3 NMe2

+ O + NO CID CID (i) 2 - NO2 - H3C

m/z 135 O m/z 167 O m/z 151 O m/z 136 A O B D

NMe3 NMe3 NMe3 NMe2

(ii) + O2 CID CID CID

- O - H3C

m/z 135 O m/z 167 O m/z 151 O m/z 136 A O B D Scheme 4.4: Description of the experiment undertaken in Figure 4.6 to conﬁrm the identity of the m/z 151 ion generated during reaction of B with NO·.

(a) Reaction of B with NO· generates m/z 151 which is subjected to CID to produce an ion at m/z 136.

(b) Subjecting B to CID resulted in generation of an m/z 151 ion. Isolating and subjecting this ion to CID results in an ion of m/z 136.

Figure 4.6: Comparative CID spectra of m/z 136 ions synthesised as described. NCE is the normalised collision energy setting in the Xcalibur software.95 The similar heights of the m/z 108 ion at the same normalised collision energy suggests that the ions are the same in both cases. Minor ions not consistent between the two groups of comparative spectra are most probably due to minor isobaric interferences. 158 GAS PHASE REACTIONS OF DISTONIC PHENYLPEROXYL RADICALS WITH NOX

NMe3 NMe3

CID (i)

O O N O m/z 181 m/z 151 E D

NMe3 NMe3 NMe2

CID (ii) +

N m/z 135 N O O A O O m/z 181 m/z 166 F Scheme 4.5: Overall scheme for CID of the isomeric (i) N,N,N-trimethyl-4-(nitrosooxy)benzaminium cation (E) and (ii) 4-N,N,N-trimethyl-4-nitrobenzaminium cation (F).

Figure 4.7: Ion-molecule reaction of 4-(N,N,N-trimethylammonium)phenoxyl radical cation (m/z 151) with NO· for 500 ms. at m/z 151. Under the conditions of this experiment it is not possible to confirm or contradict either reaction pathway. While the spectrum shown in Figure 4.5 features the m/z 197 ion it was not always measured in repeat experiments, suggesting the formation of the ion is quite sensitive to the experimental conditions. The tentative assignment of this ion is certainly consistent with the formation of the two abundant and well characterised ions at m/z 151 (D) and m/z 181 (E). The low abundance of the m/z 197 ion suggests that collisional cooling within the low pressure ion trap may not be sufficient to stabilise the ion and instead it quickly decomposes either to products or back to reactants. The m/z 151 ion may well be formed after initial reaction between B and NO· resulting in generation of ROONO which quickly decomposes to the resonance · stabilised phenoxyl radical (D) + NO2. Unfortunately, due to the inability to isolate the tentative ROONO or RONO2 molecule and undertake further characterisation, there appears no way under the conditions of this experiment to confidently assign the reactions observed to any of the three mechanisms previously described. RESULTS 159

Figure 4.8: CID of the m/z 181 ion produced by the ion-molecule reaction of 4-(N,N,N- trimethylammonium)phenoxyl radical cation (m/z 151) with NO·.

4.3.2 Reaction of the alkylperoxyl radicals with NO·

A methanolic solution of 1,4-cyclohexanedicarboxylic acid (2) was infused via negative − − electrospray ionisation to afford [M - H] and [M - 2H]2 ions at m/z 171 and 85, respectively. The m/z 85 dianion was isolated and subjected to CID resulting in generation of

−. an [M - 2H - CO2] ion at m/z 126 consistent with formation of 4-carboxylatocyclohexyl radical anion (G). Isolating this ion and allowing it to react in the presence of NO· · − − and O2 resulted in formation of [M + NO ] and [M + O2] adducts assigned as 4- nitrosocyclohexanecarboxylate anion and 4-carboxylatocyclohexylperoxyl radical anion (H) which are both described and characterised in Chapter 5.

CO2H CO2 CO2 CO2

ESI CID + O2 + NO

- 2H - CO2, e

CO2H CO2 m/z 126 O m/z 158 O 2 m/z 85 G H Scheme 4.6: Electrospray ionisation of 1,4-cyclohexanedicarboxylic acid and subsequent CID of the ion formed (m/z 85) results in formation of the 4-carboxylatocyclohexyl radical anion at m/z 126. When allowed to react in the presence of O2 the associated 4-carboxylatocyclohexylperoxyl radical anion (B) is formed at m/z 158 which is subsequently isolated and allowed to react with NO·.

· Isolation of H in the presence of O2 and NO for 8000 ms resulted in no product ions. There was a slight decrease in total ion count which may suggest formation of ions · not detected by the instrument. For example, if NO2 formed (Ea = 2.273 eV) it may, as shown Scheme 4.7(i), remove an electron from the carboxylate anion resulting in an ion at m/z 46 which would not be detected as it is below the m/z 50 low mass cutoff. This is unlikely, however, given the high electron afﬁnity of other alkylcarbonyloxyl radicals 160 GAS PHASE REACTIONS OF DISTONIC PHENYLPEROXYL RADICALS WITH NOX such as acetyloxyl and propylcarbonyloxyl radicals which are reported to have electron

309,310 – affinities of 3.25 eV. Alternatively, NO2 may be eliminated directly by attack of the carboxylate anion as shown in Scheme 4.7(ii). Overall, these data suggest that either (i) a barrier exists for the reaction of H with NO·, (ii) the ROONO adduct is not sufficiently collisionally cooled and thus decomposes back to reactants, (iii) there exists a barrier higher than the entrance channel of ROO + NO·. It seems unlikely that the 4- carboxylatocyclohexylperoxyl radical (H) is inert to NO· as previous experiments have measured n-hexyl nitrates during reaction of n-hexylperoxyl radicals and NO·.311 As such, one would expect H is large enough to accommodate formation of a nitrate, and instead it seems likely that the low pressure of the ion trap (2.5 mTorr) and helium buffer gas do not provide sufficient collisional cooling.

Figure 4.9: Isolation of 4-carboxylatocyclohexylperoxyl radical anion (m/z 158) in the presence of NO·.

O O O O O O O O NO2

+ NO + NO - (i) 2 m/z 46 O O N O O O O O m/z 158 m/z 188 H

O O O O O

+ NO O + NO - (ii) 2 O m/z 46 O O N O O O m/z 158 m/z 188 H Scheme 4.7: Putative mechanisms resulting in loss of charge of the nascent charge-tagged 3- carboxylatocyclohexyl peroxynitrite.

A methanolic solution of 1,3-adamantanedicarboxylic acid (3) was infused via negative electrospray ionisation into the mass spectrometer to afford [M + H]− and [M - RESULTS 161

− 2H]2 ions at m/z 223 and 111, respectively. As depicted in Scheme 4.8, the m/z 111

−. dianion was isolated and subjected to CID generating [M - 2H - CO2] ion at m/z 178 consistent with 3-carboxylatoadamantyl radical anion (I) formation. As depicted in · Scheme 4.8, isolation of I in the presence of NO and O2 resulted in formation of [M · − − + NO ] and [M + O2] adducts assigned as 3-nitrosoadamantanecarboxylate anion and 3-carboxylatocyclohexylperoxyl radical anion (J), respectively. The gas phase synthesis of J is consistent with the previous experiments described in Chapter 2 and by Harman and Blanksby.154,155 Subsequently, J was isolated and allowed to react in the presence · of O2 and NO , the spectrum of which is shown in Figure 4.10.

Two product ions observed at m/z 194 and m/z 224 were assigned as [M + NO· · − · · · − - NO2] (L) and [M + NO - NO2 + NO ] (M) ions, respectively. In contrast to the reaction of the phenylperoxyl radical B with NO·, there was no observable adduct between J and NO· at m/z 240 (even at 1000 times magniﬁcation) which suggests that either the putative alkoxyl radical L is formed by direct abstraction of O by NO·, or that any nascent ROONO or RONO2 intermediate generated is very short-lived due to insufﬁcient collisional cooling and decomposes quickly to either products or back to reactants.

O CO2H CO2 O

ESI CID + O2 + NO

- 2H - CO2, e CO2H CO2 CO2 CO2 3 m/z 111 m/z 178 m/z 210 I J

+ NO

CO2 m/z 208 Scheme 4.8: Electrospray ionisation of 1,3-adamantanedicarboxylic acid and subsequent CID of the ion formed (m/z 111) results in formation of the 3-carboxylatoadamantyl radical anion at m/z 178. Allowing this ion to react in the presence of O2 results in formation of 3-carboxylatoadamanatylperoxyl radical anion (C) at m/z 210 which is subsequently isolated and allowed to react in the presence of NO·.

· 4.3.3 Reaction of the phenylperoxyl radical with NO2

The 4-(N,N,N-trimethylammonium)phenylperoxyl radical cation (B) was isolated and · allowed to react in the presence of O2 and NO2, the spectrum of which is shown in Figure 4.11. The most abundant ion featured at m/z 151 and was assigned as the 4- (N,N,N-trimethylammonium)phenoxyl radical cation (D). This ion is believed to arise 162 GAS PHASE REACTIONS OF DISTONIC PHENYLPEROXYL RADICALS WITH NOX

Figure 4.10: Isolation of 3-carboxylatoadamantylperoxyl radical anion (m/z 210) in the presence of NO·.

· · from addition of NO2 to B followed by subsequent loss of nitrate radical (NO3). Ions of minor abundance were revealed by magnifying the spectrum and include ions at m/z · + · · · + 213 and 197 assigned as [M + NO2] (N) and [M + NO2 - NO3 + NO2] (O). The ion · · at m/z 181 was assigned as addition of residual NO present in the NO2 reagent gas to · · · + D [M + NO2 - NO3 + NO ] . Putative ions N and O were not isolable due to their low abundance. Further characterisation of the purported RO2NO2 molecule was therefore not explored. Its presence, however, does provide evidence to suggest the formation of the D proceeds through an RO2NO2 intermediate. The transient nature of the ions at m/z 213 (N) is consistent with the work previously undertaken by Jagiella and Zabel308 in which phenoxyl radical was detected, but not the PhO2NO2 intermediate.

NMe3 NMe3 NMe3 NMe3

m/z 197 + NO2 - NO3 + NO2 O O O O N O O O O O O N m/z 167 m/z 213 m/z 151 O BN D

NMe3

+ NO m/z 181 E

O O N Scheme 4.9: Overall scheme for the reaction of 4-(N,N,N-trimethylammonium)phenylperoxyl radical cation · (B) with NO2. CONCLUSION 163

Figure 4.11: Ion-molecule reaction of 4-(N,N,N-trimethylammonium)phenylperoxyl radical cation (m/z 167) · with NO2.

4.4 Conclusion

The aims of these experiments were to synthesis and isolate charge-tagged models of ROONO and RO2NO2 molecules in order to gain insight into the reaction · · of alkyl and arylperoxyl radicals with NO and NO2. The reaction of 4-(N,N,N- trimethylammonium)phenylperoxyl radical cation (B) with NO· resulted in formation of RO· and RONO and, in addition, appeared to result in formation of a very low · abundance of an ROONO adduct. Similarly, the reaction of B with NO2 resulted in · formation of RO and RONO2 as well as a small quantity of an ion consistent with

RO2NO2. However, in both cases, the ROONO and RO2NO2 adducts were not isolable due to their low abundance thus further characterisation of the nascent species was not possible.

Reaction of the aliphatic bridgehead 3-carboxyatoadamantylperoxyl radical anion (J) with NO· resulted in formation of both RO· and RONO species as found during reaction of the model phenylperoxyl radical, but in this case there was no observation of an ROONO adduct. Furthermore, comparison of Figures 4.5 and 4.10 reveal signiﬁcantly different RO· to RONO abundances suggesting that NO· reacts slower with the phenoxyl radical than the adamantyloxyl radical, presumably due to resonance stabilisation of the phenoxyl radical which makes further addition of NO· less favourable. Interestingly, the 4-carboxylatocyclohexylperoxyl radical anion (H) did not react at all in the presence of NO·. This may be because (i) the reaction leads to a product (presumably ROONO) which quickly degrades back to reactants before further reaction or stabilisation of 164 GAS PHASE REACTIONS OF DISTONIC PHENYLPEROXYL RADICALS WITH NOX the nascent adduct by collisional cooling, (ii) there exists a signiﬁcant barrier for the reaction between the ROO· species and NO·, (iii) there exists a barrier higher than the entrance channel of ROO· + NO· for decomposition of a nascent ROONO to further products, or (iv) the 4-carboxylatocyclohexylperoxyl radical anion adopts a structure which hinders approach and/or reaction of the incoming neutral.

Reaction of the negative charge-tagged cyclohexylperoxyl and adamantylperoxyl · · radical models with NO2 failed as addition of NO2 to the helium buffer gas resulted in their respective dicarboxylate dianion precursors being depleted by electron transfer to · · NO2. Dicarboxylate dianions tend to have a lower electron afﬁnity than NO2 (2.27 eV), for example, Ding et al. experimentally determined the adiabatic electron afﬁnity of

- – - – O2C(CH2)3CO2 and O2C(CH2)4CO2 (analogous of 1,3-adamantanedicarboxylate and 1,4-cyclohexanedicarboxylate dianion) at 0.55 and 1.01 eV, respectively.312 As such, while it was possible to form the radical anions, they were not produced in sufﬁcient abundance for subsequent ion-molecule experiments.

Overall, there was only indirect evidence to support formation of an ROONO · · · or RONO2 intermediate during reaction of peroxyl radicals with NO (ROO + NO ). In both aliphatic and aromatic models, this was supported by formation of an RO· · · · + species consistent with addition of NO to the peroxyl radical and loss of NO2 (ROO . −−→ −−→ · + . NO RO2NO RO NO2). RONO2 has previously been measured in signiﬁcant quantities ( 10%) during reaction of NO· with alkylperoxyl radicals of chain length greater than three carbons long. For example, the branching ratio of RONO2 during reaction of the 3-hexylperoxyl radical (c.f. cyclohexyl radical) with NO· was measured at 22% of the total products. Longer n-alkylperoxyl radicals had yet higher branching ratios. Reaction of tertiary species such as 2-methyl-2-butylperoxyl radical were found to produce lower abundances of RONO2 with the exception of tert-butylperoxyl radical 5 (c.f. adamantyl radical) which was measured to yield 18% RONO2. This would suggest that the models investigated in this study should be large enough to support formation and stabilisation of ROONO or RONO2. Benzene, cyclohexane and adamantane are more rigid than their n-alkane analogues and therefore the number of low energy vibrational degrees of freedom available for redistribution of reaction energy is reduced. The energy generated when NO· adds to the peroxylradicals may instead provide sufﬁcient energy for degradation of the nascent ROONO intermediate to reactants or CONCLUSION 165

products. Alternatively, the absence of ROONO or RONO2 from these experiments may be explained by considering the low pressure of the ion trap. The pressure in the ion trap is approximately 2.5 mTorr while the literature yields of RONO2 were measured at 1 atmosphere. This would suggest that third body collisional cooling is exceptionally important in stabilising the nascent ROONO intermediate.

In conclusion, the instrumentation and conditions used in these experiments were not amenable to the synthesis and isolation of ROONO, RONO2 or RO2NO2. The formation of RO· and RO· adducts during these experiments is, however, consistent with the formation and degradation of an intermediate ROONO species. As demonstrated during comparative CID of N,N,N-trimethyl-4-(nitrosooxy)benzaminium (E) and N,N,N-trimethyl-4-nitrobenzaminium (F) cations, differentiation of nitrogen oxide isomers is clearly possible using ion-trap mass spectrometry. As such, one would expect to be able to characterise putative ROONO or RONO2 products either by observation of unique fragmentation products or by comparison of collision energy to fragmentation onsets, however, this is obviously predicated on the formation and isolation of the target ion, which was not accomplished during these experiments.

a aygetydpnigo h egho h abncan rsueadtempera- and pressure chain, carbon the of length the on depending greatly vary can − ( combination radical-radical exothermic h xdto faihtchdoabn sa motn rcs ncmuto and combustion in pollution. troposphere process of important production an the is hydrocarbons aliphatic of oxidation The 5.1 ceeaeakleoy radicals. alkylperoxyl are scheme hc a btathdoe tm.Oc omd h aeo ly aiasi largely formed. is were radicals they which alkyl in of environment fate the the on dependent formed, Once atoms. hydrogen abstract may which the In radical. HO alkyl an atmosphere, of formation to leading abstraction atom degradation hydrogen however, by processes initiated both is In combustion. during observed that from differ can atmosphere the the in which hydrocarbons under of degradation pressure the Consequently, and occur. reactions temperature hydrocarbon, the on dependant process uigcmuto hr r uttd fratv pce nldn HO including species reactive of multitude a are there combustion During 03ka mol kcal 30.3 ly aiasraiyudrorato ihO with reaction undergo readily radicals Alkyl G Introduction SPAERATOSO CYCLOHEXYL OF REACTIONS PHASE AS − 1 · ) 315 edl btat yrgnao rmhdoabn ofr H form to hydrocarbons from atom hydrogen a abstracts readily ofr lyprxlrdcl.Teratvt fakleoy radicals alkylperoxyl of reactivity The radicals. alkylperoxyl form to 15,5 h erdto fhdoabn sacomplicated a is hydrocarbons of degradation The 313,314 167 e.g. h e nemdae nti reaction this in intermediates key The ∆ rxn 2 H ( 3 298 Σ − g [C ) 2 via H 5 · + arels,highly barrierless, a RADICALS O 2 −−→ C 2 · H

n HO and C HAPTER 5 OO 5 · ] 2 O. = 2 · 168 GAS PHASE REACTIONS OF CYCLOHEXYL RADICALS

ture. Rienstra-Kiracofe et al.315 undertook a critical computational evaluation of ﬁve key channels proposed in previous studies of the archetypal ethylperoxyl system (see · Scheme 5.1). Direct hydrogen abstraction by O2 to form ethane and HO2 (Reaction 5.1) was largely refuted at low temperatures (< 700 K) due to high activation barriers. Two competing mechanisms are proposed to account for the experimentally observed · ethene and HO2; isomerisation of ethylperoxyl radical by 1,5-hydrogen atom transfer to form 2-hydroperoxylethyl radical followed by C-O bond rupture (Reaction 5.4), and 1,5-hydrogen atom transfer with concommitent C-O bond rupture (Reaction 5.5). Rearrangement of the 2-hydroperoxylethyl radical and elimination of .OH can also give rise to oxirane (Reaction 5.2). 1,3-hydrogen atom transfer of the ethylperoxyl radical yields 1-hydroperoxylethyl which may similarly eject .OH to form ethanal (Reaction 5.3).

· + → + · C2H5 O2 C2H4 HO2 (5.1) · + → · → . → − + . C2H5 O2 CH3CH2OO CH2CH2OOH c CH2CH2O OH (5.2) . · + → · → → + . C2H5 O2 CH3CH2OO CH3CHOOH CH3CHO OH (5.3) · + → · → . → + · C2H5 O2 CH3CH2OO CH2CH2OOH C2H4 HO2 (5.4) · + → · → + · C2H5 O2 CH3CH2OO C2H4 HO2 (5.5)

· Scheme 5.1: Degradation pathways of the ethylperoxyl radical (CH3CH2OO ).

Reaction 5.5 is considered the primary mechanism by which alkenes are formed during alkane oxidation. This mechanism accounts for the negative temperature coefﬁ- · cients observed during oxidation of alkyl radicals at temperatures ca. 500 K. When HO2 is formed at this temperature it may abstract hydrogen from a second hydrocarbon, or · react with another HO2 radical. At 500 K the product of the chain terminating HO2 self- + reaction, H2O2 O2 , is relatively stable and the rate of oxidation decreases, however, as · the temperature increases above 800 K H2O2 can dissociate back to HO resulting in an increase in oxidation rate.316

Previous experimental and theoretical studies have shown that at higher temperatures hydrocarbons can undergo direct unimolecular bond scission to form radicals or isomerise prior to hydrogen atom transfer.317,318,319,26 Once formed, the short- lived high mass radicals undergo repeated bond scission reactions until more stable long-lived low mass radicals, such as methyl or ethyl, are formed which survive on a

320,321 timescale in which they can subsequently react with O2. This means that at high INTRODUCTION 169

temperatures (T > 1100 K), there is a relatively limited range of products. In contrast, oxidation of hydrocarbons at low temperatures is signiﬁcantly more complex. The intramolecular hydrogen atom transfer step (e.g. Reaction 5.2 and 5.4) is predicated to play an important role as the intermediary hydroperoxylalkyl radicals, commonly

. referred to as QOOH radicals, can react with O2 to generate highly oxygenated species · · in addition to their involvement in generating HO and HO2, both processes of which are important in further chain branching.322,320,321,5

Although there have been a wide range of experiments investigating the oxidation of simple alkanes, these studies were not well suited to mechanistic analysis as the reaction pathways were largely extrapolated from end product analysis. Ion-trap mass spectrometry using distonic radical ion analogues of alkyl radicals provides an elegant solution to this problem. Using this method the reactant radical can be both synthesised and isolated free of possible reactants, and sequential reactions can be monitored directly by careful addition of neutral reagents. Reaction intermediates and products may be isolated, and can be characterised by their mass, and furthermore, subjecting these isolated ions to CID and analysing product ions may aid in structure elucidation.

An ideal distonic alkyl or alkylperoxyl radical should separate the charge and unpaired electron by a rigid structure to minimise interaction between the charge and radical moieties, however, a distonic alkyl or alkylperoxyl radical ion with this characteristic structure has not previously been reported in the literature. Yu et al.150 and later Sorrilha et al.149 synthesised the β-distonic N-ethylpyridinium radical cation, which was observed to form a +32 Da adduct in the presence of O2 suggesting generation of a distonic alkylperoxyl radical. While this was the ﬁrst demonstration of the synthesis and isolation of a distonic alkylperoxyl radical, the charge group was located only two atoms away from the carbon-centred radical. In addition, no further reaction was observed in contrast to alkyl radical oxidation models which, as discussed, predict loss of HO· and · HO2 as major products suggesting the N-ethylpyridinium radical cation was not a good model of alkyl radical reactivity.

As previously described in Section 2.2.2.2, Harman and Blanksby studied the distonic 3-carboxylatoadamantyl radical anion154,155 as a model of carbon-centred bridgehead radicals. When allowed to react in the presence of O2 this radical was found to add +32 Da to form the 3-carboxylatoadamantylperoxyl radical anion. The focus in this 170 GAS PHASE REACTIONS OF CYCLOHEXYL RADICALS

case, however, was not on the conventional geometry of a simple alkyl radical, as the strain at the adamantyl bridgehead forces the radical to adopt a pyramidal geometry which deviates signiﬁcantly from the trigonal planar geometry of straight-chain carbon- centred radicals. While the adamantylperoxyl radical was found to lose HO· to form an · epoxide, no loss of HO2 was observed. The rigid structure of the adamantyl radical does, however, prevent the relatively inert carboxyate charge-tag from interacting directly with the radical centre.

In a similar vein, this chapter presents a study of the 4-carboxylatocyclohexyl radical anion as a model of the reaction of the cyclohexyl radical and its reaction with O2.

The reaction of cyclohexyl radicals with O2 has previously been studied in the context of combustion,323,324,318,325,326,26,327 as well as atmospheric chemistry.328 Unlike the adamantyl radical, the geometry of the cyclohexyl radical is comparable to that of straight chain alkyl radicals. The cyclic structure will also form a framework to hinder direct interaction between the carboxylate charge-tag and radical centre.

Pyrolysis experiments have identiﬁed a number of reaction products resulting from cyclohexyl radical oxidation.326,325 At higher temperatures bond homolysis occurs at a faster rate than O2 addition, and smaller low molecular weight alkyl radicals must 26,318,319,317 form before reaction occurs with O2. In contrast, cyclohexyl radicals have a longer lifetime in low temperature environments, and their degradation is dominated by formation of cyclohexylperoxyl radicals followed by their subsequent unimolecular degradation. Recent theoretical studies have proposed a range of reaction pathways occurring due to chain branching at low temperatures.323,329,327 From these data, a scheme describing some of the signiﬁcant pathways during low temperature oxidation has been constructed (Scheme 5.2), and will be used to compare experiments undertaken in this work.

5.2 Method

5.2.1 Mass spectrometry

Experiments were performed on a modiﬁed ThermoFisher LTQ (San Jose, CA) linear quadrupole ion trap mass spectrometer90 ﬁtted with a conventional IonMax electrospray ionisation source and operating Xcalibur 2.0 SUR1 software. Ions were generated METHOD 171

OH O O (iii) (iv) HO2 HO J - K - - L CO2 CO2 CO2

(i) O (ii) O OH O O HO

+O (v) N 2 (vi) - CO2

(vii) - E - - M CO2 O HO CO2 I CO2 (ix) OH O O O - CO2

(viii) HO O HO

R - P - - Q CO2 CO2 CO2

Scheme 5.2: Possible branching pathways during the unimolecular degradation of the 4- carboxylatocyclohexylperoxyl radical anion (I). by infusion at 3 - 5 µL min−1 of a methanolic solution (10 - 50 µM) into the electrospray ion source. In some instances, aqueous ammonia was added to raise the pH and facilitate the formation of dianions. Typical instrumental settings were: spray voltage 2.5 - 3.5 kV,capillary temperature 200 - 250 °C, sheath gas flow between 10 - 30 (arbitrary units), sweep and auxillary gas flow set at between 0 - 10 (arbitrary units). For collision induced experiments, ions were mass-selected with a window of 1 - 4 Th, using a q- parameter of 0.250 and the fragmentation energy applied was typically 10 - 45 (arbitrary units) with an excitation time was 30 ms (unless otherwise noted). Modifications to the mass spectrometer to allow the introduction of neutral gases into the ion trap region of the instrument were described in Chapter 2. Briefly, liquid and gas reagents are introduced into a flow of Ultra High Purity (UHP) helium (3 - 5 psi) via a heated

18 · septum inlet (25 - 250 °C). The reagent ﬂow (e.g., O2, NO ) is controlled using a syringe pump, while mixture is supplied via a variable leak valve to provide a total ion gauge reading equal to that obtained in normal operating mode representing an estimated trap pressure of 2.5 mTorr. The temperature of the vacuum manifold surrounding the ion trap was measured at 307 ± 1 K, which is taken as being the effective temperature 172 GAS PHASE REACTIONS OF CYCLOHEXYL RADICALS

for ion-molecule reactions observed herein.235 Reaction times of 30 - 10000 ms were set using the excitation time parameter within the control software using a fragmentation energy of 0 (arbitrary units). All spectra presented represent are an average of at least 50 scans. Reaction rate coefﬁcients were determined as previously described in Chapter 2 with collision rates for the ion-molecule reactions estimated by trajectory collision theory at 307 K. Branching ratios were calculated using the method of Grabowski and Zhang.152

5.2.2 Computational chemistry

All calculations were initially undertaken using the hybrid density functional theory B3LYP method266,330 and the 6-31+G(d) basis set with the GAUSSIAN03 suite of pro- grams.268 Additionally, accurate energies were obtained using the CBS-QB3 composite theoretical method.331,332,333,334,335,336,337,338 All stationary points on the potential energy surface were characterised as either minima (no imaginary frequencies) or transition states (one imaginary frequency) by calculation of the frequencies using analytical gradient procedures. All reported energies include unscaled zero-point energy corrections. Minima were conﬁrmed to join a transition state by intrinsic reaction coordinate calculation.339,340

5.2.3 Materials

1,1-Cyclohexanedicarboxylic acid (95%) was obtained from Matrix Scientiﬁc (Columbia, USA). trans-1,2-cyclohexane dicarboxylic acid (95%), 1,3-cyclohexane dicarboxylic acid (98%) and 1,4-cyclohexane dicarboxylic acid (99%) were purchased from Aldrich (Mil- waukee, USA) with the latter two diacids obtained as mixtures of both cis and trans

18 stereoisomers. Isotopically labeled oxygen O2 (95%) was obtained from Cambridge Isotope Laboratories (Andover, MA). Methanol (HPLC grade) and ammonia solution (28%, AR grade) were obtained from Ajax Finechem (Sydney, Australia). All commercial compounds were used without further puriﬁcation. 4-Oxocyclohexane carboxylic acid and 3,4-epoxycyclohexane carboxylic acid were prepared by literature procedures.341,342 Freshly prepared nitric oxide was prepared in a disposable syringe as previously reported.155,275 METHOD 173

5.2.3.1 Synthesis of dimethyl cyclohexane-1,4-dicarboxylate

To a stirred solution of 1,4-cyclohexanedicarboxylic acid (2.002 g, 11.6 mmol, mixture of cis and trans diastereomers) in methanol (50 mL) was added sulfuric acid (10 drops, concentrated). When addition was complete the mixture was heated at reﬂux overnight. The solvent was removed in vacuo and the residue diluted with water (50 mL) and extracted with dichloromethane (3 x 30 mL). The combined extracts were dried over magnesium sulfate and the solvent removed in vacuo to yield a colourless oil (2.604 g, 85%), which was puriﬁed by distillation under reduced pressure before further use. 1H δ δ 13 NMR (500 MHz, CDCl3) 1.43-2.48 (m, 10H), 3.68 (s, 6H). C NMR (125 MHz, CDCl3) 127.26, 51.42, 51.39, 42.23, 40.48, 27.86, 25.84.

5.2.3.2 Synthesis of dimethyl-1,4-dideuterocyclohexane-1,4-dicarboxylate n-Butyllithium (1.3 M in hexanes, 1.70 mL, 2.20 mmol) was added to a stirred solution of diisopropylamine (308 µL, 2.20 mmol) in tetrahydrofuran (2 mL) under dry nitrogen at 0 °C. The mixture was cooled to -70 °C after which dimethyl cyclohexane-1,4- dicarboxylate (100 mg, 0.50 mmol) was slowly added. The mixture was allowed to warm to room temperature over 3 h after which it was allowed to cool to -70 °C and D2O (1 mL, 45.1 mmol) added slowly. Further tetrahydrofuran (1 mL) was added to aid stirring of the mixture, after which it was allowed to warm to room temperature over 1 h. The mixture was then extracted with dichloromethane (3 x 2 mL), the combined extracts dried over magnesium sulfate and the solvent removed in vacuo to yield a colourless oil which was used without further puriﬁcation.

5.2.3.3 Synthesis of 1,4-dideuterocyclohexane-1,4-dicarboxylic acid

A solution of lithium hydroxide (148.7 mg, 3.5 mmol) in water (2.5 mL) was added to a stirred solution of dimethyl 1,4-dideuterocyclohexane-1,4-dicarboxylate (104.6 mg, 0.5 mmol) in tetrahydrofuran (2.5 mL). The mixture was allowed to stir overnight after which it was washed with dichloromethane (3 x 2 mL) and acidiﬁed to pH 1 with dilute hydrochloric acid. The mixture was then extracted with ethyl acetate (3 x 5 mL), and the combined extracts dried over magnesium sulfate and solvent removed in vacuo to yield a white powder that was used without further puriﬁcation. Electrospray ionisation

− 13 mass spectrometry gives [M-H] 173(100%), 172(84), 171(22); D2:D1:D0 ( C-isotope 174 GAS PHASE REACTIONS OF CYCLOHEXYL RADICALS

corrected) 4.5:3.8:1.

5.2.3.4 Synthesis of methyl cyclohexanecarboxylate

To a stirred solution of cyclohexanecarboxylic acid (2.102 g, 16.3 mmol) in methanol (50 mL) was added sulfuric acid (10 drops, concentrated). When addition was complete the solution was heated at reﬂux (2.5 h) after which it was allowed to cool to room temperature and left overnight. The reaction mixture was diluted with water (50 mL) and extracted with dichloromethane (3 x 30 mL). The combined extracts were dried over magnesium sulfate and the solvent removed in vacuo to yield a pale yellow oil (2.196 g,

1 99% yield), which was used without further puriﬁcation. H NMR (500 MHz, CD3OD) δ δ 13 1.11-3.59 (m, 10H), 3.59 (s, 3H) C NMR (125 MHz, CDCl3) 176.54, 51.39, 43.07, 28.98, 25.71, 25.40.

5.2.3.5 Synthesis of methyl-1-deutero-cyclohexanecarboxylate n-Butyllithium (2.4 mL, 3.10 mmol) was added to a stirred solution of diisopropylamine (434 µL, 3.1 mmol) in tetrahydrofuran (2 mL) under dry nitrogen at 0 °C. The mixture was cooled to -70 °C after which methyl cyclohexanecarboxylate (200 mg, 1.4 mmol) was slowly added. The mixture was allowed to warm to room temperature over 3 h after which it was cooled to -70 °C and D2O (1 mL, 45.1 mmol) added slowly. The mixture was allowed to warm to room temperature after which it was extracted with dichloromethane (4 x 3 mL), and the combined extracts dried over magnesium sulfate and the solvent removed in vacuo to yield a yellow oil which was used without further puriﬁcation. Electron ionisation mass spectrometry gives [M.+] 143 (100%); 142 (65%);

13 D1:D0 ( C-isotope corrected) 1.45:1.

5.3 Results

5.3.1 Gas phase synthesis and structural characterisation of the

4-carboxylatocyclohexyl radical anion

A methanolic solution of the parent 1,4-cyclohexanedicarboxylic acid was infused into an electrospray ionisation mass spectrometer yielding both [M - H]− and [M - 2H]2− ions at m/z 171 and 85, respectively (Scheme 5.3). Isolation of the dianion and application

.− of collision induced dissociation (CID) led to formation of an [M - 2 H - CO2] radical RESULTS 175

anion (E) at m/z 126 via oxidative decarboxylation as previously observed for analogous dianions.135,130,154,155 Infusion of 1,3-cyclohexanedicarboxylic acid proceeds similarly,

2− .− with CID of the [M - 2H] ion at m/z 85 yielding an [M - 2H - CO2] radical anion (F) at m/z 126. Electrospray ionisation of 1,1- and 1,2-cyclohexanedicarboxylic acid under the same conditions led to no detectable [M - 2H]2− dianions, instead the [M - H]− ion was generated exclusively. Presumably formation of a dianion is prevented due to excessive Coulombic repulsion between the two closely spaced carboxylate moieties.

CID of dicarboxylic acids has previously been shown to produce distonic radical anions in a regiospeciﬁc manner.130,154,155 4-carboxylatocyclohexyl (E) and 3- carboxylatocyclohexyl (F) radical anion would therefore be expected as the major products of CID of 1,4- and 1,3-cyclohexanedicarboxylate dianions. Unimolecular rearrangement of the nascent radical anions must also be considered. In the case of 4-carboxylatocyclohexyl radical anion (E) there are three possible isomers that could arise from hydrogen atom transfer across the cyclohexyl radical and a further ring- opened isomer that could arise from β-scission. Of these four plausible structures, three isomers (F, G and S) are distonic radical anions (although S is likely to decompose further) while one is formally cross-conjugated (H). Calculations at the CBS-QB3 level of theory suggest that H is the global minimum isomer 6.3 kcal mol−1 below the energy of E, but direct rearrangement to of E¨F via 1,4-hydrogen atom transfer is calculated to require 29.2 kcal mol−1 of energy. Optimised transition state structures for

CO2H CO2 CO2

ESI CID (a) + - 2H - CO2 - e-

CO2H CO2 m/z 126 m/z 85 E

CO2H CO2 CO2

ESI CID (b) + - 2H - CO2 CO2H CO2 - e- m/z 85 m/z 126 F

Scheme 5.3: Electrospray ionisation of (a) 1,4-cyclohexanedicarboxylic acid and (b) 1,3- cyclohexanedicarboxylic acid and subsequent CID results in oxidative decarboxylation and formation of the distonic 4-carboxylatocyclohexyl and 3-carboxylatocyclohexyl radical anions, respectively. 176 GAS PHASE REACTIONS OF CYCLOHEXYL RADICALS

these unimolecular rearrangements are displayed in Figures 5.1(a-f). The two distonic isomers F and G are essentially isoenergetic with E but rearrangement to these isomers via 1,2- and 1,3-hydrogen atom transfer is also calculated to require significant energy at 40.8 kcal mol−1 and 41.5 kcal mol−1, respectively. Consecutive hydrogen atom transfers between E and H must also overcome high energy barriers with rearrangement from F to H and G to H requiring 37.4 kcal mol−1 and 35.2 kcal mol−1, respectively. As depicted in Figure 5.1(a-e), the cyclohexyl ring must deform significantly in order to accommodate hydrogen atom transfer. The transition state for 1,5-hydrogen atom transfer requires the cyclohexane ring to isomerise to its boat conformation and required the lowest energy of the hydrogen atom transfers. The four-membered transition state associated with 1,3-hydrogen atom transfer (Figure 5.1b and Figure 5.1d) is exceptionally tight and not surprisingly has the highest barrier. The β-scission product 2-carboxylatocyclohex- 5-enyl (S) was calculated to reside 20.4 kcal mol−1 above E in addition to having a significant barrier at 27.9 kcal mol−1. S may isomerise to a low energy allylic radical, but this process still requires 35.4 kcal mol−1 and the final product is higher in energy than E at 2.4 kcal mol−1. In contrast to hydrogen atom transfer, the transition state associated with β-scission (shown in Figure 5.1f) was relatively loose and, while still above the energy of the reactants, represented the lowest barrier to isomerisation. Rearrangement of E to other isomers is therefore unlikely given that (i) any excess energy produced during oxidative decarboxylation is partitioned between the radical ion, carbon dioxide and the departing election, (ii) CID in an ion-trap instrument does not produce energetic secondary collisions, and (iii) ions within the trap are rapidly thermalised by collisions with the helium buffer gas.343

While H could not be synthesised from a dicarboxylate precursor, an alternative

– synthesis was conducted by allowing CD3O , generated by negative electrospray of

CD3OD, to react in the presence of methyl cyclohexanecarboxylate which was seeded – into the helium buffer gas (Scheme 5.5). The reaction of CD3O with methyl cyclohexanecarboxylate is expected to result in exothermic proton transfer (∆Hr xn ≈ 10 kcal mol−1, calculated from proton afﬁnity data)279 from the α-carbon of the ester. On addition of methyl cyclohexanecarboxylate into the buffer gas a peak arises at m/z 141 consistent with proton transfer from this neutral and formation of 1- methoxycarbonylcyclohexanide anion (shown in Figure 5.2). The analogous experi- RESULTS 177

(a) 1,2-hydrogen atom transfer (E¨F) (b) 1,3-hydrogen atom transfer (E¨G)

(e) 1,2-hydrogen atom transfer (G¨H) (f) β-scission (E¨S)

Figure 5.1: Optimised transition state structures calculated at the CBS-QB3 level of theory. The dashed line represents changes in the bonding at the transition state. 178 GAS PHASE REACTIONS OF CYCLOHEXYL RADICALS

CO2

β-scission H-transfer 18.4 chx5001 37.4 -CO2 CO 40.8 chx1201 CO -e CO2 2 F -0.7 2 chx701 chx201 27.9 chx2501 29.2 chx1101

H -6.3 S 20.4 E CO2 chx001 CO2 chx4801 41.5 35.4 0.0 kcal mol-1 chxl10006 chx901 chx301 35.2 chx1301 2.4 chxl10005 G -0.4 CBS-QB3 chx102 Scheme 5.4: Reaction energy and transition state energy of β-scission and hydrogen atom transfer in the 4-carboxylatocyclohexyl radical anion. Calculated at the CBS-QB3 level of theory. Text under energy is a reference to the calculation in supplementary material.

CO2CH3 CO2CH3 CO2 H(D) CID - CD3O + - CD3OH(D) - CH3 m/z 34 m/z 141 m/z 126 H

Scheme 5.5: Electrospray ionisation of d4-methanol yields d3-methoxide anion (m/z 34) which reacts with methyl cyclohexanecarboxylate by proton transfer to yield the 1-methoxycarbonylcyclohexanide anion (m/z 141). CID of the generated anion results in methyl radical loss to produce 1-carboxylatocyclohexyl radical anion. ment with methyl 1-deuterocyclohexanecarboxylate also generated an ion at m/z 141 which demonstrated that the proton is indeed removed from the α-carbon. CID of the isolated m/z 141 ion results in a small m/z 126 ion which is consistent with previous literature reporting homolytic dissociation of ester-enolate anions during CID,162 suggesting the structure of m/z 126 corresponds to H, the 1-carboxylatocyclohexyl radical anion (Scheme 5.5). This isomer is computed to be the lowest energy of the carboxylatocyclohexyl radicals at the CBS-QB3 level of theory at -6.3 kcal mol−1 below the 4-carboxylatocylohexyl radical anion (Scheme 5.4).

Ion-molecule reactions have been used in previous studies to probe the structure of putative distonic ions, making use of neutral reagents such as dimethyl disulﬁde

142 150,149 · 153 (DMDS), O2 and NO . Isomers E, F and H where subjected to reactions with these reagents to aid in their structural elucidation. DMDS was found to readily donate protons to the 1,4-cyclohexanedicarboxyate dianion, thereby decreasing the abundance RESULTS 179

Figure 5.2: (a) Full spectrum demonstrating electrospray ionisation of d4-methanol which yielded d3- methoxide anion (m/z 34). (b) On addition of methyl cyclohexanecarboxylate to the helium buffer gas an ion consistent with proton transfer to d3-methoxide anion to form 1-methoxycarbonylcyclohexanide anion (m/z 141) arises. This reaction occurs during isolation of the initial ion population so reaction time is unknown. of the dianion to an extent to which production of sufﬁcient target radical ion for isolation and further reaction was hindered. Harman and Blanksby made use of Barton ester derivatives to avoid this problem,155 but CID of the Barton ester derivative of 1,4- cyclohexanedicarboxylic acid resulted in charge driven rearrangement and elimination of 2-thioxo-2H-pyridin-1-ide, resulting in an ion at m/z 110 (data not shown), not homolytic cleavage of the nitrogen-oxygen bond and formation of E, thus DMDS could not be employed in this case.

Isolating the radical ions E, F and H in the presence of O2 resulted in the spectra − displayed in Figure 5.3. An [M + 32] ion at m/z 158 is generated during reaction of

E and F consistent with addition of O2. Two secondary ions at m/z 125 and m/z 141 · · are assigned to losses of HO2 and HO respectively and the structure of these ions will be discussed in detail later. More importantly, the ratio between the abundances of the reaction products is unquestionably different, and additionally, CID of the adduct products of m/z 158 resulted in distinct spectra with markedly different product ion abundance ratios (Figure 5.5). In contrast, reaction of H did not yield an analogous +32 Da adduct, instead a single product at m/z 60 was observed. This transformation has not been previously reported, and appears to occur due to formation of the carbonate

–· radical anion CO3 , a possible mechanism for which is outlined in Scheme 5.6(a). 180 GAS PHASE REACTIONS OF CYCLOHEXYL RADICALS

Figure 5.3: Allowing (a) 4-carboxylatocyclohexyl radical anion (E), (b) 3-carboxylatocyclohexyl radical anion (F), and (c) 1-carboxylatocyclohexyl radical anion (H) to react in the presence of background O2. (a) and (b) were isolation for 2000 ms while (c) was isolation for 200 ms. While no O2 was added during measurement of (c) there was a higher concentration (not calculated) of O2 due to use of the mixing manifold. RESULTS 181

Figure 5.4: Reaction of (a) 4-carboxylatocyclohexyl radical anion (E), (b) 3-carboxylatocyclohexyl radical anion (F), and (c) 1-carboxylatocyclohexyl radical anion (H) with NO· (Concentration of NO· during (a) and (b) was approximately 2 x 1011 molecules cm−3, while concentration of NO· during (c) was not measured). Spectrum (a) and (b) were measured by isolating the parent dianions for 100 ms while applying CID energy, as attempting isolation of the product radical anion in the presence of NO· resulted in little signal. Excitation only affects the mass selected ion, in this case m/z 85, so by increasing activation time during CID the reaction of product ions with NO· is still able to be measured. The ion at m/z 171 is proton transfer to the m/z 85 dianion resulting in their respective [M - H]− ion. 182 GAS PHASE REACTIONS OF CYCLOHEXYL RADICALS

Figure 5.5: Subjecting (a) 3-carboxylatocyclohexylperoxyl and (b) 4-carboxylatocyclohexylperoxyl (I)) radical anions to CID at 20 (normalised collision energy).

NO· is expected to undergo facile radical-radical combination reactions with carbon-centred radicals. Wenthold and Squires demonstrated that cross-conjugated α · -carboxylate acetate radical anions undergo addition of NO and elimination of CO2 . – + . −−→ - + 155,156 (overall CH2CO2 NO CH2NO CO2). It should therefore be possible to distinguish between putative distonic radical anions and their conventional analogue by allowing reaction between the radical ion and NO· to observe either radical-radical · · combination with NO (addition of 30 Da) or addition of NO and loss of CO2 (overall loss of 14 Da) for distonic or cross-conjugated radical anions, respectively. Isolation

O O O O O O C C C O O O O +O2 O (a) CO3

m/z 126 m/z 60 H O O O O O C C N NO +NO (b) CO2

m/z 126 m/z 112 H Scheme 5.6: Putative mechanisms for (a) formation of m/z 60 during reaction of 1-carboxlatocyclohexyl · radical anion with O2, and (b) formation of m/z 112 during reaction of NO with 1-carboxylatocyclohexyl radical anion. RESULTS 183

of both E and F (m/z 126) in the presence of NO· was carried out by increasing the activation time during CID of their dianion precursors. The experiment was undertaken in this manner as isolation of the E and F in the presence of NO· resulted in little signal, most probably due to the facile reaction between the radical with excess NO· and background O2 during the isolation process. As discussed in Chapter 1.2.3.2.5, excitation only affects the mass-selected ion during CID, so increasing the activation time is still an effective means of measuring the reaction of product ions in the presence of a neutral molecule. As E and F were not isolated free of their parent dianions, an [M − - H] ion at m/z 171 is present in both spectra due to proton transfer from to the [M − - 2H]2 ion (m/z 85). As shown in Figures 5.4(a) and (b), these reactions resulted in − an abundant [M + 30] ion at m/z 156 consistent with addition of NO· to form the 4- and 3-nitrosocyclohexanecarboxylate anions respectively, while reaction of H with NO· − (Figure 5.4c) was found to result in an [M - 14] ion at m/z 112 (see Scheme 5.6b). These data provide evidence to support the assignment of E and F as the distonic 4- and 3- carboxylatocyclohexyl radical anions, while H displayed similar reactivity to the acetate radical anion investigated by Wenthold and Squires, suggesting that this α-carboxylate radical is cross-conjugated, and not distonic.

All experimental and computational data presented supports the assignment of three distinct structures for the independently synthesised, isomeric m/z 126 ions. Dur- ing longer activation times in the presence of O2, an m/z 60 ion appears during isolation of 4-carboxylatocyclohexyl and 3-carboxylatocyclohexyl radical anions, however, when their respective m/z 126 ions are isolated such that they are consumed by their reaction with O2, the m/z 60 peak is at highest abundance approximately 0.5% of the dominant ion, suggesting the presence of H at very low levels and thus when isomerisation occurs it is only to a very minor extent. Importantly, this provides convincing evidence that the m/z 126 ion generated by oxidative decarboxylation of 1,4-cyclohexanedicarboxylate dianion may be conﬁdently assigned as the distonic 4-carboxylatocyclohexyl radical anion(E). 184 GAS PHASE REACTIONS OF CYCLOHEXYL RADICALS

5.3.2 Reaction of the 4-carboxylatocyclohexyl radical anion (E) with

The reaction of the 4-carboxylatocyclohexyl radical (E) with O2 was measured over trapping periods of 30 - 5000 ms. The spectra from these experiments are presented in − Figure 5.6. An [M + 32] ion at m/z 158 increases in abundance over longer trapping times. An analogous experiment in which E was allowed to react with 18O-labelled dioxygen (Figure 5.8a) resulted in a +4 Da mass shift of the [M + 32]− ion to generate − − an [M + 36] ion at m/z 162 demonstrating that the [M + 32] ion observed in the unlabelled experiment is an O2 adduct, and does not occur due to addition of an isobaric neutral such as MeOH. The m/z 158 ion is therefore assigned as the 4- carboxylatocyclohexylperoxyl radical anion (I). A rate constant for the addition of O2 10 to the 4-carboxylatocyclohexyl radical anion was measured at k2 = 1.8 × 10 with an overall reaction efﬁciency of 30%. Isolation and CID of the m/z 158 (shown in Figure 5.5b) resulted in an abundant [M - 33]− ion and minor [M - 17]− consistent with neutral

· · 18 losses of HO2 and HO , respectively. Subjecting the O2 labelled counterpart to CID (shown in Figure 5.8(b)) resulted in [M - 37]− and [M - 19]− ions displaying mass shifts of +4 Da and +2 Da relative to the unlabelled spectrum therefore conﬁrming the · · assignment of these ions as HO2 and HO neutral losses, respectively. In addition, these – data demonstrate that all losses involve extraneous O and not the CO2 charge-tag. In both cases, reformation of the m/z 126 arising due to loss of O2 to generate the parent ion E was minimal suggesting that reversible O2 loss is minimal and does not compete with formation of the products.

Branching ratios for the three major competitive pathways were calculated at 54%, · · 11% and 32% for the loss of HO2, HO and addition of O2, respectively. Minor reaction channels account for 3% of the total product ion abundance. There was no evidence of secondary products such as addition of O2 to I. This was conﬁrmed by isolating I in the presence of O2 for up to 5000 ms (data not shown) which resulted in no reaction. Generation of these three products must therefore arise due to direct reaction E with O2, and are therefore primary reaction products. Furthermore, these data suggest that the ions at m/z 158 do not contain a signiﬁcant population of long-lived hydroperoxylcyclohexyl radicals (.QOOH radicals) which have been proposed as intermediates in further RESULTS 185

Figure 5.6: Isolation of 4-carboxylatocyclohexyl radical anion (E) in the presence of background O2 for increasing reaction times between 30 and 5000 ms. chain branching,327 instead it would seem that any .QOOH radicals are transient and further reaction is unimolecular. Alternatively, the 4-carboxylatocyclohexylperoxyl radical anion could isomerise to a structure which hinders approach of neutral molecules (a possible isomer is shown in Scheme 5.7). Reaction of the isomerised would be hindered until sufﬁcient energy is applied (e.g., CID) to disturb the interaction.

O O O O O O O C C O H -44.8 -6.6 chxl1501 chxl10004 Scheme 5.7: Putative isomerisation mechanism of the 4-carboxylatocyclohexylperoxyl radical anion such that reaction of O2 is hindered. Calculated at the CBS-QB3 level of theory relative to the 4-cyclohexyl + O2 entrance channel.

· Loss of HO2 to form cyclohexene (J) can occur by direct hydrogen abstraction from

E by O2, or through addition of O2 to form I followed by a concerted 1,5-hydrogen . · atom transfer through the QOOH intermediate M and subsequent elimination of HO2 (Scheme 5.2ii). CID of I (Figure 5.5b) resulted in formation of m/z 125 ion conﬁrming · that loss of HO2 can occur from the nascent 4-carboxylatocyclohexylperoxyl radical 186 GAS PHASE REACTIONS OF CYCLOHEXYL RADICALS

(I), and further suggesting that J is produced via the concerted isomerisation and elimination pathway as previously predicted for low temperature oxidation,315 and not direct hydrogen abstraction.

The reaction product at m/z 141 was found to be a minor product with a branching ratio of 11%. This ion was also observed as a loss of HO· (17 Da) during CID of I which shows that this product is similarly formed from the peroxyl radical intermediate. Scheme 5.2 describes ﬁve separate channels previously predicted to result in loss of HO· from the cyclohexylperoxyl radical. Deuterium labelling was undertaken in order to rule out a number of these products. 1,4-dideutero-1,4-cyclohexanedicarboxylic acid was used as a precursor to the 4-carboxylato-1,4-dideuterocyclohexyl radical anion, which was allowed to react O2 (Figure 5.8c). Products ions at m/z 127 and 143 were · · assigned as loss of HO2 and loss of HO , respectively, from the 4-carboxylato-1,4- dideuterocyclohexylperoxyl radical present at m/z 160. There was a minor m/z 142 peak which would indicate loss of DO·. CID of the deuterated peroxyl radical (Figure 5.8d) resulted in a similar array of ions, but in this case there were no fragment ions indicating a deuterated neutral loss. This suggests that if hydrogen abstraction from the 1- or 4-position (Scheme 5.2ix and 5.2vii, respectively) occurs, it is only to a very minor extent, and therefore the ion population of m/z 141 contains little contribution from the 4-oxocyclohexanecarboxylate (R) and 1,4-epoxycyclohexanecarboxylate anions (P).

To account for the other three reaction channels, the ions at m/z 141 were isolated and subjected to CID (Figure 5.7a). A comparison of this spectrum to that of 4-oxocyclohexanecarboxylate (Figure 5.7b) reveals noticeable differences. The latter spectrum contains two major ions at m/z 97 and m/z 99, and while the CID spectrum measured of the isolated m/z 141 ions contains a major ion at m/z 97, there is only a very minor peak at m/z 99, and more importantly, there are a significant number of additional ions. This finding is consistent with the deuterium labelling data which suggested that there was little contribution from 4-oxocyclohexanecarboxylate. Sub- jecting 3,4-epoxycyclohexanecarboxylate (L) to CID resulted in the spectrum shown in Figure 5.7(b). While there were corresponding ions at m/z 79, 95, 97, 113 and 123, significantly, there were no ions present at m/z 43 and 61. The abundance of the corresponding fragment ions was also inconsistent given that the CID spectra were measured at the same normalised collision energy. This suggests that while 3,4- RESULTS 187

epoxycyclohexanecarboxylate may contribute to the ion population at m/z 141, an alternative isomer must also be present. Unfortunately, it was not possible to rule out contributions from 2,4-epoxycyclohexanecarboxylate (O) due to lack of an authentic precursor, but the ring opened isomer N (Scheme 5.2) can be ruled out based on calculations of the 4-carboxylatocyclohexyl + O2 potential energy surface (presented below) where the barrier for generation of this product was predicted to be higher than the energy of the reactants.

These experimental data suggest that the major products of the reaction between · · the cyclohexyl radical and O2 occur due to loss of HO2 and HO from a nascent · cyclohexylperoxyl radical intermediate. Loss of HO2 results in generation of cyclohexene while loss of HO· may result in isomeric epoxides including 3,4- and 2,4- epoxycyclohexane. Loss of HO· resulting in formation of 1-cyclohexanone or 1,4- epoxycyclohexane were ruled out based on labelling experiments.

5.3.3 Computation of the 4-carboxylatophenyl + O2 potential energy surface

Calculation of the potential energy surface for the reaction of the 4- carboxylatocyclohexyl radical anion (E) with O2 was undertaken at B3LYP/6-31+G(d) and CBS-QB3 levels of theory and the results are illustrated in Figure 5.9. CBS-QB3 has been shown to have a maximum unsigned error of 6.2 kcal mol−1 with a mean absolute deviation of 1.1 kcal mol−1 for calculation of the G2/97 test set which contains model calculations of heat of formation, ionisation potential, electron affinity and proton affinity across a broad range of molecules338 and, in addition, performed well for a the DBH24/08 database for chemical reaction barrier heights with a mean unsigned error of 1.62 kcal mol−1.344 Both energies are reported in Figure 5.9 for comparative purposes. While Knepp et al.329 found B3LYP energies to deviate significantly from higher level calculations, the relative order of transition states was found to be consistent with the higher level calculations suggesting that B3LYP was a good qualitative point of reference for comparing competing reaction channels. B3LYP/6-31+G(d) was therefore used to survey the potential energy surface while accurate energies were calculated at the CBS-QB3 level. Addition of dioxygen to the 4-carboxylatocyclohexyl radical anion (E) can give rise to two geometric isomers with the carboxylate anion and peroxyl 188 GAS PHASE REACTIONS OF CYCLOHEXYL RADICALS

Figure 5.7: CID spectra of a) ion of m/z 141 generated by reaction of 4-carboxylatocyclohexyl radical anion E with O2, b) 3,4-epoxycyclohexanecarboxylate anion, and c) 4-oxocyclohexanecarboxylate anion. RESULTS 189

18 Figure 5.8: Reaction of a) 4-carboxylatocyclohexyl radical anion E with O2, and b) CID spectrum measured of the m/z 162 ion generated in this reaction. Reaction of c) 1,4-dideutero-4-carboxylatocyclohexyl with O2, and d) CID spectrum measured of the m/z 160 ion generated in this reaction. 190 GAS PHASE REACTIONS OF CYCLOHEXYL RADICALS

radical moieties bound to the same (IZ ) or opposite (IE ) faces of the cyclohexane ring. While these diastereomeric peroxyl radicals are calculated to be within 1.7 kcal mol−1, they can result in different chemistries and different reaction energetics as illustrated in Figures 5.9(a) and (b).

Calculations predict that rearrangement of the nascent peroxyl radical (I) via hydrogen atom transfer can proceed without additional activation energy from all positions in the cyclohexane ring (Scheme 5.2ix). 1,3-hydrogen atom transfer and concerted loss of HO· results in formation of 4-oxocyclohexylcarboxylate anion (R) via a 4-centred transition state that lies either -4.8 or -7.4 kcal mol−1 below the entrance channel for rearrangement from the IZ and IE isomers, respectively. This barrier is higher than those observed of all other transition states except for the β-scission reaction resulting

−1 in NZ which lies +1.9 kcal mol above barrier. This may explain why the otherwise exothermic product channel (-69.6 kcal mol−1 below reactants) was not observed experimentally (vide supra).

· Both pathways for the observed elimination of HO2, either via a concerted mechanism (Scheme 6i) or via the intermediate 2-hydroperoxylcyclohexyl radical (K, Scheme 6iii), are found to be energetically accessible. Transition structures for concerted

· −1 elimination of HO2 from IZ and IE are located some -7.8 and -9.1 kcal mol , below the entrance channel, while the most energetic transition states for the stepwise process are found at -9.7 and -9.1 kcal mol−1. The calculations thus give no clear indication as to whether the concerted or stepwise mechanism is more likely responsible for the · experimental observation of HO2 loss. Transition states for rearrangement via 1,4-, 1,5- or 1,6-hydrogen atom transfer, are calculated to lie between -9.7 and -21.6 kcal mol−1 below the entrance channel. While all the resulting hydroperoxylcyclohexyl radicals (K, M, P Scheme 6) are found to be minima on the potential energy surface, all are further connected to energetically accessible pathways for subsequent reaction via epoxide formation with concomitant loss of the HO· radical. The accessibility of such pathways is consistent with fact that no secondary addition of O2 was observed experimentally (vide supra) as might be expected if such .QOOH isomers are present in the ion population at m/z 158.

Interestingly, the calculations do not provide a clear prediction as to the major reaction product. In particular, the 1,4-epoxycyclohexane carboxylate (QE ) is found to RESULTS 191

- 10 CO2

+ O 2 1.9 0.0 [11.5] E chx2601 chx301 0 -4.8 [4.3] chx1901 -9.7 chx1502 [-0.4] chx6301 - CO2 - -7.8 -14.7 - -10 CO2 CO2 [-4.2] [-8.8] -16.4 -16.6 -17.1 chx6101 [-1.8] JZ + HO2 [-10.3] + HO2 [-8.0] chx2301 O chx4001 chx2101 -20.7 chx4001 OH J -16.6 [-0.6] Z MZ chx6501 -20 [-10.3] -28.1 [-15.6]

chx2901 -30 chx3301 -31.9 KZ Energy (kcal/mol) [-20.7] - CO2 chx501 O chx4601 CO - Z -40 -44.8 2 -37.3 O + HO IZ [-37.5] [-24.0] chxl3502 - -42.6 CO2 LZ O chx6503 [-31.3] chx4701 OH -45.5 -45.0 - NZ -50 [-38.7] CO2 [-38.4] - O + HO CO2 O HO - 2 CO2 O + HO

-60 O - CO2 4-carboxylatocyclohexyl + O2 -69.6 cis-PES RZ + HO [-61.0] CBS-QB3 O [ B3LYP/6-31+G(d) ] -70 chx3801

- 10 CO2

+ O2

E 0.0 chxl301 0 -7.4 -9.1 [3.1] -9.1 [-6.6] chxl6601 -11.3 -11.6 [-4.2] -12.3 chxl1801 - chxl1402 [-4.2] [-5.6] - CO2 CO - CO2 -10 chxl6401 2 [-12.2] chxl6801 chxl2401 -16.4 + HO -17.2 + HO 2 KE [-10.6] 2 [-9.1] chxl1601 chxl2201 chxl3901 O chxl3901 OH -16.8 -16.8 JE JE -20 [-11.6] chxl2001 [-11.6] -21.6 -27.1 [-13.6] [-16.6]

chxl3201 chxl2801 -27.5 -30 M E [-16.8] chxl3401 P -32.6 - Energy (kcal/mol) E CO2 [-23.1] CO - -40 2

O chxl4501 OH chxl3604 O -41.1 chxl401 E LE -42.3 O [-27.8] IE -43.1 [-31.2] OH - [-37.3] CO2 - -50 CO2 - CO2 O + HO + HO O Q chxl4401 O E -55.1 O -60 [-40.0]

- CO2 - CO2 4-carboxylatocyclohexyl + O2 O + HO R -72.8 trans-PES E + HO CBS-QB3 -70 [-62.6] [ B3LYP/6-31+G(d) ] chxl7001 O

Figure 5.9: Reaction coordinate diagram for the reaction of the 4-carboxylatocyclohexyl radical anion (E) with −1 O2 where all energy values are presented as kcal mol . Addition of O2 can occur from either the face of the cyclohexyl moiety resulting in cis- and trans-isomers of the 4-carboxylatocyclohexylperoxyl radical anion. The rearrangements are calculated for both the (a) the cis-isomer IZ and (b) the trans-isomer IE . All calculations were performed at CBS-QB3 and B3LYP/6-31+G(d) level of theory (in brackets), which included unscaled ZPE corrections. 192 GAS PHASE REACTIONS OF CYCLOHEXYL RADICALS

be energetically accessible with among the lowest energy transition states for both the 1,6-hydrogen atom transfer (-17.2 kcal mol−1) and subsequent epoxide formation (-12.3 kcal mol−1). It should be noted, however, that this product can only be formed from the trans-peroxyl radical (IE , see Figure 5.9b) while both 1,2- and 1,3-epoxycyclohexane carboxylates can be formed from both cis and trans isomers (see LZ , OZ , LE , and OE in Figures 5.9a and b). Furthermore, from symmetry there are effectively four equivalent paths (2 for cis and 2 for trans) for formation of the 1,2- and 1,3-epoxycyclohexane carboxylates compared to only a single pathway for formation of the low energy 1,4- epoxyocylcohexane carboxylate (QE ). The resulting low probability for this reaction may account for the fact that no experimental evidence (based on deuterium-labelling) was obtained for this otherwise energetically competitive exit channel. Finally the calculated β-scission pathway (Scheme 5.2vi) for the 3-hydroperoxycyclohexyl radical (M) was found to have a transition state above that of the entrance channel (+1.9 kcal mol−1, Figure 5.9a). This result suggests that the ring opened aldehyde isomer is an unlikely contributor to the ion population at m/z 141.

The carboxylate charge tag is an obvious point of difference between the distonic radical anion studied here and the neutral radical system. The calculated energies for the reaction of dioxygen and the 4-carboxylatocyclohexyl radical (E) are in generally good agreement with previous computations of the analogous neutral cyclohexyl radical reaction.329,327 For example, the overall reaction energies are very similar between the carboxylate charge-tagged and neutral surfaces with values for the 1,2-, 1,3- and 1,4-

−1 epoxides (L, O and Q) calculated at -42.6 or -42.3 kcal mol (LZ and LE , respectively) −1 −1 compared with -40.4 kcal mol on the neutral surface, -37.3 or -41.1 kcal mol (OZ and −1 −1 OE , respectively) compared with -35.8 kcal mol for the neutral and -55.1 kcal mol −1 329 (QE ) compared with -53.4 kcal mol . Furthermore, the barriers to rearrangement of the 4-carboxylatocyclohexyl radical via hydrogen atom transfer range between 21.5 to 35.1 kcal mol−1 compared with 27.2 to 34.8 kcal mol−1 for rearrangement on the neutral cyclohexyl radical potential energy surface.

The relative order of the barrier heights for 1,4-, 1,5-, 1,6-hydrogen atom transfer · and 1,4-hydrogen transfer with concerted elimination of HO2 are similar to those calculated by Silke et al., namely, 1,5-hydrogen atom transfer (to form K) < 1,6 hy- < · < drogen atom transfer (to form P) concerted elimination of HO2 (to form J) 1,4- RESULTS 193

hydrogen atom transfer (to form M).327 With the carboxylate charge tag, the barrier · to concerted elimination of HO2 is slightly higher than that of the stepwise process. This also differs slightly from the ﬁndings of Knepp et al. who report a lower activation

· 329 energy for concerted elimination of HO2 with respect to 1,6-hydrogen atom transfer. The resulting hydroperoxylcyclohexyl radical isomers computed in the present study (K, M, P Scheme 6) range in energy between -27.1 kcal mol−1 and -32.6 kcal mol−1 below the entrance channel with the lowest of these minima being identiﬁed as the 1-carboxylato-4-hydroperoxylcyclohexyl radical (P, Scheme 6). This contrasts with calculations of the analogous neutral surface where 3-hydroperoxylcyclohexyl radical and 2-hydroperoxylcyclohexyl radical are calculated as the lowest energy .QOOH structures.329,327 This is not surprising however, as the presence of the carboxylate moiety in 1-carboxylato-4-hydroperoxylcyclohexyl radical (P) provides for resonance delocalisation that is absent in the neutral system.

As reported by Knepp et al., B3LYP/6-31+G(d) energies were found to deviate from those calculated at the higher CBS-QB3 level. B3LYP was found to predict similar barrier heights to CBS-QB3 with a mean unsigned deviation of 3.8 kcal mol−1 and a maximum absolute deviation of 9.4 kcal mol−1. Absolute energies were relatively poor, with a mean unsigned deviation of 8.9 kcal mol−1 and maximum absolute deviation of 20.1 kcal mol−1. More signiﬁcantly, however, was calculation of the barrier height for 1,3- hydrogen atom transfer from cyclohexylperoxyl radical (I¨F) where B3LYP calculated the transition state structure energy for the two isomers as above the entrance channel while CBS-QB3 calculated both to lie below. Overall, this suggests that higher level energy calculations were signiﬁcant in accurately representing the potential energy surface of this reaction.

· Finally, these calculations supported experimental assignments, for example, HO2 loss results in formation of cyclohexene as previously predicted, however, an exclusive mechanism for this process could not be determined. Furthermore, elucidation of the product ion population observed at m/z 141 occurring due to loss of HO· from the nascent 4-carboxylatocyclohexylperoxyl radical anion (I) was not possible as all products, including the isomeric epoxides and ketone, are predicted to be thermodynamically accessible at the CBS-QB3 level of theory. This suggests that barrier heights may not be the limiting factor in these branching pathways, and that entropy plays a 194 GAS PHASE REACTIONS OF CYCLOHEXYL RADICALS

major role in the degradation processes.

5.4 Conclusion

The distonic 4-carboxylatocyclohexyl radical anion (E) was synthesised by oxidative dicarboxylation of 1,4-cyclohexanedicarboxylate dianion and its reactivity towards O2 investigated in order to explore the applicability of using this charge-tagged model to investigate the reactivity of the neutral cyclohexyl radical. Isomeric 3- and 1- carboxylatocyclohexyl radical anions (F and H, respectively) were also independently synthesised and their structures probed and compared with that of the putative 4- carboxylatocyclohexyl radical anion by CID experiments and by their reaction with O2 and NO·. These experiments demonstrated that the charge-tagged radical anion was structurally distinct from its isomeric counterparts, and furthermore, displayed radical · reactivity, speciﬁcally radical-radical combination with O2 and NO , a characteristic trait of distonic radical ions, conﬁrming the structural assignment of E.

When (E) was allowed to react in the presence of O2, a +32 Da adduct 18 was formed which was assigned after O2 labelling experiments as the 4- carboxylatocyclohexylperoxyl radical anion (I); secondary reactions of I were not observed. Isolation of I for reaction times of up to 5000 ms resulted in no reaction. This ruled out formation of the isobaric hydroperoxylcyclohexyl radicals (.QOOH radicals) which have been proposed as intermediates in further chain branching pathways during cyclohexane oxidation. In this case, the peroxyl radical I may adopt a conformation which hinders further reaction. This would not necessarily alter CID fragmentation but could hinder approach of neutral reactants.

· · Characteristic product ions arising from loss of HO2 and HO were consistent with previous investigations of the oxidation of cyclohexane and were found to result from the degradation of the cyclohexylperoxyl radical intermediate. Labelling experiments and computation of the 4-carboxylatocyclohexyl radical anion + O2 potential energy surface were used to probe and assign the structure of the product ions. The major · HO2 loss channel was predicted to generate cyclohexene, while a speciﬁc structural assignment of ions formed from HO· loss was not possible as calculations suggested many products should be energetically accessible. Experimental data did, however, exclude formation of cyclohexanone, 1,4-epoxycyclohexane and 5-hexen-1-al, with CONCLUSION 195

HO· loss ultimately predicted to result in formation of the isomeric 1,2- and 1,3- epoxycyclohexanes.

The ability to synthesise and isolate model alkyl and alkylperoxyl radicals as described here provides a novel method to probe the gas phase reactivity of these otherwise elusive species. This approach offers promise for the study of the bimolecular reactions of peroxyl radicals as well as their gas phase photochemistry. The distonic radical anion E was found to react with O2 and gave rise to characteristic products: · · the addition of O2 (+32 Da) and neutral losses of HO (+15 Da) and HO2. In contrast, the analogous reaction with the cross-conjugated radical anion H was found to result

–· in ejection of CO3 . Such diagnostic gas phase reactions not only enable veriﬁcation of distonic character, but also the structure of the radical itself. These experiments support

149,150 previous investigations which suggest use of O2 as a diagnostic probe of distonic –· radical ions, and furthermore, suggest that the characteristic loss of CO3 arising from reaction with O2 with cross-conjugated radical anions may be used to structurally probe radical ions.

inlsuyicroaigetoycnrbtossgetn htls fCHO CO of with loss competitive that suggesting contributions entropy incorporating study tional and the of gas irradiation buffer laser ions. (ii) the trapped and of into reactions reagents ion-molecule of neutral observation of for flow spectrometer controlled mass a accommodate of To to addition modified (i) was allow ion-trap counterparts. quadrupole neutral linear LTQ their ThermoFisher of a have investigations, reactivity radicals these cyclohexyl the and model phenyl to of used models been ion radical distonic studies, these In ohOadCHO and O that both demonstrated radical reaction. phenylperoxyl isolated this the of of dissociation product induced Collision reaction major the CO as of radical loss clopentadienyl predict that long-standing predictions from computational significantly and differs which mechanistic radical, phenoxyl the product was degradation cases major both The in isolation. subsequent for gen- quantity was sufficient radical in phenylperoxyl erated the case, cationic the In intermediate. radical peroxyl a O with anion radical 4-carboxylatophenyl of reactions The spectrometry. itncpey,aaatladccoey aia oswt abxlt anion carboxylato with ions radical cyclohexyl and adamantyl phenyl, Distonic N,N,N tiehlmoimcto oiswr yteie sn o-rpmass ion-trap using synthesised were motifs cation -trimethylammonium · a els ietyfo h hnleoy aia,wt computa- a with radical, phenylperoxyl the from directly lost be may 2 osa o temperatures. low at loss S MAYADCONCLUSIONS AND UMMARY N,N,N tiehlmoimhnlrdclcto and cation radical -trimethylammoniumphenyl 2 197 eefudt rce hog omto of formation through proceed to found were 2 ognrt cy- generate to 53,54,57,58,261,55

C HAPTER · a be may 6 198 SUMMARY AND CONCLUSIONS

A study of the carboxylate charge-tagged cyclohexyl radicals uncovered a diagnostic α reaction of O2 with -carboxylates similar to the E1cb reaction of the cross-conjugated acetate radical anion with NO· observed by Wenthold and Squires (Scheme 6.1).153 The 1-carboxylatocyclohexyl radical anion, which has a similar α-carboxlyate motif, was found to react in the presence of O2 to form an ion of m/z 60. The formation of this ion is consistent with addition of O2 to the 1-carboxylatocyclohexyl radical anion followed –· by ejection of CO3 . In direct contrast, the isomeric distonic 4-carboxylatocyclohexyl radical anion underwent radical-radical combination in the presence of O2 to form the 4-carboxylatocyclohexylperoxyl radical anion. The structure of putative distonic radical anions featuring a carboxylate moiety may therefore be probed by their reaction with − O2. An [M + 32] ion is diagnostic of a distonic radical anion, while formation of a m/z 60 ion is characteristic of an α-carboxylate radical anion. The relative intensity of the m/z 60 ion to the [M + 32] ion may also be used to gauge the extent to which a carboxylate charge-tagged radical isomerises to the α-carboxlyate.

O O O O O O C C C O O O O +O2 O (a) CO3

m/z 126 m/z 60 H O O O O O C C N NO +NO (b) CO2

m/z 126 m/z 112 H Scheme 6.1: Putative mechanisms for (a) formation of m/z 60 during reaction of 1-carboxlatocyclohexyl · radical anion with O2, and (b) formation of m/z 112 during reaction of NO with 1-carboxylatocyclohexyl radical anion.

A combined computational and experimental study aimed at investigating the degradation of the cyclohexylperoxyl radical found that the distonic 4- · · carboxylatocyclohexylperoxyl radical anion model degrades by loss of HO2 and HO to form alkenes and isomeric epoxides. These data mirrored previous investigations of the neutral archetype.327,329,323 Alternative pathways leading to formation of cyclohexanone or ring-opened isomers were not observed. Subsequent isolation of the nascent cyclohexylperoxyl radical in the presence of O2 resulted in no reaction. This SUMMARY AND CONCLUSIONS 199

suggested that under the conditions of these experiments the radical does not isomerise to a hydroperoxycyclohexyl radical as has previously been proposed.

After successful production of distonic cyclohexyl, adamantyl and phenylperoxyl radical ions, experiments aimed at synthesising and isolating key ROONO and ROONO2 · · intermediates in the reaction of these models with NO and NO2 were also undertaken. Allowing the distonic cyclohexylperoxyl radical to react with NO· resulted in · no reaction, while its reaction with NO2 was hindered by electron transfer of the · parent dicarboxylate to NO2. The major products of the reaction of the N,N,N- trimethylammoniumphenylperoxyl radical cation and 3-carboxylatoadamantylperoxyl radical anion with NO· were their respective RO· and RONO species. When the phenylperoxyl radical was allowed to react with NO·, ions of low abundance consistent with an ROONO or RO2NO adduct were observed but not isolable.

The absence of a reaction between the 4-carboxylatocyclohexylperoxyl radical anion and NO· was surprising given the successful reaction between 3- carboxyloadamantylperoxyl radical anion and NO·. This could suggest that there is a barrier to the reaction, energetically or geometrically. The peroxyl radical could, conceivably, adopt a conformation or isomerise (e.g. Scheme 6.2) such that ap- · proach of extraneous O2 or NO is hindered until energy is applied (i.e. collision induced dissociation) to disturb the interaction. This would not occur in the 4- carboxylatoadamantylperoxyl radical anion due to the rigidity of the molecule, which hinders reaction between the peroxyl and carboxylate moieties.

O O O O O O O C C O H -44.8 -6.6 chxl1501 chxl10004 Scheme 6.2: Putative isomerisation mechanism of the 4-carboxylatocyclohexylperoxyl radical anion such that · reaction of O2, NO or other neutrals is hindered.

Formation and isolation of the 4-carboxylatocyclohexylperoxyl radical anion in the presence of NO· was difﬁcult. The 4-carboxylatocyclohexyl radical anion precursor · was found to undergo a facile reaction with both NO and O2. The concentration of

O2 could not be increased by further addition through the gas mixing manifold as it · · reacts with NO to form the interfering neutral NO2. As such, a limit was placed on 200 SUMMARY AND CONCLUSIONS

the amount of NO· that could be added to the buffer gas before it competed detri- mentally with formation of the peroxyl radical. The lack of reaction observed between 4-carboxylatocyclohexylperoxyl radical anion could therefore be due to an overly low concentration of NO·.

Alternatively, the absense of products be a result of the low pressure within the ion trap. Previous reports suggest that 22% of the products formed when 3-hexylperoxyl radical is allowed to react with NO· are the hexylnitrate. Collisional cooling plays an important role in stabilising the nascent cyclohexylperoxynitrite which forms during reaction of the cyclohexylperoxyl radical with NO·. Reported measurements were undertaken at 1 atm (760 Torr)5 in contrast to the experiments described in this thesis which were undertaken at 2.5 mTorr. The low pressure and inability to collisionally cool the nascent products may therefore explain the absence of products. This is similarly true for the reaction of 4-carboxylatophenyl radical anion with O2, in which a low abundance of the 4-carboxylatophenylperoxyl radical anion detected. The presence of an [M + 16] product assigned as phenoxyl radical demonstrates that the 4- carboxylatophenylperoxyl radical anion must form prior to degradation, however, there is insufﬁcient collisional cooling to stabilise the nascent peroxyl radical prior to its decomposition to products or back to reactants.

Isolation of both the 3- and 4-carboxylocyclohexyl radical anions was hindered, as while a high ion-count of the radicals was generated, using small isolation widths resulted in little or no ions. Asano et al. observed this phenomenon during isolation of leucine enkephalin and attributed it to off-resonance power absorption.345 Increasing the concentration of additional neutral reagents in many cases exacerbated this effect. While large isolation widths could in some cases could be used to isolate a smaller population of ions, some ion-molecule reactions were instead undertaken during the same isolation step as the radical anion formation stage, in which the dicarboxylic acid is subjected to CID. This alleviated the problem to some extent, but resulted in a more complicated spectrum containing ions formed due to reaction of the precursor in addition to reaction of the radical anion.

While there are distinct advantages of using FT-ICR instruments over a LIT (e.g., measurement accuracy of rates, better isolation from interfering neutral reagents due to ultra-high vacuum, resolution of spectra) the speed and simplicity of the modern SUMMARY AND CONCLUSIONS 201

LIT instrumentation and software greatly streamlines these experiments over research grade FT-ICR mass spectrometers typically employed when undertaking ion-molecule experiments. Furthermore, MSn capabilities, which offer structural elucidation of reaction products, are greatly simpliﬁed in a linear ion-trap which do not require collisional/cooling gas pump-in and pump-out cycles. This is certainly demonstrated by modern FT-ICR implementations, for example the ThermoFinnigan LTQ FT,220 where CID of ions occurs in a linear ion-trap prior to injection into the FT-ICR analyser for measurement.

Methods used to synthesise distonic radical ions in an ESI-LIT differ signiﬁcantly to those used in electron ionisation/chemical ionisation coupled FT-ICR instruments. As demonstrated in the preceding chapters there are a number of methods available with which to synthesise distonic radical ions using an ESI-LIT mass spectrometer. These include collision induced dissociation of dicarboxylic acids or sulfocarboxylic acids resulting in oxidative decarboxylation, collision induced dissociation of positive charge- tagged halides resulting in C-X bond homolysis, photodissociation of carboxylates or halides resulting in photodetachment and decarboxylation or C-X bond homolysis. In most cases, synthesis of the parent ion requires simply dissolving the molecule in MeOH with addition of NH4OH during negative electrospray. It is important to note that collision induced dissociation of halides does not necessarily result in an isomerically pure population of the desired distonic radical ion (demonstrated in these experiments and previously by Thoen et al.114). In these cases, photodissociation was required to produce the distonic radical ion in these cases. In contrast, research groups employing FT-ICR typically synthesise distonic radical ions by a three-step chemical ionisation method: (i) positive ionisation of a di-halogenated molecule to produce a radical cation; (ii) allowing the ion to react with a nucleophile resulting in substitution of the second halogen with the charge-tag; (iii) subjecting the ion to sustained off-resonance irradiation collision activated dissociation to generate the distonic radical ion. While Thoen et al. demonstrated the use of photodissociation to generate distonic radical ions in an FT-ICR mass spectrometer, this is not a commonly applied method.114

Limitations were also identiﬁed which may restrict the utility of the ESI-LIT mass spectrometer. This may include (i) the complicated nature of multiple neutral reagents in the ion-trap which leads to competing reactions with the isolated radicals. This may 202 SUMMARY AND CONCLUSIONS

limit the investigation of step-wise reactions which require different neutral reagents; (ii) the ﬁxed pressure of the ion-trap which may limit the collisional cooling of reactive or fragile product ions leading to their absence or low ion-count in a given spectrum; (iii) excitation of reactive species such as radicals during the isolation process which may result in signiﬁcant ion losses from destabilisation or degradation; (iv) the limited structural information provided when a product is in low abundance or collision induced dissociation fails due to the inherent structural stability, which ultimately results in premature ejection of the ion prior to fragmentation. These limitations, however, are not limited to the ESI-LIT and these problems also present themselves during ion- molecule reactions in FT-ICR instrumentation.

The modiﬁcations made to allow photolysis of trapped ions have now been augmented by installation of a tunable laser. This allows measurement of action spectra which describe the wavelength dependence of ion fragmentation pathways. Fur- thermore, the experiments described in this thesis provide grounds for further work. Dicarboxylate dianions, for example, are interesting species in that they contain a ’repulsive Coulomb barrier’ (RCB) resulting in an electron afﬁnity higher than their electron binding energy. There is much debate in the literature as to how the structure of dicarboxylate dianions affects the oxidative decarboxylation pathway used in this thesis to generate alkylcarboxylate radical anions. Recent work by Lai-Sheng Wang and co-workers employed photoelectron spectroscopy (PES) and velocity map imaging (VMI) to probe the structure and spectroscopy of dicarboxylate dianions.228,226 While PES and VMI are exceptional methods to probe the characteristics of the RCB, the instrumentation and methods described in the previous chapters are well suited to probing the charged decomposition products.

In all, the results presented in this thesis highlight the advantages of using a distonic radical ion approach to investigate radical reactions and their ability to model the reactivity of their neutral counterparts. Signiﬁcantly, this approach provides insight into reaction mechanisms where neutral data is unavailable or difﬁcult to obtain experimentally. BIBLIOGRAPHY

[1] IUPAC. Compendium of Chemical Terminology, 2nd ed. (the "Gold Book"). Blackwell

Scientiﬁc Publications, Oxford, 2006.

[2] T.J. Wallington, L.M. Skewes, and W.O. Siegl. Kinetics of the gas phase reaction of chlorine

atoms with a series of alkenes, alkynes and aromatic species at 295 K. J. Photochem.

Photobiol., A, 45:167–175, 1988.

[3] T.E. Graedel and W.C. Keene. Tropospheric budget of reactive chlorine. Global Bio-

geochem. Cycl., 9(1):47–77, 1995.

[4] L. Wang, J. Arey, and R. Atkinson. Reactions of chlorine atoms with a series of aromatic

hydrocarbons. Environ. Sci. Technol., 39:5302–5310, 2005.

[5] B.J. Finlayson-Pitts and Jr. Pitts, J.N. Chemistry of the Upper and Lower Atmosphere.

Academic Press, San Diego, 2000.

[6] T.J. Wallington, L.M. Skewes, W.O. Siegl, C. Wu, and S.M. Japar. Gas phase reaction of Cl

atoms with a series of oxygenated organic species at 295 K. Int. J. Chem. Kin., 20(11):867–

875, 1988.

[7] T.J. Wallington, L.M. Skewes, W.O. Siegl, and S.M. Japar. A relative rate study of the reaction

of bromine atoms with a variety of organic compounds at 295 K. Int. J. Chem. Kin.,

21(11):1069–1076, 1989.

[8] H. Togo. Advanced Free Radical Reactions for Organic Synthesis. Elsevier Ltd., Oxford,

2004.

[9] M. Gomberg. An instance of trivalent carbon: Triphenylmethyl. J. Am. Chem. Soc., 22:757–

771, 1900.

[10] B. Halliwell and J.M.C. Gutteridge. Free Radicals in Biology and Medicine. Oxford

University Press, Oxford, 3rd edition, 1999.

203 204 BIBLIOGRAPHY

[11] L.J. Marnett. Lipid peroxidation - DNA damage by malondialdehyde. Mutat. Res., 424:83–

95, 1999.

[12] S.P. Hussain, L.J. Hofseth, and C.C. Harris. Radical causes of cancer. Nat. Rev. Cancer,

3(4):276–285, 2003.

[13] H. Bartsch and J. Nair. Accumulation of lipid peroxidation-derived DNA lesions: Potential

lead markers for chemoprevention of inﬂammation-driven malignancies. Mutat. Res.,

591:34–44, 2005.

[14] U. Nair, H. Bartsch, and J. Nair. Lipid peroxidation-induced DNA damage in cancer-prone

inﬂammatory diseases: A review of published adduct types and levels in humans. Free

Radical Biol. Med., 43:1109–1120, 2007.

[15] B.J. Finlayson-Pitts and J.N. Pitts. Tropospheric air pollution: Ozone, airborne toxics,

polycyclic aromatic hydrocarbons, and particles. Science, 276(5315):1045–1052, 1997.

[16] J. Futrell. Gas phase radiolysis of n-pentane. J. Phys. Chem., 64(11):1634–1636, 1960.

[17] R.A. Back and N. Miller. The radiolysis of n-pentane and other hydrocarbons in the gas

phase. Trans. Faraday Soc., 55:911–920, 1959.

[18] F.C. Fehsenfeld, A.L. Schmeltekopf, P.D. Goldan, H.I. Schiff, and E.E. Ferguson. Thermal

energy ion-neutral reaction rates. I. Some reactions of helium ions. J. Chem. Phys.,

44(11):4087–4094, 1966.

[19] E.H. Appelman and M.A.A. Clyne. Elementary reaction kinetics of ﬂuorine atoms, FO,

and NF free radicals. In J.W. Root, editor, Fluorine-Containing Free Radicals: Kinetics

and Dynamics of Reactions, ACS symposium series 3–25. American Chemical Society,

Washington, 1978.

[20] R. Atkinson. Kinetics and mechanisms of the gas-phase reactions of the hydroxyl radical

with organic compounds under atmospheric conditions. Chem. Rev., 85:69–201, 1985.

[21] G.P. Glass and R.B. Quy. Measurement of high temperature rate constants using a

discharge ﬂow shock tube. J. Phys. Chem., 83(1):30–33, 1979.

[22] C.J. Howard. Kinetic measurements using ﬂow tubes. J. Phys. Chem., 83(1):3–9, 1979.

[23] F. Kaufman. Kinetics of elementary radical reactions in the gas phase. J. Phys. Chem.,

88(21):4909–4917, 1984. BIBLIOGRAPHY 205

[24] R.W.F.Gross. Temperature dependence of chemiluminescent reactions. I. Nitrogen after-

glow. J Chem. Phys., 48(3):1302–1305, 1968.

[25] R.S. Tranter, R. Sivaramakrishnan, K. Brezinsky, and M.D. Allendorf. High pressure, high

temperature shock tube studies of ethane pyrolysis and oxidation. Phys. Chem. Chem.

Phys., 4(11):2001–2010, 2002.

[26] J.H. Kiefer, K.S. Gupte, L.B. Harding, and S.J. Klippenstein. Shock tube and theory

investigation of cyclohexane and 1-hexene decomposition. J. Phys. Chem. A, 113:13570–

13583, 2009.

+ [27] J.T. Herbon. Shock Tube Measurements of CH3 O2 Kinetics and the Heat of For- mation of the OH Radical. PhD thesis, Stanford University, 2004. available at

http://thermosciences.stanford.edu/pdf/TSD-153.pdf.

[28] D.W. Kohn, H. Clauberg, and P.Chen. Flash pyrolysis nozzle for generation of radicals in a

supersonic jet expansion. Rev. Sci. Instrum., 63(8):4003–4005, 1992.

[29] S. Nandi, S.J. Blanksby, X. Zhang, M.R. Nimlos, D.C. Dayton, and G.B. Ellison. Polarized ˜ 2 ′′ infrared absorption spectrum of matrix-isolated methylperoxyl radicals, CH3OO X A . J. Phys. Chem. A, 106:7547–7556, 2002.

[30] X. Zhang, A.V. Friderichsen, S. Nandi, G.B. Ellison, D.E. David, J.T. McKinnon, T.G. Linde-

man, D. Dayton, and M.R. Nimlos. Intense, hyperthermal source of organic radicals for

matrix-isolation spectroscopy. Rev. Sci. Instrum., 74(6):3077–3086, 2003.

[31] P.A.Glaude, N. Marinov, Y. Koshiishi, N. Matsunaga, and M. Hori. Kinetic modeling of the

mutual oxidation of NO and larger alkanes at low temperature. Energy & Fuels, 19:1839–

1849, 2005.

[32] G. Rotzoll. Molecular beam sampling mass-spectrometric study of high-temperature

benzene oxidation. Int. J. Chem. Kinet., 17(6):637–653, 1985.

[33] R.G.W. Norrish and G. Porter. Chemical reactions produced by very high light intensities.

Nature, 164:658, 1949.

[34] G. Porter. Flash photolysis and spectroscopy - a new method for the study of free radical

reactions. Proc. R. Soc. A, 200:284–300, 1950.

[35] R.G.W. Norrish and G. Porter. Spectroscopy of the H−O explosion initiated by the ﬂash

photolysis of NO2. Proc. R. Soc. A, 210:439–449, 1952. 206 BIBLIOGRAPHY

[36] H. Thrush. The genesis of ﬂash photolysis. Photochem. Photobiol. Sci., 2:453–454, 2003.

[37] L. Lindqvist. Utilisation d’un laser à Émission ultraviolette pulsé en photolyse-éclairs:

Étude de l’état triplet de l’acridine. C. R. Hebd. Seances Acad. Sci., Ser. C, 263:852–854,

1966.

[38] D.L. Osborn, P.Zou, H. Johnsen, C.C. Hayden, C.A. Taatjes, V.D. Knyazev, S.W. North, D.S.

Peterka, M. Ahmed, and S.R. Leone. The multiplexed chemical kinetic photoionization

mass spectrometer: A new approach to isomer-resolved chemical kinetics. Rev. Sci.

Instrum., 79, 2008.

[39] T. Yu and M.C. Lin. Kinetics of the C6H5 + O2 reaction at low temperatures. J. Am. Chem. Soc., 116:9571–9576, 1994.

[40] G.M.P.Just, E.N. Sharp, S.J. Zalyubovsky, and T.A. Miller. Cavity ringdown spectroscopy of

the Ã-X˜ electronic transition of the phenyl peroxy radical. Chem. Phys. Lett., 417:378–382,

2006.

[41] D. Weldon, S. Holland, and J.C. Scaiano. Temperature dependence of the reactions of

phenyl radicals with 1,1-diphenylethylene, carbon tetrachloride, and cyclohexene. J. Org.

Chem., 61:8544–8546, 1996.

[42] T. Yu and M.C. Lin. Kinetics of the C6H5 + NO association reaction. J. Phys. Chem., 98:2105–2109, 1994.

[43] M.D. Wheeler, S.M. Newman, A.J. Orr-Ewing, and M.N.R. Ashfold. Cavity ring-down

spectroscopy. J. Chem. Soc. Faraday Trans., 94(3):337–351, 1998.

[44] P. Rupper, E.N. Sharp, G. Tarczay, and T.A. Miller. Investigation of ethyl peroxy radical

conformers via cavity ringdown spectroscopy of the Ã-X˜ electronic transition. J. Phys.

Chem. A, 111(5):832–840, 2007.

[45] S.J. Zalyubovsky, B.G. Glover, T.A. Miller, C. Hayes, J. K. Merle, and C.M. Hadad. Observa- ˜ tion of the Ã-X electronic transition of the 1- C3H7O2 and 2- C3H7O2 radicals using cavity ringdown spectroscopy. J. Phys. Chem. A, 109(7):1308–1315, 2005.

[46] B.G. Glover and T.A. Miller. Near-IR cavity ringdown spectroscopy and kinetics of the

isomers and conformers of the butyl peroxy radical. J. Phys. Chem. A, 109(49):11191–

11197, 2005. BIBLIOGRAPHY 207

[47] E.N. Sharp, P.Rupper, and T.A. Miller. Observation of the Ã-X˜ electronic transition of the

isomers and conformers of pentyl peroxy radical using cavity ringdown spectroscopy. J.

Phys. Chem. A, 112:1445–1456, 2008.

[48] P.S.Thomas, R. Chhantyal-Pun, and T.A. Miller. Observation of the Ã-X˜ electronic transi-

tions of cyclopentyl and cyclohexyl peroxy radicals via cavity ringdown spectroscopy. J.

Phys. Chem. A, 114:218–231, 2010.

[49] E. Whittle, D.A. Dows, and G.C. Pimentel. Matrix isolation method for the experimental

study of unstable species. J. Chem. Phys., 1954(11):1943–1943, 1954.

[50] M.E. Jacox. The spectroscopy of molecular reaction intermediates trapped in the solid

rare gases. Chem. Soc. Rev., 31:108–115, 2002.

[51] E.B. Jochnowitz, X. Zhang, M.R. Nimlos, B.A. Flowers, J.F. Stanton, and G.B. Ellison. − − ˜ 2 ” Infrared spectrum of the propargyl peroxyl radical, HC−C CH2OO X A . J. Phys. Chem. A, 114:1498–1507, 2010.

[52] A. Mardyukov and W. Sander. Matrix isolation and spectroscopic characterization of the

phenylperoxy radical and its rearranged products. Chem. Eur. J., 15(6):1462–1467, 2009.

[53] B.K. Carpenter. Computational predication of new mechanisms for the reactions of vinyl

and phenyl radicals with molecular oxygen. J. Am. Chem. Soc., 115:9806–9807, 1993.

[54] B.K. Carpenter. Ring opening of dioxiranylmethyl radical: A caution on the use of G2-type

ab initio mo methods for mechanistic analysis. J. Phys. Chem. A, 105(19):4585–4588, 2001.

[55] I.V.Tokmakov, G. Kim, V.V.Kislov, A.M. Mebel, and M.C. Lin. The reaction of phenyl radical

with molecular oxygen: A G2M study of the potential energy surface. J. Phys. Chem. A,

109:6114–6127, 2005.

[56] J. K. Merle and C. M. Hadad. Computational study of the oxygen initiated decomposition

of 2-oxepinoxy radical: A key intermediate in the oxidation of benzene. J. Phys. Chem. A,

108(40):8419–8433, 2004.

[57] M.J. Fadden, C. Barckholtz, and C.M. Hadad. Computational study of the unimolecular

decomposition pathways of phenylperoxy radical. J. Phys. Chem. A, 104:3004–3011, 2000.

[58] M.J. Fadden and C. M. Hadad. Unimolecular decomposition of the 2-oxepinoxy radical:

A key seven-membered ring intermediate in the thermal oxidation of benzene. J. Phys.

Chem. A, 104:8121–8130, 2000. 208 BIBLIOGRAPHY

[59] W. Henderson and J.S. McIndoe. Mass Spectrometry of Inorganic and Organometallic

Compounds. Inorganic Chemistry. John WIley & Sons, Ltd, Hoboken, NJ, 2005.

[60] J.H. Gross. Mass Spectrometry: A Textbook. Springer-Verlag, Heidelberg, 2004.

[61] A. Nier. A mass spectrometer for isotope and gas analysis. Rev. Sci. Instrum., 18(6):398–

411, 1947.

[62] J.G. Dillard. Negative ion mass spectrometry. Chem. Rev., 73:589–643, 1973.

[63] I.R. Slagle. Kinetics of free radicals produced by infrared multiphoton-induced decompo-

sitions. 1. reactions of allyl radicals with nitrogen dioxide and bromine. J. Am. Chem. Soc.,

103:149–153, 1981.

[64] V.D. Knyazev and I.R. Slagle. Kinetics of the reaction bewteen propargyl radical and

acetylene. J. Phys. Chem. A, 106:5613–5617, 2002.

[65] Y. Li and F.Qi. Recent applications of synchrotron VUV photoionization mass spectrome-

try: Insight into combustion chemistry. Acc. Chem. Res., 43(1):68–78, 2010.

[66] B. Yang, Y. Li, L. Wei, C. Huang, J. Wang, Z. Tian, R. Yang, L. Sheng, Y. Zhang, and

F. Qi. An experimental study of the premixed benzene/oxygen/argon ﬂame with tunable

synchrotron photoionization. Proc. Combust. Inst., 31:555–563, 2007.

[67] J.A. Syage, K.A. Hanold, and M.A. Hanning-Lee. Photoionization mass spectrometer, 2003.

[68] K.K. Murray, R.K. Boyd, M.N. Eberlin, G.J. Langley, L. Li, and Y. Naito. Standard deﬁnitions

of terms relating to mass spectrometry, IUPAC recommendations. Technical report,

International Union of Pure and Applied Chemistry, Analytical Chemistry Division, 2006.

[69] Jr. Ligon, W.V. Molecular analysis by mass spectrometry. Science, 205(4402):151–159, 1979.

[70] C. Dass. Fundamentals of Contemporary Mass Spectrometry. John Wiley & Sons, Inc.,

Hoboken, NJ, 2007.

[71] F.W. McLafferty. Tandem mass spectrometry. Science, 214:280–287, 1981.

[72] J.M. Laux. Mickey laux explains how a quadruple mass spectrometer works - part 1, 2009.

[73] V.H. Wysocki and H.I. Kenttämaa. Collisional activation of distonic radical cations and

their conventional isomers in quadrupole tandem mass spectrometry. J. Am. Chem. Soc.,

1990(112):5110–5116, 1990. BIBLIOGRAPHY 209

[74] E.E. Ferguson, F.C. Fehsenfeld, and A.L. Schmeltekopf. Flowing afterglow measurements

of ion-neutral reactions. Adv. At. Mol. Phys., 5:1–56, 1969.

[75] R.R. Squires. Advances in ﬂowing afterglow and selected-ion ﬂow tube techniques. Int. J.

Mass Spectrom. Ion Proc., 118/119:503–518, 1992.

[76] N.G. Adams and Smith. D. The Selected Ion Flow Tube (SIFT); a technique for studying

ion-neutral reactions. Int. J. Mass Spectrom. Ion Phys., 21:349–359, 1976.

[77] R. Bain. SIFT-MS: Gas phase chemistry - MS with a difference. Spectroscopy, 22(3):22–25,

2007.

[78] J.C. Poutsma, R.A. Seburg, L.J. Chyall, L.S. Sunderlin, B.T. Hill, J. Hu, and R.R. Squires.

Combining electrospray ionization and the ﬂowing afterglow method. Rapid Commun.

Mass Spectrom., 11:489–493, 1997.

[79] P.W. Villalta, L.G. Huey, and C.J. Howard. A temperature-dependent study of the CH3O2 + NO reaction using chemical ionization mass spectrometry. J. Phys. Chem., 99:12829–

12834, 1995.

[80] A.G. Marshall, M.B. Comisarow, and G. Parisod. Relaxation and spectra line shape in

fourier transform ion resonance spectroscopy. J. Chem. Phys., 71(11):4434–4444, 1979.

[81] E.O. Lawrence and N.E. Edlefsen. On the production of high speed protons. Science,

72(1867):376–377, 1930.

[82] M.B. Comisarow and A.G. Marshall. The early development of Fourier Transform Ion

Cyclotron Resonance (FT-ICR) spectroscopy. J. Mass Spectrom., 31(6):581–585, 1996.

[83] M.B. Comisarow and A.G. Marshall. Fourier transform ion cyclotron resonance spec-

troscopy. J. Mass Spectrom., 31:586–587, 1996.

[84] I.J. Amster. Fourier transform mass spectrometry. J. Mass Spectrom., 31:1325–1337, 1996.

[85] A.G. Marshall, C.L. Hendrickson, and G.S. Jackson. Fourier transform ion cyclotron

resonance mass spectrometry: A primer. Mass Spec. Rev., 17:1–35, 1998.

[86] C. Peltz, L. Drahos, and K. V’ekey. SORI excitation: Collisional and radiative processes. J.

Am. Soc. Mass Spectrom., 18:2119–2126, 2007.

[87] A.T. Iavarone and E.R. Williams. Collisionally activated dissociation of supercharged

proteins formed by electrospray ionization. Anal. Chem., 75:4525–4533, 2003. 210 BIBLIOGRAPHY

[88] M.W. Senko, J.P.Spelr, and F.W. McLafferty. Collisional activation of large multiply charged

ions using fourier transform mass spectrometry. Anal. Chem., 66:2801–2808, 1994.

[89] S.A. McLuckey and D.E. Goeringer. Slow heating methods in tandem mass spectrometry.

J. Mass Spectrom., 32:461–474, 1997.

[90] J.C. Schwartz, M.W. Senko, and J.E.P.Syka. A two-dimensional quadrupole ion trap mass

spectrometer. J. Am. Soc. Mass Spectrom., 13:659–669, 2002.

[91] R.E. March. An introduction to quadrupole ion trap mass spectrometry. J. Mass Spectrom.,

32:351–369, 1997.

[92] R.E. March and J.F.J. Todd. Quadrupole Ion Trap Mass Spectrometry. Chemical Analysis:

A Series of Monographs on Analytical Chemistry and Its Applications. John Wiley & Sons,

Inc., Hoboken, New Jersey, 2005.

[93] D.J. Douglas, A.J. Frank, and D. Mao. Linear ion traps in mass spectrometry. Mass

Spectrom. Rev., 24(1):1–29, 2005.

[94] S.A. McLuckey, G.L. Glish, and G.J. Van Berkel. Charge determination of product

ions formed from collision-induced dissociation of multiply protonated molecules via

ion/molecule reactions. Anal. Chem., 63(18):1971–1978, 1991.

[95] Thermo Electron Corporation. Finnigan LTQ: Harware Manual. Thermo Electron Corpo-

ration, San Jose, California, revision b edition, 2003.

[96] E.R. Badman, G.E. Patterson, J.M. Wells, R.E. Santini, and R.G. Cooks. Differential non-

destructive image current detection in a fourier transform quadrupole ion trap. J. Mass

Spectrom., 34:889–894, 1999.

[97] M. Soni, V. Frankevich, M. Nappi, R.E. Santini, J.W. Amy, and R.G. Cooks. Broad-band

fourier transform quadrupole ion trap mass spectrometry. Anal. Chem., 68:3314–3320,

1996.

[98] W. Xu, J.B. Maas, F.J. Boudreau, W.J. Chappell, and Z. Ouyang. Nondestructive ion trap

mass analysis at high pressure. Anal. Chem., 83:685–689, 2011.

[99] R.A. Zubarev, N.L. Kelleher, and F.W. McLafferty. Electron capture dissociation of multiply

charged protein cations. A nonergodic process. J. Am. Chem. Soc., 120:3265–3266, 1998.

[100] D.F.Hunt, J.J. Coon, J.E.P.Syka, and J.A. Marto. Electron transfer dissociation for biopoly-

mer sequence mass spectrometric analysis, 2009. BIBLIOGRAPHY 211

[101] F.L. Ansari, R. Qureshi, and M.L. Qureshi. Electrocyclic reactions: from fundamentals to

research. Wiley-VCH Verlag GmbH, Weinheim, 1999.

[102] H.J.P. de Lijser and D.R. Arnold. Radical ions in photochemistry. 44. The photo-NOCAS

reaction with acetonitrile as the nucleophile. J. Org. Chem., 62:8432–8438, 1997.

[103] S. Kobayashi, T. Kawamoto, S. Uehara, T. Fukuyama, and I. Ryu. Black-light-induced

radical/ionic hydroxymethylation of alkyl iodides with atmospheric CO in the presence

of tetrabutylammonium borohydride. Org. Lett., 12(7):1548–1551, 2010.

+· [104] W.J. Bouma, R.H. Nobes, and L. Radom. The methylenoxonium radical cation (CH2OH2 ): A surprisingly stable isomer of the methanol radical cation. J. Am. Chem. Soc., 104:2929–

2930, 1982.

[105] W.J. Bouma, J.K. MacLeod, and L. Radom. Experimental evidence for the existance of a

+· +· stable isomer of CH3OH : The methylenoxonium radical cation, CH2OH2 J. Am. Chem. Soc., 104:2930–2931, 1982.

[106] B.F. Yates, W.J. Bouma, and L. Radom. Detection of the prototype phosphonium

(CH2PH3), sulfonium (CH2SH2), and chloronium (CH2ClH) ylides by neutralization- reionization mass spectrometry: A theoretical predication. J. Am. Chem. Soc., 106:5805–

5808, 1984.

[107] B.F. Yates, W.J. Bouma, and L. Radom. Distonic radical cations: Guidelines for the

assessment of their stability. Tetrahedron, 42(22):6225–6234, 1986.

[108] C. Wesdemiotis, P.O. Danis, R. Feng, J. Tso, and F.W. McLafferty. Distonic oxonium and

ammonium radical cations. a neutralization-reionization and collisional activation study.

J. Am. Chem. Soc., 107:8059–8066, 1985.

[109] D.M. Tomazela, A.A. Sabino, R. Sparrapan, F.C. Gozzo, and M.N. Eberlin. Distonoid ions.

J. Am. Chem. Soc., 17:1014–1022, 2006.

[110] T.E. Munsch, L.V. Slipchenko, A.I. Krylov, and P.G. Wenthold. Reactivity and structure of

the 5-dehydro-m-xylylene anion. J. Org. Chem., 69:5735–5741, 2004.

[111] A. Adeuya, L. Yang, F.S.Amegayibor, J.J. Nash, and H.I. Kenttämaa. Reactivity of aromatic

σ,σ-biradicals towards riboses. J. Am. Chem. Soc., 17(10):1325–1334, 2006. 212 BIBLIOGRAPHY

[112] J. Hu and R.R. Squires. Distonic biradical anions. synthesis and characterization of the

3,5-dehydrophenyl and 1,3,5-trimethylenebenzene negative ions. J. Am. Chem. Soc.,

118(24):5816–5817, 1996.

[113] K.K Thoen and H.I. Kenttämaa. Synthesis and characterization of aromatic biradicals in

the gas phase: A meta-benzyne with an inert positively charged substituent and its ortho-

and para-isomers. J. Am. Chem. Soc., 119:3832–3833, 1997.

[114] K.K Thoen, J. Pérez, J.J. Ferra, and H.I. Kenttämaa. Synthesis of charged phenyl radicals

and biradicals by laser photolysis in a Fourier-transform ion cyclotron resonance mass

spectrometer. J. Am. Soc. Mass Spectrom., 9:1135–1140, 1998.

[115] L. Yang, J.J. Nash, M.J. Yurkovich, J. Zhicheng, N.R. Vinueza, and H.I. Kenttämaa. Gas-

phase reactivity of aromatic σ,σ-biradicals towards dinucleoside phosphates. Org. Lett.,

10(10):1889–1892, 2008.

[116] B.J. Jankiewicz, J.N. Reece, N.R. Vinueza, J.J. Nash, and H.I. Kenttämaa. Reactivity of the

3,4,5-tridehydropyridinium cation - an aromatic σ,σ,σ-triradical. Angew. Chem. Int. Ed.,

47(51):9860–9865, 2008.

[117] P.G. Wenthold, J. Hu, B.T. Hill, and R.R. Squires. Gas-phase negative ion chemistry of

molecular ﬂuorine. synthesis of distonic radical anions and related species. Int. J. Mass

Spectrom., 179/180:173–183, 1998.

[118] S.E. Tichy, B.T. Hill, J.L. Campbell, and H.I. Kenttämaa. Synthesis and characteriza-

tion of a distonic nitrene ion: Gas-phase reactivity of singlet and triplet N-phenyl-3-

nitrenopyridinium ion. J. Am. Chem. Soc., 123:7923–7924, 2001.

[119] C.H. DePuy, J.M. Van Doren, S. Gronert, S.R. Kass, E.L. Motell, G.B. Ellison, and V.M.

Bierbaum. Gas-phase negative-ion chemistry of diazomethane. J. Org. Chem., 54:1846–

1850, 1989.

[120] J. Hu, B.T. Hill, and R.R. Squires. New approach to carbene chemistry via distonic carbene

ions. a case of spin-forbidden proton transfer and carbene-biradical tautomerism. J. Am.

Chem. Soc., 1997:11699–11700, 1997.

[121] S. Billets, H.H. Jaffé, and F. Kaplan. Rearrangements of molecular ions of dialkyl-N-

nitrosamines. J. Am. Chem. Soc., 92(23):6964–6965, 1970.

[122] H.H. Jaffé and S. Billets. Electrical effects of free-radical groups. J. Am. Chem. Soc.,

92(23):6965–6965, 1970. BIBLIOGRAPHY 213

[123] F.W. McLafferty. Mass spectrometric analysis: Molecular rearrangements. Anal. Chem.,

31(1):82–87, 1959.

[124] J. Diekman, J.K. MacLeod, C. Djerassi, and J.D. Baldeschwieler. Determination of the

structures of the ions produced in the single and double mclafferty rearrangements by

ion cyclotron resonance spectroscopy. J. Am. Chem. Soc., 91(8):2069–2084, 1969.

[125] J.S. Smith and F.W. McLafferty. Evidence for a stepwise mechanism for the γ-hydrogen

rearrangement. Org. Mass Spectrom., 5:483–485, 1971.

·+ [126] M.L. Gross and F.W. McLafferty. Identiﬁcation of C3H6 structural isomers by ion cyclotron resonance spectroscopy. J. Am. Chem. Soc., 93(5):1267–1268, 1971.

+. [127] F.W. Crow, M.L. Gross, and M.M. Bursey. Four isomers of [C3H8O] distinguished by collisional activation. Org. Mass. Spectrom., 16(7):309–311, 1981.

[128] D.R.G. Brimage, J.D.P. Cassell, J.H. Sharp, and M.C.R. Symons. Unstable intermediates.

Part LXV. Electron spin resonance studies of radicals derived from protonated alcohols by γ . + -irradiation: the radicals R2cOH2. J. Chem. Soc. A, (17):2619–2621, 1969.

+. [129] J.L. Holmes, F.P.Lossing, J.K. Terlouw, and P.C.Burgers. The radical cation [CH2OH2] and related stable gas-phase ion-dipole complexes. J. Am. Chem. Soc., 104:2931–2932, 1982.

[130] D.R. Reed, M. Hare, and S.R. Kass. Formation of gas-phase dianions and distonic ions as

a general method for the synthesis of protected reactive intermediates. energetics of 2,3-

and 2,6-dehydronaphthalene. J. Am. Chem. Soc., 122(43):10689–10696, 2000. Notes: p.12.

[131] A.G. Harrison and K.R. Jennings. Reactions of the negative oxygen ion (O–·) with carbonyl

compounds. J. Chem. Soc. Faraday Trans., 72:1601–1608, 1976.

[132] J.H.J. Dawson, A.J. Noest, and N.M.M. Nibbering. 1,1 and 1,3 eliminations of water from

the reaction complex of O–· with 1,1,1-trideuteroacetone. Int. J. Mass Spectrom. Ion Phys.,

30:189–191, 1979.

[133] A.P.Bruins, A.J. Ferrer-Correia, A.G. Harrison, K.R. Jennings, and R.K. Mitchum. Negative

ion chemical ionization mass spectra of some aromatic compounds using O–· as the

reagent ion. Adv. Mass Spectrom., 7:355–358, 1978.

[134] K. Pius and J. Chandrasekhar. Towards stable distonic radical anions: a theoretical study.

J. Chem. Soc. Perkin Trans. II, (7):1291–1295, 1988. 214 BIBLIOGRAPHY

[135] K.W.M. Siu, G.J. Gardner, and S.S. Berman. Multiply charged ions in ionspray tandem

mass spectrometry. Org. Mass. Spectrom., 24:931–942, 1989.

[136] Y. Guo and J.J. Grabowski. Reactions of the benzyne radical anion in the gas phase, the

acidity of the phenyl radical, and the heat of formation of o-benzyne. J. Am. Chem. Soc.,

113(16):5923–5931, 1991.

[137] H.I. Kenttämaa. Long-lived distonic radical cations. Org. Mass. Spectrom., 29:1–10, 1994.

[138] S. Hammerum. Distonic radical cations in gaseous and condensed phase. Mass Spec. Rev.,

7:123–202, 1988.

[139] C. Wesdemiotis, R. Feng, E.R. Williams, and F.W. McLafferty. Neutralization-reionization

mass spectrometry (NRMS). structural information from vertical neutralization and

reionization efﬁciences. Org. Mass Spectrom., 21:689–695, 1986.

[140] R. Flammang, M. Plisnier, G. Leroy, M. Sana, M.T. Nguyen, and L.G. Vanquickenborne. The + . distonic HC (OH)OCH2 radical cation: a stable isomer of ionized methyl formate. Chem. Phys. Lett., 186(4-5):393–400, 1991.

·+ [141] B.J. Smith, M.T. Nguyen, and L. Radom. Ionized methyl formate (CH3OCHO ) and its . + distonic isomer ( CH2OC HOH). J. Am. Chem. Soc., 114:1151–1156, 1992.

[142] K.M. Stirk, J.C. Orlowski, D.T. Leeck, and H.I. Kenttämaa. The identiﬁcation of distonic

radical cations on the basis of a reaction with dimethyl disulﬁde. J. Am. Chem. Soc.,

114:8604–8606, 1992.

[143] K.M. Stirk, L.K.M. Kiminkinen, and H.I. Kenttämaa. Ion-molecule reactions of distonic

radical cations. Chem. Rev., 92:1649–1665, 1992.

[144] H.I. Kenttämaa. Ion-molecule reactions of distonic radical cations., 2005.

[145] B.J. Beasley, R.L. Smith, and H.I. Kenttämaa. A new reagent for structure-speciﬁc ion-

molecule reactions. dimethyl diselenide. J. Mass Spectrom., 30(2):384–385, 1996.

[146] K.K Thoen, B.J. Beasley, R.L. Smith, and H.I. Kenttämaa. Distinguishing conventional and

distonic radical cations by using dimethyl diselenide. J. Am. Soc. Mass Spectrom., 7:1245–

1250, 1996.

[147] E.D. Nelson, R. Li, and H.I. Kenttämaa. Reactions of tert-butyl isocyanide with distonic

radical cations. Int. J. Mass Spectrom., 185/186/187:91–96, 1999. BIBLIOGRAPHY 215

[148] C.J. Petzold, E.D. Nelson, H.A. Lardin, and H.I. Kenttämaa. Charge-site effects on the

radical reactivity of distonic ions. J. Phys. Chem. A, 106(42):9767–9775, 2002.

[149] A.E.P.M. Sorrilha, F.C. Gozzo, R.S. Pimpim, and M.N. Eberlin. Multistage pentaquadrupole

mass spectrometry for generation and characterization of gas-phase ionic species. the

+· case of the PyC2H5 isomers. J. Am. Soc. Mass. Spectrom., 7:1126–1137, 1996.

[150] S.J. Yu, C.L. Holliman, D.L. Rempel, and M.L. Gross. The β-distonic ion from the reaction

of pyridine radical cation and ethene: A demonstration of high-pressure trapping in

fourier transform mass spectrometry. J. Am. Chem. Soc., 115:9676–9682, 1993.

[151] B.T. Hill, J.C. Poutsma, L.J. Chyall, J. Hu, and R.R. Squires. Distonic ions of the "ate" class.

J. Am. Soc. Mass Spectrom., 10:896–906, 1999.

[152] J.J. Grabowski and L. Zhang. Dimethyl disulﬁde: Anion-molecule reactions in the gas

phase at 300 K. J. Am. Chem. Soc., 111(4):1193–1203, 1989.

[153] P.G.Wenthold and R.R. Squires. Gas-phase properties and reactivity of the acetate radical

anion. Determination of the C−H bond strengths in acetic acid and acetate ion. J. Am.

Chem. Soc., 116:11890–11897, 1994.

[154] D.G. Harman and S.J. Blanksby. Trapping of a tert-adamantyl peroxyl radical in the gas

phase. Chem. Commun., 8:859–861, 2006.

[155] D.G. Harman and S.J. Blanksby. Investigation of the gas phase reactivity of the 1-

adamantyl radical using a distonic radical anion approach. Org. Biomol. Chem., 5:3495–

3503, 2007.

[156] P.G.Wenthold, J. Hu, and R.R. Squires. Regioselective synthesis of biradical negative ions

in the gas phase. Generation of trimethylenemethane, m-benzyne, and p-benzyne anions.

J. Am. Chem. Soc., 116:6961–6962, 1994.

[157] R.L. Smith and H.I. Kenttämaa. A general method for the synthesis of charged phenyl

radicals in the gas phase. J. Am. Chem. Soc., 117:1393–1396, 1995.

[158] K.K. Thoen, R.L. Smith, E.D. Nousiainen, E.D. Nelson, and H.I. Kenttämaa. Charged

phenyl radicals. J. Am. Chem. Soc., 118:8669–8676, 1996. Printed copy in folder.

[159] R. Li, R.L. Smith, and H.I. Kenttämaa. Fluorine substitution enhances the reactivity of

substituted phenyl radicals toward organic hydrogen atom donors. J. Am. Chem. Soc.,

118:5056–5061, 1996. 216 BIBLIOGRAPHY

[160] K.K Thoen and H.I. Kenttämaa. Reactivity of a substituted m-benzyne biradical. J. Am.

Chem. Soc., 121:800–805, 1999. Printed copy in folder.

[161] E.D. Nelson, A. Artau, J.P. Price, S.E. Tichy, and H.I. Kenttämaa. m-benzyne reacts as an

electrophile. J. Am. Chem. Soc., 122:8781–8782, 2000.

[162] J.H. Bowie. The fragmentation of even-electron organic negative ions. Mass Spectrom.

Rev., 9(3):349, 1990.

[163] W.P.M. Mass and N. M. Nibbering. Collision-induced consecutive dissociation reactions

of the monocarboxylate anions generated from the dimethyl and diethyl esters of glutaric

acid and its 3,3-dimethyl analogue. Org. Mass Spectrom., 25(3):154–160, 1990.

[164] J.I. Brauman and K. Smyth. Photodetachment energies of negative ions by ion cy-

clotron resonance spectroscopy. electron afﬁnities of neutral radicals. J. Am. Chem. Soc.,

91(27):7778–7780, 1969.

[165] K. Smyth and J.I. Brauman. Photodetachment of electrons from phosphide ion; the

electron afﬁnity of PH2. J. Chem. Phys., 56(3):1132–1142, 1972.

[166] K. Smyth and J.I. Brauman. Photodetachment of electrons from amide and arsenide ions: · · The electron afﬁnities of NH2 and AsH2 J. Chem. Phys., 56(9):4620–4625, 1972.

[167] K. Smyth and J.I. Brauman. Photodetachment of an electron from selenide ion; the

electron afﬁnity and spin-orbit coupling constant for SeH· J. Chem. Phys., 56(12):5993–

5997, 1972.

[168] J.H. Richardson, L.M. Stephenson, and J.I. Brauman. Photodetachment of electrons from

large molecular systems: Benzyl anion. an upper limit ot he electron afﬁnity of C6H5CH2. J. Chem. Phys., 63(1):74–76, 1975.

[169] J.M. Jasinski, R.N. Rosenfeld, F.K. Meyer, and J.I. Brauman. Comparison of pulsed and

continuous-wave laser photodissociation methods. J. Am. Chem. Soc., 104:652–658, 1982.

[170] W.D. Bowers, S. Delbert, R.L. Hunter, and Jr. McIver, R.T. Fragmentation of oligopeptide

ions using ultraviolet laser radiation and fourier transform mass spectrometry. J. Am.

Chem. Soc., 106:7289–7291, 1984.

[171] W.D. Bowers, S. Delbert, and Jr. McIver, R.T. Consecutive laser-induced photodissociation

as a probe of ion structure. Anal. Chem., 58:969–972, 1986. BIBLIOGRAPHY 217

[172] T-Y. Kim, M.S. Thompson, and J.P.Reilly. Peptide photodissociation at 157nm in a linear

ion trap mass spectrometer. Rapid Commun. Mass Spectrom., 19:1657–1665, 2005.

[173] L.E. Ramírez-Arizmendi, L. Guler, J.J. Ferra, K.K Thoen, and H.I. Kenttämaa. Hydrogen

atom abstraction and addition reaction of charged phenyl radicals with aromatic sub-

strates in the gas phase. Int. J. Mass Spectrom., 210(1-3):511–520, 2001.

[174] J. Heidbrink, L.E. Ramírez-Arizmendi, K.K Thoen, L. Guler, and H.I. Kenttämaa. Polar

effects control hydrogen-abstraction reactions of charged, substituted phenyl radicals. J.

Phys. Chem. A, 105:7875–7884, 2001.

[175] T. Ly and R.R. Julian. Residue-speciﬁc radical-directed dissociation of whole proteins in

the gas phase. J. Am. Chem. Soc., 130:351–358, 2008.

[176] T. Ly and R.R. Julian. Ultraviolet photodissociation: Developments towards applications

for mass-spectrometry-based proteomics. Angew. Chem. Int. Ed., 48:7130–7137, 2009.

[177] Z. Liu and R.R. Julian. Deciphering the peptide iodination code: Inﬂuence on subsequent

gas-phase radical generation with photodissociation ESI-MS. J. Am. Soc. Mass Spectrom.,

20:965–971, 2009.

[178] J. Lee and J.J. Grabowski. Reactions of the atomic oxygen radical anion and the synthesis

of organic reactive intermediates. Chem. Rev., 92(7):1611, 1992.

[179] J. Kauw, M. Born, S. Ingemann, and N.M.M. Nibbering. Gas-phase reactions of isomeric

carbene and distonic radical anions serived from methylthioacetonitrile with dimethyl

disulﬁde. Rapid Commun. Mass Spectrom., 10:1400–1404, 1996.

[180] M. Born, S. Ingemann, and N.M.M. Nibbering. Formation and characterization of the − − – − sulfur-containing distonic radical anion, CH2 S C H CN, in the gas phase. J. Am. Soc. Mass Spectrom., 6:71–75, 1995.

[181] M. Born, J. Chamot-Rooke, S. Ingemann, and N.M.M. Nibbering. Formation of the

. – –· distonic acetate radical anion CH2CO2 in reactions of O with lactones in the gas phase. J. Mass Spectrom., 33:677–679, 1998.

[182] C.H. DePuy, V.M.Bierbaum, L.A. Flippen, J.J. Grabowski, G.K. King, and R.J. Schmitt. Gen-

eration of speciﬁc isomeric carbanions in the gas phase. J. Am. Chem. Soc., 101(21):6443–

6443, 1979. 218 BIBLIOGRAPHY

[183] G. Klass, C. Trenerry, J.C. Sheldon, and J.H. Bowie. An ab initio and ion cyclotron

resonance study of the reactions between alkoxide negative ions and silane systems.

stable adduct formation, and nucleophilic displacement reactions. Aust. J. Chem., 34:519–

29, 1981.

[184] C.H. DePuy, V.M. Bierbaum, R. Damrauer, and J.A. Soderquist. Gas-phase reactions of the

acetyl anion. J. Am. Chem. Soc., 107:3386–3387, 1985.

[185] S.R. Kass and K.M. Broadus. Reactive intermediates via Fourier transform mass spectrom-

etry. J. Phys. Org. Chem., 15(8):461–468, 2002.

[186] K.M. Broadus and S.R. Kass. The electron as a protecting group. 2. generation of benzo-

cyclobutadiene radical anion in the gas phase and an experimental determination of the

heat of formation of benzocyclobutadiene. J. Am. Chem. Soc., 122:10697–10703, 2000.

[187] K.M. Broadus and S.R. Kass. The electron as a protecting group. 3. generation of ace-

naphthyne radical anion and the determination fo the heat of formation fo a strained

cycloalkyne. J. Am. Chem. Soc., 123:4189–4196, 2001.

[188] D.R. Reed, M.C. Hare, A. Fattahi, G. Chung, M.S. Gordon, and S.R. Kass. α,2-, α,3-, and

α,4-dehydrophenol radical anions: Formation, reactivity, and energetics leading to the

heats of formation of α,2-, α,3-, and α,4-oxocyclohexadienylidene. J. Am. Chem. Soc.,

125(15):4643–4651, 2003.

[189] G.J. Van Berkel, G.L. Glish, and S.A. McLuckey. Electrospray ionization combined with ion

trap mass spectrometry. Anal. Chem., 62(13):1284–1295, 1990.

[190] Q. Zha, R.N. Hayes, T. Nishimura, G.G. Meisels, and M.L. Gross. A study of the metastable

dissociations of formate esters: A mclafferty rearrangement to a distonic radical cation. J.

Phys. Chem., 94(1286-1290):1286, 1990.

[191] M.D. Sevilla, D. Becker, C.L. Sevilla, and S. Swarts. Reactions of the methyl and ethyl

formate cation radicals: Deuterium isotope effects. J. Phys. Chem., 89:633–636, 1985.

[192] J. Linhong, L.P. Guler, G.O. Pates, and H.I. Kenttämaa. The selectivity of charged phenyl

radicals in hydrogen atom abstraction reactions with isopropanol. J. Phys. Chem. A,

112:9708–9715, 2008.

[193] S. Li, M. Fu, S.C. Habicht, G.O. Pates, J.J. Nash, and H.I. Kenttämaa. Phenyl radical-

induced damage to dipeptides. J. Org. Chem., 74:7724–7732, 2009. BIBLIOGRAPHY 219

[194] R. Braslau and M.O. Anderson. Abstraction of deuterium from dideuteroglycine by aryl

radical: A model for 1,4-benzene diradical reactions with proteins. Tetrahedron Lett.,

39(24):4227–4230, 1998.

[195] M.J. Perkins. Radical Chemistry. Ellis Horwood, New York, 1994.

[196] K. Kavita and K.D. Puspendu. Photodissociation of C6H5I, C6F5I, and related iodides in the ultraviolet. J. Chem. Phys., 117(5):2038–2044, 2002.

[197] A.G. Sage, T.A.A. Oliver, D. Murdock, M.B. Crow, G.A.D. Ritchie, J.N. Harvey, and M.N.R.

Ashfold. nσ* and πσ* excited states in aryl halide photochemistry: a comprehensive study

of the UV photodissociation dynamics of iodobenzene. Phys. Chem. Chem. Phys., Advance

Article, 2011.

[198] R.A. Durie, T. Iredale, and J.M.S. Jarvie. Absorption spectra of substances containing the

carbon-iodine bond. Part I. J. Chem. Soc. 1181–1184, 1950.

[199] T.M. Dunn and T. Iredale. Absorption spectra of substances containing the carbon-iodine

bond. Part II. J. Chem. Soc. 1592–1597, 1952.

[200] J. Oomens, A.R. Moehlig, and T.H. Morton. Infrared multiple photon dissociation (irmpd)

spectroscopy of the proton-bound dimer of 1-methylcytosine in the gas phase. J. Phys.

Chem. Lett., 1(19):2891–2897, 2010.

[201] J.L. Stephenson, Jr., M.M. Booth, J.A. Shalosky, J.R. Eyler, and R.A. Yost. Infrared multiple

photon dissociation in the quadrupole ion trap via a multipass optical arrangement. J.

Am. Soc. Mass Spectrom., 5:886–893, 1994.

[202] J.J. Wilson and J. Brodbelt. Infrared multiphoton dissociation for enhanced de novo

sequence interpretation of N-terminal sulfonated peptides in a quadrupole ion trap. Anal.

Chem., 78:6855–6862, 2006.

[203] A. Colorado, J.X. Shen, V.H. Vartanian, and J. Brodbelt. Use of Infrared Multiphoton Pho-

todissociation with SWIFT for Electrospray Ionization and Laser Desorption Applications

in a Quadrupole Ion Trap Mass Spectrometer. Anal. Chem., 68, 1996.

[204] Y. Hashimoto, H. Hasegawa, K. Yoshinari, and I. Waki. Collision-activated infrared

multiphoton dissociation in a quadrupole ion trap mass spectrometer. Anal. Chem.,

75:420–425, 2003. 220 BIBLIOGRAPHY

[205] C.L. Kalcic, T.C. Gunaratne, A.D. Jones, M. Dantus, and G.E. Reid. Femtosecond laser-

induced ionization/dissociation of protonated peptides. J. Am. Chem. Soc., 131:940–942,

2009.

[206] C.L. Kalcic, T.C. Gunaratne, G.E. Reid, A.D. Jones, and M. Dantus. Femtosecond laser

scalpel technology for proteomic mass spectrometry. Proc. SPIE Int. Soc. Opt. Eng., 7203,

2009. Art. No. 72030C.

[207] S.S. Kumaran, M.C. Su, and J.V. Michael. Thermal decomposition of iodobenzene using

I-atom absorption. Chem. Phys. Lett., 269:99–106, 1997.

[208] D.F.McMillen and D.M. Golden. Hydrocarbon bond dissociation energies. Ann. Rev. Phys.

Chem., 33:493–532, 1982.

[209] T.A. Blake, Z. Ouyang, J.M. Wiseman, Z. Toká’ts, A.J. Guymon, S. Kothari, and R.G. Cooks.

Preparative linear ion trap mass spectrometer for separation and collection of puriﬁed

proteins and peptides in arrays using ion soft landing. Anal. Chem., 76:6293–6305, 2004.

[210] T. Ly, X. Zhang, Q. Sun, B.N. Moore, Y. Tao, and R.R. Julian. Rapid, quantitative, and site

speciﬁc synthesis of biomolecular radicals from a simple photocaged precursor. Chem.

Comm., 47:2835–2837, 2011.

[211] S.M. Klein, C. Zhang, and Y.L. Jiang. Simple synthesis of fresh alkyl iodides using alcohols

and hydriodic acid. Tetrahedron Lett., 49:2638–2641, 2008.

[212] J.A. Singletary, H. Lam, and G.B. Dudley. A succinct method for preparing the Stork-Jung

vinylsilane Robinson annulation reagent. J. Org. Chem., 70:739–741, 2005.

[213] H. Finkelstein. Preparation of organic iodides from the corresponding bromides and

chlorides. Ber., 43:1528–1532, 1910.

[214] D.H. Barton, J. MacKinnon, R.N. Perchet, and C. Tse. Efﬁcient synthesis of bromides from

carboxylic acids containing a sensitive functional group: Dec-9-enyl bromide from 10-

undecenoic acid. Org. Syn., Coll. Vol. 10:237, 2004.

[215] N.P.Schepp, C.J.M. Green, and F.L.Cozens. Non-resonant two-photon photochemistry of

a barton ester, N-phenylacetyloxy-2-pyridinethione. Photochem. Photobiol. Sci., 9:110–

113, 2010. BIBLIOGRAPHY 221

[216] C. Dietlin, X. Allonas, F. Morlet-Savary, J.P. Fouassier, M. Visconti, G. Norcini, and S. Ro-

magano. Investigation of barton esters as radical photoinitiators. J. Appl. Polym. Sci.,

109(2):825–833, 2008.

[217] D. Burdi, B.M. Aveline, P.D. Wood, J. Stubbe, and R.W. Redmond. Generation of tryptophan

radical in high quantum yield from a novel amino acid analog using near-UV/visible light.

J. Am. Chem. Soc., 119(28):6457–6460, 1997.

[218] C.J. Easton, A.J. Ivory, and C.A. Smith. Nitrate esters in the generation of amino acid

radicals. J. Chem. Soc., Perkin Trans. 2 503–507, 1997.

[219] S. Wee, A. Mortimer, D. Moran, A. Wright, C.K. Barlow, R.A.J. O’Hair, L. Radom, and

C.J. Easton. Gas-phase regiocontrolled generation of charged amino acid and peptide

radicals. Chem. Commun. 4233–4235, 2006.

[220] Thermo Fisher Scientiﬁc. Finnigan LTQ FT: Harware Manual. Thermo Electron Corpora-

tion, San Jose, California, revision b edition, 2004.

[221] Thermo Fisher Scientiﬁc. Finnigan LTQ Orbitrap XL: Hardware Manual. Thermo Electron

Corporation, San Jose, California, revision b edition, 2008.

[222] Insulator Seal. Fused silica introduction, 2010.

[223] S.P. Verevkin, E.L. Krasnykh, and J.S. Wright. Thermodynamic properties of benzyl

halids: enthalpies of formation, strain enthalpies, and carbon-halogen bond dissociation

enthalpies. Phys. Chem. Chem. Phys., 5:2605–2611, 2003.

[224] B.N. Moore, S.J. Blanksby, and R.R. Julian. Ion-molecule reactions reveal facile radical

migration in peptides. Chem. Comm. 5015–5017, 2009.

[225] X.B. Wang, H.K. Woo, B. Kiran, and L.S. Wang. Observation of weak C−H ··· O hydrogen

bonding to unactivated alkanes. Angew. Chem. Int. Ed., 44:4968–4972, 2005.

[226] X.B. Wang, C.F. Ding, and L-S. Wang. Electron tunneling through the repulse coulomb

barrier in photodetachment of multply charged anions. Chem. Phys. Lett., 307:391–396,

1999.

[227] H.K. Woo, X.B. Wang, B. Kiran, and L.S. Wang. Temperature-dependent photoelectron

spectroscopy of methyl benzoate anions: Observation of steric effect in o-methyl ben-

zoate. J. Phys. Chem. A, 109:11395–11400, 2005. 222 BIBLIOGRAPHY

[228] X-P. Xing, X-B. Wang, and L-S. Wang. Photoelectron imaging of double charged anions,

- – O2C(CH2)nCO2 (n = 2-8): Observation of near 0 ev electrons due to secondary dissociative autodetachment. J. Phys. Chem. A, 114:4524–4530, 2010.

[229] G.L. Glish. Multiple stage mass spectrometry: The next generation tandem mass spec-

trometry experiment. Analyst, 119:533–537, 1994.

[230] V.L. Tal’roze and A.K. Ljubimova. Secondary processes in the ion source of a mass

spectrometer (presented by academician N. N. Semenov 27 VIII 1952) - reprinted from

Report of the Soviet Academy of Sciences, Volume LXXXVI, -N5 (1952). J. Mass Spectrom.,

33:502–504, 1998.

[231] S. Gronert, L.M. Pratt, and S. Mogali. Substituent effects in gas-phase substitutions and

eliminations: β-halo substituents. solvation reverses sn2 substituent effects. J. Am. Chem. Soc., 123:3081–3091, 2001.

[232] J. Brodbelt, S. Maleknia, J. Liou, T. Donovan, A. Mcintosh, and E. Eichmann. Analytical

applications of selective ion/molecule reaction chemistry. Proc. 39th Am. Soc. Mass Spect.

Conf. 47–48, 1991.

[233] T. Waters, R.A.J. O’Hair, and A.G. Wedd. Catalytic gas phase oxidation of methanol to

formaldehyde. J. Am. Chem. Soc., 125:3384–3396, 2003.

[234] A.E. Flores and S. Gronert. The gas-phase reactions of dianons with alkyl bromides: Direct

identiﬁcation of sn2 and E2 products. J. Am. Chem. Soc., 121:2627–2628, 1999.

[235] S. Gronert. Estimation of effective ion temperatures in a quadrupole ion trap. J. Am. Soc.

Mass Spectrom., 9:845–848, 1998.

[236] S.C. Habicht, N.R. Vinueza, E.F. Archibold, D. Penggao, and H.I. Kenttämaa. Indentiﬁca-

tion of the carboxylic acid functionality by using electrospray ionization and ion-molecule

reactions in a modiﬁed linear quadrupole ion trap mass spectrometer. Anal. Chem.,

80:3416–3421, 2008.

[237] S. Gronert. Deuterium isotope effects in gas-phase reactions of alkyl halides: Distinguish-

ing E2 and SN 2 pathways. J. Am. Chem. Soc., 113:4009–4010, 1991.

[238] T. Su and W.J. Chesnavich. Parametrization fo the ion-polar molecule collision rate

constant by trajectory calculations. J. Chem. Phys., 76:5183–5185, 1982. BIBLIOGRAPHY 223

[239] S. Enami, T. Yamanaka, and M. Kawasaki. Near-infrared cavity ring-down spectroscopic

study of the reaction of methylperoxy radical with nitrogen monoxide. Chem. Lett.,

38(1):80–81, 2009.

[240] J.R. Barker, L.L. Lohr, R.M. Shroll, and S. Reading. Modeling the organic nitrate yields

in the reaction of alkyl peroxy radicals with nitric oxide. 2. reaction simulations. J. Phys.

Chem. A, 107:7434–7444, 2003.

[241] A. Lesar, M. HodoŽcek, E. Drougas, and A.M. Kosmas. Quantum mechanical investigation

of the atmospheric reaction CH3O2 + NO. J. Phys. Chem. A, 110:7898–7903, 2006.

[242] J. Eberhard and C.J. Howard. Rate coefﬁcients for the reactions of some C3 to C5 hydrocarbon peroxy radicals with NO. J. Phys. Chem. A, 101:3360–3366, 1997.

2 ′ [243] X. Gu, F.Zhang, and R.I. Kaiser. Crossed beam reaction of the phenyl radical, (C6H5,X A ) 3Σ− 2 ′ with molecular oxygen (O2,X g ): Observation of the phenoxy radical, (C6H5O, X A ). Chem. Phys. Lett., 448:7–10, 2007.

[244] M.U. Alzueta, P. Glarborg, and K. Dam-Johansen. Experimental and kinetic modeling

study of the oxidation of benzene. J. Int. Chem. Kinet., 32(8):498–522, 2000.

[245] R. Atkinson and J. Arey. Atmospheric degradation of volatile organic compounds. Chem.

Rev., 103:4605–4638, 2003.

[246] R. Atkinson and J. Arey. Gas-phase tropospheric chemistry of biogenic volatile organic

compounds: a review. Atmos. Environ., 37(Supp. 2):S197–S219, 2003.

[247] K. Tonokura, T. Norikane, M. Koshi, Y. Nakano, S. Nakamichi, M. Goto, S. Hashimoto,

M. Kawasaki, M.P.S. Anderson, M.D. Hurley, and T.J. Wallington. Cavity ring-down study

of the visible absorption spectrum of the phenyl radical and kinetics of its reactions with

Cl, Br, Cl2, and O2. J. Phys. Chem. A, 106:5908–5917, 2002.

[248] Basic research needs for clean and efﬁcient combustion of 21st century transportation fu-

els [ http://www.sc.doe.gov/bes/reports/ﬁles/ctf_rpt.pdf ]. Technical report, Department

of Energy, 2006.

[249] Fuel standard (petrol) determination 2001 made under section 21 of the fuel quality

standards act 2000, 2001.

[250] M.M Maricq. An examination of soot composition in premixed hydrocarbon ﬂames via

laser ablation particla mass spectrometry. J. Aerosol Sci., 40:844–857, 2009. 224 BIBLIOGRAPHY

[251] H. Richter and J.B. Howard. Formation of polycyclic aromatic hydrocarbons and their

growth to soot - a review of chemical reaction pathways. Prog. Energy Combust. Sci.,

26:565–608, 2000.

[252] N.D. Marsh, E.B. Ledesma, A.K. Sandrowitz, and M.J. Wornat. Yields of polycyclic aromatic

hydrocarms form the pyrolysis of catechol [ortho-dihydroxybenzene]: Temperature and

residence time effects. Energy Fuels, 18:209–217, 2004.

[253] S.P. Karakitsios, V.K. Delis, P.A. SKassomenos, and G.A. Pilidis. Contribution to ambient

benzene concentrations in the vicinity of petrol stations: Estimation of the associated

health risk. Atmos. Environ., 41(9):1889–1902, 2007.

[254] A. Srivastava, A.E. Joseph, A. More, and S. Patil. Emissions of VOCs at urban petrol retail

distribution centres in India (Delhi and Mumbai). Environ. Monit. Assess., 109:227–242,

2005.

[255] N. Gonzalez-Flesca, S. Vardoulakis, and A. Cicolella. BTX concentrations near a stage II

implemented petrol station. Environ. Sci. Pollut. Res., 9(3):169–174, 2002.

[256] R.G. Derwent, T.J. Davies, M. Dalaney, G.J. Dollard, R.A. Field, P. Dumitrean, P.D. Nason,

B.M.R. Jones, and S.A. Pepler. Analysis and interpretation of the continuous hourly

monitoring data for 26 C2-C8 hydrocarbons at 12 univted kingdom sites during 1996. Atmos. Environ., 34(2):297–312, 2000.

[257] H.A. Khan. Benzene’s toxicity: a consolidated short review of human and animal studies.

Hum. Exp. Toxicol., 26:677–685, 2007.

[258] J.L. Durant, W.F. Busby Jr., A.L. Laﬂeur, B.W. Penman, and C.L. Crespi. Human cell mu-

tagenicity of oxygenated, nitrated and unsubstituted polycyclic aromatic hydrocarbons

associated with urban aerosols. Mutat. Res., 371:123–157, 1996.

[259] T.H. Lay, J.W. Bozzelli, and J.H. Seinfeld. Atmospheric photochemical oxidation of ben-

zene: Benzene + OH and the benzene–OH adduct (hydroxyl-2,4-cyclohexadienyl) + O2. J. Phys. Chem., 100(16):6543–6554, 1996.

[260] S.M. Kroner, M.P. DeMatteo, C.M. Hadad, and B.K. Carpenter. The gas-phase acidity of

2(3H)-oxepinone: A step toward an experimental heat of formation for the 2-oxepinoxy

radical. J. Am. Chem. Soc., 127(20):7466–7473, 2005. BIBLIOGRAPHY 225

[261] N. Sebbar, H. Bockhorn, and J.W. Bozzelli. Thermodynamic properties of the species

resulting from the phenyl radical with O2 reaction system. Int. J. Chem. Kin., 40(9):583– 604, 2008.

+ [262] D.R. Albert and H.F. Davis. Collision complex lifetimes in the reaction C6H5 −−→ + O2 C6H5O O. J. Phys. Chem. Lett., 1(7):1107–1111, 2010.

[263] A.M. Mebel and M.C. Lin. Ab initio molecular orbital calculations of C6H5O2 isomers. J. Am. Chem. Soc., 116:9577–9584, 1994.

[264] Y. Chai and L.D. Pfefferle. An experimental study of benzene oxidation at fuel-lean and

stoichiometric equivalence ratio conditions. Fuel, 77(4):313–320, 1998.

[265] A.V. Tolmachev, A.N. Vilkov, B. Bogdanov, L. Pasa-Toli´c,C.D.ˇ Masselon, and R.D. Smith.

Collisional activation of ions in RF ion traps and ion guides: The effective ion temperature

treatment. J. Am. Soc. Mass Spectrom., 15:1616–1628, 2004.

[266] A.D. Becke. A new mixing of hartree-fock and local density-functional theories. J. Chem.

Phys., 98:1372, 1993.

[267] R.G. Parr and W. Yang. Density-functional theory of the electronic structure of molecules.

Ann. Rev. Phys. Chem., 46:701–728, 1995.

[268] M.J. Frisch, G.W. Trucks, H.B. Schlegel, G.E. Scuseria, M.A. Robb, J.R. Cheeseman, J.A.

Montgomery, Jr., T. Vreven, K.N. Kudin, J.C. Burant, J.M. Millam, S.S. Iyengar, J. Tomasi,

V. Barone, B. Mennucci, M. Cossi, G. Scalmani, N. Rega, G.A. Petersson, H. Nakatsuji,

M. Hada, M. Ehara, K. Toyota, R. Fukuda, J. Hasegawa, M. Ishida, T. Nakajima, Y. Honda,

O. Kitao, H. Nakai, M. Klene, X. Li, J.E. Knox, H.P. Hratchian, J.B. Cross, V. Bakken,

C. Adamo, J. Jaramillo, R. Gomperts, R.E. Stratmann, O. Yazyev, A.J. Austin, R. Cammi,

C. Pomelli, J.W. Ochterski, P.Y. Ayala, K. Morokuma, G.A. Voth, P. Salvador, J.J. Dannen-

berg, V.G. Zakrzewski, S. Dapprich, A.D. Daniels, M.C. Strain, O. Farkas, D.K. Malick,

A.D. Rabuck, K. Raghavachari, J.B. Foresman, J.V. Ortiz, Q. Cui, A.G. Baboul, S. Clifford,

J. Cioslowski, B.B. Stefanov, G. Liu, A. Liashenko, P.Piskorz, I. Komaromi, R.L. Martin, D.J.

Fox, T. Keith, M.A. Al-Laham, C.Y. Peng, A. Nanayakkara, M. Challacombe, P.M.W. Gill,

B. Johnson, W. Chen, M.W. Wong, C. Gonzalez, and J.A. Pople. Gaussian 03, revision D.01,

2004.

[269] L.A. Curtiss, P.C.Redfern, K. Raghavachari, and J.A. Pople. Gaussian-3X (G3X) theory: Use

of improved geometries, zero-point energies, and hartree-fock basis sets. J. Chem. Phys.,

114(1):108–117, 2001. 226 BIBLIOGRAPHY

[270] L.A. Curtiss, P.C.Redfern, K. Raghavachari, and J.A. Pople. Gaussian-3X (G3X) theory using

coupled cluster and brueckner energies. Chem. Phys. Lett., 359:390–396, 2002.

[271] J.M. Hevko, S. Dua, J.H. Bowie, and M.S. Taylor. Gas phase and condensed phase sni

reactions. the competitive six and seven centre cyclisations of the 5,6-epoxyhexoxide

anion. a joint experimental and ab initio study. a comparison with SNi reactions of homologous epoxyalkoxide anions. J. Chem. Soc., Perkin Trans. 2 457–464, 1999.

[272] A.M. McAnoy, S. Dua, S.J. Blanksby, and J.H. Bowie. The loss of CO from the ortho, meta

and para forms of deprotonated methyl benzoate in the gas phase. J. Chem. Soc., Perkin

Trans. 2 1665–1673, 2000.

[273] A.M. McAnoy, M.R.L. Paine, and S.J. Blanksby. Reactions of the hydroperoxide anion with

dimethyl methylphosphonate in an ion trap mass spectrometer: evidence for a gas phase

α-effect. Org. Biomol. Chem., 6:2316–2326, 2008.

[274] P.W. Atkins. Physical Chemistry. Oxford University Press, Oxford, 4th edition, 1990.

[275] B. Mattson, M. Anderson, and S. Mattson. Microscale Gas Chemistry Book. Educational

Innovations, 3rd edition, 2003.

[276] A. Zaki and H. Fahim. Some quaternary ammonium salts and their decomposition

products. J. Chem. Soc. 270–272, 1942.

[277] Y. Huang, L. Guler, J. Heidbrink, and H.I. Kenttämaa. Reactions of charged phenyl radicals

with aliphatic amino acids in the gas phase. J. Am. Chem. Soc., 127:3973–3978, 2005.

[278] J.J. Philips, M.A. Hudspeth, and Jr. Browne, P.M. Basis set dependence of atomic spin

populations. Chem. Phys. Lett., 495:146–150, 2010.

[279] P.J. Linstrom and W.G. Mallard. NIST chemistry webbook. In NIST Standard Reference

Database. National Institute of Standards and Technology, Gaithersburg, MD, 2003.

2 [280] A. Weaver, R.B. Metz, S.E. Bradforth, and D.M. Neumark. Observation of the Ã( B2) and ˜ 2 – C( A2) states of NO2 by negative ion photoelectron spectroscopy of NO2. J. Chem. Phys., 90(3):2070–2071, 1989.

[281] P.M. Sommeling, P. Mulder, R. Louw, D. V. Avila, J. Lusztyk, and K. U. Ingold. Rate of

reaction of phenyl radicals with oxygen in solution and in the gas phase. J. Phys. Chem.,

97(32):8361–8364, 1993. BIBLIOGRAPHY 227

[282] J.F.O’Hanlon. A User’s Guide to Vacuum Technology. Wiley, New York, 1980.

[283] M. Krauss and R. Osman. Absorption spectra of ethynyl, ethenyl, and phenyl peroxyl

radicals. J. Phys. Chem., 99:11387–11391, 1995.

[284] J.A. Jafri and D.H. Phillips. Ground and lower excisted states of methyl peroxy, CH3O2, radical: A computational investigation. J. Am. Chem. Soc., 112:2586–2590, 1990.

[285] T.L. Osif and J. Heicklen. Oxidation of HCO radicals. J. Phys. Chem., 80(14):1526–1531,

1976.

[286] M. Colberg and G. Friedrichs. Room temperature and shock tube study of the reaction

HCO + O2 using the photolysis of glyoxal as an efﬁcient HCO source. J. Phys. Chem. A, 110:160–170, 2006.

[287] H.C. Yang, H.L. Chen, and J.J. Ho. Ab initio study of intramolecular hydrogen transfer in

formylperoxy radical. J. Mol. Struct. (THEOCHEM), 774:35–41, 2006.

[288] K.J. Feilberg, G.D. Billing, and M.S. Johnson. Quantum dressed classical mechanics: + −−→ + Application to the HO CO H CO2 reaction. J. Phys. Chem. A, 105:11171–11176, 2001.

[289] K.J. Feilberg, S.R. Sellevåg, C.J. Nielsen, D.W.T. Grifﬁth, and M.S. Johnson. CO + −−→ + OH CO2 H: The relative reaction rate of ﬁve CO isotopologues. Phys. Chem. Chem. Phys., 4:4687–4693, 2002.

[290] J. Vandooren, J. Peeters, and P.J.Van Tiggelen. Rate constant of the elementary reaction of

carbon monoxide with hydroxyl radical. In 15th Symp. (Int) Combust., volume 15 745–753,

Pittsburgh, 1975.

+ [291] P.J.Stimac and J.R. Barker. Non-RRKM dynamics on the CH3O2 NO reaction system. J. Phys. Chem. A, 112:2553–2562, 2008.

[292] Y. Zhao, K.N. Houk, and L.P. Olsen. Mechanisms of peroxynitrous acid and methyl per-

oxynitrite, ROONO (R = H, Me), rearrangements: A conformation-dependent homolytic

dissociation. J. Phys. Chem. A, 2004:5864–5871, 2004.

[293] L.L. Lohr, J.R. Barker, and R.M. Shroll. Modeling the organic nitrate yields in the reaction

of alkyl peroxy radicals with nitric oxide. 1. Electronic structure calculations and thermo-

chemistry. J. Phys. Chem. A, 107:7429–7433, 2003. 228 BIBLIOGRAPHY

[294] G.B. Ellison, J.M. Herbert, A.B. McCoy, J.F. Stanton, and P.G. Szalay. Unimolecular

rearrangement of trans-FONO to FNO2. a possible model system for atmospheric nitrate formation. J. Phys. Chem. A, 108:7639–7642, 2004.

[295] J.F. Arenas, F.J. Avila, J.C. Otero, D. Peláez, and J. Soto. Approach to the atmospheric

chemistry of methyl nitrate and methylperoxy nitrite. chemical mechanisms of their

formation and decomposition reactions in the gas phase. J. Phys. Chem. A, 112:249–255,

2008.

[296] A.M. Kosmas, Z. Salta, and A. Lesar. Effect of halogenation on the mechanism of the

atmospheric reactions between methylperoxy radicals and NO. a computational study.

J. Phys. Chem. A, 113:3545–3554, 2009.

[297] S. Switalla, J. Knebel, D. Ritter, N. Krug, A. Braun, and K. Sewald. Effects of acute in vitro

exposure of murine precision-cut lung slices to gaseous nitrogen dioxide and ozone in an

air-liquid interface (ALI) culture. Toxicol. Lett., 2010:117–124, 2010.

[298] P.Moldéus. Toxicity induced by nitrogen dioxide in experimental animals and isolated cell

systems. Scand. J. Work Environ. Health, 19(Supp. 2):28–36, 1993.

[299] W.F.McDonnell, H.R. Kehrl, S. Abdulsalaam, P.J.Ives, L.J. Folinsbee, R.B. Devlin, J.J. Oneil,

and D.H. Horstman. Respiratory response of humans exposed to low-levels of ozone for

6.6 hours. Arch. Environ. Health 145–150, 1991.

[300] J.A. Graham, J. Overton, and D.L. Costa. Toxicology of ozone as characterized by labora-

tory animals and extrapolated to humans. Stud. Environ. Sci., 72:465–481, 1998.

[301] T.S. Dibble. Failures and limitations of quantum chemistry for two key problems in the

atmospheric chemistry of peroxy radicals. Atmos. Environ., 42:5837–5848, 2008.

[302] V. Van den Bergh, I Vanhees, R. De Boer, F. Compernolle, and C. Vinckier. Identiﬁcation

of the oxidation products of the reaction between α-pinene and hydroxyl radicals by

gas and high-performance liquid chromatography with mass spectrometric detection. J.

Chromatogr., A, 896:135–148, 2000.

[303] J.J. Orlando, G.S. Tyndall, and T.J. Wallington. The atmospheric chemistry of alkoxy

radicals. Chem. Rev., 103:4657–4689, 2003.

[304] B.D. Bean, A.J. Mollner, S.A. Nizkorodov, G. Nair, M. Okumura, S.P. Sander, K.A. Pe-

terson, and J.S. Francisco. Cavity ringdown spectroscopy of cis-cis HOONO and the

HOONO/HONO2. J. Phys. Chem. A, 107:6974–6985, 2003. BIBLIOGRAPHY 229

[305] A.J. Mollner, S. Valluvadasan, L. Feng, M.K. Sprague, M. Okumura, D.E. Milligan, W.J. Bloss,

S.P. Sander, P.T. Martien, R.A. Harley, A.B. McCoy, and W.P.L. Carter. Rate of gas phase

association of hydroxyl radical and nitrogen dioxide. Science, 330:646–649, 2010.

[306] X. Zhang, M.R. Nimlos, G.B. Ellison, M.E. Varner, and J.F. Stanton. Infrared absorption

spectra of matrix-isolated cis, cis-HOONO and its ab initio CCSD(T) anharmonic vibra-

tional bands. J. Chem. Phys., 124, 2006.

[307] X. Zhang, M.R. Nimlos, G.B. Ellison, M.E. Varner, and J.F. Stanton. Vibrational overtone

spectrum of matrix isolated cis, cis-HOONO. J. Chem. Phys., 126:174308, 2007.

[308] S. Jagiella and F.Zabel. Reaction of phenylperoxyl radicals with NO2 at 298K. Phys. Chem. Chem. Phys., 9:5036–5051, 2007.

[309] J.S. Steill and J. Oomens. Action spectroscopy of gas-phase carboxylate anions by multiple

photon IR electron detachment/attachment. J. Phys. Chem. A., 113:4941–4946, 2009.

[310] X.B. Wang, H.K. Woo, L.S. Wang, B. Minofar, and P.Jungwirth. Determination of the elec-

tron afﬁnity of the acetyloxyl radical (CH3COO) by low-temperature anion photoelectron spectroscopy and ab initio calculations. J. Phys. Chem. A, 110:5047–5050, 2006.

[311] J. Arey, S.M. Aschmann, E.S.C. Kwok, and R. Atkinson. Alkyl nitrate, hydroxylalkyl nitrate, − and hydroxycarbonyl formation from the NOx-air photooxidations of C5 C8 n-alkanes. J. Phys. Chem. A, 105:1020–1027, 2001.

[312] C.F. Ding, X.B. Wang, and L.S. Wang. Photoelectron spectroscopy of double charged

anions: Intramolecular coulomb repulsion and solvent stabilization. J. Phys. Chem. A,

102(45):8633–8636, 1998.

[313] P.D. Lightfoot, R.A. Cox, J.N. Crowley, M. Destriau, G.D. Hayman, M.E. Jenkin, G.K.

Moortgat, and F.Zabel. Organic peroxy radicals: Kinetics, spectroscopy and tropospheric

chemistry. Atmos. Environ., 26A:1805, 1992.

[314] G.S. Tyndall, R.A. Cox, C. Granier, R. Lesclaux, G.K. Moortgat, M.J. Pilling, A.R. Ravis-

hankara, and T.J. Wallington. Atmospheric chemistry of small organic peroxy radicals.

J. Geophys. Res. D: Atmos., 106(D11):12157–12182, 2001.

+ [315] J.C. Rienstra-Kiracofe, W.D. Allen, and H.F. Schaefer, III. The C2H5 O2 reaction mechanism: High-level ab Initio characterizations. J. Phys. Chem. A, 104(44):9823–9840, 2000. 230 BIBLIOGRAPHY

[316] J.A. Miller and S.J. Klippenstein. The reaction between ethyl and molecular oxygen II:

Further analysis. Int. J. Chem. Kin., 33(11):654–668, 2001.

[317] W. Tsang. Thermal stability of cyclohexane and 1-hexene. Int. J. Chem. Kin., 10(11):1119–

1138, 1978.

[318] T.C. Brown, K.D. King, and T.T. Nguyen. Kinetics of primary processes in the pyrolysis of

cyclopentanes and cyclohexanes. J. Phys. Chem., 90:419–424, 1986.

[319] B. Sirjean, P.A.Glaude, M.F.Ruiz-Lopez, and R. Fournet. Detailed kinetic study of the ring

opening of cycloalkanes by CBS-QB3 calculations. J. Phys. Chem. A, 110:12693–12704,

2006.

[320] J. Warnatz, U. Maas, and R.W. Dibble. Combustion. Springer, Berlin; New York, 2nd

edition, 1999.

[321] J.F. Grifﬁths and J.A. Barnard. Flame and Combustion. Blackie Academic & Professional,

London, 3rd edition, 1995.

◦ ◦ [322] L. Zhu, J.W. Bozzelli, and L.M. Kardos. Thermochemical properties, ∆fH (298), S (298), ◦ and Cp (t), for n-butyl and n-pentyl hydroperoxides and the alkyl and peroxy radicals, transition states, and kinetics for intramolecular hydrogen shift reactions of the peroxy

radicals. J. Phys. Chem. A, 111:6361–6377, 2007.

[323] R.X. Fernandes, J. Zádor, L.E. Jusinski, J.A. Miller, and C.A. Taatjes. Formally direct path-

ways and low-temperature chain branching in hydrocarbon autoignition: the cyclohexyl

+ O2 reaction at high pressure. Phys. Chem. Chem. Phys., 11:1320–1327, 2009.

[324] F. Buda, B. Heyberger, R. Fournet, P. Glaude, V. Warth, and F. Battin-Leclerc. Modeling of

the gas-phase oxidation of cyclohexane. Energy & Fuels, 20:1450–1459, 2006.

[325] S.M. Handford-Styring and R.W. Walker. Arrhenius parameters for the reaction HO2 + cyclohexane between 673 and 773 K, and for H atom transfer in cyclohexylperoxyl radicals.

Phys. Chem. Chem. Phys., 3:2043–2052, 2001.

[326] S.K. Gulati and R.W. Walker. Addition of cyclohexane to slowly reacting H2-O2 mixtures at 480 °C. J. Chem. Soc. Faraday Trans. 2, 85(11):1799–1812, 1989.

[327] E.J. Silke, W.J. Pitz, C.K. Westbrook, and M. Ribaucour. Detailed chemical kinetic modeling

of cyclohexane oxidation. J. Phys. Chem. A, 111:3761–3775, 2007. BIBLIOGRAPHY 231

[328] J. Platz, J. Sehested, O.J. Nielsen, and T.J. Wallington. Atmospheric chemistry of cyclo- − · − · hexane: UV spectra of c C6H11 and (c C6H11)O2 radicals, kinetics of the reactions of − · − · (c C6H11)O2 radicals with no and NO2, and the fate of the alkoxy radical (c C6H11)O J. Phys. Chem. A, 103:2688–2695, 1999.

[329] A.M. Knepp, G. Meloni, L.E. Jusinski, C.A. Taatjes, C. Cavallotti, and S.J. Klippenstein.

Theory, measurements, and modeling of OH and HO2 formation in the reaction of

cyclohexyl radicals with O2. Phys. Chem. Chem. Phys., 9:4315–4331, 2007.

[330] C. Lee, W.T. Yang, and R.G. Parr. Development of the Colle-Salvetti correlation-energy

formula into a functional of the electron density. Phys. Rev. B: Condens. Matter, 37(2):785–

789, 1988.

[331] M.R. Nyden and G.A. Petersson. Complete basis set correlation energies. I. The asymptotic

convergence of pair natural orbital expansions. J. Chem. Phys., 75:1843–1862, 1981.

[332] G.A. Petersson, A. Bennett, T.G. Tensfeldt, M.A. Al-Laham, W.A. Shirley, and J. Mantzaris.

A complete basis model set model chemistry. I. The total energies of closed-shell atoms

and hydrides of the ﬁrst-row atoms. J. Chem. Phys., 89:2193–2218, 1988.

[333] G.A. Petersson and M.A. Al-Laham. A complete basis set model chemistry. II. Open-shell

systems and the total energies of the ﬁrst-row atoms. J. Chem. Phys., 94:6081–6090, 1991.

[334] G.A. Petersson, T.G. Tensfeldt, and J.A. Montgomery, Jr. A complete basis set model chem-

istry. III. The complete basis set-quadratic conﬁguration interaction family of methods. J.

Chem. Phys., 94:6091–6101, 1991.

[335] J.A. Montgomery, Jr., J.W. Ochterski, and G.A. Petersson. A complete basis set model

chemistry. IV. an improved atomic pair natural orbital method. J. Chem. Phys., 101:5900–

5909, 1994.

[336] J.W. Ochterski, G.A. Petersson, and J.A. Montgomery, Jr. A complete basis set model

chemistry. V. extensions to six or more heavy atoms. J. Chem. Phys., 104:2598–2619, 1996.

[337] J.A. Montgomery, Jr., M.J. Frisch, J.W. Ochterski, and G.A. Petersson. A complete basis set

model chemistry. VI. use of density functional geometries and frequencies. J. Chem. Phys.,

110:2822–2827, 1999.

[338] J.A. Montgomery, Jr., M.J. Frisch, J.W. Ochterski, and G.A. Petersson. A complete basis

set model chemistry. VII. use of the minimum population localization method. J. Chem.

Phys., 112:6532–6542, 2000. 232 BIBLIOGRAPHY

[339] C. Gonzalez and H. B. Schlegel. An improved algorithm for reaction path following. J.

Chem. Phys., 90:2154, 1989.

[340] C. Gonzalez and H. B. Schlegel. Reaction path following in mass-weighted internal

coordinates. J. Phys. Chem., 94:5523, 1990.

[341] J.L. Mateos and H. Kwart. Linear free-energy relationships among reactions occurring

on the cyclohexyl ring. bromination of C4-substituted cyclohexanones. J. Org. Chem., 37(18):2826–2830, 1972.

[342] J.A. Hirsch and V.C. Truc. General syntheses of 6- and 7-carbomethoxy-trans-1-

heteradecalins and 6- and 7-carbomethoxy-trans-2-heteradecalins. J. Org. Chem.,

51(12):2218–2227, 1986.

[343] D.M. Black, A.H. Payne, and G.L. Glish. Determination of cooling rates in a quadrupole

ion trap. J. Am. Soc. Mass Spectrom., 17(7):932–938, 2006.

[344] J. Zheng, Y. Zhao, and D.G. Truhlar. The DBH24/08 database and its use to assess

electronic structure model chemistries for chemical reaction barrier heights. J. Chem.

Theory Comput., 5:808–821, 2009.

[345] K.G. Asano, D.E. Goeringer, and S.A. McLuckey. Thermal dissociation in the quadrupole

ion trap: ions derived from leucine enkephalin. Int. J. Mass Spectrom., 185/186/187:207–

219, 1999. Appendices

233

235 A PPENDIX

A PPENDIX A A 236 APPENDIX A

CO2 CO2 CO2 CO2

-CO2 -CHO

O m/z 140 O O O O m/z 199

CO2

O O O O O O C O O O O O m/z 143 m/z 168 m/z 200 m/z 184

CO2

-CO2

O O O m/z 140 O O O

CO2 CO2

+O2 +O2 H -CO2 -H O O O O

O O O m/z 196 O O O O O O O O O m/z 216 m/z 215 m/z 228 m/z 184

-O

CO2 CO2 CO2 CO2 CO2

+O2

-CO -O

m/z 184 O O O O O m/z 216 m/z 200 m/z 212 -CO

O O

-CO -CO -O

O m/z 128 m/z 160 O m/z 144 O O O O O m/z 184 m/z 156 -H +O2 +O2

H H

H O -O O O O O O -H O -CO O O O O m/z 160 O O m/z 159 m/z 176 m/z 188 m/z 187

-H -O

H O

+O2 O O

O O O m/z 159 O O O O m/z 172 m/z 204

-O

H -H O O

O O O O m/z 188 m/z 187

Scheme A.1: Mechanisms for the formation of ions measured during the reaction of the 4-carboxylato-4’- biphenyl radical anion with O2. APPENDIX A 237

(a) 4-carboxylatophenyl radical anion

(b) 4-sulfonatophenyl radical anion

(d) 4-carboxylato-4’-biphenyl radical anion

18 Figure A.1: Isolation of charge-tagged phenyl radicals in the presence of O2. The number of asterisks (*) denotes the number of incorporated extraneous O atoms.

239 A PPENDIX

A PPENDIX B B 240 APPENDIX B

Anion Electron Afﬁnity (Ea) Reference (eV)

I– (3.059)

1.09 (1.09) phen002, phen003

O2N 1.97 phen102, phen103

CO2

3.5 (3.6) adc202, adc204

- CO2

3.5 adc2701, adc2702

- CO2 3.5 ac801, ac802

- CO2 3.6 (3.5) ac901, ac902

N 2.8 be206, be401

Table B.1: Electron afﬁnities (Ea) of the neutral (anion shown) in eV calculated at the B3LYP/6-311++G(d,p) level of theory. Values in brackets are literature values from NIST Chemistry Webbook.279 APPENDIX B 241

Ion DH298 Reference (kcal mol−1)

N CH3

47.1 tma005, tma3301

N CH3

57.2 tma3401, tma3501

N CH3

Br 46.4 tma3101, tma3201

N CH3

Cl 47.2 tma3601, tma3701

N CH3

F 46.8 tma3801, tma3901

Table B.2: Bond Dissociation Energy calculated at B3LYP/6-311++G(d,p) level of theory.

oswr eivdt eiprat h ubro cn vrgdwsicesd h low The increased. was averaged scans of number the important, abundance low be When to scans. 50 believed least were at of ions average the were measured spectra mass cases, all In for spectra. ion are mass 9] and assigning times, when + account isolation [M There into long (or taken only duly ion after trap. appear 18] usually ion + identified, [M the easily an are to these as but added arises dianions), reagent which water, neutral as and such ion contaminants, isolated common are the of reactions only will to spectrum due mass containing the arise in spectrum measured mass products reaction a Ion-molecule of ion. measurement isolated by the confirmed only is ions This other all ion-trap. with the reaction, from to ejected prior isolated is ion reactant the reactions, ion-molecule During the and instrument the of ratio impurities. of signal/noise absence high general the demonstates some This In products. absence real. the reaction show are to of used peaks was magnification abundance 1000x is low 4.10, Figure spectrum that and 4.9 mass ensure Figure example, a to for cases, in taken observed are ion steps an of before number and a used assigned, is magnification of level high a While 1 Comment Gronert] Scott [Prof. Reviewer First A PPENDIX :R C: SOS TO ESPONSE 243 C R OMMENTS EVIEWERS

A PPENDIX C 244 APPENDIX C: RESPONSE TO REVIEWERS COMMENTS

abundance ion was then conﬁrmed to have been measured in multiple scans of the averaged

18 mass spectrum. Isotope labelling, for example, use of O2 as a neutral reagent, was also used to aid in the identiﬁcation of product ions, and conﬁrm low abundance ions arose due to reaction of the reagent ion with a neutral reagent.

While mass spectra published in this thesis contain no decimal places, the mass defect (ﬁrst/second decimal place of the ion measured) has also been used as a ’check’ to conﬁrm products (whether

CID, or ion-molecule) arose due to reaction or degradation of the isolated parent ion, where the mass defect was maintained.

Comment 2

Altered capitalisation of subject headings throughout thesis to sentence case.

Comment 3

The pressure measurements have not been checked by an independent measure, however the calculated concentration of reagent delivered has been conﬁrmed by comparing measurements of known second order rate constants. Harman and Blanksby previously compared a number of reactions, now described in Chapter 2 (p95), to those reported in the literature, and found the rates measured on the current instrumentation to agree favourably.

The following text has been added to the Mass Spectrometry section of Chapter 2 (p95) to clarify this point:

To validate this approach, Harman and Blanksby compared second order rate constants for four reactions (described in Equations 2.10-2.13) measured using this instrumentation (shown in

Table 2.2) to those reported in the literature.155 In all cases, the measured values were within the experimental error estimated for each method. This demonstrated that the current instrumentation is able to provide gas phase rate measurements of reasonable accuracy (the precision of this method will be discussed in Section 2.3.7), and validated the method described for determining neutral reagent concentration within the ion trap.

Comment 4 p121: As reviewer suggested, calculated electron afﬁnity values for phenyl and 4-nitrophenyl at · the B3LYP/6-311+g(d,p) level of theory. Altered value used in text to: (Ea (NO2C6H5) = 1.97 eV, see Appendix B Table B.1).

Added both phenyl and 4-nitrophenyl calculated electron afﬁnities to Appendix B.

Added footnote in text which notes the change and clariﬁes my position: APPENDIX C: RESPONSE TO REVIEWERS COMMENTS 245

· Calculation of the adiabatic electron afﬁnity of C6H5 at the B3LYP/6-311++G(d,p) level resulted · in a value of 1.087 eV, consistent with the NIST Chemistry Webbook which places the Ea of C6H5 · at 1.0960 ±0.0060 eV. This benchmark provides conﬁdence in the value calculated for NO2C6H5 radical at 1.97 eV.

Comment 5 p135: As requested, a hard sphere collision rate has been calculated and included in Table 3.1.

This allows the estimation of a reaction efﬁciency of ca. 6%.

Added the following to the text (p134):

A reaction efﬁciency of 6% was estimated for the neutral phenyl radical using the second order rate constant reported by Yu and Lin39 and an estimated hard sphere collision rate of 1.81 x − − − 10 10 molecules 1 cm3 s 1. While there are obvious differences between the neutral and charged phenyl radical systems, the efﬁciencies calculated for 1 and 4 - 6 were surprisingly consistent with this value.

Added a footnote to explain hard sphere calculation (p134):

= πσ2 8kB T 1/2 σ2 = + 2 Hard sphere collision rate estimated using khs AB ( πµ ) , where AB (RA RB ) . Molec- ular radii for benzene (R = 1.75 Å) and O (R = 1.55 Å) were obtained from O’Hanlon.282 C6H6 2 O2

Comment 6

Pursuant to reviewers request, calculated the electron afﬁnity of the decarboxylated oxepinone radical.

Altered electron detachment section to read (p135-136):

If the electron binding energy of a minimum was lower than the reaction enthalpy of its formation then it follows that the intermediate may undergo electron detachment, losing its charge, without producing any products measurable by mass spectrometry. The electron binding energy of the 4-carboxylatophenylperoxyl and 5-carboxylatooxepinon-7-yl radical anions, intermediates of the 4-carboxylatophenyl radical anion + O2 reaction where this anomaly occurs, and 4-sulfonatophenylperoxyl and 5-sulfonatooxepinon-7-yl radical anions, where it did not, were calculated at the composite G3SX level of theory to investigate this hypothesis. The 4- − carboxylatophenylperoxyl radical anion resides -45.2 kcal mol 1 below reactants, with an elec- − tron binding energy of 91.5 kcal mol 1 (see Figure 3.20), while the rearranged 5-carboxylatooxepinon- − 7-yl radical anion, which resides in a deeper potential well -98.3 kcal mol 1 below reactants, has − an accessible electron binding energy of 93.8 kcal mol 1. In contrast, the electron binding energy of the 4-sulfonatophenylperoxyl and 5-sulfadionatooxepinon-7-yl radical anions were predicted 246 APPENDIX C: RESPONSE TO REVIEWERS COMMENTS

− − at 70 kcal mol 1 and 20.7 kcal mol 1 higher than the energy of the reactants.

It is also possible that electron attachment occurs subsequent to decarboxylation due to formation of a loosely bound anion, for example, intermediates Hc and Nc . Decarboxylation re- − − sults in products -56.5 kcal mol 1 and -58.2 kcal mol 1 below barrier, respectively, while the − − binding energies were calculated at 43.4 kcal mol 1 and 57.2 kcal mol 1, respectively (shown in Figure 3.20). This suggests that the reaction energy is sufﬁcient to drive electron detachment from these intermediates. Ejection of the sulfonato group from 5-sulfonatooxepinon- − 7-yl radical anion results in a product -5.7 kcal mol 1 below reactants, well below the energy required for electron detachment (see Figure 3.20). These predictions account not only for the curvature of the normalised pseudo-ﬁrst order decay plot, but may also explain the low yield of the cyclopentadienone Nc (compared with sulfonato charged-tagged phenyl radicals

5 and 6), as decarboxylation and electron detachment of Nc would result in a low apparent yield of this product. These calculations support the hypothesis that electron detachment occurs after rearrangement of the 3- and 4-carboxylatophenylperoxyl radical anion, which ultimately accounts for the curvature observed in the normalised pseudo-ﬁrst order decay.

Altered Figure 3.20 caption to read (p136):

Electron binding energy for the 4-carboxylatophenylperoxyl, 5-carboxylatooxepinon-7-yl and oxepinon-5-id-7-yl radical anions relative to 4-carboxylatophenyl radical anion + O2 and 4-sulfonatophenylperoxyl, 5-sulfonatooxepinon-7-yl and oxepinon-5-id-7-yl radical anions relative to 4-sulfonatophenyl +

O2. Calculations were undertaken using G3SX theory.

Comment 7

As per reviewer request, added the following to the "Errors in calculated reaction rates" section

(p98-99) in regards to mass discrimination effects:

Further error may be introduced by mass discrimination effects.93 Mass discrimination in these experiments is likely to arise due to space-charge effects and reaction of ions with residual solvent inside the ion-trap. Both effects cause curvature in pseudo-first order decay plots. Reaction of ions with the background gas will likely present as adduct ions, for example, an [M + 18] for addition of water. In general, this adduct formation is significantly slower than the reaction being measured. The use of MS3 and higher MSn experiments for kinetic measurements largely precludes space-charge effects, as subsequent MS experiments result in isolation of only part of the initial ion packet such that over-filling of the ion-trap is minimised by the time kinetic measurements are made. Rate measurements typically contained linear pseudo-first order decay curves and therefore mass discrimination effects are considered minimal. Rates have, therefore, APPENDIX C: RESPONSE TO REVIEWERS COMMENTS 247

not been corrected and any minimal effects are assumed to contribute to the overall error.

Comment 8

Undertook further calculations and added text as described in comment 6 (p134-135).

Comment 9

Additional calculations were undertaken to explore the reviewers proposed mechanism.

Updated Scheme 5.4 (p178) with additional calculated energies.

As shown in Scheme 5.4, the additional pathway stil requires signiﬁcant energy at 35.4 kcal/mol and results in a product less stable than the 4-carboxylatocyclohexyl radical (2.4 kcal/mol higher in energy).

Added to text (p176): "S may isomerise to a low energy allylic radical, but this process still − − requires 35.4 kcal mol 1 and the ﬁnal product is higher in energy than E at 2.4 kcal mol 1.

Comment 10

Added for clarity (p178): "Text under energy is a reference to the calculation archive in supplementary information."

Second reviewer [Assoc. Prof. Uta Wille]

Comment 1

Reviewed as requested. Three other people have proof-read this thesis, focussing on structure and grammar. I am hopeful that these improvements would now meet the expectations of the reviewer.

Comment 2

Applied changes as requested throughout thesis.

Comment 3

This implication has been removed as a consequence of changes for comment 4 (p5).

Comment 4

Removed ﬁrst sentence. (p5) 248 APPENDIX C: RESPONSE TO REVIEWERS COMMENTS

Comment 5

Added the text "removal of an electron", as requested, along with the corresponding Reaction

1.4. (p6)

Comment 6

Applied change as requested. (p6, second paragraph)

Comment 7

Changed to: "Radicals are, typically, highly reactive species, in many cases seeking an electron with which to reach a stable valence octet." (p6, 3rd paragraph)

Comment 8

Compounds not shown in ﬁgure were removed to simplify and the sentence rearranged so that only MDA (the example given in Figure 1.2) is the only one mentioned. (p8)

Comment 9

Altered diagram in Figure 1.2 to show resonance structure. Changed "cyclize" to "5-exo". (p8)

The use of ’ﬁshhook’ arrows is a widely adopted nomenclature in organic chemistry. From a quantum mechanical perspective, however, the movement of individual or pairs of electrons is not meaningful. Furthermore, formally twice as many ’ﬁshhook’ arrows are required for the same transformation and thus I (and I am sure others) have found this unnecessarily confusing.

I have used arrows to assist the reader in their understanding of the mechanisms presented, and thus do not ﬁnd this convention useful.

Comment 10

Removed "radical(s)" from HO·. Also updated throughout the thesis.

Comment 11

Applied change as requested. (p10, ﬁrst paragraph under 1.2.1.1)

Comment 12

· + −−→ + · −−→ · Removed H F2 HF F reaction and included F2 2 F (p10, Reactions 1.11)

Comment 13

Applied change as requested. (p11, last paragraph). APPENDIX C: RESPONSE TO REVIEWERS COMMENTS 249

Comment 14

Applied change as requested. (p11, throughout)

Comment 15

Corrected value and units. Now reads 740 nm. (p12, ﬁrst paragraph)

Comment 16

Applied change as requested. (p12, ﬁrst paragraph)

Comment 17

Altered sentence from "spectroscopy was the used" to "spectroscopy was then used". (p12, third paragraph)

Comment 18

Altered sentence as requested. (p12, under 1.2.2)

Comment 19

Altered caption to (p14, Figure 1.4): The Near-IR O-O stretching region of the methylperoxyl and phenylperoxyl radical spectrum measured by Just et al.40 Vibrations of the methylperoxyl radical

1 1 and two vibrations (120 and 14O) of the phenylperoxyl radical are indicated.

Comment 20

As per the document "Copyright guidelines for study and research - Fair Dealing" (http://www.library.uow.edu.au/copyright/UOW026682.html), material may be copied for criticism or review if properly attributed."

A review is: "1. A critical article or report, as in a periodical, on some literary work, commonly some work of recent appearance; a critique..."?" [ Beaumont J. BRIAN KELVIN DE GARIS and

MATTHEW MOORE And: NEVILLE JEFFRESS PIDLER PTY LIMITED No. G1319 of 1988 FED No.

352 Copyright 18 IPR 292 (1991) 20 IPR 605 (1990) 37 FCR 99]".

As such, the diagrams and ﬁgures attributed from journal articles and books used within this thesis fall under the Fair Dealings provisions.

Furthermore, I contacted our copyright and digitisation ofﬁcer (Noel Broadhead) who responded:

"Your assumption was correct; as long as you’ve attributed the source, you can operate under the Research/Study Fair Dealing exception, without having to seek permission.

You would only need to seek permission if you subsequently publish your thesis." 250 APPENDIX C: RESPONSE TO REVIEWERS COMMENTS

Comment 21

(p20, ﬁrst paragraph): I am unsure what the reviewer is referring here. The Australian spelling is

"photoionisation" and the capital letter should not be required.

Removed colon as requested.

Comment 22

(p21): Altered main text as requested to include deﬁnition/abreviation of FWHM.

Comment 23

Altered text as requested. (p23, third paragraph)

Comment 24

Altered text as requested. (p36, below 1.3.1)

Comment 25

This was keeping in style with rest of structures in the thesis, but altered as per examiner comment. (p36)

Comment 26

Altered size of structure as requested. (p37, Scheme 1.17)

Comment 27

Altered as requested throughout thesis.

Comment 28

Original ﬁgure as been altered. Compound numbers have been removed to remove ambiguity as requested. (p38, 1.18)

Comment 29

Changed text to: "observed that proton transfer only took place when the neutral species in the

’ reaction, R2NNO, had an equal or greater basicity than R2NNO." (p39, second paragraph)

Comment 30

Altered text from "Indeed, previous mass spectrometric studies had hinted at such an intermediate due to McLafferty rearrangements" to "Indeed, previous mass spectrometric studies had APPENDIX C: RESPONSE TO REVIEWERS COMMENTS 251

hinted at charge-spin separated species as intermediates in McLafferty rearrangements" (p39, second paragraph)

Altered text from "fragmentation" to "collisional activation" (p39)

Comment 31

Modiﬁed scheme to remove mechanism arrows. (Scheme 1.3, p40).

Removed extraneous full stop from Figure 1.19. (p42)

Comment 32

Removed mechanism. (p40, 1.3)

· Added radical dot to C2H5 in Scheme 1.3. (p40)

Altered caption to "ionised cyclopropane". This is more accurate than "cyclopropyl radical cation" as the structure of the ionised cyclopropane was never determined.

Comment 33

This was intentional. The former information is an introduction to the topic and details the formal deﬁnition of ’distonic radical ions’ and its etymology, while the latter describes the history of ’distonic radical cations’ in detail from an experimental perspective. No sentences were repeated verbatim. This was not altered. (p36 & p40).

Comment 34

Changed to consistent labelling of Schemes and Figures throughout.

Comment 35

This was correct and referring to Figure 1.19 shown on the next page (p42). The ﬁgure has been moved closer to text.

Comment 36

Modiﬁed text as requested. (p42)

Comment 37

See response to comment 9.

Comment 38

Modiﬁed as requested. (Scheme 1.7, p43) 252 APPENDIX C: RESPONSE TO REVIEWERS COMMENTS

Comment 39

Modiﬁed as requested. (p44, below 1.3.4)

Comment 40

Added text "The following section introduces mass spectrometric methods, including collisional activation and ion-molecule reactions, typically used to differentiate and structurally characterise distonic radicals." (p44, below 1.3.4)

Comment 41

Modiﬁed as requested. (p46, below Scheme 1.8)

Comment 42

In regards to radical arrows, see response to comment 9.

Added scheme labels to reaction blocks as requested. (Scheme 1.9 - Scheme 1.12, p46-47)

Comment 43

Added following text and reference in text: "Yu et al. have suggested that 1-pyridiniumethyl radical does not react with O2 due to delocalisation of the unpaired electron, whereas in the case of 2-pyridiniumethyl, the unpaired electron is isolated from the ring.150" (p47)

Comment 44

Changed Figure 1.24 to a Table, as requested. Removed compound labels from table, as requested. (p48, Table 1.2)

Comment 45

Altered text to: "Reactions of distonic radical anions with a high gas phase basicity were observed to undergo competitive proton-transfer reactions with some neutral reagents, whereas, those with a low gas phase basicity typically reacted via radical type reactions." (p48)

Comment 46

Altered text as requested. (p49)

Comment 47

Altered paragraph to: "Cross conjugation occurs when there are common molecular orbitals shared between distinct resonance structures, an example of which is shown in Scheme 1.14. As APPENDIX C: RESPONSE TO REVIEWERS COMMENTS 253

shown in Scheme 1.15, Wenthold and Squires observed an E1cb like elimination reaction that occurs after addition of NO· to both the 2-methylpropan-2-id-1-yl radical anion and acetate radical anion resulting in formation of an α-distonic alkylcarboxylate radical anion followed by subsequent ejection of a nitroso carbanion.156,153 This reactivity appears characteristic of cross- conjugated radical anions." (p49, second paragraph)

Comment 48

See response to comment 9.

Comment 49

As requested, the scheme was moved up closer to the paragraph where it was mentioned. (p50,

Scheme 1.16)

Changed size of pyridine to same size as other molecules.

−I . has been added to Scheme.

Comment 50

Altered as requested. (p51, Scheme 1.17)

In regards to radical arrows, see response to comment 9.

Comment 51

Altered text as requested. (p52, bottom paragraph).

Comment 52

Altered text as requested. (p52)

Comment 53

Changed caption to: "Photodissociation (266 nm) and sustained off-resonance irradiation collision activated dissociation (SORI-CAD) of iodinated and brominated pyridinium cations affords distonic radical cations, as demonstrated by Thoen et al.114"(p51)

Comments 54 & 55

Changed paragraph to: "The first reported regiospecific synthesis of carbanions in the gas phase was demonstrated by DePuy et al.182 This technique, known as fluorine induced desilylation, results in the nucleophilic displacement of a carbonanion from an alkylsilane by fluoride anion

F–). Around the same time, the Bowie group found hydroxide (HO–) and alkoxides (RO–) to 254 APPENDIX C: RESPONSE TO REVIEWERS COMMENTS

react similarly with alkylsilanes.183 While charge-separated radical anions were never the focus of DePuy’s research, an investigation into formation of the acetyl anion by ﬂuorine induced desilylation resulted in the formation of the acetate radical anion (shown in Scheme 1.20). The

– acetyl anion was shown to react with O2 by O atom transfer to form CH3CO2 and, in addition, · . – 184 by a concerted O atom transfer and loss of HO to yield the acetate radical anion CH2CO2. " (p52)

Comment 56

Edited ﬁgure as request, added −(H3C)3SiF to ﬁrst arrow. (p53, Scheme 1.20)

Altered second "resulting in formation of the" to "generating acetate radical anion" (p53, Scheme

1.20 caption)

Comment 57

Altered text to "resulting in formation of p-benzyne, a distonic radical anion" (p53, second paragraph)

Comment 58

This example is described later in chapter 2.

Added reference to the scheme in chapter 2 (p54): "Blanksby and co-workers have utilised this method to synthesise the distonic charge-tagged bridgehead radical, 3-carboxylatoadamantyl radical anion (described later, in Chapter 2, Scheme 2.5)"

Comment 59

Altered text to: "The method was generalised after several investigations in the literature, wherein

CID of dicarboxylate dianions led not only to the expected decarboxylation, but concomitant electron loss, resulting in formation of a radical anion. This was proposed to occur due to a lowering of the electron afﬁnity of the nascent anion formed after CO2 loss. This resulted in an increased Coulombic repulsion as the distance between the two charge centres decreased, ultimately leading to electron ejection and formation of a radical anion.130,135,189" (p54, ﬁrst paragraph)

Comment 60

Altered sentence. Moved to new paragraph which now reads: "Blanksby and co-workers have utilised this method to synthesise the distonic bridgehead radical, 3-carboxylatoadamantyl radical anion, from 1,3-adamantane dicarboxylic acid (as described later, in Chapter 2, Scheme APPENDIX C: RESPONSE TO REVIEWERS COMMENTS 255

2.5). This method allowed them to investigate the reactivity of a strained carbon-centred radical.154,155" (p54, second paragraph)

Comment 61

See response to comment 9.

Comment 62

Changed last sentence to "There were two major ﬁndings of this study: i) the charge-tag has a signiﬁcant effect on the second order rate of reaction due to perturbation of the electrophilicity or nucleophilicity of the radical moiety, and ii) the charge-tag did not affect how a radical would react with a neutral molecule, for example, whether it would undergo hydrogen atom abstraction, therefore, distonic radical ions make a good qualitative model of radical reactivity." (p55, second paragraph)

Restructured as requested.

Comment 63

Changed to "By tailoring the charge-tag and substituents one can study the effect of differing the partial charge at the radical moiety." (p55, third paragraph)

Removed "model absolute reaction kinetics".

Comment 64

Altered as requested. (p57, below 1.5)

Comment 65

Added: "The linear decay suggested generation of a single isomer, assigned as N-(3,5-dehydrophenyl)-

3-ﬂuoropyridinium ion. In contrast, when generated by SORI-CAD, the m/z 172 ion reacted with a lower efﬁciency and no product was formed at m/z 198. Furthermore, a non-linear psuedo-

ﬁrst order decay proﬁle measured in this case suggested a mixture of isomeric ions, probably due to intramolecular hydrogen atom migration."

Altered text as requested. (p61, ﬁrst paragraph)

Comment 66

Altered to "266 nm fourth harmonic generated from the output of a Nd:YAG laser". (p61, last line) The suggested "266 nm output from a Nd:YAG" is incorrect as the direct output from a

Nd:YAG laser is 1064 nm. 256 APPENDIX C: RESPONSE TO REVIEWERS COMMENTS

Comment 67

Re: Radical arrows, see comment 9.

Comment 68

Altered as requested. (p63, Figure 2.4) (p64, Scheme 2.1)

Comment 69

Changed to "means of generating distonic radical ions" (p70, below 2.2.2) and "precursors for regioselective generation of distonic radical ions" (p71, ﬁrst paragraph)

Comment 70

(p71, below 2.2.2.1): To clarify text, altered to: "photolysis product of m/z M and [TIC]0 is the total ion abundance" and "represents the yield of a product (of m/z M) compared with the initial abundance"

Changed all "currents" to "abundances" for clarity.

This equation is correct. Conversion efﬁciency measures ion retention as stated in text. The total ion abundance (i.e., the abundance of ALL ions in the trap) prior to and after photolysis is compared. Prior to photolysis, the only ion in the trap is the isolated precursor ion. After photolysis, the ions retained include remaining precursor ions and any photo-dissociation products.

If the total ion abudances do not match, then there have been ion losses, and this is what this parameter measures.

Comment 71

9 Added footnote (p73): Oxygen present in the ion-trap. Estimated at [O2] = 3 x 10 molecules − cm 3.155 Due primarly to transmission from the ionisation source, which is under atmospheric pressure. (p73)

Added clariﬁcation of laser energy "single laser pulse at 266 nm (2.4 mJ/pulse)" (p71, bottom paragraph)

The photolysis yields of all photolysis precursors were measured on the same day at the same energy at 24 mW (10Hz, 2.4mJ/pulse)

Comment 72

See response to comment 9.

See response to comment 81. APPENDIX C: RESPONSE TO REVIEWERS COMMENTS 257

Comment 73

See response to comment 9.

Resonance arrow removed as requested.

Comment 74

− Removed eV number and left in the kcalmol 1. (p83)

Added (4.667 eV). (p83, ﬁrst paragraph)

Comment 75

Altered text for clarity: "In this case, photolysis was the only method of synthesising the target radical cation D. CID resulted in generation of m/z 149 ions, the majority of which did not react in the presence of O2 (compare Figure 2.12 and Figure 2.13 where the reaction times were the same)." (p77)

Comment 76

Changed text to ’can’

Added "(shown in Figure 2.15 and Figure 2.16)" (p78)

Comment 77

Altered position of ﬁgure 2.12. See response to comment 81.

Comment 78 p78: Altered paragraph to: The 3-carboxylatoadamantyl radical anion (H) is a prototypical example of a distonic carbon-centred radical anion. Harman and Blanksby have previously demonstrated that H could be readily synthesised in the gas phase.154,155 A methanolic solution of 1,3- adamantanedicarboxylic acid (4) can be infused by negative electrospray ionisation to afford − an [M - 2H]2 ion G. When subjected to CID, G undergoes oxidative decarboxylation to yield H

154,155 (shown in Scheme 2.5). Furthermore, the reactivity of G towards O2 has previously been characterised. As discussed above, iodinated aromatic and aliphatic compounds and Barton ester derivatives are expected to absorb in the short wavelength visible and UV.3-(N-oxycarbonyl-

2(1H)-pyridinethione)adamantane-1-carboxylic acid (5) and 3-iodoadamantanecarboxylic acid

(6) were therefore logical targets for these experiments as both molecules are expected to yield

H when subject to PD (shown in Figure 2.15 and Figure 2.16). In the following experiments, a comparison of the reactivity of two radical anions formed by PD of 5 and 6 towards O2 is made 258 APPENDIX C: RESPONSE TO REVIEWERS COMMENTS

to those previously generated by CID of the dicarboxylic acid 4 to determine whether all the three precursors form the same radical anion product.

Added reference to scheme. Fixed caption so that ’5’ and ’6’ are labelled. (p80)

Removed (i) from scheme/caption.

Comment 79

Re: radical arrows, see comment 9.

Altered compound identiﬁers in schemes/captions. (p80-81 Scheme 2.6 and 2.7)

Comment 80

Altered position of ﬁgures. (p80-81 Figure 2.15 and 2.16)

Comment 81

The schemes have been combined as suggested, and now appear in Figures 2.12, 2.13, 2.14, 2.15 and 2.16 (p80-81), (p75-77)

Comment 82

Information was supplementary and interrupted ﬂow of text. Changed to footnote instead referenced from main text. (p78, p79)

Comment 83

Altered position of ﬁgure. Compound labels are indicated in the caption. (p83)

Comment 84

Altered conclusion for clarity: "While PD and CID of I resulted in relatively high photolysis yield and conversion efficiency, the highest yield of isolable ions was realised by the oxidative decarboxylation of G by CID. A high ESI and trapping efficiency afforded by use of the dicarboxylic acid 4 resulted in a significantly higher precursor ion abundance. This compensated for the low conversion efficiency usually afforded by CID of dicarboxylates and ultimately resulted in a greater abundance of the target radical anion H." (p84-85)

Comment 85

The radical dot is indicated. (p83) APPENDIX C: RESPONSE TO REVIEWERS COMMENTS 259

Comment 86

Added "(as demonstrated in Figure 2.18)." (p84)

Comment 87

Altered as requested. (p91)

Comment 88

Changed to: "In the following experiments, rate constants are measured for bimolecular reactions. In order to calculate rate constants for bimolecular reactions the concentration of at least one reactant or product must be known. Under these experimental conditions, it is not possible to determine the concentration of the reactant radical anion, or products, therefore, the reaction rates are measured under psuedo-ﬁrst order conditions, where the neutral molecule added to the helium buffer gas is in excess. To then determine the real second-order rate constant the concentration of the neutral molecule in excess must be known." (p92, below 2.3.3)

Comment 89

Fixed typo in letters: Wn to WN and Wh to WHe .

Added to text: (WN and WHe are the molecular weight of the neutral and helium, respectively) (p94)

Comment 90

Altered position of table. (p95, Table 2.1, p98, Table 2.3)

Comment 91

Altered from "ﬁrst-order rate k1" to "pseudo-ﬁrst order rate constant k1" (p96, bottom paragraph)

Comment 92

In response to reviewer comment: In ﬂow tube experiments utilising both neutrals and ions, losses are expected due to interaction of the reagents and reactants with the ﬂow tube walls. In ion-trap experiments, the ions are contained axially within a DC potential, and radially within an oscillating potential. In addition, ions are focussed towards the centre of the ion trap.90,93,91,92

As such, wall losses are not expected. As demonstrated by reports of previous experiments undertaken using ion-trap mass spectrometers with similar modiﬁcations,237,235 it is not necessary 260 APPENDIX C: RESPONSE TO REVIEWERS COMMENTS

to account for wall losses during rate measurements when using an ion-trap mass spectrometer.

Comment 93

Altered as requested. (p100,p111)

Comment 94

Altered to "If the reaction was incomplete after 6 hours and there was no evidence of its continuation, further MeI was added." (p100)

Comment 95

Altered as requested. (p101)

Comment 96

Altered as requested. (p101)

Comment 97

Changed format of NMR data, measured and added 13C NMR data for compounds not previously characterised. (p101)

Comment 98

Altered to include "real" phenyl scheme, as requested. (p104)

Fixed incorrect reaction scheme so that it correctly shows HO· addition channel:

+ · → · C6H6 HO HOC6H6

Comment 99

As I understand the grammar, both approaches are correct in this instance, and I prefer the original style and have retained it in this instance.

Comment 100

Altered as requested to: "Furthermore, potential energy surfaces describing the unimolecular degradation of both neutral and charge-tagged phenylperoxyl radical were constructed using computational methods, and a comparison of the reaction pathways made to ion-molecule reactions measured using ion-trap mass spectrometry." (p108) APPENDIX C: RESPONSE TO REVIEWERS COMMENTS 261

Comment 101

Altered sentence to "isolation potential (DC trapping potential applied to trap ions in the x- direction of the ion-trap, see Section 1.2.3.2.5 for more information) and RF drive (quadrupole

AC ﬁeld applied to trap ions radially)" (p110)

Comment 102

Added "nitric oxide(98.5%)" to material list. (p111)

Comment 103

Structure of these compounds are shown in mass spectra where they are used, see Figure 3.15.

Comment 104

Cross-referenced schemes as requested.

Compound labelling is consistent, and independent, within chapters deliberately as they will be published separately as papers.

Comment 105

Altered as requested. (p117)

Comment 106

The (*) indicates either presence of O2 or NO2 reaction products which are contaminent ions during these reactions. (p118,p119)

This is now described in captions for Figures 3.7, 3.8, 3.9 and 3.11.

Comment 107

Altered to ﬁx nomenclature. Phenide replace with benzenide as per IUPAC naming. (p120)

Nitrophenoxide is correct.

(iii) This was altered as per ﬁrst reviewers comment 4, which clariﬁes the point and contains correct nomenclature.

Comment 108

Chapter 2 was intended to illustrate the methodology, while speciﬁc precursors were introduced in the relevant chapters. 262 APPENDIX C: RESPONSE TO REVIEWERS COMMENTS

Comment 109

Fixed numbering in text.

Comment 110

Altered ChemDraw structure for m/z 136 ion as requested. (p119, Figure 3.11)

Comment 111

This was not pertinent to the discussion. These spectra were only used to demonstrate the presence of isotopic labels. Isotopic labelling in this instance is used to derive qualitative, mechanistic insight only. (p129, Figure 3.17)

Kinetic data are provided for the unlabelled case - See Table 3.1.

Comment 112

In the photoelectron spectroscopy literature, electron binding energy is used to denote the energy required to detach an electron from an anion (formally, the negative of the electron afﬁnity).

Ionisation potential or ionisation energy is used to denote the energy required to detach an electron from a neutral to form a cation. For example, these terms are deﬁned in Sengupta B.;

Curtiss, L.A.; Miller, J.R; J. Chem. Phys., 1996, 104 (24). (p135-136)

Comment 113

Changed neutral hydrocarbons to "solvents used during electrospray ionisation (such as MeOH)"

(p144, ﬁrst paragraph)

Comment 114

It was concluded that when it degrades "to products" it degrades preferentially via O loss. The

PES does not consider entropy which would prefer return to starting products (133.1 cal/molK and 127.9 cal/molK, respectively). Added: "While this process requires more energy than other degradation channels, it is entropically favourable. (p145)

∆ = Added footnote: Calculation at B3LYP/6-311++G(d,p) level of theory shows S AC 131.1 cal −1 −1 ∆ = −1 −1 mol K and SCC 127.9 cal mol K ." (p145, footnote)

Comment 115

Added M to Equation 4.3 (p148). APPENDIX C: RESPONSE TO REVIEWERS COMMENTS 263

Comment 116

See response to comment 104.

Comment 117

Removed (a)/(b)/(c) from subﬁgures (p157, Figure 4.6). Added "NCE is the normalised collision energy setting in the Xcalibur software.95" to the caption.

Comment 118

Added "As previously described in Section 2.2.2.2," to beginining of this paragraph. (p169, bottom paragraph)

Comment 119

Added 1.3 M in hexanes. (p173)

Comment 120

See response to comment 9.

Comment 121

Altered as requested. (p170)

Comment 122

Fixed Tetrahedron Lett.

Removed "pages", as requested.

References are formatted using a standard style (natbib, unsrt style). There were some articles with no issue number which were omitted. Style is described below:

Journal articles: [Number] A. Author, B. Author, C. Author. Article Title. Abbrv. J. Title., Vol(Issue):PageFrom-

PageTo, Year.

Books: [Number] A. Author, B.Author, C. Author. Book Title. Publisher, Location, Edition, Year.